12_custom_models_excercises.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "12. Implement a custom layer that performs Layer Normalization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "class NormalizationLayer(keras.layers.Layer):\n",
    "\n",
    "    def __init__(self, epsilon=0.001, **kwargs):\n",
    "        super().__init__(**kwargs)\n",
    "        self.epsilon = epsilon\n",
    "    \n",
    "    def build(self, batch_input_shape):\n",
    "        self.alpha = self.add_weight(\n",
    "            name=\"alpha\",\n",
    "            shape=batch_input_shape[-1:],\n",
    "            dtype=tf.float32,\n",
    "            trainable=True,\n",
    "            initializer=tf.ones\n",
    "            \n",
    "        )    \n",
    "        self.beta = self.add_weight(\n",
    "            name=\"beta\",\n",
    "            shape=batch_input_shape[-1:],\n",
    "            dtype=tf.float32,\n",
    "            trainable=True,\n",
    "            initializer=tf.zeros\n",
    "            \n",
    "        )    \n",
    "        super().build(batch_input_shape)\n",
    "        \n",
    "        \n",
    "    def call(self, inputs):\n",
    "        mean, variance = tf.nn.moments(inputs, axes=-1, keepdims=True)\n",
    "        std = tf.sqrt(variance + self.epsilon)    # derivative of sqrt(z) is undefined for z=0\n",
    "        return self.alpha * (inputs - mean) / std + self.beta"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tf.Tensor([4.7031407 3.0655727 7.487052  4.412018  4.158657 ], shape=(5,), dtype=float32)\n"
     ]
    }
   ],
   "source": [
    "data = tf.random.normal(shape=(5,), mean=5, stddev=1)\n",
    "print(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tf.Tensor([-0.04228865 -1.15658     1.8520375  -0.24038468 -0.41278526], shape=(5,), dtype=float32)\n"
     ]
    }
   ],
   "source": [
    "layer = tf.keras.layers.LayerNormalization(axis=0)\n",
    "output = layer(data)\n",
    "print(output)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tf.Tensor([-0.04228865 -1.15658     1.8520375  -0.24038468 -0.41278526], shape=(5,), dtype=float32)\n"
     ]
    }
   ],
   "source": [
    "my_layer = NormalizationLayer()\n",
    "my_output = my_layer(data)\n",
    "print(my_output)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "13. Train a model using a custom training loop"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "(X_train_full, y_train_full), (X_test, y_test) = keras.datasets.mnist.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Scale the features for Gradient Descent (is it really necessary?)\n",
    "X_valid, X_train = X_train_full[:5000] / 255.0, X_train_full[5000:] / 255.0\n",
    "y_valid, y_train = y_train_full[:5000], y_train_full[5000:]\n",
    "X_test = X_test / 255.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = keras.models.Sequential()\n",
    "model.add(keras.layers.Flatten(input_shape=[28, 28]))\n",
    "model.add(keras.layers.Dense(200, activation=\"relu\"))\n",
    "model.add(keras.layers.Dense(200, activation=\"relu\"))\n",
    "model.add(keras.layers.Dense(10, activation=\"softmax\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "flatten (Flatten)            (None, 784)               0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 200)               157000    \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 200)               40200     \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 10)                2010      \n",
      "=================================================================\n",
      "Total params: 199,210\n",
      "Trainable params: 199,210\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<tf.Variable 'dense/kernel:0' shape=(784, 200) dtype=float32, numpy=\n",
       " array([[ 0.07531695,  0.06428696,  0.06990725, ...,  0.055079  ,\n",
       "          0.00066985, -0.05107821],\n",
       "        [ 0.05469356,  0.05387262, -0.0416443 , ..., -0.03119045,\n",
       "          0.0102316 ,  0.07232903],\n",
       "        [-0.02286955,  0.05385177,  0.0756866 , ..., -0.02611945,\n",
       "          0.02291939,  0.0380661 ],\n",
       "        ...,\n",
       "        [ 0.03041444,  0.07093401,  0.06709787, ...,  0.03425878,\n",
       "         -0.01465827,  0.01852486],\n",
       "        [-0.06564058,  0.04024745, -0.06805687, ...,  0.06872854,\n",
       "          0.00028025, -0.06044695],\n",
       "        [-0.0453898 , -0.02092795, -0.04542916, ..., -0.04152139,\n",
       "          0.0341448 , -0.06919041]], dtype=float32)>,\n",
       " <tf.Variable 'dense/bias:0' shape=(200,) dtype=float32, numpy=\n",
       " array([0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.], dtype=float32)>,\n",
       " <tf.Variable 'dense_1/kernel:0' shape=(200, 200) dtype=float32, numpy=\n",
       " array([[-0.05358986, -0.07282169, -0.02902377, ..., -0.00638223,\n",
       "          0.00855548,  0.04686414],\n",
       "        [-0.11222805, -0.05997095, -0.08057142, ...,  0.12127394,\n",
       "         -0.06117078, -0.02701435],\n",
       "        [ 0.06351034,  0.09384722, -0.10306284, ..., -0.01485329,\n",
       "          0.10985725,  0.09992235],\n",
       "        ...,\n",
       "        [ 0.10753558,  0.00190496, -0.05303608, ...,  0.11618786,\n",
       "          0.04867258,  0.03522166],\n",
       "        [-0.02623229, -0.05522075, -0.05485192, ...,  0.03620482,\n",
       "         -0.10127217, -0.05719071],\n",
       "        [ 0.00191043,  0.10155062,  0.02649751, ..., -0.04057585,\n",
       "          0.12126162, -0.05008212]], dtype=float32)>,\n",
       " <tf.Variable 'dense_1/bias:0' shape=(200,) dtype=float32, numpy=\n",
       " array([0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,\n",
       "        0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.], dtype=float32)>,\n",
       " <tf.Variable 'dense_2/kernel:0' shape=(200, 10) dtype=float32, numpy=\n",
       " array([[-0.14755026, -0.11048468, -0.0638601 , ..., -0.00584731,\n",
       "         -0.05747922,  0.02666262],\n",
       "        [ 0.05499838,  0.11837682, -0.14282788, ...,  0.04858112,\n",
       "         -0.07401592,  0.16873437],\n",
       "        [ 0.10247996, -0.10167258,  0.03466944, ...,  0.1384592 ,\n",
       "          0.13892135,  0.14194527],\n",
       "        ...,\n",
       "        [-0.09815882, -0.05886719, -0.09074002, ...,  0.13999024,\n",
       "          0.01369716, -0.14843278],\n",
       "        [ 0.00070037, -0.13716882,  0.15620473, ..., -0.00615644,\n",
       "          0.04400617,  0.13976592],\n",
       "        [ 0.07033361,  0.07495596, -0.1359672 , ...,  0.11491573,\n",
       "         -0.04093248,  0.0851528 ]], dtype=float32)>,\n",
       " <tf.Variable 'dense_2/bias:0' shape=(10,) dtype=float32, numpy=array([0., 0., 0., 0., 0., 0., 0., 0., 0., 0.], dtype=float32)>]"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.trainable_variables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "n_epochs = 5\n",
    "batch_size = 32\n",
    "n_steps = len(X_train) // batch_size\n",
    "optimizer_lower = keras.optimizers.Nadam(learning_rate=0.01)\n",
    "optimizer_upper = keras.optimizers.Adam(learning_rate=0.001)\n",
    "loss_fn = keras.losses.sparse_categorical_crossentropy\n",
    "mean_loss = keras.metrics.Mean()\n",
    "mean_validation_loss = keras.metrics.Mean()\n",
    "metrics = [keras.metrics.SparseCategoricalAccuracy()]\n",
    "validation_metrics = [keras.metrics.SparseCategoricalAccuracy()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "def print_status_bar(iteration, total, loss, metrics=None):\n",
    "    metrics = \" - \".join([\"{}: {:.4f}\".format(m.name, m.result())\n",
    "                         for m in [loss] + (metrics or [])])\n",
    "    end = \"\" if iteration < total else \"\\n\"\n",
    "    print(\"\\r{}/{} - \".format(iteration, total) + metrics,\n",
    "          end=end)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1 / 5\n",
      "Step 1 / 1718\n",
      "32/55000 - mean: 0.0009 - sparse_categorical_accuracy: 1.0000Step 2 / 1718\n",
      "64/55000 - mean: 0.0012 - sparse_categorical_accuracy: 1.0000Step 3 / 1718\n",
      "96/55000 - mean: 0.0009 - sparse_categorical_accuracy: 1.0000Step 4 / 1718\n",
      "128/55000 - mean: 0.0069 - sparse_categorical_accuracy: 1.0000Step 5 / 1718\n",
      "160/55000 - mean: 0.0058 - sparse_categorical_accuracy: 1.0000Step 6 / 1718\n",
      "192/55000 - mean: 0.0056 - sparse_categorical_accuracy: 1.0000Step 7 / 1718\n",
      "224/55000 - mean: 0.0051 - sparse_categorical_accuracy: 1.0000Step 8 / 1718\n",
      "256/55000 - mean: 0.0045 - sparse_categorical_accuracy: 1.0000Step 9 / 1718\n",
      "288/55000 - mean: 0.0078 - sparse_categorical_accuracy: 0.9965Step 10 / 1718\n",
      "320/55000 - mean: 0.0092 - sparse_categorical_accuracy: 0.9969Step 11 / 1718\n",
      "352/55000 - mean: 0.0111 - sparse_categorical_accuracy: 0.9943Step 12 / 1718\n",
      "384/55000 - mean: 0.0103 - sparse_categorical_accuracy: 0.9948Step 13 / 1718\n",
      "416/55000 - mean: 0.0097 - sparse_categorical_accuracy: 0.9952Step 14 / 1718\n",
      "448/55000 - mean: 0.0096 - sparse_categorical_accuracy: 0.9955Step 15 / 1718\n",
      "480/55000 - mean: 0.0091 - sparse_categorical_accuracy: 0.9958Step 16 / 1718\n",
      "512/55000 - mean: 0.0085 - sparse_categorical_accuracy: 0.9961Step 17 / 1718\n",
      "544/55000 - mean: 0.0081 - sparse_categorical_accuracy: 0.9963Step 18 / 1718\n",
      "576/55000 - mean: 0.0076 - sparse_categorical_accuracy: 0.9965Step 19 / 1718\n",
      "608/55000 - mean: 0.0074 - sparse_categorical_accuracy: 0.9967Step 20 / 1718\n",
      "640/55000 - mean: 0.0095 - sparse_categorical_accuracy: 0.9953Step 21 / 1718\n",
      "672/55000 - mean: 0.0092 - sparse_categorical_accuracy: 0.9955Step 22 / 1718\n",
      "704/55000 - mean: 0.0088 - sparse_categorical_accuracy: 0.9957Step 23 / 1718\n",
      "736/55000 - mean: 0.0085 - sparse_categorical_accuracy: 0.9959Step 24 / 1718\n",
      "768/55000 - mean: 0.0103 - sparse_categorical_accuracy: 0.9948Step 25 / 1718\n",
      "800/55000 - mean: 0.0101 - sparse_categorical_accuracy: 0.9950Step 26 / 1718\n",
      "832/55000 - mean: 0.0185 - sparse_categorical_accuracy: 0.9928Step 27 / 1718\n",
      "864/55000 - mean: 0.0178 - sparse_categorical_accuracy: 0.9931Step 28 / 1718\n",
      "896/55000 - mean: 0.0172 - sparse_categorical_accuracy: 0.9933Step 29 / 1718\n",
      "928/55000 - mean: 0.0167 - sparse_categorical_accuracy: 0.9935Step 30 / 1718\n",
      "960/55000 - mean: 0.0161 - sparse_categorical_accuracy: 0.9937Step 31 / 1718\n",
      "992/55000 - mean: 0.0156 - sparse_categorical_accuracy: 0.9940Step 32 / 1718\n",
      "1024/55000 - mean: 0.0151 - sparse_categorical_accuracy: 0.9941Step 33 / 1718\n",
      "1056/55000 - mean: 0.0147 - sparse_categorical_accuracy: 0.9943Step 34 / 1718\n",
      "1088/55000 - mean: 0.0143 - sparse_categorical_accuracy: 0.9945Step 35 / 1718\n",
      "1120/55000 - mean: 0.0151 - sparse_categorical_accuracy: 0.9937Step 36 / 1718\n",
      "1152/55000 - mean: 0.0148 - sparse_categorical_accuracy: 0.9939Step 37 / 1718\n",
      "1184/55000 - mean: 0.0154 - sparse_categorical_accuracy: 0.9932Step 38 / 1718\n",
      "1216/55000 - mean: 0.0156 - sparse_categorical_accuracy: 0.9934Step 39 / 1718\n",
      "1248/55000 - mean: 0.0152 - sparse_categorical_accuracy: 0.9936Step 40 / 1718\n",
      "1280/55000 - mean: 0.0152 - sparse_categorical_accuracy: 0.9937Step 41 / 1718\n",
      "1312/55000 - mean: 0.0148 - sparse_categorical_accuracy: 0.9939Step 42 / 1718\n",
      "1344/55000 - mean: 0.0150 - sparse_categorical_accuracy: 0.9933Step 43 / 1718\n",
      "1376/55000 - mean: 0.0148 - sparse_categorical_accuracy: 0.9935Step 44 / 1718\n",
      "1408/55000 - mean: 0.0145 - sparse_categorical_accuracy: 0.9936Step 45 / 1718\n",
      "1440/55000 - mean: 0.0142 - sparse_categorical_accuracy: 0.9937Step 46 / 1718\n",
      "1472/55000 - mean: 0.0146 - sparse_categorical_accuracy: 0.9932Step 47 / 1718\n",
      "1504/55000 - mean: 0.0143 - sparse_categorical_accuracy: 0.9934Step 48 / 1718\n",
      "1536/55000 - mean: 0.0140 - sparse_categorical_accuracy: 0.9935Step 49 / 1718\n",
      "1568/55000 - mean: 0.0137 - sparse_categorical_accuracy: 0.9936Step 50 / 1718\n",
      "1600/55000 - mean: 0.0134 - sparse_categorical_accuracy: 0.9937Step 51 / 1718\n",
      "1632/55000 - mean: 0.0132 - sparse_categorical_accuracy: 0.9939Step 52 / 1718\n",
      "1664/55000 - mean: 0.0130 - sparse_categorical_accuracy: 0.9940Step 53 / 1718\n",
      "1696/55000 - mean: 0.0128 - sparse_categorical_accuracy: 0.9941Step 54 / 1718\n",
      "1728/55000 - mean: 0.0128 - sparse_categorical_accuracy: 0.9942Step 55 / 1718\n",
      "1760/55000 - mean: 0.0129 - sparse_categorical_accuracy: 0.9943Step 56 / 1718\n",
      "1792/55000 - mean: 0.0126 - sparse_categorical_accuracy: 0.9944Step 57 / 1718\n",
      "1824/55000 - mean: 0.0144 - sparse_categorical_accuracy: 0.9940Step 58 / 1718\n",
      "1856/55000 - mean: 0.0143 - sparse_categorical_accuracy: 0.9941Step 59 / 1718\n",
      "1888/55000 - mean: 0.0156 - sparse_categorical_accuracy: 0.9936Step 60 / 1718\n",
      "1920/55000 - mean: 0.0153 - sparse_categorical_accuracy: 0.9937Step 61 / 1718\n",
      "1952/55000 - mean: 0.0152 - sparse_categorical_accuracy: 0.9939Step 62 / 1718\n",
      "1984/55000 - mean: 0.0150 - sparse_categorical_accuracy: 0.9940Step 63 / 1718\n",
      "2016/55000 - mean: 0.0151 - sparse_categorical_accuracy: 0.9936Step 64 / 1718\n",
      "2048/55000 - mean: 0.0149 - sparse_categorical_accuracy: 0.9937Step 65 / 1718\n",
      "2080/55000 - mean: 0.0147 - sparse_categorical_accuracy: 0.9937Step 66 / 1718\n",
      "2112/55000 - mean: 0.0150 - sparse_categorical_accuracy: 0.9934Step 67 / 1718\n",
      "2144/55000 - mean: 0.0147 - sparse_categorical_accuracy: 0.9935Step 68 / 1718\n",
      "2176/55000 - mean: 0.0164 - sparse_categorical_accuracy: 0.9926Step 69 / 1718\n",
      "2208/55000 - mean: 0.0163 - sparse_categorical_accuracy: 0.9928Step 70 / 1718\n",
      "2240/55000 - mean: 0.0162 - sparse_categorical_accuracy: 0.9929Step 71 / 1718\n",
      "2272/55000 - mean: 0.0206 - sparse_categorical_accuracy: 0.9921Step 72 / 1718\n",
      "2304/55000 - mean: 0.0205 - sparse_categorical_accuracy: 0.9922Step 73 / 1718\n",
      "2336/55000 - mean: 0.0205 - sparse_categorical_accuracy: 0.9923Step 74 / 1718\n",
      "2368/55000 - mean: 0.0202 - sparse_categorical_accuracy: 0.9924Step 75 / 1718\n",
      "2400/55000 - mean: 0.0200 - sparse_categorical_accuracy: 0.9925Step 76 / 1718\n",
      "2432/55000 - mean: 0.0197 - sparse_categorical_accuracy: 0.9926Step 77 / 1718\n",
      "2464/55000 - mean: 0.0194 - sparse_categorical_accuracy: 0.9927Step 78 / 1718\n",
      "2496/55000 - mean: 0.0192 - sparse_categorical_accuracy: 0.9928Step 79 / 1718\n",
      "2528/55000 - mean: 0.0191 - sparse_categorical_accuracy: 0.9929Step 80 / 1718\n",
      "2560/55000 - mean: 0.0189 - sparse_categorical_accuracy: 0.9930Step 81 / 1718\n",
      "2592/55000 - mean: 0.0187 - sparse_categorical_accuracy: 0.9931Step 82 / 1718\n",
      "2624/55000 - mean: 0.0189 - sparse_categorical_accuracy: 0.9928Step 83 / 1718\n",
      "2656/55000 - mean: 0.0187 - sparse_categorical_accuracy: 0.9928Step 84 / 1718\n",
      "2688/55000 - mean: 0.0185 - sparse_categorical_accuracy: 0.9929Step 85 / 1718\n",
      "2720/55000 - mean: 0.0184 - sparse_categorical_accuracy: 0.9930Step 86 / 1718\n",
      "2752/55000 - mean: 0.0191 - sparse_categorical_accuracy: 0.9927Step 87 / 1718\n",
      "2784/55000 - mean: 0.0189 - sparse_categorical_accuracy: 0.9928Step 88 / 1718\n",
      "2816/55000 - mean: 0.0187 - sparse_categorical_accuracy: 0.9929Step 89 / 1718\n",
      "2848/55000 - mean: 0.0185 - sparse_categorical_accuracy: 0.9930Step 90 / 1718\n",
      "2880/55000 - mean: 0.0183 - sparse_categorical_accuracy: 0.9931Step 91 / 1718\n",
      "2912/55000 - mean: 0.0181 - sparse_categorical_accuracy: 0.9931Step 92 / 1718\n",
      "2944/55000 - mean: 0.0180 - sparse_categorical_accuracy: 0.9932Step 93 / 1718\n",
      "2976/55000 - mean: 0.0193 - sparse_categorical_accuracy: 0.9926Step 94 / 1718\n",
      "3008/55000 - mean: 0.0192 - sparse_categorical_accuracy: 0.9927Step 95 / 1718\n",
      "3040/55000 - mean: 0.0191 - sparse_categorical_accuracy: 0.9928Step 96 / 1718\n",
      "3072/55000 - mean: 0.0189 - sparse_categorical_accuracy: 0.9928Step 97 / 1718\n",
      "3104/55000 - mean: 0.0206 - sparse_categorical_accuracy: 0.9926Step 98 / 1718\n",
      "3136/55000 - mean: 0.0205 - sparse_categorical_accuracy: 0.9927Step 99 / 1718\n",
      "3168/55000 - mean: 0.0203 - sparse_categorical_accuracy: 0.9927Step 100 / 1718\n",
      "3200/55000 - mean: 0.0201 - sparse_categorical_accuracy: 0.9928Step 101 / 1718\n",
      "3232/55000 - mean: 0.0207 - sparse_categorical_accuracy: 0.9926Step 102 / 1718\n",
      "3264/55000 - mean: 0.0206 - sparse_categorical_accuracy: 0.9926Step 103 / 1718\n",
      "3296/55000 - mean: 0.0204 - sparse_categorical_accuracy: 0.9927Step 104 / 1718\n",
      "3328/55000 - mean: 0.0202 - sparse_categorical_accuracy: 0.9928Step 105 / 1718\n",
      "3360/55000 - mean: 0.0200 - sparse_categorical_accuracy: 0.9929Step 106 / 1718\n",
      "3392/55000 - mean: 0.0198 - sparse_categorical_accuracy: 0.9929Step 107 / 1718\n",
      "3424/55000 - mean: 0.0197 - sparse_categorical_accuracy: 0.9930Step 108 / 1718\n",
      "3456/55000 - mean: 0.0195 - sparse_categorical_accuracy: 0.9931Step 109 / 1718\n",
      "3488/55000 - mean: 0.0193 - sparse_categorical_accuracy: 0.9931Step 110 / 1718\n",
      "3520/55000 - mean: 0.0191 - sparse_categorical_accuracy: 0.9932Step 111 / 1718\n",
      "3552/55000 - mean: 0.0190 - sparse_categorical_accuracy: 0.9932Step 112 / 1718\n",
      "3584/55000 - mean: 0.0189 - sparse_categorical_accuracy: 0.9933Step 113 / 1718\n",
      "3616/55000 - mean: 0.0188 - sparse_categorical_accuracy: 0.9934Step 114 / 1718\n",
      "3648/55000 - mean: 0.0186 - sparse_categorical_accuracy: 0.9934Step 115 / 1718\n",
      "3680/55000 - mean: 0.0185 - sparse_categorical_accuracy: 0.9935Step 116 / 1718\n",
      "3712/55000 - mean: 0.0185 - sparse_categorical_accuracy: 0.9935Step 117 / 1718\n",
      "3744/55000 - mean: 0.0215 - sparse_categorical_accuracy: 0.9931Step 118 / 1718\n",
      "3776/55000 - mean: 0.0214 - sparse_categorical_accuracy: 0.9931Step 119 / 1718\n",
      "3808/55000 - mean: 0.0212 - sparse_categorical_accuracy: 0.9932Step 120 / 1718\n",
      "3840/55000 - mean: 0.0210 - sparse_categorical_accuracy: 0.9932Step 121 / 1718\n",
      "3872/55000 - mean: 0.0221 - sparse_categorical_accuracy: 0.9930Step 122 / 1718\n",
      "3904/55000 - mean: 0.0234 - sparse_categorical_accuracy: 0.9926Step 123 / 1718\n",
      "3936/55000 - mean: 0.0232 - sparse_categorical_accuracy: 0.9926Step 124 / 1718\n",
      "3968/55000 - mean: 0.0230 - sparse_categorical_accuracy: 0.9927Step 125 / 1718\n",
      "4000/55000 - mean: 0.0228 - sparse_categorical_accuracy: 0.9927Step 126 / 1718\n",
      "4032/55000 - mean: 0.0227 - sparse_categorical_accuracy: 0.9928Step 127 / 1718\n",
      "4064/55000 - mean: 0.0225 - sparse_categorical_accuracy: 0.9929Step 128 / 1718\n",
      "4096/55000 - mean: 0.0236 - sparse_categorical_accuracy: 0.9924Step 129 / 1718\n",
      "4128/55000 - mean: 0.0234 - sparse_categorical_accuracy: 0.9925Step 130 / 1718\n",
      "4160/55000 - mean: 0.0232 - sparse_categorical_accuracy: 0.9925Step 131 / 1718\n",
      "4192/55000 - mean: 0.0230 - sparse_categorical_accuracy: 0.9926Step 132 / 1718\n",
      "4224/55000 - mean: 0.0232 - sparse_categorical_accuracy: 0.9924Step 133 / 1718\n",
      "4256/55000 - mean: 0.0231 - sparse_categorical_accuracy: 0.9925Step 134 / 1718\n",
      "4288/55000 - mean: 0.0231 - sparse_categorical_accuracy: 0.9923Step 135 / 1718\n",
      "4320/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9921Step 136 / 1718\n",
      "4352/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9922Step 137 / 1718\n",
      "4384/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9922Step 138 / 1718\n",
      "4416/55000 - mean: 0.0235 - sparse_categorical_accuracy: 0.9923Step 139 / 1718\n",
      "4448/55000 - mean: 0.0235 - sparse_categorical_accuracy: 0.9924Step 140 / 1718\n",
      "4480/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9920Step 141 / 1718\n",
      "4512/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9918Step 142 / 1718\n",
      "4544/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9919Step 143 / 1718\n",
      "4576/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9919Step 144 / 1718\n",
      "4608/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9920Step 145 / 1718\n",
      "4640/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9918Step 146 / 1718\n",
      "4672/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9919Step 147 / 1718\n",
      "4704/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9919Step 148 / 1718\n",
      "4736/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9920Step 149 / 1718\n",
      "4768/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9920Step 150 / 1718\n",
      "4800/55000 - mean: 0.0236 - sparse_categorical_accuracy: 0.9921Step 151 / 1718\n",
      "4832/55000 - mean: 0.0234 - sparse_categorical_accuracy: 0.9921Step 152 / 1718\n",
      "4864/55000 - mean: 0.0233 - sparse_categorical_accuracy: 0.9922Step 153 / 1718\n",
      "4896/55000 - mean: 0.0232 - sparse_categorical_accuracy: 0.9922Step 154 / 1718\n",
      "4928/55000 - mean: 0.0230 - sparse_categorical_accuracy: 0.9923Step 155 / 1718\n",
      "4960/55000 - mean: 0.0229 - sparse_categorical_accuracy: 0.9923Step 156 / 1718\n",
      "4992/55000 - mean: 0.0228 - sparse_categorical_accuracy: 0.9924Step 157 / 1718\n",
      "5024/55000 - mean: 0.0226 - sparse_categorical_accuracy: 0.9924Step 158 / 1718\n",
      "5056/55000 - mean: 0.0227 - sparse_categorical_accuracy: 0.9923Step 159 / 1718\n",
      "5088/55000 - mean: 0.0227 - sparse_categorical_accuracy: 0.9923Step 160 / 1718\n",
      "5120/55000 - mean: 0.0230 - sparse_categorical_accuracy: 0.9922Step 161 / 1718\n",
      "5152/55000 - mean: 0.0229 - sparse_categorical_accuracy: 0.9922Step 162 / 1718\n",
      "5184/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9921Step 163 / 1718\n",
      "5216/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9921Step 164 / 1718\n",
      "5248/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9922Step 165 / 1718\n",
      "5280/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9917Step 166 / 1718\n",
      "5312/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9913Step 167 / 1718\n",
      "5344/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9914Step 168 / 1718\n",
      "5376/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9914Step 169 / 1718\n",
      "5408/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9915Step 170 / 1718\n",
      "5440/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9915Step 171 / 1718\n",
      "5472/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9916Step 172 / 1718\n",
      "5504/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9916Step 173 / 1718\n",
      "5536/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9917Step 174 / 1718\n",
      "5568/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9917Step 175 / 1718\n",
      "5600/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9918Step 176 / 1718\n",
      "5632/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9918Step 177 / 1718\n",
      "5664/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9919Step 178 / 1718\n",
      "5696/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9919Step 179 / 1718\n",
      "5728/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9920Step 180 / 1718\n",
      "5760/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9920Step 181 / 1718\n",
      "5792/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9921Step 182 / 1718\n",
      "5824/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9921Step 183 / 1718\n",
      "5856/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9920Step 184 / 1718\n",
      "5888/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9920Step 185 / 1718\n",
      "5920/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9921Step 186 / 1718\n",
      "5952/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9921Step 187 / 1718\n",
      "5984/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9921Step 188 / 1718\n",
      "6016/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9920Step 189 / 1718\n",
      "6048/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9921Step 190 / 1718\n",
      "6080/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9921Step 191 / 1718\n",
      "6112/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9921Step 192 / 1718\n",
      "6144/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9922Step 193 / 1718\n",
      "6176/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9922Step 194 / 1718\n",
      "6208/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9923Step 195 / 1718\n",
      "6240/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9921Step 196 / 1718\n",
      "6272/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9922Step 197 / 1718\n",
      "6304/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9922Step 198 / 1718\n",
      "6336/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9923Step 199 / 1718\n",
      "6368/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9921Step 200 / 1718\n",
      "6400/55000 - mean: 0.0236 - sparse_categorical_accuracy: 0.9922Step 201 / 1718\n",
      "6432/55000 - mean: 0.0235 - sparse_categorical_accuracy: 0.9922Step 202 / 1718\n",
      "6464/55000 - mean: 0.0234 - sparse_categorical_accuracy: 0.9923Step 203 / 1718\n",
      "6496/55000 - mean: 0.0233 - sparse_categorical_accuracy: 0.9923Step 204 / 1718\n",
      "6528/55000 - mean: 0.0232 - sparse_categorical_accuracy: 0.9923Step 205 / 1718\n",
      "6560/55000 - mean: 0.0231 - sparse_categorical_accuracy: 0.9924Step 206 / 1718\n",
      "6592/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9923Step 207 / 1718\n",
      "6624/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9923Step 208 / 1718\n",
      "6656/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9923Step 209 / 1718\n",
      "6688/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9924Step 210 / 1718\n",
      "6720/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9924Step 211 / 1718\n",
      "6752/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9924Step 212 / 1718\n",
      "6784/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9925Step 213 / 1718\n",
      "6816/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9925Step 214 / 1718\n",
      "6848/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9923Step 215 / 1718\n",
      "6880/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9923Step 216 / 1718\n",
      "6912/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9922Step 217 / 1718\n",
      "6944/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9922Step 218 / 1718\n",
      "6976/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9923Step 219 / 1718\n",
      "7008/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9923Step 220 / 1718\n",
      "7040/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9923Step 221 / 1718\n",
      "7072/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9922Step 222 / 1718\n",
      "7104/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9923Step 223 / 1718\n",
      "7136/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9923Step 224 / 1718\n",
      "7168/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9923Step 225 / 1718\n",
      "7200/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9922Step 226 / 1718\n",
      "7232/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9923Step 227 / 1718\n",
      "7264/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9922Step 228 / 1718\n",
      "7296/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9918Step 229 / 1718\n",
      "7328/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9917Step 230 / 1718\n",
      "7360/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9917Step 231 / 1718\n",
      "7392/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9917Step 232 / 1718\n",
      "7424/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9918Step 233 / 1718\n",
      "7456/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9918Step 234 / 1718\n",
      "7488/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9917Step 235 / 1718\n",
      "7520/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9916Step 236 / 1718\n",
      "7552/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9915Step 237 / 1718\n",
      "7584/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9914Step 238 / 1718\n",
      "7616/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9915Step 239 / 1718\n",
      "7648/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9915Step 240 / 1718\n",
      "7680/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9913Step 241 / 1718\n",
      "7712/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9912Step 242 / 1718\n",
      "7744/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9912Step 243 / 1718\n",
      "7776/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9913Step 244 / 1718\n",
      "7808/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9913Step 245 / 1718\n",
      "7840/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9912Step 246 / 1718\n",
      "7872/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9912Step 247 / 1718\n",
      "7904/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9913Step 248 / 1718\n",
      "7936/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9913Step 249 / 1718\n",
      "7968/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9912Step 250 / 1718\n",
      "8000/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9912Step 251 / 1718\n",
      "8032/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9912Step 252 / 1718\n",
      "8064/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9912Step 253 / 1718\n",
      "8096/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9912Step 254 / 1718\n",
      "8128/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9913Step 255 / 1718\n",
      "8160/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9912Step 256 / 1718\n",
      "8192/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9911Step 257 / 1718\n",
      "8224/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 258 / 1718\n",
      "8256/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9910Step 259 / 1718\n",
      "8288/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9911Step 260 / 1718\n",
      "8320/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9911Step 261 / 1718\n",
      "8352/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 262 / 1718\n",
      "8384/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9911Step 263 / 1718\n",
      "8416/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9911Step 264 / 1718\n",
      "8448/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9910Step 265 / 1718\n",
      "8480/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 266 / 1718\n",
      "8512/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9908Step 267 / 1718\n",
      "8544/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 268 / 1718\n",
      "8576/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9907Step 269 / 1718\n",
      "8608/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9907Step 270 / 1718\n",
      "8640/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9907Step 271 / 1718\n",
      "8672/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9907Step 272 / 1718\n",
      "8704/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9907Step 273 / 1718\n",
      "8736/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9907Step 274 / 1718\n",
      "8768/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9906Step 275 / 1718\n",
      "8800/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9907Step 276 / 1718\n",
      "8832/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9907Step 277 / 1718\n",
      "8864/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9907Step 278 / 1718\n",
      "8896/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9907Step 279 / 1718\n",
      "8928/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9907Step 280 / 1718\n",
      "8960/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9907Step 281 / 1718\n",
      "8992/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9908Step 282 / 1718\n",
      "9024/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9908Step 283 / 1718\n",
      "9056/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9908Step 284 / 1718\n",
      "9088/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9909Step 285 / 1718\n",
      "9120/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9908Step 286 / 1718\n",
      "9152/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9908Step 287 / 1718\n",
      "9184/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9909Step 288 / 1718\n",
      "9216/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9909Step 289 / 1718\n",
      "9248/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9909Step 290 / 1718\n",
      "9280/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9909Step 291 / 1718\n",
      "9312/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9909Step 292 / 1718\n",
      "9344/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9909Step 293 / 1718\n",
      "9376/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9909Step 294 / 1718\n",
      "9408/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9909Step 295 / 1718\n",
      "9440/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9909Step 296 / 1718\n",
      "9472/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9909Step 297 / 1718\n",
      "9504/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9908Step 298 / 1718\n",
      "9536/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9908Step 299 / 1718\n",
      "9568/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9908Step 300 / 1718\n",
      "9600/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9908Step 301 / 1718\n",
      "9632/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9909Step 302 / 1718\n",
      "9664/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9908Step 303 / 1718\n",
      "9696/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9908Step 304 / 1718\n",
      "9728/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9909Step 305 / 1718\n",
      "9760/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 306 / 1718\n",
      "9792/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9908Step 307 / 1718\n",
      "9824/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9908Step 308 / 1718\n",
      "9856/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9909Step 309 / 1718\n",
      "9888/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9909Step 310 / 1718\n",
      "9920/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9909Step 311 / 1718\n",
      "9952/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9910Step 312 / 1718\n",
      "9984/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9909Step 313 / 1718\n",
      "10016/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9909Step 314 / 1718\n",
      "10048/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9909Step 315 / 1718\n",
      "10080/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9910Step 316 / 1718\n",
      "10112/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9910Step 317 / 1718\n",
      "10144/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9910Step 318 / 1718\n",
      "10176/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9911Step 319 / 1718\n",
      "10208/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9911Step 320 / 1718\n",
      "10240/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9911Step 321 / 1718\n",
      "10272/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9911Step 322 / 1718\n",
      "10304/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9912Step 323 / 1718\n",
      "10336/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9911Step 324 / 1718\n",
      "10368/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 325 / 1718\n",
      "10400/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9911Step 326 / 1718\n",
      "10432/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 327 / 1718\n",
      "10464/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 328 / 1718\n",
      "10496/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 329 / 1718\n",
      "10528/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9911Step 330 / 1718\n",
      "10560/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 331 / 1718\n",
      "10592/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 332 / 1718\n",
      "10624/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9911Step 333 / 1718\n",
      "10656/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9911Step 334 / 1718\n",
      "10688/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 335 / 1718\n",
      "10720/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9911Step 336 / 1718\n",
      "10752/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 337 / 1718\n",
      "10784/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 338 / 1718\n",
      "10816/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 339 / 1718\n",
      "10848/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 340 / 1718\n",
      "10880/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9909Step 341 / 1718\n",
      "10912/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 342 / 1718\n",
      "10944/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9909Step 343 / 1718\n",
      "10976/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9909Step 344 / 1718\n",
      "11008/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9909Step 345 / 1718\n",
      "11040/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 346 / 1718\n",
      "11072/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 347 / 1718\n",
      "11104/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 348 / 1718\n",
      "11136/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 349 / 1718\n",
      "11168/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 350 / 1718\n",
      "11200/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 351 / 1718\n",
      "11232/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 352 / 1718\n",
      "11264/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9910Step 353 / 1718\n",
      "11296/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9911Step 354 / 1718\n",
      "11328/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9911Step 355 / 1718\n",
      "11360/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9911Step 356 / 1718\n",
      "11392/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9911Step 357 / 1718\n",
      "11424/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9912Step 358 / 1718\n",
      "11456/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9911Step 359 / 1718\n",
      "11488/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9911Step 360 / 1718\n",
      "11520/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9911Step 361 / 1718\n",
      "11552/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9911Step 362 / 1718\n",
      "11584/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9911Step 363 / 1718\n",
      "11616/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 364 / 1718\n",
      "11648/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9912Step 365 / 1718\n",
      "11680/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9912Step 366 / 1718\n",
      "11712/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9912Step 367 / 1718\n",
      "11744/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9911Step 368 / 1718\n",
      "11776/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9911Step 369 / 1718\n",
      "11808/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9911Step 370 / 1718\n",
      "11840/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 371 / 1718\n",
      "11872/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9912Step 372 / 1718\n",
      "11904/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9912Step 373 / 1718\n",
      "11936/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9912Step 374 / 1718\n",
      "11968/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9912Step 375 / 1718\n",
      "12000/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9912Step 376 / 1718\n",
      "12032/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9913Step 377 / 1718\n",
      "12064/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9913Step 378 / 1718\n",
      "12096/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9913Step 379 / 1718\n",
      "12128/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9913Step 380 / 1718\n",
      "12160/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9914Step 381 / 1718\n",
      "12192/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9914Step 382 / 1718\n",
      "12224/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9912Step 383 / 1718\n",
      "12256/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9912Step 384 / 1718\n",
      "12288/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9912Step 385 / 1718\n",
      "12320/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9912Step 386 / 1718\n",
      "12352/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9913Step 387 / 1718\n",
      "12384/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9913Step 388 / 1718\n",
      "12416/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9911Step 389 / 1718\n",
      "12448/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9911Step 390 / 1718\n",
      "12480/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9911Step 391 / 1718\n",
      "12512/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9911Step 392 / 1718\n",
      "12544/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9912Step 393 / 1718\n",
      "12576/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9912Step 394 / 1718\n",
      "12608/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9912Step 395 / 1718\n",
      "12640/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9912Step 396 / 1718\n",
      "12672/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9912Step 397 / 1718\n",
      "12704/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9912Step 398 / 1718\n",
      "12736/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9911Step 399 / 1718\n",
      "12768/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9911Step 400 / 1718\n",
      "12800/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9911Step 401 / 1718\n",
      "12832/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9911Step 402 / 1718\n",
      "12864/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9911Step 403 / 1718\n",
      "12896/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9912Step 404 / 1718\n",
      "12928/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9912Step 405 / 1718\n",
      "12960/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9912Step 406 / 1718\n",
      "12992/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9912Step 407 / 1718\n",
      "13024/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9912Step 408 / 1718\n",
      "13056/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9913Step 409 / 1718\n",
      "13088/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9913Step 410 / 1718\n",
      "13120/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9913Step 411 / 1718\n",
      "13152/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9913Step 412 / 1718\n",
      "13184/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9913Step 413 / 1718\n",
      "13216/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9912Step 414 / 1718\n",
      "13248/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9912Step 415 / 1718\n",
      "13280/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9912Step 416 / 1718\n",
      "13312/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9912Step 417 / 1718\n",
      "13344/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9912Step 418 / 1718\n",
      "13376/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9912Step 419 / 1718\n",
      "13408/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9912Step 420 / 1718\n",
      "13440/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9911Step 421 / 1718\n",
      "13472/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9912Step 422 / 1718\n",
      "13504/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9912Step 423 / 1718\n",
      "13536/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9912Step 424 / 1718\n",
      "13568/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9912Step 425 / 1718\n",
      "13600/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9912Step 426 / 1718\n",
      "13632/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9913Step 427 / 1718\n",
      "13664/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9913Step 428 / 1718\n",
      "13696/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9913Step 429 / 1718\n",
      "13728/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9913Step 430 / 1718\n",
      "13760/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9914Step 431 / 1718\n",
      "13792/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9914Step 432 / 1718\n",
      "13824/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9914Step 433 / 1718\n",
      "13856/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9914Step 434 / 1718\n",
      "13888/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9914Step 435 / 1718\n",
      "13920/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9915Step 436 / 1718\n",
      "13952/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9915Step 437 / 1718\n",
      "13984/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9915Step 438 / 1718\n",
      "14016/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9915Step 439 / 1718\n",
      "14048/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9915Step 440 / 1718\n",
      "14080/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9915Step 441 / 1718\n",
      "14112/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9916Step 442 / 1718\n",
      "14144/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9916Step 443 / 1718\n",
      "14176/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9916Step 444 / 1718\n",
      "14208/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9916Step 445 / 1718\n",
      "14240/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9916Step 446 / 1718\n",
      "14272/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9915Step 447 / 1718\n",
      "14304/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9915Step 448 / 1718\n",
      "14336/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9916Step 449 / 1718\n",
      "14368/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9916Step 450 / 1718\n",
      "14400/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9916Step 451 / 1718\n",
      "14432/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9915Step 452 / 1718\n",
      "14464/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9916Step 453 / 1718\n",
      "14496/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9916Step 454 / 1718\n",
      "14528/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9916Step 455 / 1718\n",
      "14560/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9916Step 456 / 1718\n",
      "14592/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9916Step 457 / 1718\n",
      "14624/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9917Step 458 / 1718\n",
      "14656/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9917Step 459 / 1718\n",
      "14688/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9917Step 460 / 1718\n",
      "14720/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9917Step 461 / 1718\n",
      "14752/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9917Step 462 / 1718\n",
      "14784/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9917Step 463 / 1718\n",
      "14816/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9918Step 464 / 1718\n",
      "14848/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9918Step 465 / 1718\n",
      "14880/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9918Step 466 / 1718\n",
      "14912/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9917Step 467 / 1718\n",
      "14944/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9917Step 468 / 1718\n",
      "14976/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9917Step 469 / 1718\n",
      "15008/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9917Step 470 / 1718\n",
      "15040/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9918Step 471 / 1718\n",
      "15072/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9918Step 472 / 1718\n",
      "15104/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9918Step 473 / 1718\n",
      "15136/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9918Step 474 / 1718\n",
      "15168/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9918Step 475 / 1718\n",
      "15200/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9918Step 476 / 1718\n",
      "15232/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9919Step 477 / 1718\n",
      "15264/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9919Step 478 / 1718\n",
      "15296/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9919Step 479 / 1718\n",
      "15328/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9919Step 480 / 1718\n",
      "15360/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9919Step 481 / 1718\n",
      "15392/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9918Step 482 / 1718\n",
      "15424/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9918Step 483 / 1718\n",
      "15456/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9918Step 484 / 1718\n",
      "15488/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9919Step 485 / 1718\n",
      "15520/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9919Step 486 / 1718\n",
      "15552/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9919Step 487 / 1718\n",
      "15584/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9919Step 488 / 1718\n",
      "15616/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9919Step 489 / 1718\n",
      "15648/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9919Step 490 / 1718\n",
      "15680/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9919Step 491 / 1718\n",
      "15712/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9919Step 492 / 1718\n",
      "15744/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9919Step 493 / 1718\n",
      "15776/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9919Step 494 / 1718\n",
      "15808/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9920Step 495 / 1718\n",
      "15840/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9920Step 496 / 1718\n",
      "15872/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9920Step 497 / 1718\n",
      "15904/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9920Step 498 / 1718\n",
      "15936/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9920Step 499 / 1718\n",
      "15968/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9920Step 500 / 1718\n",
      "16000/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9920Step 501 / 1718\n",
      "16032/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9920Step 502 / 1718\n",
      "16064/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9920Step 503 / 1718\n",
      "16096/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9920Step 504 / 1718\n",
      "16128/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9921Step 505 / 1718\n",
      "16160/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9921Step 506 / 1718\n",
      "16192/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9921Step 507 / 1718\n",
      "16224/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9921Step 508 / 1718\n",
      "16256/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9921Step 509 / 1718\n",
      "16288/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9921Step 510 / 1718\n",
      "16320/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9922Step 511 / 1718\n",
      "16352/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9922Step 512 / 1718\n",
      "16384/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9922Step 513 / 1718\n",
      "16416/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9922Step 514 / 1718\n",
      "16448/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9922Step 515 / 1718\n",
      "16480/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9922Step 516 / 1718\n",
      "16512/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9922Step 517 / 1718\n",
      "16544/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9923Step 518 / 1718\n",
      "16576/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9923Step 519 / 1718\n",
      "16608/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9922Step 520 / 1718\n",
      "16640/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9922Step 521 / 1718\n",
      "16672/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9922Step 522 / 1718\n",
      "16704/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9922Step 523 / 1718\n",
      "16736/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9922Step 524 / 1718\n",
      "16768/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9922Step 525 / 1718\n",
      "16800/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9923Step 526 / 1718\n",
      "16832/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9923Step 527 / 1718\n",
      "16864/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9923Step 528 / 1718\n",
      "16896/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9923Step 529 / 1718\n",
      "16928/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9923Step 530 / 1718\n",
      "16960/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9923Step 531 / 1718\n",
      "16992/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9922Step 532 / 1718\n",
      "17024/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9922Step 533 / 1718\n",
      "17056/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9922Step 534 / 1718\n",
      "17088/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9922Step 535 / 1718\n",
      "17120/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9922Step 536 / 1718\n",
      "17152/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9922Step 537 / 1718\n",
      "17184/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9922Step 538 / 1718\n",
      "17216/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9922Step 539 / 1718\n",
      "17248/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9922Step 540 / 1718\n",
      "17280/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9922Step 541 / 1718\n",
      "17312/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9922Step 542 / 1718\n",
      "17344/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9922Step 543 / 1718\n",
      "17376/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9922Step 544 / 1718\n",
      "17408/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9922Step 545 / 1718\n",
      "17440/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9923Step 546 / 1718\n",
      "17472/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9923Step 547 / 1718\n",
      "17504/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9921Step 548 / 1718\n",
      "17536/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9921Step 549 / 1718\n",
      "17568/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9921Step 550 / 1718\n",
      "17600/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9920Step 551 / 1718\n",
      "17632/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9921Step 552 / 1718\n",
      "17664/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9921Step 553 / 1718\n",
      "17696/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9921Step 554 / 1718\n",
      "17728/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9920Step 555 / 1718\n",
      "17760/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9921Step 556 / 1718\n",
      "17792/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9921Step 557 / 1718\n",
      "17824/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9921Step 558 / 1718\n",
      "17856/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9921Step 559 / 1718\n",
      "17888/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9921Step 560 / 1718\n",
      "17920/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9921Step 561 / 1718\n",
      "17952/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9921Step 562 / 1718\n",
      "17984/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9922Step 563 / 1718\n",
      "18016/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9922Step 564 / 1718\n",
      "18048/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9921Step 565 / 1718\n",
      "18080/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9921Step 566 / 1718\n",
      "18112/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9919Step 567 / 1718\n",
      "18144/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9920Step 568 / 1718\n",
      "18176/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9920Step 569 / 1718\n",
      "18208/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9919Step 570 / 1718\n",
      "18240/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9919Step 571 / 1718\n",
      "18272/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9919Step 572 / 1718\n",
      "18304/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9919Step 573 / 1718\n",
      "18336/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9919Step 574 / 1718\n",
      "18368/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9919Step 575 / 1718\n",
      "18400/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9920Step 576 / 1718\n",
      "18432/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9919Step 577 / 1718\n",
      "18464/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9919Step 578 / 1718\n",
      "18496/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9919Step 579 / 1718\n",
      "18528/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9920Step 580 / 1718\n",
      "18560/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9920Step 581 / 1718\n",
      "18592/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9920Step 582 / 1718\n",
      "18624/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9920Step 583 / 1718\n",
      "18656/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9920Step 584 / 1718\n",
      "18688/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9920Step 585 / 1718\n",
      "18720/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9920Step 586 / 1718\n",
      "18752/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9921Step 587 / 1718\n",
      "18784/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9921Step 588 / 1718\n",
      "18816/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9921Step 589 / 1718\n",
      "18848/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9921Step 590 / 1718\n",
      "18880/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9921Step 591 / 1718\n",
      "18912/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9921Step 592 / 1718\n",
      "18944/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9921Step 593 / 1718\n",
      "18976/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9920Step 594 / 1718\n",
      "19008/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9921Step 595 / 1718\n",
      "19040/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9921Step 596 / 1718\n",
      "19072/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9920Step 597 / 1718\n",
      "19104/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9920Step 598 / 1718\n",
      "19136/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9921Step 599 / 1718\n",
      "19168/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9921Step 600 / 1718\n",
      "19200/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9921Step 601 / 1718\n",
      "19232/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9921Step 602 / 1718\n",
      "19264/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9921Step 603 / 1718\n",
      "19296/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9920Step 604 / 1718\n",
      "19328/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9920Step 605 / 1718\n",
      "19360/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9920Step 606 / 1718\n",
      "19392/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9921Step 607 / 1718\n",
      "19424/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9921Step 608 / 1718\n",
      "19456/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9920Step 609 / 1718\n",
      "19488/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9920Step 610 / 1718\n",
      "19520/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9920Step 611 / 1718\n",
      "19552/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9920Step 612 / 1718\n",
      "19584/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9920Step 613 / 1718\n",
      "19616/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9920Step 614 / 1718\n",
      "19648/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9920Step 615 / 1718\n",
      "19680/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9920Step 616 / 1718\n",
      "19712/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9920Step 617 / 1718\n",
      "19744/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9920Step 618 / 1718\n",
      "19776/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9920Step 619 / 1718\n",
      "19808/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9920Step 620 / 1718\n",
      "19840/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9920Step 621 / 1718\n",
      "19872/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9919Step 622 / 1718\n",
      "19904/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9920Step 623 / 1718\n",
      "19936/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9919Step 624 / 1718\n",
      "19968/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9919Step 625 / 1718\n",
      "20000/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9919Step 626 / 1718\n",
      "20032/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9919Step 627 / 1718\n",
      "20064/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9919Step 628 / 1718\n",
      "20096/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9919Step 629 / 1718\n",
      "20128/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9918Step 630 / 1718\n",
      "20160/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9918Step 631 / 1718\n",
      "20192/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9918Step 632 / 1718\n",
      "20224/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9918Step 633 / 1718\n",
      "20256/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9918Step 634 / 1718\n",
      "20288/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9917Step 635 / 1718\n",
      "20320/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9917Step 636 / 1718\n",
      "20352/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9917Step 637 / 1718\n",
      "20384/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9918Step 638 / 1718\n",
      "20416/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9917Step 639 / 1718\n",
      "20448/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9917Step 640 / 1718\n",
      "20480/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9917Step 641 / 1718\n",
      "20512/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9917Step 642 / 1718\n",
      "20544/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9917Step 643 / 1718\n",
      "20576/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9916Step 644 / 1718\n",
      "20608/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9917Step 645 / 1718\n",
      "20640/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9917Step 646 / 1718\n",
      "20672/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9917Step 647 / 1718\n",
      "20704/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9916Step 648 / 1718\n",
      "20736/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9916Step 649 / 1718\n",
      "20768/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9916Step 650 / 1718\n",
      "20800/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9916Step 651 / 1718\n",
      "20832/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9916Step 652 / 1718\n",
      "20864/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9916Step 653 / 1718\n",
      "20896/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9916Step 654 / 1718\n",
      "20928/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9916Step 655 / 1718\n",
      "20960/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9916Step 656 / 1718\n",
      "20992/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9916Step 657 / 1718\n",
      "21024/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9916Step 658 / 1718\n",
      "21056/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9916Step 659 / 1718\n",
      "21088/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9917Step 660 / 1718\n",
      "21120/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9917Step 661 / 1718\n",
      "21152/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9917Step 662 / 1718\n",
      "21184/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9916Step 663 / 1718\n",
      "21216/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9917Step 664 / 1718\n",
      "21248/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9917Step 665 / 1718\n",
      "21280/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9917Step 666 / 1718\n",
      "21312/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9917Step 667 / 1718\n",
      "21344/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9917Step 668 / 1718\n",
      "21376/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9917Step 669 / 1718\n",
      "21408/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9917Step 670 / 1718\n",
      "21440/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9917Step 671 / 1718\n",
      "21472/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9918Step 672 / 1718\n",
      "21504/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9917Step 673 / 1718\n",
      "21536/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9917Step 674 / 1718\n",
      "21568/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9917Step 675 / 1718\n",
      "21600/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9917Step 676 / 1718\n",
      "21632/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9917Step 677 / 1718\n",
      "21664/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9917Step 678 / 1718\n",
      "21696/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9917Step 679 / 1718\n",
      "21728/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9917Step 680 / 1718\n",
      "21760/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9916Step 681 / 1718\n",
      "21792/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9915Step 682 / 1718\n",
      "21824/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 683 / 1718\n",
      "21856/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9915Step 684 / 1718\n",
      "21888/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9915Step 685 / 1718\n",
      "21920/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9915Step 686 / 1718\n",
      "21952/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 687 / 1718\n",
      "21984/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 688 / 1718\n",
      "22016/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 689 / 1718\n",
      "22048/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 690 / 1718\n",
      "22080/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9915Step 691 / 1718\n",
      "22112/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9915Step 692 / 1718\n",
      "22144/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9915Step 693 / 1718\n",
      "22176/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 694 / 1718\n",
      "22208/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9915Step 695 / 1718\n",
      "22240/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9914Step 696 / 1718\n",
      "22272/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 697 / 1718\n",
      "22304/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 698 / 1718\n",
      "22336/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 699 / 1718\n",
      "22368/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 700 / 1718\n",
      "22400/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 701 / 1718\n",
      "22432/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 702 / 1718\n",
      "22464/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9915Step 703 / 1718\n",
      "22496/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 704 / 1718\n",
      "22528/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 705 / 1718\n",
      "22560/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 706 / 1718\n",
      "22592/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9915Step 707 / 1718\n",
      "22624/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9915Step 708 / 1718\n",
      "22656/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9915Step 709 / 1718\n",
      "22688/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9915Step 710 / 1718\n",
      "22720/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9915Step 711 / 1718\n",
      "22752/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9915Step 712 / 1718\n",
      "22784/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9915Step 713 / 1718\n",
      "22816/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9915Step 714 / 1718\n",
      "22848/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9915Step 715 / 1718\n",
      "22880/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9915Step 716 / 1718\n",
      "22912/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 717 / 1718\n",
      "22944/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 718 / 1718\n",
      "22976/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 719 / 1718\n",
      "23008/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9914Step 720 / 1718\n",
      "23040/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9914Step 721 / 1718\n",
      "23072/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 722 / 1718\n",
      "23104/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9914Step 723 / 1718\n",
      "23136/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9914Step 724 / 1718\n",
      "23168/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 725 / 1718\n",
      "23200/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 726 / 1718\n",
      "23232/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 727 / 1718\n",
      "23264/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 728 / 1718\n",
      "23296/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 729 / 1718\n",
      "23328/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 730 / 1718\n",
      "23360/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9913Step 731 / 1718\n",
      "23392/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9913Step 732 / 1718\n",
      "23424/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9913Step 733 / 1718\n",
      "23456/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 734 / 1718\n",
      "23488/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 735 / 1718\n",
      "23520/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 736 / 1718\n",
      "23552/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 737 / 1718\n",
      "23584/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 738 / 1718\n",
      "23616/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 739 / 1718\n",
      "23648/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 740 / 1718\n",
      "23680/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9913Step 741 / 1718\n",
      "23712/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9913Step 742 / 1718\n",
      "23744/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9913Step 743 / 1718\n",
      "23776/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9913Step 744 / 1718\n",
      "23808/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9913Step 745 / 1718\n",
      "23840/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9913Step 746 / 1718\n",
      "23872/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9913Step 747 / 1718\n",
      "23904/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9913Step 748 / 1718\n",
      "23936/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9913Step 749 / 1718\n",
      "23968/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9913Step 750 / 1718\n",
      "24000/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 751 / 1718\n",
      "24032/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 752 / 1718\n",
      "24064/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9914Step 753 / 1718\n",
      "24096/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9914Step 754 / 1718\n",
      "24128/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9914Step 755 / 1718\n",
      "24160/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 756 / 1718\n",
      "24192/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9914Step 757 / 1718\n",
      "24224/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 758 / 1718\n",
      "24256/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9914Step 759 / 1718\n",
      "24288/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9914Step 760 / 1718\n",
      "24320/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9914Step 761 / 1718\n",
      "24352/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9914Step 762 / 1718\n",
      "24384/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 763 / 1718\n",
      "24416/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9914Step 764 / 1718\n",
      "24448/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9914Step 765 / 1718\n",
      "24480/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9914Step 766 / 1718\n",
      "24512/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9914Step 767 / 1718\n",
      "24544/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9914Step 768 / 1718\n",
      "24576/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 769 / 1718\n",
      "24608/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9914Step 770 / 1718\n",
      "24640/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 771 / 1718\n",
      "24672/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 772 / 1718\n",
      "24704/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 773 / 1718\n",
      "24736/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 774 / 1718\n",
      "24768/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9914Step 775 / 1718\n",
      "24800/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9914Step 776 / 1718\n",
      "24832/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9914Step 777 / 1718\n",
      "24864/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9914Step 778 / 1718\n",
      "24896/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9914Step 779 / 1718\n",
      "24928/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9914Step 780 / 1718\n",
      "24960/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9914Step 781 / 1718\n",
      "24992/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 782 / 1718\n",
      "25024/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 783 / 1718\n",
      "25056/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 784 / 1718\n",
      "25088/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 785 / 1718\n",
      "25120/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 786 / 1718\n",
      "25152/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 787 / 1718\n",
      "25184/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 788 / 1718\n",
      "25216/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 789 / 1718\n",
      "25248/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 790 / 1718\n",
      "25280/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 791 / 1718\n",
      "25312/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 792 / 1718\n",
      "25344/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9914Step 793 / 1718\n",
      "25376/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9914Step 794 / 1718\n",
      "25408/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9914Step 795 / 1718\n",
      "25440/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 796 / 1718\n",
      "25472/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 797 / 1718\n",
      "25504/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9915Step 798 / 1718\n",
      "25536/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 799 / 1718\n",
      "25568/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 800 / 1718\n",
      "25600/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9913Step 801 / 1718\n",
      "25632/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9913Step 802 / 1718\n",
      "25664/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9913Step 803 / 1718\n",
      "25696/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9913Step 804 / 1718\n",
      "25728/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9913Step 805 / 1718\n",
      "25760/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9913Step 806 / 1718\n",
      "25792/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9913Step 807 / 1718\n",
      "25824/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9913Step 808 / 1718\n",
      "25856/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9913Step 809 / 1718\n",
      "25888/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9913Step 810 / 1718\n",
      "25920/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9912Step 811 / 1718\n",
      "25952/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9912Step 812 / 1718\n",
      "25984/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9912Step 813 / 1718\n",
      "26016/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9912Step 814 / 1718\n",
      "26048/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9912Step 815 / 1718\n",
      "26080/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9912Step 816 / 1718\n",
      "26112/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9912Step 817 / 1718\n",
      "26144/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 818 / 1718\n",
      "26176/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 819 / 1718\n",
      "26208/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 820 / 1718\n",
      "26240/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9912Step 821 / 1718\n",
      "26272/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 822 / 1718\n",
      "26304/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9911Step 823 / 1718\n",
      "26336/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9911Step 824 / 1718\n",
      "26368/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9911Step 825 / 1718\n",
      "26400/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9911Step 826 / 1718\n",
      "26432/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9911Step 827 / 1718\n",
      "26464/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9912Step 828 / 1718\n",
      "26496/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9912Step 829 / 1718\n",
      "26528/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 830 / 1718\n",
      "26560/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9912Step 831 / 1718\n",
      "26592/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9912Step 832 / 1718\n",
      "26624/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 833 / 1718\n",
      "26656/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 834 / 1718\n",
      "26688/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 835 / 1718\n",
      "26720/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 836 / 1718\n",
      "26752/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 837 / 1718\n",
      "26784/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 838 / 1718\n",
      "26816/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 839 / 1718\n",
      "26848/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 840 / 1718\n",
      "26880/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9912Step 841 / 1718\n",
      "26912/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9912Step 842 / 1718\n",
      "26944/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9912Step 843 / 1718\n",
      "26976/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9912Step 844 / 1718\n",
      "27008/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9912Step 845 / 1718\n",
      "27040/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 846 / 1718\n",
      "27072/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 847 / 1718\n",
      "27104/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9912Step 848 / 1718\n",
      "27136/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9912Step 849 / 1718\n",
      "27168/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9912Step 850 / 1718\n",
      "27200/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9912Step 851 / 1718\n",
      "27232/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9912Step 852 / 1718\n",
      "27264/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9912Step 853 / 1718\n",
      "27296/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 854 / 1718\n",
      "27328/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9913Step 855 / 1718\n",
      "27360/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9912Step 856 / 1718\n",
      "27392/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9912Step 857 / 1718\n",
      "27424/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 858 / 1718\n",
      "27456/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9913Step 859 / 1718\n",
      "27488/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9913Step 860 / 1718\n",
      "27520/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 861 / 1718\n",
      "27552/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9913Step 862 / 1718\n",
      "27584/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 863 / 1718\n",
      "27616/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 864 / 1718\n",
      "27648/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 865 / 1718\n",
      "27680/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 866 / 1718\n",
      "27712/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 867 / 1718\n",
      "27744/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 868 / 1718\n",
      "27776/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 869 / 1718\n",
      "27808/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 870 / 1718\n",
      "27840/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 871 / 1718\n",
      "27872/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 872 / 1718\n",
      "27904/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9913Step 873 / 1718\n",
      "27936/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 874 / 1718\n",
      "27968/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 875 / 1718\n",
      "28000/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 876 / 1718\n",
      "28032/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 877 / 1718\n",
      "28064/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 878 / 1718\n",
      "28096/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 879 / 1718\n",
      "28128/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 880 / 1718\n",
      "28160/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 881 / 1718\n",
      "28192/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 882 / 1718\n",
      "28224/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 883 / 1718\n",
      "28256/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 884 / 1718\n",
      "28288/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 885 / 1718\n",
      "28320/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 886 / 1718\n",
      "28352/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 887 / 1718\n",
      "28384/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 888 / 1718\n",
      "28416/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 889 / 1718\n",
      "28448/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 890 / 1718\n",
      "28480/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 891 / 1718\n",
      "28512/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 892 / 1718\n",
      "28544/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 893 / 1718\n",
      "28576/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 894 / 1718\n",
      "28608/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 895 / 1718\n",
      "28640/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 896 / 1718\n",
      "28672/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 897 / 1718\n",
      "28704/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 898 / 1718\n",
      "28736/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 899 / 1718\n",
      "28768/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 900 / 1718\n",
      "28800/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9914Step 901 / 1718\n",
      "28832/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9914Step 902 / 1718\n",
      "28864/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9914Step 903 / 1718\n",
      "28896/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 904 / 1718\n",
      "28928/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 905 / 1718\n",
      "28960/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 906 / 1718\n",
      "28992/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9914Step 907 / 1718\n",
      "29024/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9915Step 908 / 1718\n",
      "29056/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9915Step 909 / 1718\n",
      "29088/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9915Step 910 / 1718\n",
      "29120/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 911 / 1718\n",
      "29152/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 912 / 1718\n",
      "29184/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 913 / 1718\n",
      "29216/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9915Step 914 / 1718\n",
      "29248/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9915Step 915 / 1718\n",
      "29280/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9915Step 916 / 1718\n",
      "29312/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9915Step 917 / 1718\n",
      "29344/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9915Step 918 / 1718\n",
      "29376/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9915Step 919 / 1718\n",
      "29408/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9915Step 920 / 1718\n",
      "29440/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9915Step 921 / 1718\n",
      "29472/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9915Step 922 / 1718\n",
      "29504/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9915Step 923 / 1718\n",
      "29536/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9915Step 924 / 1718\n",
      "29568/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9915Step 925 / 1718\n",
      "29600/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9915Step 926 / 1718\n",
      "29632/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9915Step 927 / 1718\n",
      "29664/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9915Step 928 / 1718\n",
      "29696/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9915Step 929 / 1718\n",
      "29728/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9915Step 930 / 1718\n",
      "29760/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9915Step 931 / 1718\n",
      "29792/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 932 / 1718\n",
      "29824/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 933 / 1718\n",
      "29856/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9914Step 934 / 1718\n",
      "29888/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9914Step 935 / 1718\n",
      "29920/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9914Step 936 / 1718\n",
      "29952/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 937 / 1718\n",
      "29984/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9914Step 938 / 1718\n",
      "30016/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9914Step 939 / 1718\n",
      "30048/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9914Step 940 / 1718\n",
      "30080/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9915Step 941 / 1718\n",
      "30112/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 942 / 1718\n",
      "30144/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9914Step 943 / 1718\n",
      "30176/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9915Step 944 / 1718\n",
      "30208/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 945 / 1718\n",
      "30240/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 946 / 1718\n",
      "30272/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 947 / 1718\n",
      "30304/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 948 / 1718\n",
      "30336/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 949 / 1718\n",
      "30368/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 950 / 1718\n",
      "30400/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9915Step 951 / 1718\n",
      "30432/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9915Step 952 / 1718\n",
      "30464/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9915Step 953 / 1718\n",
      "30496/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9915Step 954 / 1718\n",
      "30528/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9915Step 955 / 1718\n",
      "30560/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9915Step 956 / 1718\n",
      "30592/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9915Step 957 / 1718\n",
      "30624/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9915Step 958 / 1718\n",
      "30656/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 959 / 1718\n",
      "30688/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 960 / 1718\n",
      "30720/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 961 / 1718\n",
      "30752/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 962 / 1718\n",
      "30784/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9915Step 963 / 1718\n",
      "30816/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9915Step 964 / 1718\n",
      "30848/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9915Step 965 / 1718\n",
      "30880/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9915Step 966 / 1718\n",
      "30912/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9915Step 967 / 1718\n",
      "30944/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9915Step 968 / 1718\n",
      "30976/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9915Step 969 / 1718\n",
      "31008/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9916Step 970 / 1718\n",
      "31040/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9916Step 971 / 1718\n",
      "31072/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9915Step 972 / 1718\n",
      "31104/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9915Step 973 / 1718\n",
      "31136/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9915Step 974 / 1718\n",
      "31168/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9915Step 975 / 1718\n",
      "31200/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9915Step 976 / 1718\n",
      "31232/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9915Step 977 / 1718\n",
      "31264/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9915Step 978 / 1718\n",
      "31296/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9915Step 979 / 1718\n",
      "31328/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9915Step 980 / 1718\n",
      "31360/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9915Step 981 / 1718\n",
      "31392/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9915Step 982 / 1718\n",
      "31424/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9914Step 983 / 1718\n",
      "31456/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 984 / 1718\n",
      "31488/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 985 / 1718\n",
      "31520/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 986 / 1718\n",
      "31552/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 987 / 1718\n",
      "31584/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 988 / 1718\n",
      "31616/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 989 / 1718\n",
      "31648/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 990 / 1718\n",
      "31680/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 991 / 1718\n",
      "31712/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 992 / 1718\n",
      "31744/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 993 / 1718\n",
      "31776/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 994 / 1718\n",
      "31808/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 995 / 1718\n",
      "31840/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 996 / 1718\n",
      "31872/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 997 / 1718\n",
      "31904/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 998 / 1718\n",
      "31936/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 999 / 1718\n",
      "31968/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 1000 / 1718\n",
      "32000/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 1001 / 1718\n",
      "32032/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1002 / 1718\n",
      "32064/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1003 / 1718\n",
      "32096/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9912Step 1004 / 1718\n",
      "32128/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 1005 / 1718\n",
      "32160/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 1006 / 1718\n",
      "32192/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 1007 / 1718\n",
      "32224/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1008 / 1718\n",
      "32256/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1009 / 1718\n",
      "32288/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1010 / 1718\n",
      "32320/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1011 / 1718\n",
      "32352/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1012 / 1718\n",
      "32384/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1013 / 1718\n",
      "32416/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 1014 / 1718\n",
      "32448/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1015 / 1718\n",
      "32480/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 1016 / 1718\n",
      "32512/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 1017 / 1718\n",
      "32544/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1018 / 1718\n",
      "32576/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1019 / 1718\n",
      "32608/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1020 / 1718\n",
      "32640/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1021 / 1718\n",
      "32672/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1022 / 1718\n",
      "32704/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 1023 / 1718\n",
      "32736/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1024 / 1718\n",
      "32768/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1025 / 1718\n",
      "32800/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1026 / 1718\n",
      "32832/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1027 / 1718\n",
      "32864/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1028 / 1718\n",
      "32896/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1029 / 1718\n",
      "32928/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1030 / 1718\n",
      "32960/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1031 / 1718\n",
      "32992/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1032 / 1718\n",
      "33024/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1033 / 1718\n",
      "33056/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1034 / 1718\n",
      "33088/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9912Step 1035 / 1718\n",
      "33120/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9912Step 1036 / 1718\n",
      "33152/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9912Step 1037 / 1718\n",
      "33184/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9912Step 1038 / 1718\n",
      "33216/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9912Step 1039 / 1718\n",
      "33248/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9912Step 1040 / 1718\n",
      "33280/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1041 / 1718\n",
      "33312/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9912Step 1042 / 1718\n",
      "33344/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9912Step 1043 / 1718\n",
      "33376/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1044 / 1718\n",
      "33408/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9911Step 1045 / 1718\n",
      "33440/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9911Step 1046 / 1718\n",
      "33472/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1047 / 1718\n",
      "33504/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1048 / 1718\n",
      "33536/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1049 / 1718\n",
      "33568/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1050 / 1718\n",
      "33600/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9912Step 1051 / 1718\n",
      "33632/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1052 / 1718\n",
      "33664/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1053 / 1718\n",
      "33696/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1054 / 1718\n",
      "33728/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1055 / 1718\n",
      "33760/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1056 / 1718\n",
      "33792/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1057 / 1718\n",
      "33824/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9911Step 1058 / 1718\n",
      "33856/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9911Step 1059 / 1718\n",
      "33888/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9911Step 1060 / 1718\n",
      "33920/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1061 / 1718\n",
      "33952/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1062 / 1718\n",
      "33984/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9911Step 1063 / 1718\n",
      "34016/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1064 / 1718\n",
      "34048/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9912Step 1065 / 1718\n",
      "34080/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9912Step 1066 / 1718\n",
      "34112/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9912Step 1067 / 1718\n",
      "34144/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9912Step 1068 / 1718\n",
      "34176/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9912Step 1069 / 1718\n",
      "34208/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1070 / 1718\n",
      "34240/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9912Step 1071 / 1718\n",
      "34272/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9912Step 1072 / 1718\n",
      "34304/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9912Step 1073 / 1718\n",
      "34336/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9911Step 1074 / 1718\n",
      "34368/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1075 / 1718\n",
      "34400/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1076 / 1718\n",
      "34432/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9911Step 1077 / 1718\n",
      "34464/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1078 / 1718\n",
      "34496/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1079 / 1718\n",
      "34528/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1080 / 1718\n",
      "34560/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9912Step 1081 / 1718\n",
      "34592/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9912Step 1082 / 1718\n",
      "34624/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9912Step 1083 / 1718\n",
      "34656/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9912Step 1084 / 1718\n",
      "34688/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9912Step 1085 / 1718\n",
      "34720/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9912Step 1086 / 1718\n",
      "34752/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9912Step 1087 / 1718\n",
      "34784/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9912Step 1088 / 1718\n",
      "34816/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9912Step 1089 / 1718\n",
      "34848/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9912Step 1090 / 1718\n",
      "34880/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9912Step 1091 / 1718\n",
      "34912/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9912Step 1092 / 1718\n",
      "34944/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9912Step 1093 / 1718\n",
      "34976/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9912Step 1094 / 1718\n",
      "35008/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9912Step 1095 / 1718\n",
      "35040/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9912Step 1096 / 1718\n",
      "35072/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9912Step 1097 / 1718\n",
      "35104/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 1098 / 1718\n",
      "35136/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 1099 / 1718\n",
      "35168/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 1100 / 1718\n",
      "35200/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9912Step 1101 / 1718\n",
      "35232/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 1102 / 1718\n",
      "35264/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1103 / 1718\n",
      "35296/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1104 / 1718\n",
      "35328/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1105 / 1718\n",
      "35360/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1106 / 1718\n",
      "35392/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1107 / 1718\n",
      "35424/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9913Step 1108 / 1718\n",
      "35456/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9913Step 1109 / 1718\n",
      "35488/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9913Step 1110 / 1718\n",
      "35520/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9913Step 1111 / 1718\n",
      "35552/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9913Step 1112 / 1718\n",
      "35584/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9913Step 1113 / 1718\n",
      "35616/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9913Step 1114 / 1718\n",
      "35648/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9913Step 1115 / 1718\n",
      "35680/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1116 / 1718\n",
      "35712/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1117 / 1718\n",
      "35744/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1118 / 1718\n",
      "35776/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1119 / 1718\n",
      "35808/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1120 / 1718\n",
      "35840/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 1121 / 1718\n",
      "35872/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1122 / 1718\n",
      "35904/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 1123 / 1718\n",
      "35936/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9912Step 1124 / 1718\n",
      "35968/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9912Step 1125 / 1718\n",
      "36000/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9912Step 1126 / 1718\n",
      "36032/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9912Step 1127 / 1718\n",
      "36064/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9912Step 1128 / 1718\n",
      "36096/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9912Step 1129 / 1718\n",
      "36128/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9912Step 1130 / 1718\n",
      "36160/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9912Step 1131 / 1718\n",
      "36192/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9912Step 1132 / 1718\n",
      "36224/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9912Step 1133 / 1718\n",
      "36256/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9912Step 1134 / 1718\n",
      "36288/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9912Step 1135 / 1718\n",
      "36320/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9912Step 1136 / 1718\n",
      "36352/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1137 / 1718\n",
      "36384/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1138 / 1718\n",
      "36416/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1139 / 1718\n",
      "36448/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1140 / 1718\n",
      "36480/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9913Step 1141 / 1718\n",
      "36512/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9913Step 1142 / 1718\n",
      "36544/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1143 / 1718\n",
      "36576/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1144 / 1718\n",
      "36608/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1145 / 1718\n",
      "36640/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1146 / 1718\n",
      "36672/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1147 / 1718\n",
      "36704/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9913Step 1148 / 1718\n",
      "36736/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9913Step 1149 / 1718\n",
      "36768/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9913Step 1150 / 1718\n",
      "36800/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9913Step 1151 / 1718\n",
      "36832/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9913Step 1152 / 1718\n",
      "36864/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9913Step 1153 / 1718\n",
      "36896/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 1154 / 1718\n",
      "36928/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 1155 / 1718\n",
      "36960/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1156 / 1718\n",
      "36992/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 1157 / 1718\n",
      "37024/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1158 / 1718\n",
      "37056/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1159 / 1718\n",
      "37088/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1160 / 1718\n",
      "37120/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1161 / 1718\n",
      "37152/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1162 / 1718\n",
      "37184/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 1163 / 1718\n",
      "37216/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 1164 / 1718\n",
      "37248/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9914Step 1165 / 1718\n",
      "37280/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9914Step 1166 / 1718\n",
      "37312/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 1167 / 1718\n",
      "37344/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 1168 / 1718\n",
      "37376/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 1169 / 1718\n",
      "37408/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 1170 / 1718\n",
      "37440/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1171 / 1718\n",
      "37472/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 1172 / 1718\n",
      "37504/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 1173 / 1718\n",
      "37536/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 1174 / 1718\n",
      "37568/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1175 / 1718\n",
      "37600/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9914Step 1176 / 1718\n",
      "37632/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 1177 / 1718\n",
      "37664/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1178 / 1718\n",
      "37696/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9913Step 1179 / 1718\n",
      "37728/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9913Step 1180 / 1718\n",
      "37760/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9913Step 1181 / 1718\n",
      "37792/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9913Step 1182 / 1718\n",
      "37824/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9913Step 1183 / 1718\n",
      "37856/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9913Step 1184 / 1718\n",
      "37888/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9913Step 1185 / 1718\n",
      "37920/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9913Step 1186 / 1718\n",
      "37952/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9913Step 1187 / 1718\n",
      "37984/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9913Step 1188 / 1718\n",
      "38016/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9912Step 1189 / 1718\n",
      "38048/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9912Step 1190 / 1718\n",
      "38080/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9912Step 1191 / 1718\n",
      "38112/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9912Step 1192 / 1718\n",
      "38144/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9912Step 1193 / 1718\n",
      "38176/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9912Step 1194 / 1718\n",
      "38208/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9912Step 1195 / 1718\n",
      "38240/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9912Step 1196 / 1718\n",
      "38272/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9912Step 1197 / 1718\n",
      "38304/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9912Step 1198 / 1718\n",
      "38336/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9912Step 1199 / 1718\n",
      "38368/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9911Step 1200 / 1718\n",
      "38400/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9911Step 1201 / 1718\n",
      "38432/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9912Step 1202 / 1718\n",
      "38464/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9912Step 1203 / 1718\n",
      "38496/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9912Step 1204 / 1718\n",
      "38528/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9912Step 1205 / 1718\n",
      "38560/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9912Step 1206 / 1718\n",
      "38592/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9912Step 1207 / 1718\n",
      "38624/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9912Step 1208 / 1718\n",
      "38656/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9912Step 1209 / 1718\n",
      "38688/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9911Step 1210 / 1718\n",
      "38720/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9911Step 1211 / 1718\n",
      "38752/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9911Step 1212 / 1718\n",
      "38784/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9912Step 1213 / 1718\n",
      "38816/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9912Step 1214 / 1718\n",
      "38848/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9912Step 1215 / 1718\n",
      "38880/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9912Step 1216 / 1718\n",
      "38912/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9912Step 1217 / 1718\n",
      "38944/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9912Step 1218 / 1718\n",
      "38976/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9911Step 1219 / 1718\n",
      "39008/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9912Step 1220 / 1718\n",
      "39040/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9912Step 1221 / 1718\n",
      "39072/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9912Step 1222 / 1718\n",
      "39104/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9912Step 1223 / 1718\n",
      "39136/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9912Step 1224 / 1718\n",
      "39168/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9912Step 1225 / 1718\n",
      "39200/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9912Step 1226 / 1718\n",
      "39232/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9912Step 1227 / 1718\n",
      "39264/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9912Step 1228 / 1718\n",
      "39296/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9912Step 1229 / 1718\n",
      "39328/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9912Step 1230 / 1718\n",
      "39360/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9912Step 1231 / 1718\n",
      "39392/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9911Step 1232 / 1718\n",
      "39424/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9911Step 1233 / 1718\n",
      "39456/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9911Step 1234 / 1718\n",
      "39488/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9911Step 1235 / 1718\n",
      "39520/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9911Step 1236 / 1718\n",
      "39552/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9911Step 1237 / 1718\n",
      "39584/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9911Step 1238 / 1718\n",
      "39616/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9911Step 1239 / 1718\n",
      "39648/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9911Step 1240 / 1718\n",
      "39680/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9911Step 1241 / 1718\n",
      "39712/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9911Step 1242 / 1718\n",
      "39744/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9911Step 1243 / 1718\n",
      "39776/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9911Step 1244 / 1718\n",
      "39808/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9911Step 1245 / 1718\n",
      "39840/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9911Step 1246 / 1718\n",
      "39872/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9911Step 1247 / 1718\n",
      "39904/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9912Step 1248 / 1718\n",
      "39936/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9912Step 1249 / 1718\n",
      "39968/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9911Step 1250 / 1718\n",
      "40000/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9911Step 1251 / 1718\n",
      "40032/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9911Step 1252 / 1718\n",
      "40064/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9911Step 1253 / 1718\n",
      "40096/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9911Step 1254 / 1718\n",
      "40128/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9911Step 1255 / 1718\n",
      "40160/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9911Step 1256 / 1718\n",
      "40192/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9911Step 1257 / 1718\n",
      "40224/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9911Step 1258 / 1718\n",
      "40256/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9911Step 1259 / 1718\n",
      "40288/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9910Step 1260 / 1718\n",
      "40320/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9910Step 1261 / 1718\n",
      "40352/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9911Step 1262 / 1718\n",
      "40384/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9910Step 1263 / 1718\n",
      "40416/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9910Step 1264 / 1718\n",
      "40448/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9911Step 1265 / 1718\n",
      "40480/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9911Step 1266 / 1718\n",
      "40512/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9911Step 1267 / 1718\n",
      "40544/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9911Step 1268 / 1718\n",
      "40576/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9911Step 1269 / 1718\n",
      "40608/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9911Step 1270 / 1718\n",
      "40640/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9910Step 1271 / 1718\n",
      "40672/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9910Step 1272 / 1718\n",
      "40704/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9910Step 1273 / 1718\n",
      "40736/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9910Step 1274 / 1718\n",
      "40768/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9910Step 1275 / 1718\n",
      "40800/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9910Step 1276 / 1718\n",
      "40832/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9910Step 1277 / 1718\n",
      "40864/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9910Step 1278 / 1718\n",
      "40896/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9910Step 1279 / 1718\n",
      "40928/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9910Step 1280 / 1718\n",
      "40960/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9910Step 1281 / 1718\n",
      "40992/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9910Step 1282 / 1718\n",
      "41024/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9910Step 1283 / 1718\n",
      "41056/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9910Step 1284 / 1718\n",
      "41088/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9910Step 1285 / 1718\n",
      "41120/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9910Step 1286 / 1718\n",
      "41152/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9910Step 1287 / 1718\n",
      "41184/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9910Step 1288 / 1718\n",
      "41216/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9910Step 1289 / 1718\n",
      "41248/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9909Step 1290 / 1718\n",
      "41280/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9909Step 1291 / 1718\n",
      "41312/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9909Step 1292 / 1718\n",
      "41344/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9909Step 1293 / 1718\n",
      "41376/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9909Step 1294 / 1718\n",
      "41408/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9909Step 1295 / 1718\n",
      "41440/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9910Step 1296 / 1718\n",
      "41472/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9909Step 1297 / 1718\n",
      "41504/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9909Step 1298 / 1718\n",
      "41536/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9909Step 1299 / 1718\n",
      "41568/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9910Step 1300 / 1718\n",
      "41600/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9910Step 1301 / 1718\n",
      "41632/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9909Step 1302 / 1718\n",
      "41664/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9910Step 1303 / 1718\n",
      "41696/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9910Step 1304 / 1718\n",
      "41728/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9910Step 1305 / 1718\n",
      "41760/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9910Step 1306 / 1718\n",
      "41792/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9910Step 1307 / 1718\n",
      "41824/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9910Step 1308 / 1718\n",
      "41856/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9910Step 1309 / 1718\n",
      "41888/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9909Step 1310 / 1718\n",
      "41920/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9909Step 1311 / 1718\n",
      "41952/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9909Step 1312 / 1718\n",
      "41984/55000 - mean: 0.0323 - sparse_categorical_accuracy: 0.9909Step 1313 / 1718\n",
      "42016/55000 - mean: 0.0323 - sparse_categorical_accuracy: 0.9909Step 1314 / 1718\n",
      "42048/55000 - mean: 0.0323 - sparse_categorical_accuracy: 0.9909Step 1315 / 1718\n",
      "42080/55000 - mean: 0.0323 - sparse_categorical_accuracy: 0.9909Step 1316 / 1718\n",
      "42112/55000 - mean: 0.0323 - sparse_categorical_accuracy: 0.9909Step 1317 / 1718\n",
      "42144/55000 - mean: 0.0323 - sparse_categorical_accuracy: 0.9909Step 1318 / 1718\n",
      "42176/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9909Step 1319 / 1718\n",
      "42208/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9909Step 1320 / 1718\n",
      "42240/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9909Step 1321 / 1718\n",
      "42272/55000 - mean: 0.0326 - sparse_categorical_accuracy: 0.9908Step 1322 / 1718\n",
      "42304/55000 - mean: 0.0326 - sparse_categorical_accuracy: 0.9909Step 1323 / 1718\n",
      "42336/55000 - mean: 0.0326 - sparse_categorical_accuracy: 0.9908Step 1324 / 1718\n",
      "42368/55000 - mean: 0.0326 - sparse_categorical_accuracy: 0.9908Step 1325 / 1718\n",
      "42400/55000 - mean: 0.0325 - sparse_categorical_accuracy: 0.9908Step 1326 / 1718\n",
      "42432/55000 - mean: 0.0325 - sparse_categorical_accuracy: 0.9909Step 1327 / 1718\n",
      "42464/55000 - mean: 0.0326 - sparse_categorical_accuracy: 0.9908Step 1328 / 1718\n",
      "42496/55000 - mean: 0.0326 - sparse_categorical_accuracy: 0.9908Step 1329 / 1718\n",
      "42528/55000 - mean: 0.0327 - sparse_categorical_accuracy: 0.9908Step 1330 / 1718\n",
      "42560/55000 - mean: 0.0326 - sparse_categorical_accuracy: 0.9908Step 1331 / 1718\n",
      "42592/55000 - mean: 0.0326 - sparse_categorical_accuracy: 0.9908Step 1332 / 1718\n",
      "42624/55000 - mean: 0.0327 - sparse_categorical_accuracy: 0.9908Step 1333 / 1718\n",
      "42656/55000 - mean: 0.0328 - sparse_categorical_accuracy: 0.9908Step 1334 / 1718\n",
      "42688/55000 - mean: 0.0328 - sparse_categorical_accuracy: 0.9908Step 1335 / 1718\n",
      "42720/55000 - mean: 0.0328 - sparse_categorical_accuracy: 0.9908Step 1336 / 1718\n",
      "42752/55000 - mean: 0.0328 - sparse_categorical_accuracy: 0.9908Step 1337 / 1718\n",
      "42784/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9907Step 1338 / 1718\n",
      "42816/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9908Step 1339 / 1718\n",
      "42848/55000 - mean: 0.0329 - sparse_categorical_accuracy: 0.9908Step 1340 / 1718\n",
      "42880/55000 - mean: 0.0329 - sparse_categorical_accuracy: 0.9908Step 1341 / 1718\n",
      "42912/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9907Step 1342 / 1718\n",
      "42944/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9907Step 1343 / 1718\n",
      "42976/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9907Step 1344 / 1718\n",
      "43008/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9907Step 1345 / 1718\n",
      "43040/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9908Step 1346 / 1718\n",
      "43072/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9907Step 1347 / 1718\n",
      "43104/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9907Step 1348 / 1718\n",
      "43136/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9907Step 1349 / 1718\n",
      "43168/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9907Step 1350 / 1718\n",
      "43200/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9907Step 1351 / 1718\n",
      "43232/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9907Step 1352 / 1718\n",
      "43264/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9908Step 1353 / 1718\n",
      "43296/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9907Step 1354 / 1718\n",
      "43328/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9907Step 1355 / 1718\n",
      "43360/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9908Step 1356 / 1718\n",
      "43392/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9908Step 1357 / 1718\n",
      "43424/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9908Step 1358 / 1718\n",
      "43456/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9908Step 1359 / 1718\n",
      "43488/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9908Step 1360 / 1718\n",
      "43520/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9908Step 1361 / 1718\n",
      "43552/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9907Step 1362 / 1718\n",
      "43584/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9907Step 1363 / 1718\n",
      "43616/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9907Step 1364 / 1718\n",
      "43648/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9907Step 1365 / 1718\n",
      "43680/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9907Step 1366 / 1718\n",
      "43712/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9907Step 1367 / 1718\n",
      "43744/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9907Step 1368 / 1718\n",
      "43776/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9907Step 1369 / 1718\n",
      "43808/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9907Step 1370 / 1718\n",
      "43840/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9907Step 1371 / 1718\n",
      "43872/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9907Step 1372 / 1718\n",
      "43904/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9907Step 1373 / 1718\n",
      "43936/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9907Step 1374 / 1718\n",
      "43968/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9907Step 1375 / 1718\n",
      "44000/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9906Step 1376 / 1718\n",
      "44032/55000 - mean: 0.0338 - sparse_categorical_accuracy: 0.9906Step 1377 / 1718\n",
      "44064/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9906Step 1378 / 1718\n",
      "44096/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9906Step 1379 / 1718\n",
      "44128/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9906Step 1380 / 1718\n",
      "44160/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9906Step 1381 / 1718\n",
      "44192/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9907Step 1382 / 1718\n",
      "44224/55000 - mean: 0.0338 - sparse_categorical_accuracy: 0.9906Step 1383 / 1718\n",
      "44256/55000 - mean: 0.0338 - sparse_categorical_accuracy: 0.9906Step 1384 / 1718\n",
      "44288/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9906Step 1385 / 1718\n",
      "44320/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9906Step 1386 / 1718\n",
      "44352/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9906Step 1387 / 1718\n",
      "44384/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9906Step 1388 / 1718\n",
      "44416/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9906Step 1389 / 1718\n",
      "44448/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9906Step 1390 / 1718\n",
      "44480/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9906Step 1391 / 1718\n",
      "44512/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9906Step 1392 / 1718\n",
      "44544/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9906Step 1393 / 1718\n",
      "44576/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9906Step 1394 / 1718\n",
      "44608/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9906Step 1395 / 1718\n",
      "44640/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9906Step 1396 / 1718\n",
      "44672/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9906Step 1397 / 1718\n",
      "44704/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9906Step 1398 / 1718\n",
      "44736/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9906Step 1399 / 1718\n",
      "44768/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9906Step 1400 / 1718\n",
      "44800/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9906Step 1401 / 1718\n",
      "44832/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9905Step 1402 / 1718\n",
      "44864/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9905Step 1403 / 1718\n",
      "44896/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9905Step 1404 / 1718\n",
      "44928/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9905Step 1405 / 1718\n",
      "44960/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9905Step 1406 / 1718\n",
      "44992/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9905Step 1407 / 1718\n",
      "45024/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9905Step 1408 / 1718\n",
      "45056/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9905Step 1409 / 1718\n",
      "45088/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9905Step 1410 / 1718\n",
      "45120/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9905Step 1411 / 1718\n",
      "45152/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9905Step 1412 / 1718\n",
      "45184/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9905Step 1413 / 1718\n",
      "45216/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9905Step 1414 / 1718\n",
      "45248/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9905Step 1415 / 1718\n",
      "45280/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9905Step 1416 / 1718\n",
      "45312/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9905Step 1417 / 1718\n",
      "45344/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9905Step 1418 / 1718\n",
      "45376/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9905Step 1419 / 1718\n",
      "45408/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9905Step 1420 / 1718\n",
      "45440/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9904Step 1421 / 1718\n",
      "45472/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9904Step 1422 / 1718\n",
      "45504/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9904Step 1423 / 1718\n",
      "45536/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9904Step 1424 / 1718\n",
      "45568/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9904Step 1425 / 1718\n",
      "45600/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9904Step 1426 / 1718\n",
      "45632/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9904Step 1427 / 1718\n",
      "45664/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9904Step 1428 / 1718\n",
      "45696/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9904Step 1429 / 1718\n",
      "45728/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9904Step 1430 / 1718\n",
      "45760/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9904Step 1431 / 1718\n",
      "45792/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9904Step 1432 / 1718\n",
      "45824/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9904Step 1433 / 1718\n",
      "45856/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9903Step 1434 / 1718\n",
      "45888/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9903Step 1435 / 1718\n",
      "45920/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9903Step 1436 / 1718\n",
      "45952/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9903Step 1437 / 1718\n",
      "45984/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9903Step 1438 / 1718\n",
      "46016/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9903Step 1439 / 1718\n",
      "46048/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9903Step 1440 / 1718\n",
      "46080/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9903Step 1441 / 1718\n",
      "46112/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9903Step 1442 / 1718\n",
      "46144/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9904Step 1443 / 1718\n",
      "46176/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9904Step 1444 / 1718\n",
      "46208/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9904Step 1445 / 1718\n",
      "46240/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9904Step 1446 / 1718\n",
      "46272/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9904Step 1447 / 1718\n",
      "46304/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9903Step 1448 / 1718\n",
      "46336/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9904Step 1449 / 1718\n",
      "46368/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9904Step 1450 / 1718\n",
      "46400/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9904Step 1451 / 1718\n",
      "46432/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9904Step 1452 / 1718\n",
      "46464/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9904Step 1453 / 1718\n",
      "46496/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9904Step 1454 / 1718\n",
      "46528/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9904Step 1455 / 1718\n",
      "46560/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9904Step 1456 / 1718\n",
      "46592/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9904Step 1457 / 1718\n",
      "46624/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9904Step 1458 / 1718\n",
      "46656/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9904Step 1459 / 1718\n",
      "46688/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9904Step 1460 / 1718\n",
      "46720/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9904Step 1461 / 1718\n",
      "46752/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9904Step 1462 / 1718\n",
      "46784/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9904Step 1463 / 1718\n",
      "46816/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9904Step 1464 / 1718\n",
      "46848/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9904Step 1465 / 1718\n",
      "46880/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9904Step 1466 / 1718\n",
      "46912/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9904Step 1467 / 1718\n",
      "46944/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9904Step 1468 / 1718\n",
      "46976/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9903Step 1469 / 1718\n",
      "47008/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9903Step 1470 / 1718\n",
      "47040/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9903Step 1471 / 1718\n",
      "47072/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9903Step 1472 / 1718\n",
      "47104/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9903Step 1473 / 1718\n",
      "47136/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9903Step 1474 / 1718\n",
      "47168/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9903Step 1475 / 1718\n",
      "47200/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9903Step 1476 / 1718\n",
      "47232/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9903Step 1477 / 1718\n",
      "47264/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9903Step 1478 / 1718\n",
      "47296/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9903Step 1479 / 1718\n",
      "47328/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9903Step 1480 / 1718\n",
      "47360/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9903Step 1481 / 1718\n",
      "47392/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9903Step 1482 / 1718\n",
      "47424/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9903Step 1483 / 1718\n",
      "47456/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9903Step 1484 / 1718\n",
      "47488/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9903Step 1485 / 1718\n",
      "47520/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9903Step 1486 / 1718\n",
      "47552/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9903Step 1487 / 1718\n",
      "47584/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9903Step 1488 / 1718\n",
      "47616/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9903Step 1489 / 1718\n",
      "47648/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9903Step 1490 / 1718\n",
      "47680/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9904Step 1491 / 1718\n",
      "47712/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9903Step 1492 / 1718\n",
      "47744/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9903Step 1493 / 1718\n",
      "47776/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9903Step 1494 / 1718\n",
      "47808/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9903Step 1495 / 1718\n",
      "47840/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9903Step 1496 / 1718\n",
      "47872/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9903Step 1497 / 1718\n",
      "47904/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9903Step 1498 / 1718\n",
      "47936/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9902Step 1499 / 1718\n",
      "47968/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9902Step 1500 / 1718\n",
      "48000/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9902Step 1501 / 1718\n",
      "48032/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9902Step 1502 / 1718\n",
      "48064/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1503 / 1718\n",
      "48096/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1504 / 1718\n",
      "48128/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1505 / 1718\n",
      "48160/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1506 / 1718\n",
      "48192/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1507 / 1718\n",
      "48224/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9902Step 1508 / 1718\n",
      "48256/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1509 / 1718\n",
      "48288/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1510 / 1718\n",
      "48320/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1511 / 1718\n",
      "48352/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1512 / 1718\n",
      "48384/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1513 / 1718\n",
      "48416/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1514 / 1718\n",
      "48448/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9902Step 1515 / 1718\n",
      "48480/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9902Step 1516 / 1718\n",
      "48512/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9902Step 1517 / 1718\n",
      "48544/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9903Step 1518 / 1718\n",
      "48576/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1519 / 1718\n",
      "48608/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1520 / 1718\n",
      "48640/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1521 / 1718\n",
      "48672/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1522 / 1718\n",
      "48704/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9902Step 1523 / 1718\n",
      "48736/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1524 / 1718\n",
      "48768/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1525 / 1718\n",
      "48800/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1526 / 1718\n",
      "48832/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9902Step 1527 / 1718\n",
      "48864/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1528 / 1718\n",
      "48896/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9902Step 1529 / 1718\n",
      "48928/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9902Step 1530 / 1718\n",
      "48960/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9902Step 1531 / 1718\n",
      "48992/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9902Step 1532 / 1718\n",
      "49024/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1533 / 1718\n",
      "49056/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1534 / 1718\n",
      "49088/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1535 / 1718\n",
      "49120/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1536 / 1718\n",
      "49152/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1537 / 1718\n",
      "49184/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1538 / 1718\n",
      "49216/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1539 / 1718\n",
      "49248/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9902Step 1540 / 1718\n",
      "49280/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9902Step 1541 / 1718\n",
      "49312/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1542 / 1718\n",
      "49344/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1543 / 1718\n",
      "49376/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1544 / 1718\n",
      "49408/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1545 / 1718\n",
      "49440/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1546 / 1718\n",
      "49472/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1547 / 1718\n",
      "49504/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1548 / 1718\n",
      "49536/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1549 / 1718\n",
      "49568/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1550 / 1718\n",
      "49600/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1551 / 1718\n",
      "49632/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9901Step 1552 / 1718\n",
      "49664/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9902Step 1553 / 1718\n",
      "49696/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9902Step 1554 / 1718\n",
      "49728/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9902Step 1555 / 1718\n",
      "49760/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9902Step 1556 / 1718\n",
      "49792/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 1557 / 1718\n",
      "49824/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9902Step 1558 / 1718\n",
      "49856/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9902Step 1559 / 1718\n",
      "49888/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9902Step 1560 / 1718\n",
      "49920/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9902Step 1561 / 1718\n",
      "49952/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1562 / 1718\n",
      "49984/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1563 / 1718\n",
      "50016/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1564 / 1718\n",
      "50048/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1565 / 1718\n",
      "50080/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9902Step 1566 / 1718\n",
      "50112/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1567 / 1718\n",
      "50144/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1568 / 1718\n",
      "50176/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1569 / 1718\n",
      "50208/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1570 / 1718\n",
      "50240/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1571 / 1718\n",
      "50272/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1572 / 1718\n",
      "50304/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1573 / 1718\n",
      "50336/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1574 / 1718\n",
      "50368/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 1575 / 1718\n",
      "50400/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 1576 / 1718\n",
      "50432/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9901Step 1577 / 1718\n",
      "50464/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9901Step 1578 / 1718\n",
      "50496/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 1579 / 1718\n",
      "50528/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9901Step 1580 / 1718\n",
      "50560/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9901Step 1581 / 1718\n",
      "50592/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9901Step 1582 / 1718\n",
      "50624/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9901Step 1583 / 1718\n",
      "50656/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9901Step 1584 / 1718\n",
      "50688/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 1585 / 1718\n",
      "50720/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 1586 / 1718\n",
      "50752/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 1587 / 1718\n",
      "50784/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 1588 / 1718\n",
      "50816/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 1589 / 1718\n",
      "50848/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1590 / 1718\n",
      "50880/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1591 / 1718\n",
      "50912/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1592 / 1718\n",
      "50944/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1593 / 1718\n",
      "50976/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1594 / 1718\n",
      "51008/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1595 / 1718\n",
      "51040/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1596 / 1718\n",
      "51072/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1597 / 1718\n",
      "51104/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1598 / 1718\n",
      "51136/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1599 / 1718\n",
      "51168/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9902Step 1600 / 1718\n",
      "51200/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9902Step 1601 / 1718\n",
      "51232/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1602 / 1718\n",
      "51264/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1603 / 1718\n",
      "51296/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1604 / 1718\n",
      "51328/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1605 / 1718\n",
      "51360/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1606 / 1718\n",
      "51392/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9902Step 1607 / 1718\n",
      "51424/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9902Step 1608 / 1718\n",
      "51456/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9902Step 1609 / 1718\n",
      "51488/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1610 / 1718\n",
      "51520/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1611 / 1718\n",
      "51552/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1612 / 1718\n",
      "51584/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1613 / 1718\n",
      "51616/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1614 / 1718\n",
      "51648/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1615 / 1718\n",
      "51680/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 1616 / 1718\n",
      "51712/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 1617 / 1718\n",
      "51744/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 1618 / 1718\n",
      "51776/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 1619 / 1718\n",
      "51808/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 1620 / 1718\n",
      "51840/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1621 / 1718\n",
      "51872/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 1622 / 1718\n",
      "51904/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 1623 / 1718\n",
      "51936/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1624 / 1718\n",
      "51968/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 1625 / 1718\n",
      "52000/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1626 / 1718\n",
      "52032/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1627 / 1718\n",
      "52064/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1628 / 1718\n",
      "52096/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1629 / 1718\n",
      "52128/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1630 / 1718\n",
      "52160/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1631 / 1718\n",
      "52192/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1632 / 1718\n",
      "52224/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1633 / 1718\n",
      "52256/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9901Step 1634 / 1718\n",
      "52288/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 1635 / 1718\n",
      "52320/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1636 / 1718\n",
      "52352/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1637 / 1718\n",
      "52384/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1638 / 1718\n",
      "52416/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1639 / 1718\n",
      "52448/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1640 / 1718\n",
      "52480/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9900Step 1641 / 1718\n",
      "52512/55000 - mean: 0.0364 - sparse_categorical_accuracy: 0.9900Step 1642 / 1718\n",
      "52544/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9900Step 1643 / 1718\n",
      "52576/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9900Step 1644 / 1718\n",
      "52608/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9900Step 1645 / 1718\n",
      "52640/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9900Step 1646 / 1718\n",
      "52672/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9900Step 1647 / 1718\n",
      "52704/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9900Step 1648 / 1718\n",
      "52736/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1649 / 1718\n",
      "52768/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1650 / 1718\n",
      "52800/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1651 / 1718\n",
      "52832/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1652 / 1718\n",
      "52864/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1653 / 1718\n",
      "52896/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1654 / 1718\n",
      "52928/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1655 / 1718\n",
      "52960/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9900Step 1656 / 1718\n",
      "52992/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1657 / 1718\n",
      "53024/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1658 / 1718\n",
      "53056/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1659 / 1718\n",
      "53088/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1660 / 1718\n",
      "53120/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1661 / 1718\n",
      "53152/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9900Step 1662 / 1718\n",
      "53184/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9900Step 1663 / 1718\n",
      "53216/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9900Step 1664 / 1718\n",
      "53248/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9900Step 1665 / 1718\n",
      "53280/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9900Step 1666 / 1718\n",
      "53312/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9900Step 1667 / 1718\n",
      "53344/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9900Step 1668 / 1718\n",
      "53376/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9900Step 1669 / 1718\n",
      "53408/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9900Step 1670 / 1718\n",
      "53440/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9900Step 1671 / 1718\n",
      "53472/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9900Step 1672 / 1718\n",
      "53504/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9900Step 1673 / 1718\n",
      "53536/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9900Step 1674 / 1718\n",
      "53568/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9900Step 1675 / 1718\n",
      "53600/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9900Step 1676 / 1718\n",
      "53632/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9900Step 1677 / 1718\n",
      "53664/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9900Step 1678 / 1718\n",
      "53696/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9900Step 1679 / 1718\n",
      "53728/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9900Step 1680 / 1718\n",
      "53760/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9900Step 1681 / 1718\n",
      "53792/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9900Step 1682 / 1718\n",
      "53824/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9900Step 1683 / 1718\n",
      "53856/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9900Step 1684 / 1718\n",
      "53888/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9900Step 1685 / 1718\n",
      "53920/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9900Step 1686 / 1718\n",
      "53952/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9900Step 1687 / 1718\n",
      "53984/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9901Step 1688 / 1718\n",
      "54016/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9901Step 1689 / 1718\n",
      "54048/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9901Step 1690 / 1718\n",
      "54080/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9901Step 1691 / 1718\n",
      "54112/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9901Step 1692 / 1718\n",
      "54144/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9901Step 1693 / 1718\n",
      "54176/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9901Step 1694 / 1718\n",
      "54208/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9901Step 1695 / 1718\n",
      "54240/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9901Step 1696 / 1718\n",
      "54272/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9901Step 1697 / 1718\n",
      "54304/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9901Step 1698 / 1718\n",
      "54336/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9901Step 1699 / 1718\n",
      "54368/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9901Step 1700 / 1718\n",
      "54400/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9901Step 1701 / 1718\n",
      "54432/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9901Step 1702 / 1718\n",
      "54464/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9901Step 1703 / 1718\n",
      "54496/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9901Step 1704 / 1718\n",
      "54528/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9901Step 1705 / 1718\n",
      "54560/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9901Step 1706 / 1718\n",
      "54592/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9901Step 1707 / 1718\n",
      "54624/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9901Step 1708 / 1718\n",
      "54656/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9901Step 1709 / 1718\n",
      "54688/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9901Step 1710 / 1718\n",
      "54720/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9901Step 1711 / 1718\n",
      "54752/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9900Step 1712 / 1718\n",
      "54784/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9901Step 1713 / 1718\n",
      "54816/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9901Step 1714 / 1718\n",
      "54848/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9901Step 1715 / 1718\n",
      "54880/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9901Step 1716 / 1718\n",
      "54912/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9901Step 1717 / 1718\n",
      "54944/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9901Step 1718 / 1718\n",
      "55000/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9901\n",
      "5000/5000 - mean: 0.1457 - sparse_categorical_accuracy: 0.9746\n",
      "Epoch 2 / 5\n",
      "Step 1 / 1718\n",
      "32/55000 - mean: 0.0000 - sparse_categorical_accuracy: 1.0000Step 2 / 1718\n",
      "64/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9844Step 3 / 1718\n",
      "96/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9896Step 4 / 1718\n",
      "128/55000 - mean: 0.0192 - sparse_categorical_accuracy: 0.9922Step 5 / 1718\n",
      "160/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9875Step 6 / 1718\n",
      "192/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9896Step 7 / 1718\n",
      "224/55000 - mean: 0.0367 - sparse_categorical_accuracy: 0.9866Step 8 / 1718\n",
      "256/55000 - mean: 0.0323 - sparse_categorical_accuracy: 0.9883Step 9 / 1718\n",
      "288/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9896Step 10 / 1718\n",
      "320/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9875Step 11 / 1718\n",
      "352/55000 - mean: 0.0422 - sparse_categorical_accuracy: 0.9858Step 12 / 1718\n",
      "384/55000 - mean: 0.0434 - sparse_categorical_accuracy: 0.9844Step 13 / 1718\n",
      "416/55000 - mean: 0.0419 - sparse_categorical_accuracy: 0.9832Step 14 / 1718\n",
      "448/55000 - mean: 0.0396 - sparse_categorical_accuracy: 0.9844Step 15 / 1718\n",
      "480/55000 - mean: 0.0389 - sparse_categorical_accuracy: 0.9833Step 16 / 1718\n",
      "512/55000 - mean: 0.0458 - sparse_categorical_accuracy: 0.9824Step 17 / 1718\n",
      "544/55000 - mean: 0.0433 - sparse_categorical_accuracy: 0.9835Step 18 / 1718\n",
      "576/55000 - mean: 0.0409 - sparse_categorical_accuracy: 0.9844Step 19 / 1718\n",
      "608/55000 - mean: 0.0418 - sparse_categorical_accuracy: 0.9836Step 20 / 1718\n",
      "640/55000 - mean: 0.0420 - sparse_categorical_accuracy: 0.9828Step 21 / 1718\n",
      "672/55000 - mean: 0.0400 - sparse_categorical_accuracy: 0.9836Step 22 / 1718\n",
      "704/55000 - mean: 0.0459 - sparse_categorical_accuracy: 0.9815Step 23 / 1718\n",
      "736/55000 - mean: 0.0466 - sparse_categorical_accuracy: 0.9810Step 24 / 1718\n",
      "768/55000 - mean: 0.0447 - sparse_categorical_accuracy: 0.9818Step 25 / 1718\n",
      "800/55000 - mean: 0.0430 - sparse_categorical_accuracy: 0.9825Step 26 / 1718\n",
      "832/55000 - mean: 0.0415 - sparse_categorical_accuracy: 0.9832Step 27 / 1718\n",
      "864/55000 - mean: 0.0414 - sparse_categorical_accuracy: 0.9826Step 28 / 1718\n",
      "896/55000 - mean: 0.0399 - sparse_categorical_accuracy: 0.9833Step 29 / 1718\n",
      "928/55000 - mean: 0.0397 - sparse_categorical_accuracy: 0.9828Step 30 / 1718\n",
      "960/55000 - mean: 0.0383 - sparse_categorical_accuracy: 0.9833Step 31 / 1718\n",
      "992/55000 - mean: 0.0372 - sparse_categorical_accuracy: 0.9839Step 32 / 1718\n",
      "1024/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9844Step 33 / 1718\n",
      "1056/55000 - mean: 0.0371 - sparse_categorical_accuracy: 0.9839Step 34 / 1718\n",
      "1088/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9844Step 35 / 1718\n",
      "1120/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9848Step 36 / 1718\n",
      "1152/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9852Step 37 / 1718\n",
      "1184/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9856Step 38 / 1718\n",
      "1216/55000 - mean: 0.0327 - sparse_categorical_accuracy: 0.9860Step 39 / 1718\n",
      "1248/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9864Step 40 / 1718\n",
      "1280/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9859Step 41 / 1718\n",
      "1312/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9863Step 42 / 1718\n",
      "1344/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9859Step 43 / 1718\n",
      "1376/55000 - mean: 0.0327 - sparse_categorical_accuracy: 0.9862Step 44 / 1718\n",
      "1408/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9865Step 45 / 1718\n",
      "1440/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9868Step 46 / 1718\n",
      "1472/55000 - mean: 0.0324 - sparse_categorical_accuracy: 0.9864Step 47 / 1718\n",
      "1504/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9867Step 48 / 1718\n",
      "1536/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9863Step 49 / 1718\n",
      "1568/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9866Step 50 / 1718\n",
      "1600/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9869Step 51 / 1718\n",
      "1632/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9871Step 52 / 1718\n",
      "1664/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9874Step 53 / 1718\n",
      "1696/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9876Step 54 / 1718\n",
      "1728/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9878Step 55 / 1718\n",
      "1760/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9875Step 56 / 1718\n",
      "1792/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9877Step 57 / 1718\n",
      "1824/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9874Step 58 / 1718\n",
      "1856/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9871Step 59 / 1718\n",
      "1888/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9873Step 60 / 1718\n",
      "1920/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9870Step 61 / 1718\n",
      "1952/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9872Step 62 / 1718\n",
      "1984/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9874Step 63 / 1718\n",
      "2016/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9871Step 64 / 1718\n",
      "2048/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9873Step 65 / 1718\n",
      "2080/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9875Step 66 / 1718\n",
      "2112/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9872Step 67 / 1718\n",
      "2144/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9874Step 68 / 1718\n",
      "2176/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9876Step 69 / 1718\n",
      "2208/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9878Step 70 / 1718\n",
      "2240/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9879Step 71 / 1718\n",
      "2272/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9881Step 72 / 1718\n",
      "2304/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9883Step 73 / 1718\n",
      "2336/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9880Step 74 / 1718\n",
      "2368/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9882Step 75 / 1718\n",
      "2400/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9883Step 76 / 1718\n",
      "2432/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9885Step 77 / 1718\n",
      "2464/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9886Step 78 / 1718\n",
      "2496/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9888Step 79 / 1718\n",
      "2528/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9889Step 80 / 1718\n",
      "2560/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9887Step 81 / 1718\n",
      "2592/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9888Step 82 / 1718\n",
      "2624/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9889Step 83 / 1718\n",
      "2656/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9891Step 84 / 1718\n",
      "2688/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9892Step 85 / 1718\n",
      "2720/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9893Step 86 / 1718\n",
      "2752/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9895Step 87 / 1718\n",
      "2784/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9896Step 88 / 1718\n",
      "2816/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9890Step 89 / 1718\n",
      "2848/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9891Step 90 / 1718\n",
      "2880/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9892Step 91 / 1718\n",
      "2912/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9894Step 92 / 1718\n",
      "2944/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9895Step 93 / 1718\n",
      "2976/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9896Step 94 / 1718\n",
      "3008/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9897Step 95 / 1718\n",
      "3040/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9898Step 96 / 1718\n",
      "3072/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9899Step 97 / 1718\n",
      "3104/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9897Step 98 / 1718\n",
      "3136/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9898Step 99 / 1718\n",
      "3168/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9899Step 100 / 1718\n",
      "3200/55000 - mean: 0.0235 - sparse_categorical_accuracy: 0.9900Step 101 / 1718\n",
      "3232/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9898Step 102 / 1718\n",
      "3264/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9899Step 103 / 1718\n",
      "3296/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9900Step 104 / 1718\n",
      "3328/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9901Step 105 / 1718\n",
      "3360/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9902Step 106 / 1718\n",
      "3392/55000 - mean: 0.0236 - sparse_categorical_accuracy: 0.9903Step 107 / 1718\n",
      "3424/55000 - mean: 0.0234 - sparse_categorical_accuracy: 0.9904Step 108 / 1718\n",
      "3456/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9902Step 109 / 1718\n",
      "3488/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9903Step 110 / 1718\n",
      "3520/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9903Step 111 / 1718\n",
      "3552/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9904Step 112 / 1718\n",
      "3584/55000 - mean: 0.0236 - sparse_categorical_accuracy: 0.9905Step 113 / 1718\n",
      "3616/55000 - mean: 0.0234 - sparse_categorical_accuracy: 0.9906Step 114 / 1718\n",
      "3648/55000 - mean: 0.0233 - sparse_categorical_accuracy: 0.9907Step 115 / 1718\n",
      "3680/55000 - mean: 0.0231 - sparse_categorical_accuracy: 0.9908Step 116 / 1718\n",
      "3712/55000 - mean: 0.0232 - sparse_categorical_accuracy: 0.9906Step 117 / 1718\n",
      "3744/55000 - mean: 0.0231 - sparse_categorical_accuracy: 0.9907Step 118 / 1718\n",
      "3776/55000 - mean: 0.0231 - sparse_categorical_accuracy: 0.9907Step 119 / 1718\n",
      "3808/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9903Step 120 / 1718\n",
      "3840/55000 - mean: 0.0235 - sparse_categorical_accuracy: 0.9904Step 121 / 1718\n",
      "3872/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9899Step 122 / 1718\n",
      "3904/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9898Step 123 / 1718\n",
      "3936/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9898Step 124 / 1718\n",
      "3968/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9899Step 125 / 1718\n",
      "4000/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9900Step 126 / 1718\n",
      "4032/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9901Step 127 / 1718\n",
      "4064/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9902Step 128 / 1718\n",
      "4096/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9902Step 129 / 1718\n",
      "4128/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9901Step 130 / 1718\n",
      "4160/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9901Step 131 / 1718\n",
      "4192/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9902Step 132 / 1718\n",
      "4224/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9901Step 133 / 1718\n",
      "4256/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9901Step 134 / 1718\n",
      "4288/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9902Step 135 / 1718\n",
      "4320/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9900Step 136 / 1718\n",
      "4352/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9901Step 137 / 1718\n",
      "4384/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9902Step 138 / 1718\n",
      "4416/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9903Step 139 / 1718\n",
      "4448/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9903Step 140 / 1718\n",
      "4480/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9902Step 141 / 1718\n",
      "4512/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9902Step 142 / 1718\n",
      "4544/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9903Step 143 / 1718\n",
      "4576/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9904Step 144 / 1718\n",
      "4608/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9905Step 145 / 1718\n",
      "4640/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9905Step 146 / 1718\n",
      "4672/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9906Step 147 / 1718\n",
      "4704/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9906Step 148 / 1718\n",
      "4736/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9907Step 149 / 1718\n",
      "4768/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9908Step 150 / 1718\n",
      "4800/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9908Step 151 / 1718\n",
      "4832/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9909Step 152 / 1718\n",
      "4864/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9910Step 153 / 1718\n",
      "4896/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9910Step 154 / 1718\n",
      "4928/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9911Step 155 / 1718\n",
      "4960/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9911Step 156 / 1718\n",
      "4992/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9910Step 157 / 1718\n",
      "5024/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9910Step 158 / 1718\n",
      "5056/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9911Step 159 / 1718\n",
      "5088/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9912Step 160 / 1718\n",
      "5120/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9910Step 161 / 1718\n",
      "5152/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9909Step 162 / 1718\n",
      "5184/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9909Step 163 / 1718\n",
      "5216/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9910Step 164 / 1718\n",
      "5248/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9909Step 165 / 1718\n",
      "5280/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9905Step 166 / 1718\n",
      "5312/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9904Step 167 / 1718\n",
      "5344/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9903Step 168 / 1718\n",
      "5376/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9903Step 169 / 1718\n",
      "5408/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9904Step 170 / 1718\n",
      "5440/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9904Step 171 / 1718\n",
      "5472/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9905Step 172 / 1718\n",
      "5504/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9906Step 173 / 1718\n",
      "5536/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9906Step 174 / 1718\n",
      "5568/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9903Step 175 / 1718\n",
      "5600/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9904Step 176 / 1718\n",
      "5632/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9904Step 177 / 1718\n",
      "5664/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9905Step 178 / 1718\n",
      "5696/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9905Step 179 / 1718\n",
      "5728/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9906Step 180 / 1718\n",
      "5760/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9905Step 181 / 1718\n",
      "5792/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9905Step 182 / 1718\n",
      "5824/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9906Step 183 / 1718\n",
      "5856/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9906Step 184 / 1718\n",
      "5888/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9907Step 185 / 1718\n",
      "5920/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9907Step 186 / 1718\n",
      "5952/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9906Step 187 / 1718\n",
      "5984/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9906Step 188 / 1718\n",
      "6016/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9905Step 189 / 1718\n",
      "6048/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9906Step 190 / 1718\n",
      "6080/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9906Step 191 / 1718\n",
      "6112/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9907Step 192 / 1718\n",
      "6144/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9907Step 193 / 1718\n",
      "6176/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9908Step 194 / 1718\n",
      "6208/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9908Step 195 / 1718\n",
      "6240/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9909Step 196 / 1718\n",
      "6272/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9909Step 197 / 1718\n",
      "6304/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9910Step 198 / 1718\n",
      "6336/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9910Step 199 / 1718\n",
      "6368/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9910Step 200 / 1718\n",
      "6400/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9911Step 201 / 1718\n",
      "6432/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9911Step 202 / 1718\n",
      "6464/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9912Step 203 / 1718\n",
      "6496/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9912Step 204 / 1718\n",
      "6528/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9911Step 205 / 1718\n",
      "6560/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9912Step 206 / 1718\n",
      "6592/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9910Step 207 / 1718\n",
      "6624/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9908Step 208 / 1718\n",
      "6656/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9908Step 209 / 1718\n",
      "6688/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9909Step 210 / 1718\n",
      "6720/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9909Step 211 / 1718\n",
      "6752/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9910Step 212 / 1718\n",
      "6784/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9909Step 213 / 1718\n",
      "6816/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9909Step 214 / 1718\n",
      "6848/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9909Step 215 / 1718\n",
      "6880/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9910Step 216 / 1718\n",
      "6912/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9909Step 217 / 1718\n",
      "6944/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9909Step 218 / 1718\n",
      "6976/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9908Step 219 / 1718\n",
      "7008/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9909Step 220 / 1718\n",
      "7040/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9909Step 221 / 1718\n",
      "7072/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9910Step 222 / 1718\n",
      "7104/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9910Step 223 / 1718\n",
      "7136/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9910Step 224 / 1718\n",
      "7168/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9911Step 225 / 1718\n",
      "7200/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9911Step 226 / 1718\n",
      "7232/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9912Step 227 / 1718\n",
      "7264/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9912Step 228 / 1718\n",
      "7296/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9911Step 229 / 1718\n",
      "7328/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9911Step 230 / 1718\n",
      "7360/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9912Step 231 / 1718\n",
      "7392/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9912Step 232 / 1718\n",
      "7424/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9912Step 233 / 1718\n",
      "7456/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9913Step 234 / 1718\n",
      "7488/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9913Step 235 / 1718\n",
      "7520/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9914Step 236 / 1718\n",
      "7552/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9914Step 237 / 1718\n",
      "7584/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9913Step 238 / 1718\n",
      "7616/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9913Step 239 / 1718\n",
      "7648/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9914Step 240 / 1718\n",
      "7680/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9914Step 241 / 1718\n",
      "7712/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9914Step 242 / 1718\n",
      "7744/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9913Step 243 / 1718\n",
      "7776/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9914Step 244 / 1718\n",
      "7808/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9913Step 245 / 1718\n",
      "7840/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9912Step 246 / 1718\n",
      "7872/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9912Step 247 / 1718\n",
      "7904/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9913Step 248 / 1718\n",
      "7936/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9913Step 249 / 1718\n",
      "7968/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9913Step 250 / 1718\n",
      "8000/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9914Step 251 / 1718\n",
      "8032/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9913Step 252 / 1718\n",
      "8064/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9912Step 253 / 1718\n",
      "8096/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9912Step 254 / 1718\n",
      "8128/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9913Step 255 / 1718\n",
      "8160/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9913Step 256 / 1718\n",
      "8192/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9912Step 257 / 1718\n",
      "8224/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9912Step 258 / 1718\n",
      "8256/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9913Step 259 / 1718\n",
      "8288/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9912Step 260 / 1718\n",
      "8320/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9912Step 261 / 1718\n",
      "8352/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9913Step 262 / 1718\n",
      "8384/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9913Step 263 / 1718\n",
      "8416/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9913Step 264 / 1718\n",
      "8448/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9914Step 265 / 1718\n",
      "8480/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9914Step 266 / 1718\n",
      "8512/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9914Step 267 / 1718\n",
      "8544/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9915Step 268 / 1718\n",
      "8576/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9914Step 269 / 1718\n",
      "8608/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9914Step 270 / 1718\n",
      "8640/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9914Step 271 / 1718\n",
      "8672/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9915Step 272 / 1718\n",
      "8704/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9913Step 273 / 1718\n",
      "8736/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9913Step 274 / 1718\n",
      "8768/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9913Step 275 / 1718\n",
      "8800/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9914Step 276 / 1718\n",
      "8832/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9914Step 277 / 1718\n",
      "8864/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9914Step 278 / 1718\n",
      "8896/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9913Step 279 / 1718\n",
      "8928/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9914Step 280 / 1718\n",
      "8960/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9913Step 281 / 1718\n",
      "8992/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9913Step 282 / 1718\n",
      "9024/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9914Step 283 / 1718\n",
      "9056/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9914Step 284 / 1718\n",
      "9088/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9913Step 285 / 1718\n",
      "9120/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9913Step 286 / 1718\n",
      "9152/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9913Step 287 / 1718\n",
      "9184/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9913Step 288 / 1718\n",
      "9216/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9913Step 289 / 1718\n",
      "9248/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9912Step 290 / 1718\n",
      "9280/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9913Step 291 / 1718\n",
      "9312/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9913Step 292 / 1718\n",
      "9344/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9913Step 293 / 1718\n",
      "9376/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9914Step 294 / 1718\n",
      "9408/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9914Step 295 / 1718\n",
      "9440/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9914Step 296 / 1718\n",
      "9472/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9914Step 297 / 1718\n",
      "9504/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9915Step 298 / 1718\n",
      "9536/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9915Step 299 / 1718\n",
      "9568/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9915Step 300 / 1718\n",
      "9600/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9916Step 301 / 1718\n",
      "9632/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9916Step 302 / 1718\n",
      "9664/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9916Step 303 / 1718\n",
      "9696/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9916Step 304 / 1718\n",
      "9728/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9916Step 305 / 1718\n",
      "9760/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9916Step 306 / 1718\n",
      "9792/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9916Step 307 / 1718\n",
      "9824/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9916Step 308 / 1718\n",
      "9856/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9916Step 309 / 1718\n",
      "9888/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9916Step 310 / 1718\n",
      "9920/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9916Step 311 / 1718\n",
      "9952/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9917Step 312 / 1718\n",
      "9984/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9916Step 313 / 1718\n",
      "10016/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9916Step 314 / 1718\n",
      "10048/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9916Step 315 / 1718\n",
      "10080/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9917Step 316 / 1718\n",
      "10112/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9916Step 317 / 1718\n",
      "10144/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9916Step 318 / 1718\n",
      "10176/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9916Step 319 / 1718\n",
      "10208/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9917Step 320 / 1718\n",
      "10240/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9917Step 321 / 1718\n",
      "10272/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9917Step 322 / 1718\n",
      "10304/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9918Step 323 / 1718\n",
      "10336/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9917Step 324 / 1718\n",
      "10368/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9917Step 325 / 1718\n",
      "10400/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9916Step 326 / 1718\n",
      "10432/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9917Step 327 / 1718\n",
      "10464/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9917Step 328 / 1718\n",
      "10496/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9917Step 329 / 1718\n",
      "10528/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9916Step 330 / 1718\n",
      "10560/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9917Step 331 / 1718\n",
      "10592/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9916Step 332 / 1718\n",
      "10624/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9915Step 333 / 1718\n",
      "10656/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9916Step 334 / 1718\n",
      "10688/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9916Step 335 / 1718\n",
      "10720/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9916Step 336 / 1718\n",
      "10752/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9913Step 337 / 1718\n",
      "10784/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9906Step 338 / 1718\n",
      "10816/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9907Step 339 / 1718\n",
      "10848/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9906Step 340 / 1718\n",
      "10880/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9906Step 341 / 1718\n",
      "10912/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9906Step 342 / 1718\n",
      "10944/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9906Step 343 / 1718\n",
      "10976/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9906Step 344 / 1718\n",
      "11008/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9906Step 345 / 1718\n",
      "11040/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9906Step 346 / 1718\n",
      "11072/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9905Step 347 / 1718\n",
      "11104/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9905Step 348 / 1718\n",
      "11136/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9905Step 349 / 1718\n",
      "11168/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9905Step 350 / 1718\n",
      "11200/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9905Step 351 / 1718\n",
      "11232/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9906Step 352 / 1718\n",
      "11264/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9906Step 353 / 1718\n",
      "11296/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9905Step 354 / 1718\n",
      "11328/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9906Step 355 / 1718\n",
      "11360/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9906Step 356 / 1718\n",
      "11392/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9905Step 357 / 1718\n",
      "11424/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9905Step 358 / 1718\n",
      "11456/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9906Step 359 / 1718\n",
      "11488/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9906Step 360 / 1718\n",
      "11520/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9905Step 361 / 1718\n",
      "11552/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9905Step 362 / 1718\n",
      "11584/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9904Step 363 / 1718\n",
      "11616/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9904Step 364 / 1718\n",
      "11648/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9905Step 365 / 1718\n",
      "11680/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9903Step 366 / 1718\n",
      "11712/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9904Step 367 / 1718\n",
      "11744/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9904Step 368 / 1718\n",
      "11776/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9904Step 369 / 1718\n",
      "11808/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9904Step 370 / 1718\n",
      "11840/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9905Step 371 / 1718\n",
      "11872/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9903Step 372 / 1718\n",
      "11904/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9903Step 373 / 1718\n",
      "11936/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9902Step 374 / 1718\n",
      "11968/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9902Step 375 / 1718\n",
      "12000/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9902Step 376 / 1718\n",
      "12032/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9903Step 377 / 1718\n",
      "12064/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9902Step 378 / 1718\n",
      "12096/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9902Step 379 / 1718\n",
      "12128/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9903Step 380 / 1718\n",
      "12160/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9903Step 381 / 1718\n",
      "12192/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9903Step 382 / 1718\n",
      "12224/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9903Step 383 / 1718\n",
      "12256/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9902Step 384 / 1718\n",
      "12288/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9902Step 385 / 1718\n",
      "12320/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9903Step 386 / 1718\n",
      "12352/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9903Step 387 / 1718\n",
      "12384/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9903Step 388 / 1718\n",
      "12416/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9903Step 389 / 1718\n",
      "12448/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9904Step 390 / 1718\n",
      "12480/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9903Step 391 / 1718\n",
      "12512/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9903Step 392 / 1718\n",
      "12544/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9903Step 393 / 1718\n",
      "12576/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9903Step 394 / 1718\n",
      "12608/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9902Step 395 / 1718\n",
      "12640/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9903Step 396 / 1718\n",
      "12672/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9903Step 397 / 1718\n",
      "12704/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9903Step 398 / 1718\n",
      "12736/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9903Step 399 / 1718\n",
      "12768/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9904Step 400 / 1718\n",
      "12800/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9904Step 401 / 1718\n",
      "12832/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9903Step 402 / 1718\n",
      "12864/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9903Step 403 / 1718\n",
      "12896/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9903Step 404 / 1718\n",
      "12928/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9903Step 405 / 1718\n",
      "12960/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9904Step 406 / 1718\n",
      "12992/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9904Step 407 / 1718\n",
      "13024/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9904Step 408 / 1718\n",
      "13056/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9904Step 409 / 1718\n",
      "13088/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9904Step 410 / 1718\n",
      "13120/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9905Step 411 / 1718\n",
      "13152/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9904Step 412 / 1718\n",
      "13184/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9904Step 413 / 1718\n",
      "13216/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9905Step 414 / 1718\n",
      "13248/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9905Step 415 / 1718\n",
      "13280/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9905Step 416 / 1718\n",
      "13312/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9905Step 417 / 1718\n",
      "13344/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9906Step 418 / 1718\n",
      "13376/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9905Step 419 / 1718\n",
      "13408/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9905Step 420 / 1718\n",
      "13440/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9905Step 421 / 1718\n",
      "13472/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9905Step 422 / 1718\n",
      "13504/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9905Step 423 / 1718\n",
      "13536/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9905Step 424 / 1718\n",
      "13568/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9906Step 425 / 1718\n",
      "13600/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9905Step 426 / 1718\n",
      "13632/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9905Step 427 / 1718\n",
      "13664/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9906Step 428 / 1718\n",
      "13696/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9906Step 429 / 1718\n",
      "13728/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9906Step 430 / 1718\n",
      "13760/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9906Step 431 / 1718\n",
      "13792/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9906Step 432 / 1718\n",
      "13824/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9906Step 433 / 1718\n",
      "13856/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9906Step 434 / 1718\n",
      "13888/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9906Step 435 / 1718\n",
      "13920/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9907Step 436 / 1718\n",
      "13952/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9907Step 437 / 1718\n",
      "13984/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9907Step 438 / 1718\n",
      "14016/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9907Step 439 / 1718\n",
      "14048/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9907Step 440 / 1718\n",
      "14080/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9907Step 441 / 1718\n",
      "14112/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9907Step 442 / 1718\n",
      "14144/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9907Step 443 / 1718\n",
      "14176/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9907Step 444 / 1718\n",
      "14208/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9907Step 445 / 1718\n",
      "14240/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9907Step 446 / 1718\n",
      "14272/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9907Step 447 / 1718\n",
      "14304/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9907Step 448 / 1718\n",
      "14336/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9906Step 449 / 1718\n",
      "14368/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9905Step 450 / 1718\n",
      "14400/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9905Step 451 / 1718\n",
      "14432/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9905Step 452 / 1718\n",
      "14464/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9905Step 453 / 1718\n",
      "14496/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9905Step 454 / 1718\n",
      "14528/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9904Step 455 / 1718\n",
      "14560/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9905Step 456 / 1718\n",
      "14592/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9905Step 457 / 1718\n",
      "14624/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9905Step 458 / 1718\n",
      "14656/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9905Step 459 / 1718\n",
      "14688/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9905Step 460 / 1718\n",
      "14720/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9906Step 461 / 1718\n",
      "14752/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9906Step 462 / 1718\n",
      "14784/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9906Step 463 / 1718\n",
      "14816/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9906Step 464 / 1718\n",
      "14848/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9906Step 465 / 1718\n",
      "14880/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9906Step 466 / 1718\n",
      "14912/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9906Step 467 / 1718\n",
      "14944/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9906Step 468 / 1718\n",
      "14976/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9906Step 469 / 1718\n",
      "15008/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9906Step 470 / 1718\n",
      "15040/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9906Step 471 / 1718\n",
      "15072/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9905Step 472 / 1718\n",
      "15104/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9905Step 473 / 1718\n",
      "15136/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9906Step 474 / 1718\n",
      "15168/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9906Step 475 / 1718\n",
      "15200/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9905Step 476 / 1718\n",
      "15232/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9905Step 477 / 1718\n",
      "15264/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9905Step 478 / 1718\n",
      "15296/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9905Step 479 / 1718\n",
      "15328/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9905Step 480 / 1718\n",
      "15360/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9904Step 481 / 1718\n",
      "15392/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9904Step 482 / 1718\n",
      "15424/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9905Step 483 / 1718\n",
      "15456/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9904Step 484 / 1718\n",
      "15488/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9904Step 485 / 1718\n",
      "15520/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9905Step 486 / 1718\n",
      "15552/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9904Step 487 / 1718\n",
      "15584/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9904Step 488 / 1718\n",
      "15616/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9905Step 489 / 1718\n",
      "15648/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9905Step 490 / 1718\n",
      "15680/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9905Step 491 / 1718\n",
      "15712/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9905Step 492 / 1718\n",
      "15744/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9905Step 493 / 1718\n",
      "15776/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9905Step 494 / 1718\n",
      "15808/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9905Step 495 / 1718\n",
      "15840/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9905Step 496 / 1718\n",
      "15872/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9905Step 497 / 1718\n",
      "15904/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9906Step 498 / 1718\n",
      "15936/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9906Step 499 / 1718\n",
      "15968/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9905Step 500 / 1718\n",
      "16000/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9906Step 501 / 1718\n",
      "16032/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9905Step 502 / 1718\n",
      "16064/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9905Step 503 / 1718\n",
      "16096/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9905Step 504 / 1718\n",
      "16128/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9905Step 505 / 1718\n",
      "16160/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9905Step 506 / 1718\n",
      "16192/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9906Step 507 / 1718\n",
      "16224/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9906Step 508 / 1718\n",
      "16256/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9906Step 509 / 1718\n",
      "16288/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9906Step 510 / 1718\n",
      "16320/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9906Step 511 / 1718\n",
      "16352/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9906Step 512 / 1718\n",
      "16384/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9907Step 513 / 1718\n",
      "16416/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9907Step 514 / 1718\n",
      "16448/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9907Step 515 / 1718\n",
      "16480/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9907Step 516 / 1718\n",
      "16512/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9907Step 517 / 1718\n",
      "16544/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9908Step 518 / 1718\n",
      "16576/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9908Step 519 / 1718\n",
      "16608/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9907Step 520 / 1718\n",
      "16640/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9907Step 521 / 1718\n",
      "16672/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9908Step 522 / 1718\n",
      "16704/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9908Step 523 / 1718\n",
      "16736/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9908Step 524 / 1718\n",
      "16768/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9908Step 525 / 1718\n",
      "16800/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9908Step 526 / 1718\n",
      "16832/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 527 / 1718\n",
      "16864/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 528 / 1718\n",
      "16896/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9909Step 529 / 1718\n",
      "16928/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9909Step 530 / 1718\n",
      "16960/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9909Step 531 / 1718\n",
      "16992/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9909Step 532 / 1718\n",
      "17024/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9910Step 533 / 1718\n",
      "17056/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9910Step 534 / 1718\n",
      "17088/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 535 / 1718\n",
      "17120/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 536 / 1718\n",
      "17152/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 537 / 1718\n",
      "17184/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9909Step 538 / 1718\n",
      "17216/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9909Step 539 / 1718\n",
      "17248/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9909Step 540 / 1718\n",
      "17280/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 541 / 1718\n",
      "17312/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 542 / 1718\n",
      "17344/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9909Step 543 / 1718\n",
      "17376/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 544 / 1718\n",
      "17408/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 545 / 1718\n",
      "17440/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 546 / 1718\n",
      "17472/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9910Step 547 / 1718\n",
      "17504/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 548 / 1718\n",
      "17536/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 549 / 1718\n",
      "17568/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 550 / 1718\n",
      "17600/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9909Step 551 / 1718\n",
      "17632/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 552 / 1718\n",
      "17664/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 553 / 1718\n",
      "17696/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9909Step 554 / 1718\n",
      "17728/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9909Step 555 / 1718\n",
      "17760/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9909Step 556 / 1718\n",
      "17792/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9909Step 557 / 1718\n",
      "17824/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9909Step 558 / 1718\n",
      "17856/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9909Step 559 / 1718\n",
      "17888/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9909Step 560 / 1718\n",
      "17920/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9909Step 561 / 1718\n",
      "17952/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9909Step 562 / 1718\n",
      "17984/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9909Step 563 / 1718\n",
      "18016/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 564 / 1718\n",
      "18048/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 565 / 1718\n",
      "18080/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 566 / 1718\n",
      "18112/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 567 / 1718\n",
      "18144/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 568 / 1718\n",
      "18176/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 569 / 1718\n",
      "18208/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 570 / 1718\n",
      "18240/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 571 / 1718\n",
      "18272/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 572 / 1718\n",
      "18304/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 573 / 1718\n",
      "18336/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 574 / 1718\n",
      "18368/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9910Step 575 / 1718\n",
      "18400/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9910Step 576 / 1718\n",
      "18432/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9909Step 577 / 1718\n",
      "18464/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 578 / 1718\n",
      "18496/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9909Step 579 / 1718\n",
      "18528/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9909Step 580 / 1718\n",
      "18560/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9909Step 581 / 1718\n",
      "18592/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9909Step 582 / 1718\n",
      "18624/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9909Step 583 / 1718\n",
      "18656/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9909Step 584 / 1718\n",
      "18688/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 585 / 1718\n",
      "18720/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9909Step 586 / 1718\n",
      "18752/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9909Step 587 / 1718\n",
      "18784/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9909Step 588 / 1718\n",
      "18816/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9909Step 589 / 1718\n",
      "18848/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9909Step 590 / 1718\n",
      "18880/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9909Step 591 / 1718\n",
      "18912/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9909Step 592 / 1718\n",
      "18944/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9909Step 593 / 1718\n",
      "18976/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9909Step 594 / 1718\n",
      "19008/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9909Step 595 / 1718\n",
      "19040/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9909Step 596 / 1718\n",
      "19072/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 597 / 1718\n",
      "19104/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 598 / 1718\n",
      "19136/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 599 / 1718\n",
      "19168/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9909Step 600 / 1718\n",
      "19200/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9909Step 601 / 1718\n",
      "19232/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9908Step 602 / 1718\n",
      "19264/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9908Step 603 / 1718\n",
      "19296/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9908Step 604 / 1718\n",
      "19328/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9908Step 605 / 1718\n",
      "19360/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9909Step 606 / 1718\n",
      "19392/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9909Step 607 / 1718\n",
      "19424/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9909Step 608 / 1718\n",
      "19456/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9909Step 609 / 1718\n",
      "19488/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 610 / 1718\n",
      "19520/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 611 / 1718\n",
      "19552/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 612 / 1718\n",
      "19584/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 613 / 1718\n",
      "19616/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 614 / 1718\n",
      "19648/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 615 / 1718\n",
      "19680/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 616 / 1718\n",
      "19712/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 617 / 1718\n",
      "19744/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 618 / 1718\n",
      "19776/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 619 / 1718\n",
      "19808/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 620 / 1718\n",
      "19840/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9909Step 621 / 1718\n",
      "19872/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9909Step 622 / 1718\n",
      "19904/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9910Step 623 / 1718\n",
      "19936/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9910Step 624 / 1718\n",
      "19968/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 625 / 1718\n",
      "20000/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 626 / 1718\n",
      "20032/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 627 / 1718\n",
      "20064/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 628 / 1718\n",
      "20096/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 629 / 1718\n",
      "20128/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 630 / 1718\n",
      "20160/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 631 / 1718\n",
      "20192/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 632 / 1718\n",
      "20224/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 633 / 1718\n",
      "20256/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 634 / 1718\n",
      "20288/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 635 / 1718\n",
      "20320/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 636 / 1718\n",
      "20352/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 637 / 1718\n",
      "20384/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 638 / 1718\n",
      "20416/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 639 / 1718\n",
      "20448/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 640 / 1718\n",
      "20480/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9911Step 641 / 1718\n",
      "20512/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 642 / 1718\n",
      "20544/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9910Step 643 / 1718\n",
      "20576/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9911Step 644 / 1718\n",
      "20608/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9911Step 645 / 1718\n",
      "20640/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9911Step 646 / 1718\n",
      "20672/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 647 / 1718\n",
      "20704/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 648 / 1718\n",
      "20736/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 649 / 1718\n",
      "20768/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 650 / 1718\n",
      "20800/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 651 / 1718\n",
      "20832/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9909Step 652 / 1718\n",
      "20864/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 653 / 1718\n",
      "20896/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 654 / 1718\n",
      "20928/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9909Step 655 / 1718\n",
      "20960/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9909Step 656 / 1718\n",
      "20992/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9909Step 657 / 1718\n",
      "21024/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9910Step 658 / 1718\n",
      "21056/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 659 / 1718\n",
      "21088/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9909Step 660 / 1718\n",
      "21120/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 661 / 1718\n",
      "21152/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 662 / 1718\n",
      "21184/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9909Step 663 / 1718\n",
      "21216/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9910Step 664 / 1718\n",
      "21248/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9910Step 665 / 1718\n",
      "21280/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9910Step 666 / 1718\n",
      "21312/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9909Step 667 / 1718\n",
      "21344/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9909Step 668 / 1718\n",
      "21376/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9909Step 669 / 1718\n",
      "21408/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9908Step 670 / 1718\n",
      "21440/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9908Step 671 / 1718\n",
      "21472/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9907Step 672 / 1718\n",
      "21504/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9907Step 673 / 1718\n",
      "21536/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9907Step 674 / 1718\n",
      "21568/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9907Step 675 / 1718\n",
      "21600/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9907Step 676 / 1718\n",
      "21632/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9907Step 677 / 1718\n",
      "21664/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9907Step 678 / 1718\n",
      "21696/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9906Step 679 / 1718\n",
      "21728/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9907Step 680 / 1718\n",
      "21760/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9907Step 681 / 1718\n",
      "21792/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9907Step 682 / 1718\n",
      "21824/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9907Step 683 / 1718\n",
      "21856/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9906Step 684 / 1718\n",
      "21888/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9906Step 685 / 1718\n",
      "21920/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9906Step 686 / 1718\n",
      "21952/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9906Step 687 / 1718\n",
      "21984/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9906Step 688 / 1718\n",
      "22016/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9906Step 689 / 1718\n",
      "22048/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9907Step 690 / 1718\n",
      "22080/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9907Step 691 / 1718\n",
      "22112/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9907Step 692 / 1718\n",
      "22144/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9907Step 693 / 1718\n",
      "22176/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9906Step 694 / 1718\n",
      "22208/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9906Step 695 / 1718\n",
      "22240/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9906Step 696 / 1718\n",
      "22272/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9907Step 697 / 1718\n",
      "22304/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9906Step 698 / 1718\n",
      "22336/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9906Step 699 / 1718\n",
      "22368/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9907Step 700 / 1718\n",
      "22400/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9907Step 701 / 1718\n",
      "22432/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9907Step 702 / 1718\n",
      "22464/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9907Step 703 / 1718\n",
      "22496/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9907Step 704 / 1718\n",
      "22528/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9907Step 705 / 1718\n",
      "22560/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9907Step 706 / 1718\n",
      "22592/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9907Step 707 / 1718\n",
      "22624/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9908Step 708 / 1718\n",
      "22656/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9907Step 709 / 1718\n",
      "22688/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9907Step 710 / 1718\n",
      "22720/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9908Step 711 / 1718\n",
      "22752/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9908Step 712 / 1718\n",
      "22784/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9908Step 713 / 1718\n",
      "22816/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9908Step 714 / 1718\n",
      "22848/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9907Step 715 / 1718\n",
      "22880/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9907Step 716 / 1718\n",
      "22912/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9907Step 717 / 1718\n",
      "22944/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9907Step 718 / 1718\n",
      "22976/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9907Step 719 / 1718\n",
      "23008/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9907Step 720 / 1718\n",
      "23040/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9908Step 721 / 1718\n",
      "23072/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9908Step 722 / 1718\n",
      "23104/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9907Step 723 / 1718\n",
      "23136/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9908Step 724 / 1718\n",
      "23168/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9908Step 725 / 1718\n",
      "23200/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9908Step 726 / 1718\n",
      "23232/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9908Step 727 / 1718\n",
      "23264/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9908Step 728 / 1718\n",
      "23296/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9908Step 729 / 1718\n",
      "23328/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9908Step 730 / 1718\n",
      "23360/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9908Step 731 / 1718\n",
      "23392/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9908Step 732 / 1718\n",
      "23424/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9908Step 733 / 1718\n",
      "23456/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9908Step 734 / 1718\n",
      "23488/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9908Step 735 / 1718\n",
      "23520/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9908Step 736 / 1718\n",
      "23552/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9908Step 737 / 1718\n",
      "23584/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9908Step 738 / 1718\n",
      "23616/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9909Step 739 / 1718\n",
      "23648/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9908Step 740 / 1718\n",
      "23680/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9908Step 741 / 1718\n",
      "23712/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9908Step 742 / 1718\n",
      "23744/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9908Step 743 / 1718\n",
      "23776/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9908Step 744 / 1718\n",
      "23808/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9908Step 745 / 1718\n",
      "23840/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9909Step 746 / 1718\n",
      "23872/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9909Step 747 / 1718\n",
      "23904/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9909Step 748 / 1718\n",
      "23936/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9909Step 749 / 1718\n",
      "23968/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9909Step 750 / 1718\n",
      "24000/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9909Step 751 / 1718\n",
      "24032/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9908Step 752 / 1718\n",
      "24064/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9909Step 753 / 1718\n",
      "24096/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9909Step 754 / 1718\n",
      "24128/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9909Step 755 / 1718\n",
      "24160/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9909Step 756 / 1718\n",
      "24192/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9909Step 757 / 1718\n",
      "24224/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9909Step 758 / 1718\n",
      "24256/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9909Step 759 / 1718\n",
      "24288/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9909Step 760 / 1718\n",
      "24320/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9908Step 761 / 1718\n",
      "24352/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9908Step 762 / 1718\n",
      "24384/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9908Step 763 / 1718\n",
      "24416/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9908Step 764 / 1718\n",
      "24448/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9908Step 765 / 1718\n",
      "24480/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9908Step 766 / 1718\n",
      "24512/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9908Step 767 / 1718\n",
      "24544/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9908Step 768 / 1718\n",
      "24576/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9908Step 769 / 1718\n",
      "24608/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9909Step 770 / 1718\n",
      "24640/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9909Step 771 / 1718\n",
      "24672/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9909Step 772 / 1718\n",
      "24704/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9909Step 773 / 1718\n",
      "24736/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9909Step 774 / 1718\n",
      "24768/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9909Step 775 / 1718\n",
      "24800/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9909Step 776 / 1718\n",
      "24832/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9909Step 777 / 1718\n",
      "24864/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9909Step 778 / 1718\n",
      "24896/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9909Step 779 / 1718\n",
      "24928/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9909Step 780 / 1718\n",
      "24960/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9909Step 781 / 1718\n",
      "24992/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9910Step 782 / 1718\n",
      "25024/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9910Step 783 / 1718\n",
      "25056/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9910Step 784 / 1718\n",
      "25088/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9910Step 785 / 1718\n",
      "25120/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9910Step 786 / 1718\n",
      "25152/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9909Step 787 / 1718\n",
      "25184/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9909Step 788 / 1718\n",
      "25216/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9910Step 789 / 1718\n",
      "25248/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9910Step 790 / 1718\n",
      "25280/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9910Step 791 / 1718\n",
      "25312/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9910Step 792 / 1718\n",
      "25344/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9910Step 793 / 1718\n",
      "25376/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9910Step 794 / 1718\n",
      "25408/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9910Step 795 / 1718\n",
      "25440/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9910Step 796 / 1718\n",
      "25472/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9909Step 797 / 1718\n",
      "25504/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9909Step 798 / 1718\n",
      "25536/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9910Step 799 / 1718\n",
      "25568/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9910Step 800 / 1718\n",
      "25600/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9910Step 801 / 1718\n",
      "25632/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9909Step 802 / 1718\n",
      "25664/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9910Step 803 / 1718\n",
      "25696/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9910Step 804 / 1718\n",
      "25728/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9909Step 805 / 1718\n",
      "25760/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9910Step 806 / 1718\n",
      "25792/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9909Step 807 / 1718\n",
      "25824/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9909Step 808 / 1718\n",
      "25856/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9909Step 809 / 1718\n",
      "25888/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9910Step 810 / 1718\n",
      "25920/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9910Step 811 / 1718\n",
      "25952/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 812 / 1718\n",
      "25984/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 813 / 1718\n",
      "26016/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 814 / 1718\n",
      "26048/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 815 / 1718\n",
      "26080/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 816 / 1718\n",
      "26112/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 817 / 1718\n",
      "26144/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 818 / 1718\n",
      "26176/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9911Step 819 / 1718\n",
      "26208/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 820 / 1718\n",
      "26240/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 821 / 1718\n",
      "26272/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 822 / 1718\n",
      "26304/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 823 / 1718\n",
      "26336/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 824 / 1718\n",
      "26368/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 825 / 1718\n",
      "26400/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 826 / 1718\n",
      "26432/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9910Step 827 / 1718\n",
      "26464/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9910Step 828 / 1718\n",
      "26496/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 829 / 1718\n",
      "26528/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 830 / 1718\n",
      "26560/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 831 / 1718\n",
      "26592/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 832 / 1718\n",
      "26624/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 833 / 1718\n",
      "26656/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 834 / 1718\n",
      "26688/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 835 / 1718\n",
      "26720/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 836 / 1718\n",
      "26752/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9909Step 837 / 1718\n",
      "26784/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 838 / 1718\n",
      "26816/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 839 / 1718\n",
      "26848/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 840 / 1718\n",
      "26880/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 841 / 1718\n",
      "26912/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9909Step 842 / 1718\n",
      "26944/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9909Step 843 / 1718\n",
      "26976/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9909Step 844 / 1718\n",
      "27008/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 845 / 1718\n",
      "27040/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 846 / 1718\n",
      "27072/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 847 / 1718\n",
      "27104/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 848 / 1718\n",
      "27136/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 849 / 1718\n",
      "27168/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 850 / 1718\n",
      "27200/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 851 / 1718\n",
      "27232/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 852 / 1718\n",
      "27264/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 853 / 1718\n",
      "27296/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 854 / 1718\n",
      "27328/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 855 / 1718\n",
      "27360/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 856 / 1718\n",
      "27392/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9910Step 857 / 1718\n",
      "27424/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9910Step 858 / 1718\n",
      "27456/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9910Step 859 / 1718\n",
      "27488/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9910Step 860 / 1718\n",
      "27520/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9910Step 861 / 1718\n",
      "27552/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 862 / 1718\n",
      "27584/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9909Step 863 / 1718\n",
      "27616/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 864 / 1718\n",
      "27648/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 865 / 1718\n",
      "27680/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 866 / 1718\n",
      "27712/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 867 / 1718\n",
      "27744/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 868 / 1718\n",
      "27776/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 869 / 1718\n",
      "27808/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 870 / 1718\n",
      "27840/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 871 / 1718\n",
      "27872/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9909Step 872 / 1718\n",
      "27904/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9909Step 873 / 1718\n",
      "27936/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9909Step 874 / 1718\n",
      "27968/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9910Step 875 / 1718\n",
      "28000/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9910Step 876 / 1718\n",
      "28032/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 877 / 1718\n",
      "28064/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 878 / 1718\n",
      "28096/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 879 / 1718\n",
      "28128/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 880 / 1718\n",
      "28160/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 881 / 1718\n",
      "28192/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 882 / 1718\n",
      "28224/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 883 / 1718\n",
      "28256/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 884 / 1718\n",
      "28288/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9910Step 885 / 1718\n",
      "28320/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9909Step 886 / 1718\n",
      "28352/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9909Step 887 / 1718\n",
      "28384/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9909Step 888 / 1718\n",
      "28416/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9909Step 889 / 1718\n",
      "28448/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9909Step 890 / 1718\n",
      "28480/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9909Step 891 / 1718\n",
      "28512/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9909Step 892 / 1718\n",
      "28544/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9909Step 893 / 1718\n",
      "28576/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 894 / 1718\n",
      "28608/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 895 / 1718\n",
      "28640/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9909Step 896 / 1718\n",
      "28672/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 897 / 1718\n",
      "28704/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 898 / 1718\n",
      "28736/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9909Step 899 / 1718\n",
      "28768/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 900 / 1718\n",
      "28800/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 901 / 1718\n",
      "28832/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9909Step 902 / 1718\n",
      "28864/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9909Step 903 / 1718\n",
      "28896/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9909Step 904 / 1718\n",
      "28928/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9909Step 905 / 1718\n",
      "28960/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 906 / 1718\n",
      "28992/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 907 / 1718\n",
      "29024/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 908 / 1718\n",
      "29056/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 909 / 1718\n",
      "29088/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9910Step 910 / 1718\n",
      "29120/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9910Step 911 / 1718\n",
      "29152/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9910Step 912 / 1718\n",
      "29184/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 913 / 1718\n",
      "29216/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 914 / 1718\n",
      "29248/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9910Step 915 / 1718\n",
      "29280/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9911Step 916 / 1718\n",
      "29312/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9911Step 917 / 1718\n",
      "29344/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9911Step 918 / 1718\n",
      "29376/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9911Step 919 / 1718\n",
      "29408/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 920 / 1718\n",
      "29440/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9910Step 921 / 1718\n",
      "29472/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9910Step 922 / 1718\n",
      "29504/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9911Step 923 / 1718\n",
      "29536/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9911Step 924 / 1718\n",
      "29568/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 925 / 1718\n",
      "29600/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 926 / 1718\n",
      "29632/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9911Step 927 / 1718\n",
      "29664/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 928 / 1718\n",
      "29696/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 929 / 1718\n",
      "29728/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 930 / 1718\n",
      "29760/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 931 / 1718\n",
      "29792/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 932 / 1718\n",
      "29824/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 933 / 1718\n",
      "29856/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 934 / 1718\n",
      "29888/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 935 / 1718\n",
      "29920/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 936 / 1718\n",
      "29952/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 937 / 1718\n",
      "29984/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9910Step 938 / 1718\n",
      "30016/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 939 / 1718\n",
      "30048/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 940 / 1718\n",
      "30080/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 941 / 1718\n",
      "30112/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 942 / 1718\n",
      "30144/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 943 / 1718\n",
      "30176/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 944 / 1718\n",
      "30208/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9910Step 945 / 1718\n",
      "30240/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9910Step 946 / 1718\n",
      "30272/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 947 / 1718\n",
      "30304/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9910Step 948 / 1718\n",
      "30336/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 949 / 1718\n",
      "30368/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 950 / 1718\n",
      "30400/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 951 / 1718\n",
      "30432/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 952 / 1718\n",
      "30464/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9910Step 953 / 1718\n",
      "30496/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 954 / 1718\n",
      "30528/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9910Step 955 / 1718\n",
      "30560/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9910Step 956 / 1718\n",
      "30592/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9910Step 957 / 1718\n",
      "30624/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9911Step 958 / 1718\n",
      "30656/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9911Step 959 / 1718\n",
      "30688/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9911Step 960 / 1718\n",
      "30720/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9911Step 961 / 1718\n",
      "30752/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9911Step 962 / 1718\n",
      "30784/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9911Step 963 / 1718\n",
      "30816/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9911Step 964 / 1718\n",
      "30848/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9911Step 965 / 1718\n",
      "30880/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9911Step 966 / 1718\n",
      "30912/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9911Step 967 / 1718\n",
      "30944/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 968 / 1718\n",
      "30976/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 969 / 1718\n",
      "31008/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 970 / 1718\n",
      "31040/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 971 / 1718\n",
      "31072/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 972 / 1718\n",
      "31104/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9912Step 973 / 1718\n",
      "31136/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 974 / 1718\n",
      "31168/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 975 / 1718\n",
      "31200/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 976 / 1718\n",
      "31232/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9911Step 977 / 1718\n",
      "31264/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9911Step 978 / 1718\n",
      "31296/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9911Step 979 / 1718\n",
      "31328/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9911Step 980 / 1718\n",
      "31360/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9911Step 981 / 1718\n",
      "31392/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9911Step 982 / 1718\n",
      "31424/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9912Step 983 / 1718\n",
      "31456/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9912Step 984 / 1718\n",
      "31488/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9912Step 985 / 1718\n",
      "31520/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9912Step 986 / 1718\n",
      "31552/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9912Step 987 / 1718\n",
      "31584/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9912Step 988 / 1718\n",
      "31616/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9912Step 989 / 1718\n",
      "31648/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9912Step 990 / 1718\n",
      "31680/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9912Step 991 / 1718\n",
      "31712/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9912Step 992 / 1718\n",
      "31744/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9912Step 993 / 1718\n",
      "31776/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9912Step 994 / 1718\n",
      "31808/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9912Step 995 / 1718\n",
      "31840/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9912Step 996 / 1718\n",
      "31872/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9912Step 997 / 1718\n",
      "31904/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9912Step 998 / 1718\n",
      "31936/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9912Step 999 / 1718\n",
      "31968/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9912Step 1000 / 1718\n",
      "32000/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9912Step 1001 / 1718\n",
      "32032/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9912Step 1002 / 1718\n",
      "32064/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9911Step 1003 / 1718\n",
      "32096/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9912Step 1004 / 1718\n",
      "32128/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9912Step 1005 / 1718\n",
      "32160/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9911Step 1006 / 1718\n",
      "32192/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9911Step 1007 / 1718\n",
      "32224/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9912Step 1008 / 1718\n",
      "32256/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9912Step 1009 / 1718\n",
      "32288/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9912Step 1010 / 1718\n",
      "32320/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9912Step 1011 / 1718\n",
      "32352/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9912Step 1012 / 1718\n",
      "32384/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9912Step 1013 / 1718\n",
      "32416/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9912Step 1014 / 1718\n",
      "32448/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9912Step 1015 / 1718\n",
      "32480/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9912Step 1016 / 1718\n",
      "32512/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9912Step 1017 / 1718\n",
      "32544/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9912Step 1018 / 1718\n",
      "32576/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9912Step 1019 / 1718\n",
      "32608/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9912Step 1020 / 1718\n",
      "32640/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9912Step 1021 / 1718\n",
      "32672/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9912Step 1022 / 1718\n",
      "32704/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9913Step 1023 / 1718\n",
      "32736/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9913Step 1024 / 1718\n",
      "32768/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9913Step 1025 / 1718\n",
      "32800/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9913Step 1026 / 1718\n",
      "32832/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9913Step 1027 / 1718\n",
      "32864/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9913Step 1028 / 1718\n",
      "32896/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9913Step 1029 / 1718\n",
      "32928/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9913Step 1030 / 1718\n",
      "32960/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9913Step 1031 / 1718\n",
      "32992/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9913Step 1032 / 1718\n",
      "33024/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9913Step 1033 / 1718\n",
      "33056/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9913Step 1034 / 1718\n",
      "33088/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9913Step 1035 / 1718\n",
      "33120/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9913Step 1036 / 1718\n",
      "33152/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9913Step 1037 / 1718\n",
      "33184/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9914Step 1038 / 1718\n",
      "33216/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9914Step 1039 / 1718\n",
      "33248/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9914Step 1040 / 1718\n",
      "33280/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9914Step 1041 / 1718\n",
      "33312/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9914Step 1042 / 1718\n",
      "33344/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9914Step 1043 / 1718\n",
      "33376/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9914Step 1044 / 1718\n",
      "33408/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9914Step 1045 / 1718\n",
      "33440/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9914Step 1046 / 1718\n",
      "33472/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9914Step 1047 / 1718\n",
      "33504/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9914Step 1048 / 1718\n",
      "33536/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9914Step 1049 / 1718\n",
      "33568/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9914Step 1050 / 1718\n",
      "33600/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9914Step 1051 / 1718\n",
      "33632/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9914Step 1052 / 1718\n",
      "33664/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9914Step 1053 / 1718\n",
      "33696/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9914Step 1054 / 1718\n",
      "33728/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9914Step 1055 / 1718\n",
      "33760/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9914Step 1056 / 1718\n",
      "33792/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9914Step 1057 / 1718\n",
      "33824/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9914Step 1058 / 1718\n",
      "33856/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9914Step 1059 / 1718\n",
      "33888/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9914Step 1060 / 1718\n",
      "33920/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9914Step 1061 / 1718\n",
      "33952/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9914Step 1062 / 1718\n",
      "33984/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9914Step 1063 / 1718\n",
      "34016/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9914Step 1064 / 1718\n",
      "34048/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9914Step 1065 / 1718\n",
      "34080/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9914Step 1066 / 1718\n",
      "34112/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9914Step 1067 / 1718\n",
      "34144/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9914Step 1068 / 1718\n",
      "34176/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9914Step 1069 / 1718\n",
      "34208/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9914Step 1070 / 1718\n",
      "34240/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9914Step 1071 / 1718\n",
      "34272/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9914Step 1072 / 1718\n",
      "34304/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9914Step 1073 / 1718\n",
      "34336/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9914Step 1074 / 1718\n",
      "34368/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1075 / 1718\n",
      "34400/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9913Step 1076 / 1718\n",
      "34432/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1077 / 1718\n",
      "34464/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1078 / 1718\n",
      "34496/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1079 / 1718\n",
      "34528/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1080 / 1718\n",
      "34560/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1081 / 1718\n",
      "34592/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1082 / 1718\n",
      "34624/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1083 / 1718\n",
      "34656/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1084 / 1718\n",
      "34688/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1085 / 1718\n",
      "34720/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1086 / 1718\n",
      "34752/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1087 / 1718\n",
      "34784/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1088 / 1718\n",
      "34816/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1089 / 1718\n",
      "34848/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1090 / 1718\n",
      "34880/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1091 / 1718\n",
      "34912/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1092 / 1718\n",
      "34944/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1093 / 1718\n",
      "34976/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 1094 / 1718\n",
      "35008/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 1095 / 1718\n",
      "35040/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1096 / 1718\n",
      "35072/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1097 / 1718\n",
      "35104/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 1098 / 1718\n",
      "35136/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 1099 / 1718\n",
      "35168/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9913Step 1100 / 1718\n",
      "35200/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 1101 / 1718\n",
      "35232/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 1102 / 1718\n",
      "35264/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 1103 / 1718\n",
      "35296/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1104 / 1718\n",
      "35328/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1105 / 1718\n",
      "35360/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1106 / 1718\n",
      "35392/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1107 / 1718\n",
      "35424/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1108 / 1718\n",
      "35456/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1109 / 1718\n",
      "35488/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1110 / 1718\n",
      "35520/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1111 / 1718\n",
      "35552/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1112 / 1718\n",
      "35584/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1113 / 1718\n",
      "35616/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1114 / 1718\n",
      "35648/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1115 / 1718\n",
      "35680/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1116 / 1718\n",
      "35712/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1117 / 1718\n",
      "35744/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1118 / 1718\n",
      "35776/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1119 / 1718\n",
      "35808/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1120 / 1718\n",
      "35840/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1121 / 1718\n",
      "35872/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1122 / 1718\n",
      "35904/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1123 / 1718\n",
      "35936/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1124 / 1718\n",
      "35968/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9913Step 1125 / 1718\n",
      "36000/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1126 / 1718\n",
      "36032/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1127 / 1718\n",
      "36064/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1128 / 1718\n",
      "36096/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1129 / 1718\n",
      "36128/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1130 / 1718\n",
      "36160/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1131 / 1718\n",
      "36192/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9913Step 1132 / 1718\n",
      "36224/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1133 / 1718\n",
      "36256/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1134 / 1718\n",
      "36288/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9913Step 1135 / 1718\n",
      "36320/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9913Step 1136 / 1718\n",
      "36352/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 1137 / 1718\n",
      "36384/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9912Step 1138 / 1718\n",
      "36416/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9912Step 1139 / 1718\n",
      "36448/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9912Step 1140 / 1718\n",
      "36480/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 1141 / 1718\n",
      "36512/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 1142 / 1718\n",
      "36544/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9912Step 1143 / 1718\n",
      "36576/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1144 / 1718\n",
      "36608/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1145 / 1718\n",
      "36640/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9912Step 1146 / 1718\n",
      "36672/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 1147 / 1718\n",
      "36704/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 1148 / 1718\n",
      "36736/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 1149 / 1718\n",
      "36768/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 1150 / 1718\n",
      "36800/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 1151 / 1718\n",
      "36832/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9912Step 1152 / 1718\n",
      "36864/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9912Step 1153 / 1718\n",
      "36896/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9912Step 1154 / 1718\n",
      "36928/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9912Step 1155 / 1718\n",
      "36960/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1156 / 1718\n",
      "36992/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1157 / 1718\n",
      "37024/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9911Step 1158 / 1718\n",
      "37056/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1159 / 1718\n",
      "37088/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1160 / 1718\n",
      "37120/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1161 / 1718\n",
      "37152/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1162 / 1718\n",
      "37184/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1163 / 1718\n",
      "37216/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1164 / 1718\n",
      "37248/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1165 / 1718\n",
      "37280/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9911Step 1166 / 1718\n",
      "37312/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1167 / 1718\n",
      "37344/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1168 / 1718\n",
      "37376/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1169 / 1718\n",
      "37408/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1170 / 1718\n",
      "37440/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1171 / 1718\n",
      "37472/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1172 / 1718\n",
      "37504/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1173 / 1718\n",
      "37536/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1174 / 1718\n",
      "37568/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9910Step 1175 / 1718\n",
      "37600/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9910Step 1176 / 1718\n",
      "37632/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9910Step 1177 / 1718\n",
      "37664/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9910Step 1178 / 1718\n",
      "37696/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9910Step 1179 / 1718\n",
      "37728/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9910Step 1180 / 1718\n",
      "37760/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1181 / 1718\n",
      "37792/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9911Step 1182 / 1718\n",
      "37824/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9911Step 1183 / 1718\n",
      "37856/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9910Step 1184 / 1718\n",
      "37888/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9911Step 1185 / 1718\n",
      "37920/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9911Step 1186 / 1718\n",
      "37952/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9911Step 1187 / 1718\n",
      "37984/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9911Step 1188 / 1718\n",
      "38016/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9911Step 1189 / 1718\n",
      "38048/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9911Step 1190 / 1718\n",
      "38080/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9911Step 1191 / 1718\n",
      "38112/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9911Step 1192 / 1718\n",
      "38144/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9911Step 1193 / 1718\n",
      "38176/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9911Step 1194 / 1718\n",
      "38208/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9911Step 1195 / 1718\n",
      "38240/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9911Step 1196 / 1718\n",
      "38272/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9911Step 1197 / 1718\n",
      "38304/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9911Step 1198 / 1718\n",
      "38336/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1199 / 1718\n",
      "38368/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1200 / 1718\n",
      "38400/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1201 / 1718\n",
      "38432/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1202 / 1718\n",
      "38464/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1203 / 1718\n",
      "38496/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1204 / 1718\n",
      "38528/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1205 / 1718\n",
      "38560/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1206 / 1718\n",
      "38592/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1207 / 1718\n",
      "38624/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9911Step 1208 / 1718\n",
      "38656/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9911Step 1209 / 1718\n",
      "38688/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1210 / 1718\n",
      "38720/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1211 / 1718\n",
      "38752/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9911Step 1212 / 1718\n",
      "38784/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9911Step 1213 / 1718\n",
      "38816/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9911Step 1214 / 1718\n",
      "38848/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9911Step 1215 / 1718\n",
      "38880/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1216 / 1718\n",
      "38912/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9911Step 1217 / 1718\n",
      "38944/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9911Step 1218 / 1718\n",
      "38976/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9911Step 1219 / 1718\n",
      "39008/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9911Step 1220 / 1718\n",
      "39040/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9911Step 1221 / 1718\n",
      "39072/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9911Step 1222 / 1718\n",
      "39104/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9911Step 1223 / 1718\n",
      "39136/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9911Step 1224 / 1718\n",
      "39168/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9911Step 1225 / 1718\n",
      "39200/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9911Step 1226 / 1718\n",
      "39232/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9912Step 1227 / 1718\n",
      "39264/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9912Step 1228 / 1718\n",
      "39296/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9912Step 1229 / 1718\n",
      "39328/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9912Step 1230 / 1718\n",
      "39360/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9912Step 1231 / 1718\n",
      "39392/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9912Step 1232 / 1718\n",
      "39424/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9912Step 1233 / 1718\n",
      "39456/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9911Step 1234 / 1718\n",
      "39488/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1235 / 1718\n",
      "39520/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1236 / 1718\n",
      "39552/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9912Step 1237 / 1718\n",
      "39584/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9912Step 1238 / 1718\n",
      "39616/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1239 / 1718\n",
      "39648/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1240 / 1718\n",
      "39680/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9912Step 1241 / 1718\n",
      "39712/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 1242 / 1718\n",
      "39744/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 1243 / 1718\n",
      "39776/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 1244 / 1718\n",
      "39808/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 1245 / 1718\n",
      "39840/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1246 / 1718\n",
      "39872/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1247 / 1718\n",
      "39904/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1248 / 1718\n",
      "39936/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1249 / 1718\n",
      "39968/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1250 / 1718\n",
      "40000/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1251 / 1718\n",
      "40032/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9912Step 1252 / 1718\n",
      "40064/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1253 / 1718\n",
      "40096/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 1254 / 1718\n",
      "40128/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 1255 / 1718\n",
      "40160/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 1256 / 1718\n",
      "40192/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 1257 / 1718\n",
      "40224/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 1258 / 1718\n",
      "40256/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9912Step 1259 / 1718\n",
      "40288/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1260 / 1718\n",
      "40320/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1261 / 1718\n",
      "40352/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9912Step 1262 / 1718\n",
      "40384/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1263 / 1718\n",
      "40416/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9911Step 1264 / 1718\n",
      "40448/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1265 / 1718\n",
      "40480/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1266 / 1718\n",
      "40512/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1267 / 1718\n",
      "40544/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1268 / 1718\n",
      "40576/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1269 / 1718\n",
      "40608/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1270 / 1718\n",
      "40640/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1271 / 1718\n",
      "40672/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1272 / 1718\n",
      "40704/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1273 / 1718\n",
      "40736/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1274 / 1718\n",
      "40768/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1275 / 1718\n",
      "40800/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1276 / 1718\n",
      "40832/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1277 / 1718\n",
      "40864/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1278 / 1718\n",
      "40896/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1279 / 1718\n",
      "40928/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1280 / 1718\n",
      "40960/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1281 / 1718\n",
      "40992/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1282 / 1718\n",
      "41024/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1283 / 1718\n",
      "41056/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1284 / 1718\n",
      "41088/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1285 / 1718\n",
      "41120/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1286 / 1718\n",
      "41152/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1287 / 1718\n",
      "41184/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1288 / 1718\n",
      "41216/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9911Step 1289 / 1718\n",
      "41248/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9911Step 1290 / 1718\n",
      "41280/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9911Step 1291 / 1718\n",
      "41312/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9910Step 1292 / 1718\n",
      "41344/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9911Step 1293 / 1718\n",
      "41376/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9911Step 1294 / 1718\n",
      "41408/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9911Step 1295 / 1718\n",
      "41440/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1296 / 1718\n",
      "41472/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1297 / 1718\n",
      "41504/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1298 / 1718\n",
      "41536/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1299 / 1718\n",
      "41568/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1300 / 1718\n",
      "41600/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1301 / 1718\n",
      "41632/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1302 / 1718\n",
      "41664/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1303 / 1718\n",
      "41696/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9911Step 1304 / 1718\n",
      "41728/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9911Step 1305 / 1718\n",
      "41760/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9910Step 1306 / 1718\n",
      "41792/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9911Step 1307 / 1718\n",
      "41824/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9910Step 1308 / 1718\n",
      "41856/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9910Step 1309 / 1718\n",
      "41888/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9910Step 1310 / 1718\n",
      "41920/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9910Step 1311 / 1718\n",
      "41952/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9910Step 1312 / 1718\n",
      "41984/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9910Step 1313 / 1718\n",
      "42016/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9911Step 1314 / 1718\n",
      "42048/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9910Step 1315 / 1718\n",
      "42080/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9910Step 1316 / 1718\n",
      "42112/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9910Step 1317 / 1718\n",
      "42144/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9910Step 1318 / 1718\n",
      "42176/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9910Step 1319 / 1718\n",
      "42208/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9910Step 1320 / 1718\n",
      "42240/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9910Step 1321 / 1718\n",
      "42272/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1322 / 1718\n",
      "42304/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1323 / 1718\n",
      "42336/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1324 / 1718\n",
      "42368/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1325 / 1718\n",
      "42400/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1326 / 1718\n",
      "42432/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1327 / 1718\n",
      "42464/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1328 / 1718\n",
      "42496/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1329 / 1718\n",
      "42528/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1330 / 1718\n",
      "42560/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1331 / 1718\n",
      "42592/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9910Step 1332 / 1718\n",
      "42624/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9910Step 1333 / 1718\n",
      "42656/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1334 / 1718\n",
      "42688/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9910Step 1335 / 1718\n",
      "42720/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9909Step 1336 / 1718\n",
      "42752/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9909Step 1337 / 1718\n",
      "42784/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9909Step 1338 / 1718\n",
      "42816/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9909Step 1339 / 1718\n",
      "42848/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9909Step 1340 / 1718\n",
      "42880/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1341 / 1718\n",
      "42912/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1342 / 1718\n",
      "42944/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1343 / 1718\n",
      "42976/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9909Step 1344 / 1718\n",
      "43008/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1345 / 1718\n",
      "43040/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1346 / 1718\n",
      "43072/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1347 / 1718\n",
      "43104/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1348 / 1718\n",
      "43136/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1349 / 1718\n",
      "43168/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1350 / 1718\n",
      "43200/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1351 / 1718\n",
      "43232/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1352 / 1718\n",
      "43264/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1353 / 1718\n",
      "43296/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1354 / 1718\n",
      "43328/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1355 / 1718\n",
      "43360/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1356 / 1718\n",
      "43392/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1357 / 1718\n",
      "43424/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1358 / 1718\n",
      "43456/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1359 / 1718\n",
      "43488/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1360 / 1718\n",
      "43520/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1361 / 1718\n",
      "43552/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1362 / 1718\n",
      "43584/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1363 / 1718\n",
      "43616/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9909Step 1364 / 1718\n",
      "43648/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1365 / 1718\n",
      "43680/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1366 / 1718\n",
      "43712/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1367 / 1718\n",
      "43744/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1368 / 1718\n",
      "43776/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1369 / 1718\n",
      "43808/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1370 / 1718\n",
      "43840/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1371 / 1718\n",
      "43872/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1372 / 1718\n",
      "43904/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1373 / 1718\n",
      "43936/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1374 / 1718\n",
      "43968/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1375 / 1718\n",
      "44000/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1376 / 1718\n",
      "44032/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1377 / 1718\n",
      "44064/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9909Step 1378 / 1718\n",
      "44096/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1379 / 1718\n",
      "44128/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1380 / 1718\n",
      "44160/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1381 / 1718\n",
      "44192/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1382 / 1718\n",
      "44224/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1383 / 1718\n",
      "44256/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1384 / 1718\n",
      "44288/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1385 / 1718\n",
      "44320/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1386 / 1718\n",
      "44352/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1387 / 1718\n",
      "44384/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1388 / 1718\n",
      "44416/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1389 / 1718\n",
      "44448/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1390 / 1718\n",
      "44480/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1391 / 1718\n",
      "44512/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1392 / 1718\n",
      "44544/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1393 / 1718\n",
      "44576/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1394 / 1718\n",
      "44608/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1395 / 1718\n",
      "44640/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1396 / 1718\n",
      "44672/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9910Step 1397 / 1718\n",
      "44704/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9910Step 1398 / 1718\n",
      "44736/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9910Step 1399 / 1718\n",
      "44768/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9910Step 1400 / 1718\n",
      "44800/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9910Step 1401 / 1718\n",
      "44832/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1402 / 1718\n",
      "44864/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1403 / 1718\n",
      "44896/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1404 / 1718\n",
      "44928/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1405 / 1718\n",
      "44960/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1406 / 1718\n",
      "44992/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9909Step 1407 / 1718\n",
      "45024/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9909Step 1408 / 1718\n",
      "45056/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9909Step 1409 / 1718\n",
      "45088/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1410 / 1718\n",
      "45120/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1411 / 1718\n",
      "45152/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1412 / 1718\n",
      "45184/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1413 / 1718\n",
      "45216/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1414 / 1718\n",
      "45248/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1415 / 1718\n",
      "45280/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1416 / 1718\n",
      "45312/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1417 / 1718\n",
      "45344/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1418 / 1718\n",
      "45376/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1419 / 1718\n",
      "45408/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1420 / 1718\n",
      "45440/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1421 / 1718\n",
      "45472/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1422 / 1718\n",
      "45504/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1423 / 1718\n",
      "45536/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1424 / 1718\n",
      "45568/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1425 / 1718\n",
      "45600/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1426 / 1718\n",
      "45632/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9909Step 1427 / 1718\n",
      "45664/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1428 / 1718\n",
      "45696/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1429 / 1718\n",
      "45728/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1430 / 1718\n",
      "45760/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1431 / 1718\n",
      "45792/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1432 / 1718\n",
      "45824/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1433 / 1718\n",
      "45856/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1434 / 1718\n",
      "45888/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1435 / 1718\n",
      "45920/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1436 / 1718\n",
      "45952/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1437 / 1718\n",
      "45984/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1438 / 1718\n",
      "46016/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1439 / 1718\n",
      "46048/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1440 / 1718\n",
      "46080/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1441 / 1718\n",
      "46112/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1442 / 1718\n",
      "46144/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1443 / 1718\n",
      "46176/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1444 / 1718\n",
      "46208/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1445 / 1718\n",
      "46240/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9910Step 1446 / 1718\n",
      "46272/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1447 / 1718\n",
      "46304/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1448 / 1718\n",
      "46336/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1449 / 1718\n",
      "46368/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1450 / 1718\n",
      "46400/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9909Step 1451 / 1718\n",
      "46432/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1452 / 1718\n",
      "46464/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1453 / 1718\n",
      "46496/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9909Step 1454 / 1718\n",
      "46528/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9909Step 1455 / 1718\n",
      "46560/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9909Step 1456 / 1718\n",
      "46592/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9909Step 1457 / 1718\n",
      "46624/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9909Step 1458 / 1718\n",
      "46656/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1459 / 1718\n",
      "46688/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1460 / 1718\n",
      "46720/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1461 / 1718\n",
      "46752/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1462 / 1718\n",
      "46784/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1463 / 1718\n",
      "46816/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1464 / 1718\n",
      "46848/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1465 / 1718\n",
      "46880/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1466 / 1718\n",
      "46912/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1467 / 1718\n",
      "46944/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1468 / 1718\n",
      "46976/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1469 / 1718\n",
      "47008/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1470 / 1718\n",
      "47040/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1471 / 1718\n",
      "47072/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1472 / 1718\n",
      "47104/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1473 / 1718\n",
      "47136/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9910Step 1474 / 1718\n",
      "47168/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9910Step 1475 / 1718\n",
      "47200/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9910Step 1476 / 1718\n",
      "47232/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9910Step 1477 / 1718\n",
      "47264/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1478 / 1718\n",
      "47296/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1479 / 1718\n",
      "47328/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1480 / 1718\n",
      "47360/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9910Step 1481 / 1718\n",
      "47392/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9911Step 1482 / 1718\n",
      "47424/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9910Step 1483 / 1718\n",
      "47456/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9910Step 1484 / 1718\n",
      "47488/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9910Step 1485 / 1718\n",
      "47520/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9910Step 1486 / 1718\n",
      "47552/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9910Step 1487 / 1718\n",
      "47584/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9910Step 1488 / 1718\n",
      "47616/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9910Step 1489 / 1718\n",
      "47648/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9910Step 1490 / 1718\n",
      "47680/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9910Step 1491 / 1718\n",
      "47712/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9911Step 1492 / 1718\n",
      "47744/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9910Step 1493 / 1718\n",
      "47776/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9910Step 1494 / 1718\n",
      "47808/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9910Step 1495 / 1718\n",
      "47840/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9910Step 1496 / 1718\n",
      "47872/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9910Step 1497 / 1718\n",
      "47904/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9910Step 1498 / 1718\n",
      "47936/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9910Step 1499 / 1718\n",
      "47968/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9910Step 1500 / 1718\n",
      "48000/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9910Step 1501 / 1718\n",
      "48032/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9910Step 1502 / 1718\n",
      "48064/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9910Step 1503 / 1718\n",
      "48096/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9910Step 1504 / 1718\n",
      "48128/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1505 / 1718\n",
      "48160/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1506 / 1718\n",
      "48192/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1507 / 1718\n",
      "48224/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1508 / 1718\n",
      "48256/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1509 / 1718\n",
      "48288/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1510 / 1718\n",
      "48320/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1511 / 1718\n",
      "48352/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1512 / 1718\n",
      "48384/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1513 / 1718\n",
      "48416/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1514 / 1718\n",
      "48448/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1515 / 1718\n",
      "48480/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1516 / 1718\n",
      "48512/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1517 / 1718\n",
      "48544/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9910Step 1518 / 1718\n",
      "48576/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1519 / 1718\n",
      "48608/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1520 / 1718\n",
      "48640/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1521 / 1718\n",
      "48672/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9909Step 1522 / 1718\n",
      "48704/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1523 / 1718\n",
      "48736/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9909Step 1524 / 1718\n",
      "48768/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9909Step 1525 / 1718\n",
      "48800/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9909Step 1526 / 1718\n",
      "48832/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9909Step 1527 / 1718\n",
      "48864/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1528 / 1718\n",
      "48896/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1529 / 1718\n",
      "48928/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1530 / 1718\n",
      "48960/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1531 / 1718\n",
      "48992/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1532 / 1718\n",
      "49024/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1533 / 1718\n",
      "49056/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9909Step 1534 / 1718\n",
      "49088/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9909Step 1535 / 1718\n",
      "49120/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9908Step 1536 / 1718\n",
      "49152/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9908Step 1537 / 1718\n",
      "49184/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9909Step 1538 / 1718\n",
      "49216/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9909Step 1539 / 1718\n",
      "49248/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1540 / 1718\n",
      "49280/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1541 / 1718\n",
      "49312/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1542 / 1718\n",
      "49344/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1543 / 1718\n",
      "49376/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1544 / 1718\n",
      "49408/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9909Step 1545 / 1718\n",
      "49440/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1546 / 1718\n",
      "49472/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1547 / 1718\n",
      "49504/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9908Step 1548 / 1718\n",
      "49536/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1549 / 1718\n",
      "49568/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1550 / 1718\n",
      "49600/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1551 / 1718\n",
      "49632/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1552 / 1718\n",
      "49664/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1553 / 1718\n",
      "49696/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1554 / 1718\n",
      "49728/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1555 / 1718\n",
      "49760/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1556 / 1718\n",
      "49792/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1557 / 1718\n",
      "49824/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1558 / 1718\n",
      "49856/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1559 / 1718\n",
      "49888/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1560 / 1718\n",
      "49920/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1561 / 1718\n",
      "49952/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1562 / 1718\n",
      "49984/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1563 / 1718\n",
      "50016/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1564 / 1718\n",
      "50048/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1565 / 1718\n",
      "50080/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1566 / 1718\n",
      "50112/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1567 / 1718\n",
      "50144/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1568 / 1718\n",
      "50176/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1569 / 1718\n",
      "50208/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1570 / 1718\n",
      "50240/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1571 / 1718\n",
      "50272/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1572 / 1718\n",
      "50304/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1573 / 1718\n",
      "50336/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9908Step 1574 / 1718\n",
      "50368/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9908Step 1575 / 1718\n",
      "50400/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9908Step 1576 / 1718\n",
      "50432/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9908Step 1577 / 1718\n",
      "50464/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9908Step 1578 / 1718\n",
      "50496/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9908Step 1579 / 1718\n",
      "50528/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9908Step 1580 / 1718\n",
      "50560/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9908Step 1581 / 1718\n",
      "50592/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9908Step 1582 / 1718\n",
      "50624/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1583 / 1718\n",
      "50656/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1584 / 1718\n",
      "50688/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1585 / 1718\n",
      "50720/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1586 / 1718\n",
      "50752/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1587 / 1718\n",
      "50784/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1588 / 1718\n",
      "50816/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9908Step 1589 / 1718\n",
      "50848/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1590 / 1718\n",
      "50880/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1591 / 1718\n",
      "50912/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9908Step 1592 / 1718\n",
      "50944/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9909Step 1593 / 1718\n",
      "50976/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9909Step 1594 / 1718\n",
      "51008/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9909Step 1595 / 1718\n",
      "51040/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9909Step 1596 / 1718\n",
      "51072/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9909Step 1597 / 1718\n",
      "51104/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9909Step 1598 / 1718\n",
      "51136/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9909Step 1599 / 1718\n",
      "51168/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9909Step 1600 / 1718\n",
      "51200/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9909Step 1601 / 1718\n",
      "51232/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1602 / 1718\n",
      "51264/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1603 / 1718\n",
      "51296/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1604 / 1718\n",
      "51328/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1605 / 1718\n",
      "51360/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1606 / 1718\n",
      "51392/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1607 / 1718\n",
      "51424/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1608 / 1718\n",
      "51456/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1609 / 1718\n",
      "51488/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9909Step 1610 / 1718\n",
      "51520/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1611 / 1718\n",
      "51552/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1612 / 1718\n",
      "51584/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1613 / 1718\n",
      "51616/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1614 / 1718\n",
      "51648/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1615 / 1718\n",
      "51680/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1616 / 1718\n",
      "51712/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1617 / 1718\n",
      "51744/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1618 / 1718\n",
      "51776/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1619 / 1718\n",
      "51808/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1620 / 1718\n",
      "51840/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9909Step 1621 / 1718\n",
      "51872/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9909Step 1622 / 1718\n",
      "51904/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9909Step 1623 / 1718\n",
      "51936/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1624 / 1718\n",
      "51968/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1625 / 1718\n",
      "52000/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1626 / 1718\n",
      "52032/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1627 / 1718\n",
      "52064/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1628 / 1718\n",
      "52096/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1629 / 1718\n",
      "52128/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1630 / 1718\n",
      "52160/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1631 / 1718\n",
      "52192/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1632 / 1718\n",
      "52224/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9910Step 1633 / 1718\n",
      "52256/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9910Step 1634 / 1718\n",
      "52288/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9910Step 1635 / 1718\n",
      "52320/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1636 / 1718\n",
      "52352/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1637 / 1718\n",
      "52384/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1638 / 1718\n",
      "52416/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9910Step 1639 / 1718\n",
      "52448/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9910Step 1640 / 1718\n",
      "52480/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9910Step 1641 / 1718\n",
      "52512/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9910Step 1642 / 1718\n",
      "52544/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9910Step 1643 / 1718\n",
      "52576/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9910Step 1644 / 1718\n",
      "52608/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1645 / 1718\n",
      "52640/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1646 / 1718\n",
      "52672/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1647 / 1718\n",
      "52704/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1648 / 1718\n",
      "52736/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9910Step 1649 / 1718\n",
      "52768/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9910Step 1650 / 1718\n",
      "52800/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9910Step 1651 / 1718\n",
      "52832/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 1652 / 1718\n",
      "52864/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 1653 / 1718\n",
      "52896/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 1654 / 1718\n",
      "52928/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 1655 / 1718\n",
      "52960/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 1656 / 1718\n",
      "52992/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 1657 / 1718\n",
      "53024/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 1658 / 1718\n",
      "53056/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 1659 / 1718\n",
      "53088/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 1660 / 1718\n",
      "53120/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9910Step 1661 / 1718\n",
      "53152/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 1662 / 1718\n",
      "53184/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 1663 / 1718\n",
      "53216/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 1664 / 1718\n",
      "53248/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 1665 / 1718\n",
      "53280/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 1666 / 1718\n",
      "53312/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 1667 / 1718\n",
      "53344/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1668 / 1718\n",
      "53376/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1669 / 1718\n",
      "53408/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1670 / 1718\n",
      "53440/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1671 / 1718\n",
      "53472/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1672 / 1718\n",
      "53504/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1673 / 1718\n",
      "53536/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9911Step 1674 / 1718\n",
      "53568/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9911Step 1675 / 1718\n",
      "53600/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9911Step 1676 / 1718\n",
      "53632/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9911Step 1677 / 1718\n",
      "53664/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1678 / 1718\n",
      "53696/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 1679 / 1718\n",
      "53728/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1680 / 1718\n",
      "53760/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1681 / 1718\n",
      "53792/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1682 / 1718\n",
      "53824/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1683 / 1718\n",
      "53856/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9911Step 1684 / 1718\n",
      "53888/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9911Step 1685 / 1718\n",
      "53920/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1686 / 1718\n",
      "53952/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 1687 / 1718\n",
      "53984/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9911Step 1688 / 1718\n",
      "54016/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9911Step 1689 / 1718\n",
      "54048/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9910Step 1690 / 1718\n",
      "54080/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9911Step 1691 / 1718\n",
      "54112/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9911Step 1692 / 1718\n",
      "54144/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9911Step 1693 / 1718\n",
      "54176/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9911Step 1694 / 1718\n",
      "54208/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 1695 / 1718\n",
      "54240/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 1696 / 1718\n",
      "54272/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 1697 / 1718\n",
      "54304/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 1698 / 1718\n",
      "54336/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 1699 / 1718\n",
      "54368/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9911Step 1700 / 1718\n",
      "54400/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9911Step 1701 / 1718\n",
      "54432/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9911Step 1702 / 1718\n",
      "54464/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9911Step 1703 / 1718\n",
      "54496/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 1704 / 1718\n",
      "54528/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 1705 / 1718\n",
      "54560/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 1706 / 1718\n",
      "54592/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 1707 / 1718\n",
      "54624/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 1708 / 1718\n",
      "54656/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9911Step 1709 / 1718\n",
      "54688/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 1710 / 1718\n",
      "54720/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 1711 / 1718\n",
      "54752/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9911Step 1712 / 1718\n",
      "54784/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9911Step 1713 / 1718\n",
      "54816/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 1714 / 1718\n",
      "54848/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 1715 / 1718\n",
      "54880/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 1716 / 1718\n",
      "54912/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 1717 / 1718\n",
      "54944/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 1718 / 1718\n",
      "55000/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911\n",
      "5000/5000 - mean: 0.1258 - sparse_categorical_accuracy: 0.9764\n",
      "Epoch 3 / 5\n",
      "Step 1 / 1718\n",
      "32/55000 - mean: 0.0036 - sparse_categorical_accuracy: 1.0000Step 2 / 1718\n",
      "64/55000 - mean: 0.0172 - sparse_categorical_accuracy: 0.9844Step 3 / 1718\n",
      "96/55000 - mean: 0.0144 - sparse_categorical_accuracy: 0.9896Step 4 / 1718\n",
      "128/55000 - mean: 0.0382 - sparse_categorical_accuracy: 0.9844Step 5 / 1718\n",
      "160/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9875Step 6 / 1718\n",
      "192/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9896Step 7 / 1718\n",
      "224/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9911Step 8 / 1718\n",
      "256/55000 - mean: 0.0229 - sparse_categorical_accuracy: 0.9922Step 9 / 1718\n",
      "288/55000 - mean: 0.0215 - sparse_categorical_accuracy: 0.9931Step 10 / 1718\n",
      "320/55000 - mean: 0.0193 - sparse_categorical_accuracy: 0.9937Step 11 / 1718\n",
      "352/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9886Step 12 / 1718\n",
      "384/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9896Step 13 / 1718\n",
      "416/55000 - mean: 0.0225 - sparse_categorical_accuracy: 0.9904Step 14 / 1718\n",
      "448/55000 - mean: 0.0213 - sparse_categorical_accuracy: 0.9911Step 15 / 1718\n",
      "480/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9875Step 16 / 1718\n",
      "512/55000 - mean: 0.0327 - sparse_categorical_accuracy: 0.9883Step 17 / 1718\n",
      "544/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9890Step 18 / 1718\n",
      "576/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9896Step 19 / 1718\n",
      "608/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9901Step 20 / 1718\n",
      "640/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9906Step 21 / 1718\n",
      "672/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9911Step 22 / 1718\n",
      "704/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9915Step 23 / 1718\n",
      "736/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9918Step 24 / 1718\n",
      "768/55000 - mean: 0.0235 - sparse_categorical_accuracy: 0.9922Step 25 / 1718\n",
      "800/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9925Step 26 / 1718\n",
      "832/55000 - mean: 0.0231 - sparse_categorical_accuracy: 0.9928Step 27 / 1718\n",
      "864/55000 - mean: 0.0222 - sparse_categorical_accuracy: 0.9931Step 28 / 1718\n",
      "896/55000 - mean: 0.0227 - sparse_categorical_accuracy: 0.9922Step 29 / 1718\n",
      "928/55000 - mean: 0.0222 - sparse_categorical_accuracy: 0.9925Step 30 / 1718\n",
      "960/55000 - mean: 0.0214 - sparse_categorical_accuracy: 0.9927Step 31 / 1718\n",
      "992/55000 - mean: 0.0208 - sparse_categorical_accuracy: 0.9929Step 32 / 1718\n",
      "1024/55000 - mean: 0.0202 - sparse_categorical_accuracy: 0.9932Step 33 / 1718\n",
      "1056/55000 - mean: 0.0196 - sparse_categorical_accuracy: 0.9934Step 34 / 1718\n",
      "1088/55000 - mean: 0.0190 - sparse_categorical_accuracy: 0.9936Step 35 / 1718\n",
      "1120/55000 - mean: 0.0185 - sparse_categorical_accuracy: 0.9937Step 36 / 1718\n",
      "1152/55000 - mean: 0.0180 - sparse_categorical_accuracy: 0.9939Step 37 / 1718\n",
      "1184/55000 - mean: 0.0178 - sparse_categorical_accuracy: 0.9941Step 38 / 1718\n",
      "1216/55000 - mean: 0.0189 - sparse_categorical_accuracy: 0.9934Step 39 / 1718\n",
      "1248/55000 - mean: 0.0184 - sparse_categorical_accuracy: 0.9936Step 40 / 1718\n",
      "1280/55000 - mean: 0.0183 - sparse_categorical_accuracy: 0.9937Step 41 / 1718\n",
      "1312/55000 - mean: 0.0179 - sparse_categorical_accuracy: 0.9939Step 42 / 1718\n",
      "1344/55000 - mean: 0.0183 - sparse_categorical_accuracy: 0.9933Step 43 / 1718\n",
      "1376/55000 - mean: 0.0180 - sparse_categorical_accuracy: 0.9935Step 44 / 1718\n",
      "1408/55000 - mean: 0.0175 - sparse_categorical_accuracy: 0.9936Step 45 / 1718\n",
      "1440/55000 - mean: 0.0172 - sparse_categorical_accuracy: 0.9937Step 46 / 1718\n",
      "1472/55000 - mean: 0.0171 - sparse_categorical_accuracy: 0.9939Step 47 / 1718\n",
      "1504/55000 - mean: 0.0167 - sparse_categorical_accuracy: 0.9940Step 48 / 1718\n",
      "1536/55000 - mean: 0.0164 - sparse_categorical_accuracy: 0.9941Step 49 / 1718\n",
      "1568/55000 - mean: 0.0160 - sparse_categorical_accuracy: 0.9943Step 50 / 1718\n",
      "1600/55000 - mean: 0.0157 - sparse_categorical_accuracy: 0.9944Step 51 / 1718\n",
      "1632/55000 - mean: 0.0154 - sparse_categorical_accuracy: 0.9945Step 52 / 1718\n",
      "1664/55000 - mean: 0.0152 - sparse_categorical_accuracy: 0.9946Step 53 / 1718\n",
      "1696/55000 - mean: 0.0149 - sparse_categorical_accuracy: 0.9947Step 54 / 1718\n",
      "1728/55000 - mean: 0.0148 - sparse_categorical_accuracy: 0.9948Step 55 / 1718\n",
      "1760/55000 - mean: 0.0146 - sparse_categorical_accuracy: 0.9949Step 56 / 1718\n",
      "1792/55000 - mean: 0.0144 - sparse_categorical_accuracy: 0.9950Step 57 / 1718\n",
      "1824/55000 - mean: 0.0151 - sparse_categorical_accuracy: 0.9945Step 58 / 1718\n",
      "1856/55000 - mean: 0.0148 - sparse_categorical_accuracy: 0.9946Step 59 / 1718\n",
      "1888/55000 - mean: 0.0167 - sparse_categorical_accuracy: 0.9942Step 60 / 1718\n",
      "1920/55000 - mean: 0.0170 - sparse_categorical_accuracy: 0.9937Step 61 / 1718\n",
      "1952/55000 - mean: 0.0170 - sparse_categorical_accuracy: 0.9939Step 62 / 1718\n",
      "1984/55000 - mean: 0.0168 - sparse_categorical_accuracy: 0.9940Step 63 / 1718\n",
      "2016/55000 - mean: 0.0166 - sparse_categorical_accuracy: 0.9940Step 64 / 1718\n",
      "2048/55000 - mean: 0.0168 - sparse_categorical_accuracy: 0.9941Step 65 / 1718\n",
      "2080/55000 - mean: 0.0170 - sparse_categorical_accuracy: 0.9942Step 66 / 1718\n",
      "2112/55000 - mean: 0.0172 - sparse_categorical_accuracy: 0.9938Step 67 / 1718\n",
      "2144/55000 - mean: 0.0170 - sparse_categorical_accuracy: 0.9939Step 68 / 1718\n",
      "2176/55000 - mean: 0.0173 - sparse_categorical_accuracy: 0.9936Step 69 / 1718\n",
      "2208/55000 - mean: 0.0170 - sparse_categorical_accuracy: 0.9937Step 70 / 1718\n",
      "2240/55000 - mean: 0.0179 - sparse_categorical_accuracy: 0.9933Step 71 / 1718\n",
      "2272/55000 - mean: 0.0180 - sparse_categorical_accuracy: 0.9934Step 72 / 1718\n",
      "2304/55000 - mean: 0.0185 - sparse_categorical_accuracy: 0.9931Step 73 / 1718\n",
      "2336/55000 - mean: 0.0183 - sparse_categorical_accuracy: 0.9932Step 74 / 1718\n",
      "2368/55000 - mean: 0.0180 - sparse_categorical_accuracy: 0.9932Step 75 / 1718\n",
      "2400/55000 - mean: 0.0178 - sparse_categorical_accuracy: 0.9933Step 76 / 1718\n",
      "2432/55000 - mean: 0.0177 - sparse_categorical_accuracy: 0.9934Step 77 / 1718\n",
      "2464/55000 - mean: 0.0174 - sparse_categorical_accuracy: 0.9935Step 78 / 1718\n",
      "2496/55000 - mean: 0.0172 - sparse_categorical_accuracy: 0.9936Step 79 / 1718\n",
      "2528/55000 - mean: 0.0174 - sparse_categorical_accuracy: 0.9933Step 80 / 1718\n",
      "2560/55000 - mean: 0.0173 - sparse_categorical_accuracy: 0.9934Step 81 / 1718\n",
      "2592/55000 - mean: 0.0172 - sparse_categorical_accuracy: 0.9934Step 82 / 1718\n",
      "2624/55000 - mean: 0.0170 - sparse_categorical_accuracy: 0.9935Step 83 / 1718\n",
      "2656/55000 - mean: 0.0169 - sparse_categorical_accuracy: 0.9936Step 84 / 1718\n",
      "2688/55000 - mean: 0.0167 - sparse_categorical_accuracy: 0.9937Step 85 / 1718\n",
      "2720/55000 - mean: 0.0165 - sparse_categorical_accuracy: 0.9937Step 86 / 1718\n",
      "2752/55000 - mean: 0.0163 - sparse_categorical_accuracy: 0.9938Step 87 / 1718\n",
      "2784/55000 - mean: 0.0161 - sparse_categorical_accuracy: 0.9939Step 88 / 1718\n",
      "2816/55000 - mean: 0.0172 - sparse_categorical_accuracy: 0.9933Step 89 / 1718\n",
      "2848/55000 - mean: 0.0170 - sparse_categorical_accuracy: 0.9933Step 90 / 1718\n",
      "2880/55000 - mean: 0.0169 - sparse_categorical_accuracy: 0.9934Step 91 / 1718\n",
      "2912/55000 - mean: 0.0168 - sparse_categorical_accuracy: 0.9935Step 92 / 1718\n",
      "2944/55000 - mean: 0.0173 - sparse_categorical_accuracy: 0.9929Step 93 / 1718\n",
      "2976/55000 - mean: 0.0172 - sparse_categorical_accuracy: 0.9929Step 94 / 1718\n",
      "3008/55000 - mean: 0.0170 - sparse_categorical_accuracy: 0.9930Step 95 / 1718\n",
      "3040/55000 - mean: 0.0168 - sparse_categorical_accuracy: 0.9931Step 96 / 1718\n",
      "3072/55000 - mean: 0.0166 - sparse_categorical_accuracy: 0.9932Step 97 / 1718\n",
      "3104/55000 - mean: 0.0165 - sparse_categorical_accuracy: 0.9932Step 98 / 1718\n",
      "3136/55000 - mean: 0.0170 - sparse_categorical_accuracy: 0.9930Step 99 / 1718\n",
      "3168/55000 - mean: 0.0168 - sparse_categorical_accuracy: 0.9931Step 100 / 1718\n",
      "3200/55000 - mean: 0.0174 - sparse_categorical_accuracy: 0.9928Step 101 / 1718\n",
      "3232/55000 - mean: 0.0202 - sparse_categorical_accuracy: 0.9923Step 102 / 1718\n",
      "3264/55000 - mean: 0.0200 - sparse_categorical_accuracy: 0.9923Step 103 / 1718\n",
      "3296/55000 - mean: 0.0211 - sparse_categorical_accuracy: 0.9921Step 104 / 1718\n",
      "3328/55000 - mean: 0.0209 - sparse_categorical_accuracy: 0.9922Step 105 / 1718\n",
      "3360/55000 - mean: 0.0208 - sparse_categorical_accuracy: 0.9923Step 106 / 1718\n",
      "3392/55000 - mean: 0.0207 - sparse_categorical_accuracy: 0.9923Step 107 / 1718\n",
      "3424/55000 - mean: 0.0205 - sparse_categorical_accuracy: 0.9924Step 108 / 1718\n",
      "3456/55000 - mean: 0.0204 - sparse_categorical_accuracy: 0.9925Step 109 / 1718\n",
      "3488/55000 - mean: 0.0212 - sparse_categorical_accuracy: 0.9923Step 110 / 1718\n",
      "3520/55000 - mean: 0.0212 - sparse_categorical_accuracy: 0.9923Step 111 / 1718\n",
      "3552/55000 - mean: 0.0210 - sparse_categorical_accuracy: 0.9924Step 112 / 1718\n",
      "3584/55000 - mean: 0.0208 - sparse_categorical_accuracy: 0.9925Step 113 / 1718\n",
      "3616/55000 - mean: 0.0206 - sparse_categorical_accuracy: 0.9925Step 114 / 1718\n",
      "3648/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9918Step 115 / 1718\n",
      "3680/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9918Step 116 / 1718\n",
      "3712/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9919Step 117 / 1718\n",
      "3744/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9917Step 118 / 1718\n",
      "3776/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9918Step 119 / 1718\n",
      "3808/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9919Step 120 / 1718\n",
      "3840/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9919Step 121 / 1718\n",
      "3872/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9917Step 122 / 1718\n",
      "3904/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9915Step 123 / 1718\n",
      "3936/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9914Step 124 / 1718\n",
      "3968/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9912Step 125 / 1718\n",
      "4000/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9912Step 126 / 1718\n",
      "4032/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9913Step 127 / 1718\n",
      "4064/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9914Step 128 / 1718\n",
      "4096/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9915Step 129 / 1718\n",
      "4128/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9915Step 130 / 1718\n",
      "4160/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9916Step 131 / 1718\n",
      "4192/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9917Step 132 / 1718\n",
      "4224/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9912Step 133 / 1718\n",
      "4256/55000 - mean: 0.0329 - sparse_categorical_accuracy: 0.9913Step 134 / 1718\n",
      "4288/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9911Step 135 / 1718\n",
      "4320/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9912Step 136 / 1718\n",
      "4352/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9913Step 137 / 1718\n",
      "4384/55000 - mean: 0.0329 - sparse_categorical_accuracy: 0.9913Step 138 / 1718\n",
      "4416/55000 - mean: 0.0326 - sparse_categorical_accuracy: 0.9914Step 139 / 1718\n",
      "4448/55000 - mean: 0.0347 - sparse_categorical_accuracy: 0.9912Step 140 / 1718\n",
      "4480/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9913Step 141 / 1718\n",
      "4512/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9911Step 142 / 1718\n",
      "4544/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9912Step 143 / 1718\n",
      "4576/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9913Step 144 / 1718\n",
      "4608/55000 - mean: 0.0338 - sparse_categorical_accuracy: 0.9913Step 145 / 1718\n",
      "4640/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9914Step 146 / 1718\n",
      "4672/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9912Step 147 / 1718\n",
      "4704/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9913Step 148 / 1718\n",
      "4736/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9913Step 149 / 1718\n",
      "4768/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9914Step 150 / 1718\n",
      "4800/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9912Step 151 / 1718\n",
      "4832/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9911Step 152 / 1718\n",
      "4864/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9912Step 153 / 1718\n",
      "4896/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9912Step 154 / 1718\n",
      "4928/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9913Step 155 / 1718\n",
      "4960/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9911Step 156 / 1718\n",
      "4992/55000 - mean: 0.0329 - sparse_categorical_accuracy: 0.9912Step 157 / 1718\n",
      "5024/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9908Step 158 / 1718\n",
      "5056/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9909Step 159 / 1718\n",
      "5088/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9910Step 160 / 1718\n",
      "5120/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9908Step 161 / 1718\n",
      "5152/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9909Step 162 / 1718\n",
      "5184/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9909Step 163 / 1718\n",
      "5216/55000 - mean: 0.0329 - sparse_categorical_accuracy: 0.9910Step 164 / 1718\n",
      "5248/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9909Step 165 / 1718\n",
      "5280/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9907Step 166 / 1718\n",
      "5312/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9908Step 167 / 1718\n",
      "5344/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9906Step 168 / 1718\n",
      "5376/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9907Step 169 / 1718\n",
      "5408/55000 - mean: 0.0338 - sparse_categorical_accuracy: 0.9908Step 170 / 1718\n",
      "5440/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9908Step 171 / 1718\n",
      "5472/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9909Step 172 / 1718\n",
      "5504/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9909Step 173 / 1718\n",
      "5536/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9908Step 174 / 1718\n",
      "5568/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9908Step 175 / 1718\n",
      "5600/55000 - mean: 0.0329 - sparse_categorical_accuracy: 0.9909Step 176 / 1718\n",
      "5632/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9908Step 177 / 1718\n",
      "5664/55000 - mean: 0.0329 - sparse_categorical_accuracy: 0.9908Step 178 / 1718\n",
      "5696/55000 - mean: 0.0327 - sparse_categorical_accuracy: 0.9909Step 179 / 1718\n",
      "5728/55000 - mean: 0.0327 - sparse_categorical_accuracy: 0.9909Step 180 / 1718\n",
      "5760/55000 - mean: 0.0327 - sparse_categorical_accuracy: 0.9908Step 181 / 1718\n",
      "5792/55000 - mean: 0.0325 - sparse_categorical_accuracy: 0.9908Step 182 / 1718\n",
      "5824/55000 - mean: 0.0323 - sparse_categorical_accuracy: 0.9909Step 183 / 1718\n",
      "5856/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9909Step 184 / 1718\n",
      "5888/55000 - mean: 0.0323 - sparse_categorical_accuracy: 0.9908Step 185 / 1718\n",
      "5920/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9909Step 186 / 1718\n",
      "5952/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9909Step 187 / 1718\n",
      "5984/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9910Step 188 / 1718\n",
      "6016/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9910Step 189 / 1718\n",
      "6048/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9911Step 190 / 1718\n",
      "6080/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9911Step 191 / 1718\n",
      "6112/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9912Step 192 / 1718\n",
      "6144/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9912Step 193 / 1718\n",
      "6176/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9913Step 194 / 1718\n",
      "6208/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9911Step 195 / 1718\n",
      "6240/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9910Step 196 / 1718\n",
      "6272/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9911Step 197 / 1718\n",
      "6304/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9910Step 198 / 1718\n",
      "6336/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9910Step 199 / 1718\n",
      "6368/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 200 / 1718\n",
      "6400/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9908Step 201 / 1718\n",
      "6432/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9908Step 202 / 1718\n",
      "6464/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9909Step 203 / 1718\n",
      "6496/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9909Step 204 / 1718\n",
      "6528/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 205 / 1718\n",
      "6560/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9909Step 206 / 1718\n",
      "6592/55000 - mean: 0.0326 - sparse_categorical_accuracy: 0.9907Step 207 / 1718\n",
      "6624/55000 - mean: 0.0324 - sparse_categorical_accuracy: 0.9908Step 208 / 1718\n",
      "6656/55000 - mean: 0.0324 - sparse_categorical_accuracy: 0.9908Step 209 / 1718\n",
      "6688/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9909Step 210 / 1718\n",
      "6720/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9905Step 211 / 1718\n",
      "6752/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9905Step 212 / 1718\n",
      "6784/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9906Step 213 / 1718\n",
      "6816/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9906Step 214 / 1718\n",
      "6848/55000 - mean: 0.0329 - sparse_categorical_accuracy: 0.9907Step 215 / 1718\n",
      "6880/55000 - mean: 0.0328 - sparse_categorical_accuracy: 0.9907Step 216 / 1718\n",
      "6912/55000 - mean: 0.0327 - sparse_categorical_accuracy: 0.9907Step 217 / 1718\n",
      "6944/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9906Step 218 / 1718\n",
      "6976/55000 - mean: 0.0329 - sparse_categorical_accuracy: 0.9907Step 219 / 1718\n",
      "7008/55000 - mean: 0.0328 - sparse_categorical_accuracy: 0.9907Step 220 / 1718\n",
      "7040/55000 - mean: 0.0326 - sparse_categorical_accuracy: 0.9908Step 221 / 1718\n",
      "7072/55000 - mean: 0.0325 - sparse_categorical_accuracy: 0.9908Step 222 / 1718\n",
      "7104/55000 - mean: 0.0323 - sparse_categorical_accuracy: 0.9909Step 223 / 1718\n",
      "7136/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9909Step 224 / 1718\n",
      "7168/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9909Step 225 / 1718\n",
      "7200/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9910Step 226 / 1718\n",
      "7232/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9910Step 227 / 1718\n",
      "7264/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9911Step 228 / 1718\n",
      "7296/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9911Step 229 / 1718\n",
      "7328/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9911Step 230 / 1718\n",
      "7360/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9912Step 231 / 1718\n",
      "7392/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9911Step 232 / 1718\n",
      "7424/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9911Step 233 / 1718\n",
      "7456/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9911Step 234 / 1718\n",
      "7488/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9912Step 235 / 1718\n",
      "7520/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9912Step 236 / 1718\n",
      "7552/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9913Step 237 / 1718\n",
      "7584/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9913Step 238 / 1718\n",
      "7616/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9913Step 239 / 1718\n",
      "7648/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9914Step 240 / 1718\n",
      "7680/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9914Step 241 / 1718\n",
      "7712/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9914Step 242 / 1718\n",
      "7744/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9915Step 243 / 1718\n",
      "7776/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9915Step 244 / 1718\n",
      "7808/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9915Step 245 / 1718\n",
      "7840/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9912Step 246 / 1718\n",
      "7872/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9912Step 247 / 1718\n",
      "7904/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9913Step 248 / 1718\n",
      "7936/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9912Step 249 / 1718\n",
      "7968/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 250 / 1718\n",
      "8000/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9910Step 251 / 1718\n",
      "8032/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9909Step 252 / 1718\n",
      "8064/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9908Step 253 / 1718\n",
      "8096/55000 - mean: 0.0323 - sparse_categorical_accuracy: 0.9907Step 254 / 1718\n",
      "8128/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9908Step 255 / 1718\n",
      "8160/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9908Step 256 / 1718\n",
      "8192/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9908Step 257 / 1718\n",
      "8224/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9909Step 258 / 1718\n",
      "8256/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9909Step 259 / 1718\n",
      "8288/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9910Step 260 / 1718\n",
      "8320/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9910Step 261 / 1718\n",
      "8352/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9910Step 262 / 1718\n",
      "8384/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9911Step 263 / 1718\n",
      "8416/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9911Step 264 / 1718\n",
      "8448/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9910Step 265 / 1718\n",
      "8480/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 266 / 1718\n",
      "8512/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9911Step 267 / 1718\n",
      "8544/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9910Step 268 / 1718\n",
      "8576/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9910Step 269 / 1718\n",
      "8608/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9911Step 270 / 1718\n",
      "8640/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9911Step 271 / 1718\n",
      "8672/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9911Step 272 / 1718\n",
      "8704/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9912Step 273 / 1718\n",
      "8736/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9912Step 274 / 1718\n",
      "8768/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9910Step 275 / 1718\n",
      "8800/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9910Step 276 / 1718\n",
      "8832/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9909Step 277 / 1718\n",
      "8864/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9910Step 278 / 1718\n",
      "8896/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9910Step 279 / 1718\n",
      "8928/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9910Step 280 / 1718\n",
      "8960/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9911Step 281 / 1718\n",
      "8992/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9911Step 282 / 1718\n",
      "9024/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9911Step 283 / 1718\n",
      "9056/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9912Step 284 / 1718\n",
      "9088/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9911Step 285 / 1718\n",
      "9120/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9911Step 286 / 1718\n",
      "9152/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9911Step 287 / 1718\n",
      "9184/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9912Step 288 / 1718\n",
      "9216/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9912Step 289 / 1718\n",
      "9248/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9912Step 290 / 1718\n",
      "9280/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9913Step 291 / 1718\n",
      "9312/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9912Step 292 / 1718\n",
      "9344/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9912Step 293 / 1718\n",
      "9376/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9913Step 294 / 1718\n",
      "9408/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9913Step 295 / 1718\n",
      "9440/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9913Step 296 / 1718\n",
      "9472/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9913Step 297 / 1718\n",
      "9504/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9914Step 298 / 1718\n",
      "9536/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9914Step 299 / 1718\n",
      "9568/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9913Step 300 / 1718\n",
      "9600/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9914Step 301 / 1718\n",
      "9632/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9913Step 302 / 1718\n",
      "9664/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9913Step 303 / 1718\n",
      "9696/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9912Step 304 / 1718\n",
      "9728/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9912Step 305 / 1718\n",
      "9760/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9912Step 306 / 1718\n",
      "9792/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9912Step 307 / 1718\n",
      "9824/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9912Step 308 / 1718\n",
      "9856/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9913Step 309 / 1718\n",
      "9888/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9913Step 310 / 1718\n",
      "9920/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9913Step 311 / 1718\n",
      "9952/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9914Step 312 / 1718\n",
      "9984/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9914Step 313 / 1718\n",
      "10016/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9913Step 314 / 1718\n",
      "10048/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9912Step 315 / 1718\n",
      "10080/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9912Step 316 / 1718\n",
      "10112/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9912Step 317 / 1718\n",
      "10144/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9912Step 318 / 1718\n",
      "10176/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9913Step 319 / 1718\n",
      "10208/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9913Step 320 / 1718\n",
      "10240/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9913Step 321 / 1718\n",
      "10272/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9912Step 322 / 1718\n",
      "10304/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9913Step 323 / 1718\n",
      "10336/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9913Step 324 / 1718\n",
      "10368/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9913Step 325 / 1718\n",
      "10400/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9913Step 326 / 1718\n",
      "10432/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9913Step 327 / 1718\n",
      "10464/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9913Step 328 / 1718\n",
      "10496/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9913Step 329 / 1718\n",
      "10528/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9913Step 330 / 1718\n",
      "10560/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9913Step 331 / 1718\n",
      "10592/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9913Step 332 / 1718\n",
      "10624/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9912Step 333 / 1718\n",
      "10656/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9913Step 334 / 1718\n",
      "10688/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9913Step 335 / 1718\n",
      "10720/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9913Step 336 / 1718\n",
      "10752/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9914Step 337 / 1718\n",
      "10784/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9913Step 338 / 1718\n",
      "10816/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9912Step 339 / 1718\n",
      "10848/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9912Step 340 / 1718\n",
      "10880/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9912Step 341 / 1718\n",
      "10912/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9912Step 342 / 1718\n",
      "10944/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9912Step 343 / 1718\n",
      "10976/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9913Step 344 / 1718\n",
      "11008/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9913Step 345 / 1718\n",
      "11040/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9913Step 346 / 1718\n",
      "11072/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9913Step 347 / 1718\n",
      "11104/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9914Step 348 / 1718\n",
      "11136/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9914Step 349 / 1718\n",
      "11168/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9914Step 350 / 1718\n",
      "11200/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9913Step 351 / 1718\n",
      "11232/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9914Step 352 / 1718\n",
      "11264/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9914Step 353 / 1718\n",
      "11296/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9914Step 354 / 1718\n",
      "11328/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9913Step 355 / 1718\n",
      "11360/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9914Step 356 / 1718\n",
      "11392/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9914Step 357 / 1718\n",
      "11424/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9914Step 358 / 1718\n",
      "11456/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9914Step 359 / 1718\n",
      "11488/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9915Step 360 / 1718\n",
      "11520/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9915Step 361 / 1718\n",
      "11552/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9915Step 362 / 1718\n",
      "11584/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9915Step 363 / 1718\n",
      "11616/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9916Step 364 / 1718\n",
      "11648/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9916Step 365 / 1718\n",
      "11680/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9916Step 366 / 1718\n",
      "11712/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9915Step 367 / 1718\n",
      "11744/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9916Step 368 / 1718\n",
      "11776/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9916Step 369 / 1718\n",
      "11808/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9916Step 370 / 1718\n",
      "11840/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9916Step 371 / 1718\n",
      "11872/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9917Step 372 / 1718\n",
      "11904/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9917Step 373 / 1718\n",
      "11936/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9917Step 374 / 1718\n",
      "11968/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9917Step 375 / 1718\n",
      "12000/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9917Step 376 / 1718\n",
      "12032/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9917Step 377 / 1718\n",
      "12064/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9917Step 378 / 1718\n",
      "12096/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9917Step 379 / 1718\n",
      "12128/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9918Step 380 / 1718\n",
      "12160/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9918Step 381 / 1718\n",
      "12192/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9917Step 382 / 1718\n",
      "12224/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9917Step 383 / 1718\n",
      "12256/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9918Step 384 / 1718\n",
      "12288/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9918Step 385 / 1718\n",
      "12320/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9918Step 386 / 1718\n",
      "12352/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9918Step 387 / 1718\n",
      "12384/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9918Step 388 / 1718\n",
      "12416/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9918Step 389 / 1718\n",
      "12448/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9918Step 390 / 1718\n",
      "12480/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9918Step 391 / 1718\n",
      "12512/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9918Step 392 / 1718\n",
      "12544/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9919Step 393 / 1718\n",
      "12576/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9919Step 394 / 1718\n",
      "12608/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9918Step 395 / 1718\n",
      "12640/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9919Step 396 / 1718\n",
      "12672/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9919Step 397 / 1718\n",
      "12704/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9918Step 398 / 1718\n",
      "12736/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9918Step 399 / 1718\n",
      "12768/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9918Step 400 / 1718\n",
      "12800/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9918Step 401 / 1718\n",
      "12832/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9918Step 402 / 1718\n",
      "12864/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9918Step 403 / 1718\n",
      "12896/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9918Step 404 / 1718\n",
      "12928/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9918Step 405 / 1718\n",
      "12960/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9917Step 406 / 1718\n",
      "12992/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9918Step 407 / 1718\n",
      "13024/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9918Step 408 / 1718\n",
      "13056/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9918Step 409 / 1718\n",
      "13088/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9918Step 410 / 1718\n",
      "13120/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9918Step 411 / 1718\n",
      "13152/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9919Step 412 / 1718\n",
      "13184/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9919Step 413 / 1718\n",
      "13216/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9919Step 414 / 1718\n",
      "13248/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9919Step 415 / 1718\n",
      "13280/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9919Step 416 / 1718\n",
      "13312/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9919Step 417 / 1718\n",
      "13344/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9919Step 418 / 1718\n",
      "13376/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9919Step 419 / 1718\n",
      "13408/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9919Step 420 / 1718\n",
      "13440/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9920Step 421 / 1718\n",
      "13472/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9920Step 422 / 1718\n",
      "13504/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9919Step 423 / 1718\n",
      "13536/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9919Step 424 / 1718\n",
      "13568/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9919Step 425 / 1718\n",
      "13600/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9919Step 426 / 1718\n",
      "13632/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9919Step 427 / 1718\n",
      "13664/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9919Step 428 / 1718\n",
      "13696/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9918Step 429 / 1718\n",
      "13728/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9918Step 430 / 1718\n",
      "13760/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9919Step 431 / 1718\n",
      "13792/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9919Step 432 / 1718\n",
      "13824/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9919Step 433 / 1718\n",
      "13856/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9919Step 434 / 1718\n",
      "13888/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 435 / 1718\n",
      "13920/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9919Step 436 / 1718\n",
      "13952/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9919Step 437 / 1718\n",
      "13984/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9919Step 438 / 1718\n",
      "14016/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9919Step 439 / 1718\n",
      "14048/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9920Step 440 / 1718\n",
      "14080/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9920Step 441 / 1718\n",
      "14112/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9920Step 442 / 1718\n",
      "14144/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9919Step 443 / 1718\n",
      "14176/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9920Step 444 / 1718\n",
      "14208/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9920Step 445 / 1718\n",
      "14240/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9919Step 446 / 1718\n",
      "14272/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9919Step 447 / 1718\n",
      "14304/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9920Step 448 / 1718\n",
      "14336/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9919Step 449 / 1718\n",
      "14368/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9919Step 450 / 1718\n",
      "14400/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9919Step 451 / 1718\n",
      "14432/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9919Step 452 / 1718\n",
      "14464/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9919Step 453 / 1718\n",
      "14496/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9919Step 454 / 1718\n",
      "14528/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9919Step 455 / 1718\n",
      "14560/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9919Step 456 / 1718\n",
      "14592/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9919Step 457 / 1718\n",
      "14624/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9919Step 458 / 1718\n",
      "14656/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9919Step 459 / 1718\n",
      "14688/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9920Step 460 / 1718\n",
      "14720/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9920Step 461 / 1718\n",
      "14752/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9920Step 462 / 1718\n",
      "14784/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9920Step 463 / 1718\n",
      "14816/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9920Step 464 / 1718\n",
      "14848/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9921Step 465 / 1718\n",
      "14880/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9920Step 466 / 1718\n",
      "14912/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9920Step 467 / 1718\n",
      "14944/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9920Step 468 / 1718\n",
      "14976/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9921Step 469 / 1718\n",
      "15008/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9921Step 470 / 1718\n",
      "15040/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9921Step 471 / 1718\n",
      "15072/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9920Step 472 / 1718\n",
      "15104/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9921Step 473 / 1718\n",
      "15136/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9921Step 474 / 1718\n",
      "15168/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9921Step 475 / 1718\n",
      "15200/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9921Step 476 / 1718\n",
      "15232/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9920Step 477 / 1718\n",
      "15264/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9920Step 478 / 1718\n",
      "15296/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9920Step 479 / 1718\n",
      "15328/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9920Step 480 / 1718\n",
      "15360/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9921Step 481 / 1718\n",
      "15392/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9921Step 482 / 1718\n",
      "15424/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9921Step 483 / 1718\n",
      "15456/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9921Step 484 / 1718\n",
      "15488/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9921Step 485 / 1718\n",
      "15520/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9921Step 486 / 1718\n",
      "15552/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9922Step 487 / 1718\n",
      "15584/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9922Step 488 / 1718\n",
      "15616/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9922Step 489 / 1718\n",
      "15648/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9922Step 490 / 1718\n",
      "15680/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9922Step 491 / 1718\n",
      "15712/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9922Step 492 / 1718\n",
      "15744/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9922Step 493 / 1718\n",
      "15776/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9922Step 494 / 1718\n",
      "15808/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9922Step 495 / 1718\n",
      "15840/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9922Step 496 / 1718\n",
      "15872/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9923Step 497 / 1718\n",
      "15904/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9923Step 498 / 1718\n",
      "15936/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9923Step 499 / 1718\n",
      "15968/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9923Step 500 / 1718\n",
      "16000/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9923Step 501 / 1718\n",
      "16032/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9923Step 502 / 1718\n",
      "16064/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9923Step 503 / 1718\n",
      "16096/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9923Step 504 / 1718\n",
      "16128/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9923Step 505 / 1718\n",
      "16160/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9923Step 506 / 1718\n",
      "16192/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9923Step 507 / 1718\n",
      "16224/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9924Step 508 / 1718\n",
      "16256/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9924Step 509 / 1718\n",
      "16288/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9924Step 510 / 1718\n",
      "16320/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9924Step 511 / 1718\n",
      "16352/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9924Step 512 / 1718\n",
      "16384/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9924Step 513 / 1718\n",
      "16416/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9924Step 514 / 1718\n",
      "16448/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9924Step 515 / 1718\n",
      "16480/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9924Step 516 / 1718\n",
      "16512/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9924Step 517 / 1718\n",
      "16544/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9924Step 518 / 1718\n",
      "16576/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9925Step 519 / 1718\n",
      "16608/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9925Step 520 / 1718\n",
      "16640/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9925Step 521 / 1718\n",
      "16672/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9925Step 522 / 1718\n",
      "16704/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9925Step 523 / 1718\n",
      "16736/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9925Step 524 / 1718\n",
      "16768/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9925Step 525 / 1718\n",
      "16800/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9926Step 526 / 1718\n",
      "16832/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9926Step 527 / 1718\n",
      "16864/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9926Step 528 / 1718\n",
      "16896/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9925Step 529 / 1718\n",
      "16928/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9926Step 530 / 1718\n",
      "16960/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9926Step 531 / 1718\n",
      "16992/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9925Step 532 / 1718\n",
      "17024/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9925Step 533 / 1718\n",
      "17056/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9926Step 534 / 1718\n",
      "17088/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9926Step 535 / 1718\n",
      "17120/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9926Step 536 / 1718\n",
      "17152/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9926Step 537 / 1718\n",
      "17184/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9926Step 538 / 1718\n",
      "17216/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9926Step 539 / 1718\n",
      "17248/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9926Step 540 / 1718\n",
      "17280/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9927Step 541 / 1718\n",
      "17312/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9927Step 542 / 1718\n",
      "17344/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9927Step 543 / 1718\n",
      "17376/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9927Step 544 / 1718\n",
      "17408/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9927Step 545 / 1718\n",
      "17440/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9927Step 546 / 1718\n",
      "17472/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9927Step 547 / 1718\n",
      "17504/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9927Step 548 / 1718\n",
      "17536/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9927Step 549 / 1718\n",
      "17568/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9927Step 550 / 1718\n",
      "17600/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9927Step 551 / 1718\n",
      "17632/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9927Step 552 / 1718\n",
      "17664/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9928Step 553 / 1718\n",
      "17696/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9928Step 554 / 1718\n",
      "17728/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9928Step 555 / 1718\n",
      "17760/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9928Step 556 / 1718\n",
      "17792/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9927Step 557 / 1718\n",
      "17824/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9928Step 558 / 1718\n",
      "17856/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9928Step 559 / 1718\n",
      "17888/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 560 / 1718\n",
      "17920/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 561 / 1718\n",
      "17952/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 562 / 1718\n",
      "17984/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 563 / 1718\n",
      "18016/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 564 / 1718\n",
      "18048/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 565 / 1718\n",
      "18080/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 566 / 1718\n",
      "18112/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 567 / 1718\n",
      "18144/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 568 / 1718\n",
      "18176/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 569 / 1718\n",
      "18208/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 570 / 1718\n",
      "18240/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 571 / 1718\n",
      "18272/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 572 / 1718\n",
      "18304/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 573 / 1718\n",
      "18336/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 574 / 1718\n",
      "18368/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 575 / 1718\n",
      "18400/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 576 / 1718\n",
      "18432/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 577 / 1718\n",
      "18464/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9930Step 578 / 1718\n",
      "18496/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 579 / 1718\n",
      "18528/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 580 / 1718\n",
      "18560/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9930Step 581 / 1718\n",
      "18592/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 582 / 1718\n",
      "18624/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 583 / 1718\n",
      "18656/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 584 / 1718\n",
      "18688/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 585 / 1718\n",
      "18720/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 586 / 1718\n",
      "18752/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 587 / 1718\n",
      "18784/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 588 / 1718\n",
      "18816/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 589 / 1718\n",
      "18848/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 590 / 1718\n",
      "18880/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 591 / 1718\n",
      "18912/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 592 / 1718\n",
      "18944/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 593 / 1718\n",
      "18976/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 594 / 1718\n",
      "19008/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 595 / 1718\n",
      "19040/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 596 / 1718\n",
      "19072/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9928Step 597 / 1718\n",
      "19104/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9928Step 598 / 1718\n",
      "19136/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9928Step 599 / 1718\n",
      "19168/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 600 / 1718\n",
      "19200/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 601 / 1718\n",
      "19232/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 602 / 1718\n",
      "19264/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 603 / 1718\n",
      "19296/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9928Step 604 / 1718\n",
      "19328/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 605 / 1718\n",
      "19360/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 606 / 1718\n",
      "19392/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9929Step 607 / 1718\n",
      "19424/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9929Step 608 / 1718\n",
      "19456/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9929Step 609 / 1718\n",
      "19488/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9929Step 610 / 1718\n",
      "19520/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9929Step 611 / 1718\n",
      "19552/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9929Step 612 / 1718\n",
      "19584/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9929Step 613 / 1718\n",
      "19616/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9929Step 614 / 1718\n",
      "19648/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9929Step 615 / 1718\n",
      "19680/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9929Step 616 / 1718\n",
      "19712/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9929Step 617 / 1718\n",
      "19744/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9930Step 618 / 1718\n",
      "19776/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9930Step 619 / 1718\n",
      "19808/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9930Step 620 / 1718\n",
      "19840/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9929Step 621 / 1718\n",
      "19872/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9930Step 622 / 1718\n",
      "19904/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9930Step 623 / 1718\n",
      "19936/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9929Step 624 / 1718\n",
      "19968/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9929Step 625 / 1718\n",
      "20000/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9930Step 626 / 1718\n",
      "20032/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9930Step 627 / 1718\n",
      "20064/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9930Step 628 / 1718\n",
      "20096/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9930Step 629 / 1718\n",
      "20128/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9929Step 630 / 1718\n",
      "20160/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9930Step 631 / 1718\n",
      "20192/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9930Step 632 / 1718\n",
      "20224/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9930Step 633 / 1718\n",
      "20256/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9930Step 634 / 1718\n",
      "20288/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9930Step 635 / 1718\n",
      "20320/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9930Step 636 / 1718\n",
      "20352/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9930Step 637 / 1718\n",
      "20384/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9930Step 638 / 1718\n",
      "20416/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9930Step 639 / 1718\n",
      "20448/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9930Step 640 / 1718\n",
      "20480/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9930Step 641 / 1718\n",
      "20512/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9930Step 642 / 1718\n",
      "20544/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9930Step 643 / 1718\n",
      "20576/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9930Step 644 / 1718\n",
      "20608/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9930Step 645 / 1718\n",
      "20640/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9930Step 646 / 1718\n",
      "20672/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9930Step 647 / 1718\n",
      "20704/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9930Step 648 / 1718\n",
      "20736/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9931Step 649 / 1718\n",
      "20768/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9931Step 650 / 1718\n",
      "20800/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9931Step 651 / 1718\n",
      "20832/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9930Step 652 / 1718\n",
      "20864/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9931Step 653 / 1718\n",
      "20896/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9931Step 654 / 1718\n",
      "20928/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9931Step 655 / 1718\n",
      "20960/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9930Step 656 / 1718\n",
      "20992/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9930Step 657 / 1718\n",
      "21024/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9931Step 658 / 1718\n",
      "21056/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9931Step 659 / 1718\n",
      "21088/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9931Step 660 / 1718\n",
      "21120/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9931Step 661 / 1718\n",
      "21152/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9931Step 662 / 1718\n",
      "21184/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9930Step 663 / 1718\n",
      "21216/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9930Step 664 / 1718\n",
      "21248/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9930Step 665 / 1718\n",
      "21280/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9930Step 666 / 1718\n",
      "21312/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9931Step 667 / 1718\n",
      "21344/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9931Step 668 / 1718\n",
      "21376/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9931Step 669 / 1718\n",
      "21408/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9931Step 670 / 1718\n",
      "21440/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9931Step 671 / 1718\n",
      "21472/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9931Step 672 / 1718\n",
      "21504/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9931Step 673 / 1718\n",
      "21536/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9930Step 674 / 1718\n",
      "21568/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9930Step 675 / 1718\n",
      "21600/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9930Step 676 / 1718\n",
      "21632/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9930Step 677 / 1718\n",
      "21664/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 678 / 1718\n",
      "21696/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 679 / 1718\n",
      "21728/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 680 / 1718\n",
      "21760/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 681 / 1718\n",
      "21792/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 682 / 1718\n",
      "21824/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 683 / 1718\n",
      "21856/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 684 / 1718\n",
      "21888/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 685 / 1718\n",
      "21920/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 686 / 1718\n",
      "21952/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9928Step 687 / 1718\n",
      "21984/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 688 / 1718\n",
      "22016/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 689 / 1718\n",
      "22048/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 690 / 1718\n",
      "22080/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9928Step 691 / 1718\n",
      "22112/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 692 / 1718\n",
      "22144/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 693 / 1718\n",
      "22176/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 694 / 1718\n",
      "22208/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 695 / 1718\n",
      "22240/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 696 / 1718\n",
      "22272/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9928Step 697 / 1718\n",
      "22304/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 698 / 1718\n",
      "22336/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 699 / 1718\n",
      "22368/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 700 / 1718\n",
      "22400/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 701 / 1718\n",
      "22432/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 702 / 1718\n",
      "22464/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 703 / 1718\n",
      "22496/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 704 / 1718\n",
      "22528/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 705 / 1718\n",
      "22560/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 706 / 1718\n",
      "22592/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 707 / 1718\n",
      "22624/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 708 / 1718\n",
      "22656/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 709 / 1718\n",
      "22688/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 710 / 1718\n",
      "22720/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 711 / 1718\n",
      "22752/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 712 / 1718\n",
      "22784/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 713 / 1718\n",
      "22816/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 714 / 1718\n",
      "22848/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 715 / 1718\n",
      "22880/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 716 / 1718\n",
      "22912/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 717 / 1718\n",
      "22944/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 718 / 1718\n",
      "22976/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9927Step 719 / 1718\n",
      "23008/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9927Step 720 / 1718\n",
      "23040/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 721 / 1718\n",
      "23072/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 722 / 1718\n",
      "23104/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 723 / 1718\n",
      "23136/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 724 / 1718\n",
      "23168/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9927Step 725 / 1718\n",
      "23200/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 726 / 1718\n",
      "23232/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 727 / 1718\n",
      "23264/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 728 / 1718\n",
      "23296/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9927Step 729 / 1718\n",
      "23328/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 730 / 1718\n",
      "23360/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9927Step 731 / 1718\n",
      "23392/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9927Step 732 / 1718\n",
      "23424/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9927Step 733 / 1718\n",
      "23456/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9928Step 734 / 1718\n",
      "23488/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9928Step 735 / 1718\n",
      "23520/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9928Step 736 / 1718\n",
      "23552/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 737 / 1718\n",
      "23584/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 738 / 1718\n",
      "23616/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 739 / 1718\n",
      "23648/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 740 / 1718\n",
      "23680/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 741 / 1718\n",
      "23712/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9928Step 742 / 1718\n",
      "23744/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9928Step 743 / 1718\n",
      "23776/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9928Step 744 / 1718\n",
      "23808/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 745 / 1718\n",
      "23840/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 746 / 1718\n",
      "23872/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9928Step 747 / 1718\n",
      "23904/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9928Step 748 / 1718\n",
      "23936/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9928Step 749 / 1718\n",
      "23968/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9928Step 750 / 1718\n",
      "24000/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9928Step 751 / 1718\n",
      "24032/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 752 / 1718\n",
      "24064/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 753 / 1718\n",
      "24096/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 754 / 1718\n",
      "24128/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 755 / 1718\n",
      "24160/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 756 / 1718\n",
      "24192/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 757 / 1718\n",
      "24224/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 758 / 1718\n",
      "24256/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 759 / 1718\n",
      "24288/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 760 / 1718\n",
      "24320/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 761 / 1718\n",
      "24352/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 762 / 1718\n",
      "24384/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 763 / 1718\n",
      "24416/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 764 / 1718\n",
      "24448/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 765 / 1718\n",
      "24480/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 766 / 1718\n",
      "24512/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 767 / 1718\n",
      "24544/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 768 / 1718\n",
      "24576/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 769 / 1718\n",
      "24608/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9926Step 770 / 1718\n",
      "24640/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9926Step 771 / 1718\n",
      "24672/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9926Step 772 / 1718\n",
      "24704/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9926Step 773 / 1718\n",
      "24736/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9926Step 774 / 1718\n",
      "24768/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 775 / 1718\n",
      "24800/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 776 / 1718\n",
      "24832/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 777 / 1718\n",
      "24864/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 778 / 1718\n",
      "24896/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 779 / 1718\n",
      "24928/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 780 / 1718\n",
      "24960/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 781 / 1718\n",
      "24992/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 782 / 1718\n",
      "25024/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 783 / 1718\n",
      "25056/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 784 / 1718\n",
      "25088/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 785 / 1718\n",
      "25120/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 786 / 1718\n",
      "25152/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 787 / 1718\n",
      "25184/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 788 / 1718\n",
      "25216/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 789 / 1718\n",
      "25248/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 790 / 1718\n",
      "25280/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 791 / 1718\n",
      "25312/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 792 / 1718\n",
      "25344/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 793 / 1718\n",
      "25376/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 794 / 1718\n",
      "25408/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 795 / 1718\n",
      "25440/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 796 / 1718\n",
      "25472/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 797 / 1718\n",
      "25504/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 798 / 1718\n",
      "25536/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 799 / 1718\n",
      "25568/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 800 / 1718\n",
      "25600/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 801 / 1718\n",
      "25632/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 802 / 1718\n",
      "25664/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 803 / 1718\n",
      "25696/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 804 / 1718\n",
      "25728/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 805 / 1718\n",
      "25760/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 806 / 1718\n",
      "25792/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 807 / 1718\n",
      "25824/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 808 / 1718\n",
      "25856/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 809 / 1718\n",
      "25888/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9926Step 810 / 1718\n",
      "25920/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9926Step 811 / 1718\n",
      "25952/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9926Step 812 / 1718\n",
      "25984/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9926Step 813 / 1718\n",
      "26016/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9926Step 814 / 1718\n",
      "26048/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9926Step 815 / 1718\n",
      "26080/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9926Step 816 / 1718\n",
      "26112/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9926Step 817 / 1718\n",
      "26144/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9926Step 818 / 1718\n",
      "26176/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9926Step 819 / 1718\n",
      "26208/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9926Step 820 / 1718\n",
      "26240/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9926Step 821 / 1718\n",
      "26272/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9926Step 822 / 1718\n",
      "26304/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9926Step 823 / 1718\n",
      "26336/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9926Step 824 / 1718\n",
      "26368/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9926Step 825 / 1718\n",
      "26400/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9926Step 826 / 1718\n",
      "26432/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9926Step 827 / 1718\n",
      "26464/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9926Step 828 / 1718\n",
      "26496/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 829 / 1718\n",
      "26528/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 830 / 1718\n",
      "26560/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 831 / 1718\n",
      "26592/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 832 / 1718\n",
      "26624/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 833 / 1718\n",
      "26656/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 834 / 1718\n",
      "26688/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 835 / 1718\n",
      "26720/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9926Step 836 / 1718\n",
      "26752/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9926Step 837 / 1718\n",
      "26784/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9926Step 838 / 1718\n",
      "26816/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9927Step 839 / 1718\n",
      "26848/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9927Step 840 / 1718\n",
      "26880/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9927Step 841 / 1718\n",
      "26912/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9927Step 842 / 1718\n",
      "26944/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9927Step 843 / 1718\n",
      "26976/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9927Step 844 / 1718\n",
      "27008/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9927Step 845 / 1718\n",
      "27040/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9926Step 846 / 1718\n",
      "27072/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9926Step 847 / 1718\n",
      "27104/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9927Step 848 / 1718\n",
      "27136/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9927Step 849 / 1718\n",
      "27168/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9927Step 850 / 1718\n",
      "27200/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9927Step 851 / 1718\n",
      "27232/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9927Step 852 / 1718\n",
      "27264/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9927Step 853 / 1718\n",
      "27296/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9926Step 854 / 1718\n",
      "27328/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9926Step 855 / 1718\n",
      "27360/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9926Step 856 / 1718\n",
      "27392/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9926Step 857 / 1718\n",
      "27424/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9925Step 858 / 1718\n",
      "27456/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9925Step 859 / 1718\n",
      "27488/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9925Step 860 / 1718\n",
      "27520/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9925Step 861 / 1718\n",
      "27552/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9925Step 862 / 1718\n",
      "27584/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9925Step 863 / 1718\n",
      "27616/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9925Step 864 / 1718\n",
      "27648/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 865 / 1718\n",
      "27680/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9925Step 866 / 1718\n",
      "27712/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9925Step 867 / 1718\n",
      "27744/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 868 / 1718\n",
      "27776/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9925Step 869 / 1718\n",
      "27808/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 870 / 1718\n",
      "27840/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 871 / 1718\n",
      "27872/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 872 / 1718\n",
      "27904/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 873 / 1718\n",
      "27936/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 874 / 1718\n",
      "27968/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 875 / 1718\n",
      "28000/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 876 / 1718\n",
      "28032/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 877 / 1718\n",
      "28064/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9926Step 878 / 1718\n",
      "28096/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9926Step 879 / 1718\n",
      "28128/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9926Step 880 / 1718\n",
      "28160/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 881 / 1718\n",
      "28192/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9926Step 882 / 1718\n",
      "28224/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9926Step 883 / 1718\n",
      "28256/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9926Step 884 / 1718\n",
      "28288/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9925Step 885 / 1718\n",
      "28320/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 886 / 1718\n",
      "28352/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9926Step 887 / 1718\n",
      "28384/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9926Step 888 / 1718\n",
      "28416/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 889 / 1718\n",
      "28448/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 890 / 1718\n",
      "28480/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 891 / 1718\n",
      "28512/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9925Step 892 / 1718\n",
      "28544/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 893 / 1718\n",
      "28576/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 894 / 1718\n",
      "28608/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 895 / 1718\n",
      "28640/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 896 / 1718\n",
      "28672/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 897 / 1718\n",
      "28704/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 898 / 1718\n",
      "28736/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 899 / 1718\n",
      "28768/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9925Step 900 / 1718\n",
      "28800/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9925Step 901 / 1718\n",
      "28832/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9925Step 902 / 1718\n",
      "28864/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9926Step 903 / 1718\n",
      "28896/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9926Step 904 / 1718\n",
      "28928/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9926Step 905 / 1718\n",
      "28960/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9926Step 906 / 1718\n",
      "28992/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9926Step 907 / 1718\n",
      "29024/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9926Step 908 / 1718\n",
      "29056/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 909 / 1718\n",
      "29088/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 910 / 1718\n",
      "29120/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 911 / 1718\n",
      "29152/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9925Step 912 / 1718\n",
      "29184/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 913 / 1718\n",
      "29216/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 914 / 1718\n",
      "29248/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9925Step 915 / 1718\n",
      "29280/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 916 / 1718\n",
      "29312/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 917 / 1718\n",
      "29344/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9925Step 918 / 1718\n",
      "29376/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9925Step 919 / 1718\n",
      "29408/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9926Step 920 / 1718\n",
      "29440/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9926Step 921 / 1718\n",
      "29472/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9926Step 922 / 1718\n",
      "29504/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9926Step 923 / 1718\n",
      "29536/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9926Step 924 / 1718\n",
      "29568/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9926Step 925 / 1718\n",
      "29600/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9926Step 926 / 1718\n",
      "29632/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9926Step 927 / 1718\n",
      "29664/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9926Step 928 / 1718\n",
      "29696/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9926Step 929 / 1718\n",
      "29728/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9926Step 930 / 1718\n",
      "29760/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9926Step 931 / 1718\n",
      "29792/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9926Step 932 / 1718\n",
      "29824/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9925Step 933 / 1718\n",
      "29856/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9925Step 934 / 1718\n",
      "29888/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9925Step 935 / 1718\n",
      "29920/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9925Step 936 / 1718\n",
      "29952/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9925Step 937 / 1718\n",
      "29984/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9925Step 938 / 1718\n",
      "30016/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9925Step 939 / 1718\n",
      "30048/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9925Step 940 / 1718\n",
      "30080/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9925Step 941 / 1718\n",
      "30112/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9925Step 942 / 1718\n",
      "30144/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9925Step 943 / 1718\n",
      "30176/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9925Step 944 / 1718\n",
      "30208/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9926Step 945 / 1718\n",
      "30240/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9925Step 946 / 1718\n",
      "30272/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9925Step 947 / 1718\n",
      "30304/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9925Step 948 / 1718\n",
      "30336/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9926Step 949 / 1718\n",
      "30368/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9926Step 950 / 1718\n",
      "30400/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9926Step 951 / 1718\n",
      "30432/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9925Step 952 / 1718\n",
      "30464/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9925Step 953 / 1718\n",
      "30496/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9926Step 954 / 1718\n",
      "30528/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9926Step 955 / 1718\n",
      "30560/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9926Step 956 / 1718\n",
      "30592/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9925Step 957 / 1718\n",
      "30624/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9926Step 958 / 1718\n",
      "30656/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9925Step 959 / 1718\n",
      "30688/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9925Step 960 / 1718\n",
      "30720/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9925Step 961 / 1718\n",
      "30752/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9925Step 962 / 1718\n",
      "30784/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9925Step 963 / 1718\n",
      "30816/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9925Step 964 / 1718\n",
      "30848/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9925Step 965 / 1718\n",
      "30880/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9926Step 966 / 1718\n",
      "30912/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9926Step 967 / 1718\n",
      "30944/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9926Step 968 / 1718\n",
      "30976/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9926Step 969 / 1718\n",
      "31008/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9926Step 970 / 1718\n",
      "31040/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9926Step 971 / 1718\n",
      "31072/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9926Step 972 / 1718\n",
      "31104/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9926Step 973 / 1718\n",
      "31136/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9925Step 974 / 1718\n",
      "31168/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9926Step 975 / 1718\n",
      "31200/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9925Step 976 / 1718\n",
      "31232/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9925Step 977 / 1718\n",
      "31264/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9925Step 978 / 1718\n",
      "31296/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9925Step 979 / 1718\n",
      "31328/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9925Step 980 / 1718\n",
      "31360/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9925Step 981 / 1718\n",
      "31392/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9925Step 982 / 1718\n",
      "31424/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9925Step 983 / 1718\n",
      "31456/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9925Step 984 / 1718\n",
      "31488/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9925Step 985 / 1718\n",
      "31520/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9925Step 986 / 1718\n",
      "31552/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9926Step 987 / 1718\n",
      "31584/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 988 / 1718\n",
      "31616/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9925Step 989 / 1718\n",
      "31648/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 990 / 1718\n",
      "31680/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 991 / 1718\n",
      "31712/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 992 / 1718\n",
      "31744/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 993 / 1718\n",
      "31776/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 994 / 1718\n",
      "31808/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 995 / 1718\n",
      "31840/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 996 / 1718\n",
      "31872/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 997 / 1718\n",
      "31904/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9925Step 998 / 1718\n",
      "31936/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9925Step 999 / 1718\n",
      "31968/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1000 / 1718\n",
      "32000/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1001 / 1718\n",
      "32032/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1002 / 1718\n",
      "32064/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9924Step 1003 / 1718\n",
      "32096/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9924Step 1004 / 1718\n",
      "32128/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9924Step 1005 / 1718\n",
      "32160/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9924Step 1006 / 1718\n",
      "32192/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9924Step 1007 / 1718\n",
      "32224/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1008 / 1718\n",
      "32256/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1009 / 1718\n",
      "32288/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1010 / 1718\n",
      "32320/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1011 / 1718\n",
      "32352/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1012 / 1718\n",
      "32384/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9924Step 1013 / 1718\n",
      "32416/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9924Step 1014 / 1718\n",
      "32448/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1015 / 1718\n",
      "32480/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9924Step 1016 / 1718\n",
      "32512/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9924Step 1017 / 1718\n",
      "32544/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1018 / 1718\n",
      "32576/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1019 / 1718\n",
      "32608/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1020 / 1718\n",
      "32640/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1021 / 1718\n",
      "32672/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9924Step 1022 / 1718\n",
      "32704/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9924Step 1023 / 1718\n",
      "32736/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9924Step 1024 / 1718\n",
      "32768/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9924Step 1025 / 1718\n",
      "32800/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9924Step 1026 / 1718\n",
      "32832/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1027 / 1718\n",
      "32864/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9924Step 1028 / 1718\n",
      "32896/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9924Step 1029 / 1718\n",
      "32928/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9924Step 1030 / 1718\n",
      "32960/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9924Step 1031 / 1718\n",
      "32992/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9924Step 1032 / 1718\n",
      "33024/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1033 / 1718\n",
      "33056/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1034 / 1718\n",
      "33088/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9924Step 1035 / 1718\n",
      "33120/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1036 / 1718\n",
      "33152/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1037 / 1718\n",
      "33184/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1038 / 1718\n",
      "33216/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1039 / 1718\n",
      "33248/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1040 / 1718\n",
      "33280/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1041 / 1718\n",
      "33312/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1042 / 1718\n",
      "33344/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1043 / 1718\n",
      "33376/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9924Step 1044 / 1718\n",
      "33408/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9923Step 1045 / 1718\n",
      "33440/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9923Step 1046 / 1718\n",
      "33472/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1047 / 1718\n",
      "33504/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1048 / 1718\n",
      "33536/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9923Step 1049 / 1718\n",
      "33568/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9923Step 1050 / 1718\n",
      "33600/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1051 / 1718\n",
      "33632/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1052 / 1718\n",
      "33664/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9924Step 1053 / 1718\n",
      "33696/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9923Step 1054 / 1718\n",
      "33728/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9923Step 1055 / 1718\n",
      "33760/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9923Step 1056 / 1718\n",
      "33792/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9923Step 1057 / 1718\n",
      "33824/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9923Step 1058 / 1718\n",
      "33856/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9923Step 1059 / 1718\n",
      "33888/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9924Step 1060 / 1718\n",
      "33920/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9924Step 1061 / 1718\n",
      "33952/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9924Step 1062 / 1718\n",
      "33984/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9924Step 1063 / 1718\n",
      "34016/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9924Step 1064 / 1718\n",
      "34048/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9924Step 1065 / 1718\n",
      "34080/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9924Step 1066 / 1718\n",
      "34112/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9924Step 1067 / 1718\n",
      "34144/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9924Step 1068 / 1718\n",
      "34176/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9924Step 1069 / 1718\n",
      "34208/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9924Step 1070 / 1718\n",
      "34240/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9924Step 1071 / 1718\n",
      "34272/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9924Step 1072 / 1718\n",
      "34304/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9924Step 1073 / 1718\n",
      "34336/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9924Step 1074 / 1718\n",
      "34368/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9924Step 1075 / 1718\n",
      "34400/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9924Step 1076 / 1718\n",
      "34432/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9923Step 1077 / 1718\n",
      "34464/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9923Step 1078 / 1718\n",
      "34496/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9923Step 1079 / 1718\n",
      "34528/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9923Step 1080 / 1718\n",
      "34560/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9923Step 1081 / 1718\n",
      "34592/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9923Step 1082 / 1718\n",
      "34624/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9923Step 1083 / 1718\n",
      "34656/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9923Step 1084 / 1718\n",
      "34688/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9923Step 1085 / 1718\n",
      "34720/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9923Step 1086 / 1718\n",
      "34752/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9923Step 1087 / 1718\n",
      "34784/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9923Step 1088 / 1718\n",
      "34816/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9923Step 1089 / 1718\n",
      "34848/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9923Step 1090 / 1718\n",
      "34880/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9923Step 1091 / 1718\n",
      "34912/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9923Step 1092 / 1718\n",
      "34944/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9923Step 1093 / 1718\n",
      "34976/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9923Step 1094 / 1718\n",
      "35008/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9923Step 1095 / 1718\n",
      "35040/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9923Step 1096 / 1718\n",
      "35072/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9923Step 1097 / 1718\n",
      "35104/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9923Step 1098 / 1718\n",
      "35136/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9923Step 1099 / 1718\n",
      "35168/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9923Step 1100 / 1718\n",
      "35200/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9923Step 1101 / 1718\n",
      "35232/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9923Step 1102 / 1718\n",
      "35264/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9923Step 1103 / 1718\n",
      "35296/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9923Step 1104 / 1718\n",
      "35328/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9923Step 1105 / 1718\n",
      "35360/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9923Step 1106 / 1718\n",
      "35392/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9923Step 1107 / 1718\n",
      "35424/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9923Step 1108 / 1718\n",
      "35456/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9924Step 1109 / 1718\n",
      "35488/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9923Step 1110 / 1718\n",
      "35520/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9923Step 1111 / 1718\n",
      "35552/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9923Step 1112 / 1718\n",
      "35584/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9923Step 1113 / 1718\n",
      "35616/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9923Step 1114 / 1718\n",
      "35648/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9923Step 1115 / 1718\n",
      "35680/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9923Step 1116 / 1718\n",
      "35712/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9923Step 1117 / 1718\n",
      "35744/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9923Step 1118 / 1718\n",
      "35776/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9923Step 1119 / 1718\n",
      "35808/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9923Step 1120 / 1718\n",
      "35840/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9924Step 1121 / 1718\n",
      "35872/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1122 / 1718\n",
      "35904/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1123 / 1718\n",
      "35936/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1124 / 1718\n",
      "35968/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1125 / 1718\n",
      "36000/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1126 / 1718\n",
      "36032/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9924Step 1127 / 1718\n",
      "36064/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9924Step 1128 / 1718\n",
      "36096/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1129 / 1718\n",
      "36128/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9924Step 1130 / 1718\n",
      "36160/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9924Step 1131 / 1718\n",
      "36192/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9924Step 1132 / 1718\n",
      "36224/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9924Step 1133 / 1718\n",
      "36256/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9924Step 1134 / 1718\n",
      "36288/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9924Step 1135 / 1718\n",
      "36320/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9923Step 1136 / 1718\n",
      "36352/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9923Step 1137 / 1718\n",
      "36384/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9923Step 1138 / 1718\n",
      "36416/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9923Step 1139 / 1718\n",
      "36448/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9923Step 1140 / 1718\n",
      "36480/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9923Step 1141 / 1718\n",
      "36512/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9923Step 1142 / 1718\n",
      "36544/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9923Step 1143 / 1718\n",
      "36576/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9923Step 1144 / 1718\n",
      "36608/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9923Step 1145 / 1718\n",
      "36640/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9923Step 1146 / 1718\n",
      "36672/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9923Step 1147 / 1718\n",
      "36704/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9923Step 1148 / 1718\n",
      "36736/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9923Step 1149 / 1718\n",
      "36768/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9923Step 1150 / 1718\n",
      "36800/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9923Step 1151 / 1718\n",
      "36832/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9923Step 1152 / 1718\n",
      "36864/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9923Step 1153 / 1718\n",
      "36896/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9923Step 1154 / 1718\n",
      "36928/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9923Step 1155 / 1718\n",
      "36960/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9923Step 1156 / 1718\n",
      "36992/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9923Step 1157 / 1718\n",
      "37024/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9924Step 1158 / 1718\n",
      "37056/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1159 / 1718\n",
      "37088/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1160 / 1718\n",
      "37120/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9923Step 1161 / 1718\n",
      "37152/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9923Step 1162 / 1718\n",
      "37184/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9923Step 1163 / 1718\n",
      "37216/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9923Step 1164 / 1718\n",
      "37248/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9923Step 1165 / 1718\n",
      "37280/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1166 / 1718\n",
      "37312/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1167 / 1718\n",
      "37344/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9924Step 1168 / 1718\n",
      "37376/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9924Step 1169 / 1718\n",
      "37408/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9924Step 1170 / 1718\n",
      "37440/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1171 / 1718\n",
      "37472/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1172 / 1718\n",
      "37504/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1173 / 1718\n",
      "37536/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1174 / 1718\n",
      "37568/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9924Step 1175 / 1718\n",
      "37600/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9924Step 1176 / 1718\n",
      "37632/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1177 / 1718\n",
      "37664/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1178 / 1718\n",
      "37696/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1179 / 1718\n",
      "37728/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9924Step 1180 / 1718\n",
      "37760/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9923Step 1181 / 1718\n",
      "37792/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9924Step 1182 / 1718\n",
      "37824/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1183 / 1718\n",
      "37856/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1184 / 1718\n",
      "37888/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1185 / 1718\n",
      "37920/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1186 / 1718\n",
      "37952/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1187 / 1718\n",
      "37984/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1188 / 1718\n",
      "38016/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1189 / 1718\n",
      "38048/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9924Step 1190 / 1718\n",
      "38080/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9924Step 1191 / 1718\n",
      "38112/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9924Step 1192 / 1718\n",
      "38144/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1193 / 1718\n",
      "38176/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1194 / 1718\n",
      "38208/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1195 / 1718\n",
      "38240/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1196 / 1718\n",
      "38272/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9924Step 1197 / 1718\n",
      "38304/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9924Step 1198 / 1718\n",
      "38336/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9924Step 1199 / 1718\n",
      "38368/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9924Step 1200 / 1718\n",
      "38400/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9924Step 1201 / 1718\n",
      "38432/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9925Step 1202 / 1718\n",
      "38464/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9925Step 1203 / 1718\n",
      "38496/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9925Step 1204 / 1718\n",
      "38528/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9925Step 1205 / 1718\n",
      "38560/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9925Step 1206 / 1718\n",
      "38592/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9925Step 1207 / 1718\n",
      "38624/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9925Step 1208 / 1718\n",
      "38656/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9925Step 1209 / 1718\n",
      "38688/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9925Step 1210 / 1718\n",
      "38720/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9924Step 1211 / 1718\n",
      "38752/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9924Step 1212 / 1718\n",
      "38784/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9924Step 1213 / 1718\n",
      "38816/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9925Step 1214 / 1718\n",
      "38848/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9925Step 1215 / 1718\n",
      "38880/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9925Step 1216 / 1718\n",
      "38912/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9925Step 1217 / 1718\n",
      "38944/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9925Step 1218 / 1718\n",
      "38976/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9925Step 1219 / 1718\n",
      "39008/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9925Step 1220 / 1718\n",
      "39040/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9925Step 1221 / 1718\n",
      "39072/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9925Step 1222 / 1718\n",
      "39104/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9925Step 1223 / 1718\n",
      "39136/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9925Step 1224 / 1718\n",
      "39168/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1225 / 1718\n",
      "39200/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1226 / 1718\n",
      "39232/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1227 / 1718\n",
      "39264/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9925Step 1228 / 1718\n",
      "39296/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1229 / 1718\n",
      "39328/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1230 / 1718\n",
      "39360/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1231 / 1718\n",
      "39392/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1232 / 1718\n",
      "39424/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1233 / 1718\n",
      "39456/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1234 / 1718\n",
      "39488/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9925Step 1235 / 1718\n",
      "39520/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9925Step 1236 / 1718\n",
      "39552/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9925Step 1237 / 1718\n",
      "39584/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1238 / 1718\n",
      "39616/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1239 / 1718\n",
      "39648/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1240 / 1718\n",
      "39680/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1241 / 1718\n",
      "39712/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9925Step 1242 / 1718\n",
      "39744/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9925Step 1243 / 1718\n",
      "39776/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9925Step 1244 / 1718\n",
      "39808/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9925Step 1245 / 1718\n",
      "39840/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9925Step 1246 / 1718\n",
      "39872/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9925Step 1247 / 1718\n",
      "39904/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1248 / 1718\n",
      "39936/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1249 / 1718\n",
      "39968/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1250 / 1718\n",
      "40000/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1251 / 1718\n",
      "40032/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9925Step 1252 / 1718\n",
      "40064/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1253 / 1718\n",
      "40096/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1254 / 1718\n",
      "40128/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1255 / 1718\n",
      "40160/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1256 / 1718\n",
      "40192/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1257 / 1718\n",
      "40224/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1258 / 1718\n",
      "40256/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9925Step 1259 / 1718\n",
      "40288/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9924Step 1260 / 1718\n",
      "40320/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9924Step 1261 / 1718\n",
      "40352/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1262 / 1718\n",
      "40384/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1263 / 1718\n",
      "40416/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1264 / 1718\n",
      "40448/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1265 / 1718\n",
      "40480/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1266 / 1718\n",
      "40512/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9924Step 1267 / 1718\n",
      "40544/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9924Step 1268 / 1718\n",
      "40576/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9924Step 1269 / 1718\n",
      "40608/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9924Step 1270 / 1718\n",
      "40640/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9924Step 1271 / 1718\n",
      "40672/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1272 / 1718\n",
      "40704/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1273 / 1718\n",
      "40736/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1274 / 1718\n",
      "40768/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1275 / 1718\n",
      "40800/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9924Step 1276 / 1718\n",
      "40832/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9924Step 1277 / 1718\n",
      "40864/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1278 / 1718\n",
      "40896/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9924Step 1279 / 1718\n",
      "40928/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9924Step 1280 / 1718\n",
      "40960/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9924Step 1281 / 1718\n",
      "40992/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9923Step 1282 / 1718\n",
      "41024/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9923Step 1283 / 1718\n",
      "41056/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9924Step 1284 / 1718\n",
      "41088/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9923Step 1285 / 1718\n",
      "41120/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1286 / 1718\n",
      "41152/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9923Step 1287 / 1718\n",
      "41184/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9923Step 1288 / 1718\n",
      "41216/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9923Step 1289 / 1718\n",
      "41248/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1290 / 1718\n",
      "41280/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9923Step 1291 / 1718\n",
      "41312/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9923Step 1292 / 1718\n",
      "41344/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9923Step 1293 / 1718\n",
      "41376/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1294 / 1718\n",
      "41408/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1295 / 1718\n",
      "41440/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1296 / 1718\n",
      "41472/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1297 / 1718\n",
      "41504/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1298 / 1718\n",
      "41536/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1299 / 1718\n",
      "41568/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1300 / 1718\n",
      "41600/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1301 / 1718\n",
      "41632/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1302 / 1718\n",
      "41664/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1303 / 1718\n",
      "41696/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1304 / 1718\n",
      "41728/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1305 / 1718\n",
      "41760/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1306 / 1718\n",
      "41792/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1307 / 1718\n",
      "41824/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1308 / 1718\n",
      "41856/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1309 / 1718\n",
      "41888/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1310 / 1718\n",
      "41920/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1311 / 1718\n",
      "41952/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1312 / 1718\n",
      "41984/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1313 / 1718\n",
      "42016/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1314 / 1718\n",
      "42048/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1315 / 1718\n",
      "42080/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1316 / 1718\n",
      "42112/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1317 / 1718\n",
      "42144/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1318 / 1718\n",
      "42176/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1319 / 1718\n",
      "42208/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1320 / 1718\n",
      "42240/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1321 / 1718\n",
      "42272/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1322 / 1718\n",
      "42304/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1323 / 1718\n",
      "42336/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1324 / 1718\n",
      "42368/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1325 / 1718\n",
      "42400/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1326 / 1718\n",
      "42432/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1327 / 1718\n",
      "42464/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9923Step 1328 / 1718\n",
      "42496/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1329 / 1718\n",
      "42528/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1330 / 1718\n",
      "42560/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1331 / 1718\n",
      "42592/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1332 / 1718\n",
      "42624/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1333 / 1718\n",
      "42656/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1334 / 1718\n",
      "42688/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1335 / 1718\n",
      "42720/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9923Step 1336 / 1718\n",
      "42752/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9923Step 1337 / 1718\n",
      "42784/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1338 / 1718\n",
      "42816/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1339 / 1718\n",
      "42848/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1340 / 1718\n",
      "42880/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1341 / 1718\n",
      "42912/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1342 / 1718\n",
      "42944/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1343 / 1718\n",
      "42976/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1344 / 1718\n",
      "43008/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1345 / 1718\n",
      "43040/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1346 / 1718\n",
      "43072/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1347 / 1718\n",
      "43104/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1348 / 1718\n",
      "43136/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1349 / 1718\n",
      "43168/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1350 / 1718\n",
      "43200/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1351 / 1718\n",
      "43232/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1352 / 1718\n",
      "43264/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9922Step 1353 / 1718\n",
      "43296/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1354 / 1718\n",
      "43328/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1355 / 1718\n",
      "43360/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1356 / 1718\n",
      "43392/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1357 / 1718\n",
      "43424/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1358 / 1718\n",
      "43456/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1359 / 1718\n",
      "43488/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9922Step 1360 / 1718\n",
      "43520/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9922Step 1361 / 1718\n",
      "43552/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9922Step 1362 / 1718\n",
      "43584/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1363 / 1718\n",
      "43616/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1364 / 1718\n",
      "43648/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1365 / 1718\n",
      "43680/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1366 / 1718\n",
      "43712/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1367 / 1718\n",
      "43744/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1368 / 1718\n",
      "43776/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1369 / 1718\n",
      "43808/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1370 / 1718\n",
      "43840/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1371 / 1718\n",
      "43872/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1372 / 1718\n",
      "43904/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1373 / 1718\n",
      "43936/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1374 / 1718\n",
      "43968/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1375 / 1718\n",
      "44000/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1376 / 1718\n",
      "44032/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1377 / 1718\n",
      "44064/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1378 / 1718\n",
      "44096/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1379 / 1718\n",
      "44128/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1380 / 1718\n",
      "44160/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1381 / 1718\n",
      "44192/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1382 / 1718\n",
      "44224/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1383 / 1718\n",
      "44256/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1384 / 1718\n",
      "44288/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1385 / 1718\n",
      "44320/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1386 / 1718\n",
      "44352/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9922Step 1387 / 1718\n",
      "44384/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1388 / 1718\n",
      "44416/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1389 / 1718\n",
      "44448/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1390 / 1718\n",
      "44480/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1391 / 1718\n",
      "44512/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1392 / 1718\n",
      "44544/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9922Step 1393 / 1718\n",
      "44576/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9922Step 1394 / 1718\n",
      "44608/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1395 / 1718\n",
      "44640/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1396 / 1718\n",
      "44672/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1397 / 1718\n",
      "44704/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1398 / 1718\n",
      "44736/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1399 / 1718\n",
      "44768/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1400 / 1718\n",
      "44800/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1401 / 1718\n",
      "44832/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1402 / 1718\n",
      "44864/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1403 / 1718\n",
      "44896/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1404 / 1718\n",
      "44928/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1405 / 1718\n",
      "44960/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1406 / 1718\n",
      "44992/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1407 / 1718\n",
      "45024/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1408 / 1718\n",
      "45056/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1409 / 1718\n",
      "45088/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1410 / 1718\n",
      "45120/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1411 / 1718\n",
      "45152/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1412 / 1718\n",
      "45184/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1413 / 1718\n",
      "45216/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1414 / 1718\n",
      "45248/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1415 / 1718\n",
      "45280/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1416 / 1718\n",
      "45312/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1417 / 1718\n",
      "45344/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1418 / 1718\n",
      "45376/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1419 / 1718\n",
      "45408/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1420 / 1718\n",
      "45440/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1421 / 1718\n",
      "45472/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1422 / 1718\n",
      "45504/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1423 / 1718\n",
      "45536/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1424 / 1718\n",
      "45568/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1425 / 1718\n",
      "45600/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1426 / 1718\n",
      "45632/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1427 / 1718\n",
      "45664/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1428 / 1718\n",
      "45696/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9922Step 1429 / 1718\n",
      "45728/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9922Step 1430 / 1718\n",
      "45760/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9922Step 1431 / 1718\n",
      "45792/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9922Step 1432 / 1718\n",
      "45824/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1433 / 1718\n",
      "45856/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9922Step 1434 / 1718\n",
      "45888/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9922Step 1435 / 1718\n",
      "45920/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9922Step 1436 / 1718\n",
      "45952/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1437 / 1718\n",
      "45984/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1438 / 1718\n",
      "46016/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1439 / 1718\n",
      "46048/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1440 / 1718\n",
      "46080/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9923Step 1441 / 1718\n",
      "46112/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1442 / 1718\n",
      "46144/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1443 / 1718\n",
      "46176/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1444 / 1718\n",
      "46208/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1445 / 1718\n",
      "46240/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1446 / 1718\n",
      "46272/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9922Step 1447 / 1718\n",
      "46304/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9922Step 1448 / 1718\n",
      "46336/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9923Step 1449 / 1718\n",
      "46368/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1450 / 1718\n",
      "46400/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1451 / 1718\n",
      "46432/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1452 / 1718\n",
      "46464/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1453 / 1718\n",
      "46496/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1454 / 1718\n",
      "46528/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1455 / 1718\n",
      "46560/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1456 / 1718\n",
      "46592/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1457 / 1718\n",
      "46624/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1458 / 1718\n",
      "46656/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1459 / 1718\n",
      "46688/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1460 / 1718\n",
      "46720/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1461 / 1718\n",
      "46752/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1462 / 1718\n",
      "46784/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9923Step 1463 / 1718\n",
      "46816/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9923Step 1464 / 1718\n",
      "46848/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9923Step 1465 / 1718\n",
      "46880/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9923Step 1466 / 1718\n",
      "46912/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9923Step 1467 / 1718\n",
      "46944/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9923Step 1468 / 1718\n",
      "46976/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9923Step 1469 / 1718\n",
      "47008/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9923Step 1470 / 1718\n",
      "47040/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9923Step 1471 / 1718\n",
      "47072/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9923Step 1472 / 1718\n",
      "47104/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9923Step 1473 / 1718\n",
      "47136/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9923Step 1474 / 1718\n",
      "47168/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9923Step 1475 / 1718\n",
      "47200/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9923Step 1476 / 1718\n",
      "47232/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9923Step 1477 / 1718\n",
      "47264/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9923Step 1478 / 1718\n",
      "47296/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9922Step 1479 / 1718\n",
      "47328/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9922Step 1480 / 1718\n",
      "47360/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1481 / 1718\n",
      "47392/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1482 / 1718\n",
      "47424/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9923Step 1483 / 1718\n",
      "47456/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9922Step 1484 / 1718\n",
      "47488/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9922Step 1485 / 1718\n",
      "47520/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9922Step 1486 / 1718\n",
      "47552/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1487 / 1718\n",
      "47584/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9922Step 1488 / 1718\n",
      "47616/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9921Step 1489 / 1718\n",
      "47648/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1490 / 1718\n",
      "47680/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1491 / 1718\n",
      "47712/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1492 / 1718\n",
      "47744/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9921Step 1493 / 1718\n",
      "47776/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1494 / 1718\n",
      "47808/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9922Step 1495 / 1718\n",
      "47840/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9921Step 1496 / 1718\n",
      "47872/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1497 / 1718\n",
      "47904/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9921Step 1498 / 1718\n",
      "47936/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9921Step 1499 / 1718\n",
      "47968/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9921Step 1500 / 1718\n",
      "48000/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9921Step 1501 / 1718\n",
      "48032/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9921Step 1502 / 1718\n",
      "48064/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9921Step 1503 / 1718\n",
      "48096/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9921Step 1504 / 1718\n",
      "48128/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9921Step 1505 / 1718\n",
      "48160/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1506 / 1718\n",
      "48192/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1507 / 1718\n",
      "48224/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1508 / 1718\n",
      "48256/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1509 / 1718\n",
      "48288/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1510 / 1718\n",
      "48320/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1511 / 1718\n",
      "48352/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1512 / 1718\n",
      "48384/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1513 / 1718\n",
      "48416/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9921Step 1514 / 1718\n",
      "48448/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1515 / 1718\n",
      "48480/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1516 / 1718\n",
      "48512/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1517 / 1718\n",
      "48544/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1518 / 1718\n",
      "48576/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1519 / 1718\n",
      "48608/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1520 / 1718\n",
      "48640/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9921Step 1521 / 1718\n",
      "48672/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9921Step 1522 / 1718\n",
      "48704/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9921Step 1523 / 1718\n",
      "48736/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9921Step 1524 / 1718\n",
      "48768/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9921Step 1525 / 1718\n",
      "48800/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9921Step 1526 / 1718\n",
      "48832/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9921Step 1527 / 1718\n",
      "48864/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1528 / 1718\n",
      "48896/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9921Step 1529 / 1718\n",
      "48928/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9920Step 1530 / 1718\n",
      "48960/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9920Step 1531 / 1718\n",
      "48992/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9920Step 1532 / 1718\n",
      "49024/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9920Step 1533 / 1718\n",
      "49056/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9920Step 1534 / 1718\n",
      "49088/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9921Step 1535 / 1718\n",
      "49120/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9921Step 1536 / 1718\n",
      "49152/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9921Step 1537 / 1718\n",
      "49184/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9921Step 1538 / 1718\n",
      "49216/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9921Step 1539 / 1718\n",
      "49248/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9921Step 1540 / 1718\n",
      "49280/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9920Step 1541 / 1718\n",
      "49312/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9921Step 1542 / 1718\n",
      "49344/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9921Step 1543 / 1718\n",
      "49376/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1544 / 1718\n",
      "49408/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1545 / 1718\n",
      "49440/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1546 / 1718\n",
      "49472/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1547 / 1718\n",
      "49504/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9921Step 1548 / 1718\n",
      "49536/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9921Step 1549 / 1718\n",
      "49568/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9921Step 1550 / 1718\n",
      "49600/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9921Step 1551 / 1718\n",
      "49632/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9921Step 1552 / 1718\n",
      "49664/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9921Step 1553 / 1718\n",
      "49696/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9921Step 1554 / 1718\n",
      "49728/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9921Step 1555 / 1718\n",
      "49760/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1556 / 1718\n",
      "49792/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1557 / 1718\n",
      "49824/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1558 / 1718\n",
      "49856/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1559 / 1718\n",
      "49888/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1560 / 1718\n",
      "49920/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1561 / 1718\n",
      "49952/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1562 / 1718\n",
      "49984/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1563 / 1718\n",
      "50016/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9921Step 1564 / 1718\n",
      "50048/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9921Step 1565 / 1718\n",
      "50080/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1566 / 1718\n",
      "50112/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1567 / 1718\n",
      "50144/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1568 / 1718\n",
      "50176/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1569 / 1718\n",
      "50208/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1570 / 1718\n",
      "50240/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1571 / 1718\n",
      "50272/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1572 / 1718\n",
      "50304/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9921Step 1573 / 1718\n",
      "50336/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1574 / 1718\n",
      "50368/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1575 / 1718\n",
      "50400/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1576 / 1718\n",
      "50432/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9921Step 1577 / 1718\n",
      "50464/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1578 / 1718\n",
      "50496/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9921Step 1579 / 1718\n",
      "50528/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9920Step 1580 / 1718\n",
      "50560/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9920Step 1581 / 1718\n",
      "50592/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9921Step 1582 / 1718\n",
      "50624/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9920Step 1583 / 1718\n",
      "50656/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9920Step 1584 / 1718\n",
      "50688/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1585 / 1718\n",
      "50720/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9921Step 1586 / 1718\n",
      "50752/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9920Step 1587 / 1718\n",
      "50784/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9920Step 1588 / 1718\n",
      "50816/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9920Step 1589 / 1718\n",
      "50848/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9920Step 1590 / 1718\n",
      "50880/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1591 / 1718\n",
      "50912/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1592 / 1718\n",
      "50944/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1593 / 1718\n",
      "50976/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1594 / 1718\n",
      "51008/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1595 / 1718\n",
      "51040/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1596 / 1718\n",
      "51072/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1597 / 1718\n",
      "51104/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1598 / 1718\n",
      "51136/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1599 / 1718\n",
      "51168/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1600 / 1718\n",
      "51200/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9921Step 1601 / 1718\n",
      "51232/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1602 / 1718\n",
      "51264/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1603 / 1718\n",
      "51296/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1604 / 1718\n",
      "51328/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1605 / 1718\n",
      "51360/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1606 / 1718\n",
      "51392/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1607 / 1718\n",
      "51424/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1608 / 1718\n",
      "51456/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1609 / 1718\n",
      "51488/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1610 / 1718\n",
      "51520/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1611 / 1718\n",
      "51552/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1612 / 1718\n",
      "51584/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1613 / 1718\n",
      "51616/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1614 / 1718\n",
      "51648/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1615 / 1718\n",
      "51680/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1616 / 1718\n",
      "51712/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1617 / 1718\n",
      "51744/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1618 / 1718\n",
      "51776/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1619 / 1718\n",
      "51808/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1620 / 1718\n",
      "51840/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1621 / 1718\n",
      "51872/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1622 / 1718\n",
      "51904/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1623 / 1718\n",
      "51936/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1624 / 1718\n",
      "51968/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1625 / 1718\n",
      "52000/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1626 / 1718\n",
      "52032/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1627 / 1718\n",
      "52064/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1628 / 1718\n",
      "52096/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1629 / 1718\n",
      "52128/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1630 / 1718\n",
      "52160/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1631 / 1718\n",
      "52192/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9920Step 1632 / 1718\n",
      "52224/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9920Step 1633 / 1718\n",
      "52256/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9920Step 1634 / 1718\n",
      "52288/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9920Step 1635 / 1718\n",
      "52320/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1636 / 1718\n",
      "52352/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1637 / 1718\n",
      "52384/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1638 / 1718\n",
      "52416/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1639 / 1718\n",
      "52448/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1640 / 1718\n",
      "52480/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1641 / 1718\n",
      "52512/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1642 / 1718\n",
      "52544/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1643 / 1718\n",
      "52576/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1644 / 1718\n",
      "52608/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1645 / 1718\n",
      "52640/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1646 / 1718\n",
      "52672/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1647 / 1718\n",
      "52704/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1648 / 1718\n",
      "52736/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9918Step 1649 / 1718\n",
      "52768/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9919Step 1650 / 1718\n",
      "52800/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1651 / 1718\n",
      "52832/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1652 / 1718\n",
      "52864/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1653 / 1718\n",
      "52896/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1654 / 1718\n",
      "52928/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9918Step 1655 / 1718\n",
      "52960/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9918Step 1656 / 1718\n",
      "52992/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9918Step 1657 / 1718\n",
      "53024/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9918Step 1658 / 1718\n",
      "53056/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9918Step 1659 / 1718\n",
      "53088/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9918Step 1660 / 1718\n",
      "53120/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9918Step 1661 / 1718\n",
      "53152/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9918Step 1662 / 1718\n",
      "53184/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9918Step 1663 / 1718\n",
      "53216/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9918Step 1664 / 1718\n",
      "53248/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9918Step 1665 / 1718\n",
      "53280/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9918Step 1666 / 1718\n",
      "53312/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9918Step 1667 / 1718\n",
      "53344/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9918Step 1668 / 1718\n",
      "53376/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9919Step 1669 / 1718\n",
      "53408/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9919Step 1670 / 1718\n",
      "53440/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9919Step 1671 / 1718\n",
      "53472/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9919Step 1672 / 1718\n",
      "53504/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1673 / 1718\n",
      "53536/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1674 / 1718\n",
      "53568/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1675 / 1718\n",
      "53600/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1676 / 1718\n",
      "53632/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1677 / 1718\n",
      "53664/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1678 / 1718\n",
      "53696/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1679 / 1718\n",
      "53728/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1680 / 1718\n",
      "53760/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1681 / 1718\n",
      "53792/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1682 / 1718\n",
      "53824/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1683 / 1718\n",
      "53856/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1684 / 1718\n",
      "53888/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1685 / 1718\n",
      "53920/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1686 / 1718\n",
      "53952/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1687 / 1718\n",
      "53984/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1688 / 1718\n",
      "54016/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1689 / 1718\n",
      "54048/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1690 / 1718\n",
      "54080/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1691 / 1718\n",
      "54112/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1692 / 1718\n",
      "54144/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1693 / 1718\n",
      "54176/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1694 / 1718\n",
      "54208/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1695 / 1718\n",
      "54240/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1696 / 1718\n",
      "54272/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1697 / 1718\n",
      "54304/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1698 / 1718\n",
      "54336/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1699 / 1718\n",
      "54368/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1700 / 1718\n",
      "54400/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1701 / 1718\n",
      "54432/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1702 / 1718\n",
      "54464/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1703 / 1718\n",
      "54496/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1704 / 1718\n",
      "54528/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1705 / 1718\n",
      "54560/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1706 / 1718\n",
      "54592/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1707 / 1718\n",
      "54624/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1708 / 1718\n",
      "54656/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1709 / 1718\n",
      "54688/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1710 / 1718\n",
      "54720/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1711 / 1718\n",
      "54752/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1712 / 1718\n",
      "54784/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1713 / 1718\n",
      "54816/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1714 / 1718\n",
      "54848/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1715 / 1718\n",
      "54880/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9919Step 1716 / 1718\n",
      "54912/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9920Step 1717 / 1718\n",
      "54944/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9919Step 1718 / 1718\n",
      "55000/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9919\n",
      "5000/5000 - mean: 0.1604 - sparse_categorical_accuracy: 0.9736\n",
      "Epoch 4 / 5\n",
      "Step 1 / 1718\n",
      "32/55000 - mean: 0.0395 - sparse_categorical_accuracy: 0.9688Step 2 / 1718\n",
      "64/55000 - mean: 0.1310 - sparse_categorical_accuracy: 0.9688Step 3 / 1718\n",
      "96/55000 - mean: 0.1141 - sparse_categorical_accuracy: 0.9688Step 4 / 1718\n",
      "128/55000 - mean: 0.0904 - sparse_categorical_accuracy: 0.9766Step 5 / 1718\n",
      "160/55000 - mean: 0.0725 - sparse_categorical_accuracy: 0.9812Step 6 / 1718\n",
      "192/55000 - mean: 0.0643 - sparse_categorical_accuracy: 0.9792Step 7 / 1718\n",
      "224/55000 - mean: 0.0735 - sparse_categorical_accuracy: 0.9777Step 8 / 1718\n",
      "256/55000 - mean: 0.0644 - sparse_categorical_accuracy: 0.9805Step 9 / 1718\n",
      "288/55000 - mean: 0.0579 - sparse_categorical_accuracy: 0.9826Step 10 / 1718\n",
      "320/55000 - mean: 0.0522 - sparse_categorical_accuracy: 0.9844Step 11 / 1718\n",
      "352/55000 - mean: 0.0487 - sparse_categorical_accuracy: 0.9858Step 12 / 1718\n",
      "384/55000 - mean: 0.0453 - sparse_categorical_accuracy: 0.9870Step 13 / 1718\n",
      "416/55000 - mean: 0.0502 - sparse_categorical_accuracy: 0.9856Step 14 / 1718\n",
      "448/55000 - mean: 0.0478 - sparse_categorical_accuracy: 0.9866Step 15 / 1718\n",
      "480/55000 - mean: 0.0446 - sparse_categorical_accuracy: 0.9875Step 16 / 1718\n",
      "512/55000 - mean: 0.0418 - sparse_categorical_accuracy: 0.9883Step 17 / 1718\n",
      "544/55000 - mean: 0.0394 - sparse_categorical_accuracy: 0.9890Step 18 / 1718\n",
      "576/55000 - mean: 0.0373 - sparse_categorical_accuracy: 0.9896Step 19 / 1718\n",
      "608/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9901Step 20 / 1718\n",
      "640/55000 - mean: 0.0472 - sparse_categorical_accuracy: 0.9891Step 21 / 1718\n",
      "672/55000 - mean: 0.0526 - sparse_categorical_accuracy: 0.9881Step 22 / 1718\n",
      "704/55000 - mean: 0.0502 - sparse_categorical_accuracy: 0.9886Step 23 / 1718\n",
      "736/55000 - mean: 0.0493 - sparse_categorical_accuracy: 0.9878Step 24 / 1718\n",
      "768/55000 - mean: 0.0597 - sparse_categorical_accuracy: 0.9870Step 25 / 1718\n",
      "800/55000 - mean: 0.0574 - sparse_categorical_accuracy: 0.9875Step 26 / 1718\n",
      "832/55000 - mean: 0.0617 - sparse_categorical_accuracy: 0.9868Step 27 / 1718\n",
      "864/55000 - mean: 0.0595 - sparse_categorical_accuracy: 0.9873Step 28 / 1718\n",
      "896/55000 - mean: 0.0573 - sparse_categorical_accuracy: 0.9877Step 29 / 1718\n",
      "928/55000 - mean: 0.0568 - sparse_categorical_accuracy: 0.9871Step 30 / 1718\n",
      "960/55000 - mean: 0.0549 - sparse_categorical_accuracy: 0.9875Step 31 / 1718\n",
      "992/55000 - mean: 0.0531 - sparse_categorical_accuracy: 0.9879Step 32 / 1718\n",
      "1024/55000 - mean: 0.0515 - sparse_categorical_accuracy: 0.9883Step 33 / 1718\n",
      "1056/55000 - mean: 0.0499 - sparse_categorical_accuracy: 0.9886Step 34 / 1718\n",
      "1088/55000 - mean: 0.0485 - sparse_categorical_accuracy: 0.9890Step 35 / 1718\n",
      "1120/55000 - mean: 0.0537 - sparse_categorical_accuracy: 0.9866Step 36 / 1718\n",
      "1152/55000 - mean: 0.0522 - sparse_categorical_accuracy: 0.9870Step 37 / 1718\n",
      "1184/55000 - mean: 0.0509 - sparse_categorical_accuracy: 0.9873Step 38 / 1718\n",
      "1216/55000 - mean: 0.0500 - sparse_categorical_accuracy: 0.9877Step 39 / 1718\n",
      "1248/55000 - mean: 0.0487 - sparse_categorical_accuracy: 0.9880Step 40 / 1718\n",
      "1280/55000 - mean: 0.0477 - sparse_categorical_accuracy: 0.9883Step 41 / 1718\n",
      "1312/55000 - mean: 0.0466 - sparse_categorical_accuracy: 0.9886Step 42 / 1718\n",
      "1344/55000 - mean: 0.0455 - sparse_categorical_accuracy: 0.9888Step 43 / 1718\n",
      "1376/55000 - mean: 0.0445 - sparse_categorical_accuracy: 0.9891Step 44 / 1718\n",
      "1408/55000 - mean: 0.0438 - sparse_categorical_accuracy: 0.9893Step 45 / 1718\n",
      "1440/55000 - mean: 0.0451 - sparse_categorical_accuracy: 0.9889Step 46 / 1718\n",
      "1472/55000 - mean: 0.0444 - sparse_categorical_accuracy: 0.9891Step 47 / 1718\n",
      "1504/55000 - mean: 0.0436 - sparse_categorical_accuracy: 0.9894Step 48 / 1718\n",
      "1536/55000 - mean: 0.0435 - sparse_categorical_accuracy: 0.9889Step 49 / 1718\n",
      "1568/55000 - mean: 0.0426 - sparse_categorical_accuracy: 0.9892Step 50 / 1718\n",
      "1600/55000 - mean: 0.0430 - sparse_categorical_accuracy: 0.9887Step 51 / 1718\n",
      "1632/55000 - mean: 0.0422 - sparse_categorical_accuracy: 0.9890Step 52 / 1718\n",
      "1664/55000 - mean: 0.0415 - sparse_categorical_accuracy: 0.9892Step 53 / 1718\n",
      "1696/55000 - mean: 0.0407 - sparse_categorical_accuracy: 0.9894Step 54 / 1718\n",
      "1728/55000 - mean: 0.0400 - sparse_categorical_accuracy: 0.9896Step 55 / 1718\n",
      "1760/55000 - mean: 0.0404 - sparse_categorical_accuracy: 0.9892Step 56 / 1718\n",
      "1792/55000 - mean: 0.0397 - sparse_categorical_accuracy: 0.9894Step 57 / 1718\n",
      "1824/55000 - mean: 0.0398 - sparse_categorical_accuracy: 0.9896Step 58 / 1718\n",
      "1856/55000 - mean: 0.0409 - sparse_categorical_accuracy: 0.9892Step 59 / 1718\n",
      "1888/55000 - mean: 0.0422 - sparse_categorical_accuracy: 0.9889Step 60 / 1718\n",
      "1920/55000 - mean: 0.0415 - sparse_categorical_accuracy: 0.9891Step 61 / 1718\n",
      "1952/55000 - mean: 0.0437 - sparse_categorical_accuracy: 0.9887Step 62 / 1718\n",
      "1984/55000 - mean: 0.0435 - sparse_categorical_accuracy: 0.9884Step 63 / 1718\n",
      "2016/55000 - mean: 0.0442 - sparse_categorical_accuracy: 0.9881Step 64 / 1718\n",
      "2048/55000 - mean: 0.0438 - sparse_categorical_accuracy: 0.9883Step 65 / 1718\n",
      "2080/55000 - mean: 0.0432 - sparse_categorical_accuracy: 0.9885Step 66 / 1718\n",
      "2112/55000 - mean: 0.0458 - sparse_categorical_accuracy: 0.9882Step 67 / 1718\n",
      "2144/55000 - mean: 0.0453 - sparse_categorical_accuracy: 0.9883Step 68 / 1718\n",
      "2176/55000 - mean: 0.0447 - sparse_categorical_accuracy: 0.9885Step 69 / 1718\n",
      "2208/55000 - mean: 0.0444 - sparse_categorical_accuracy: 0.9882Step 70 / 1718\n",
      "2240/55000 - mean: 0.0438 - sparse_categorical_accuracy: 0.9884Step 71 / 1718\n",
      "2272/55000 - mean: 0.0448 - sparse_categorical_accuracy: 0.9881Step 72 / 1718\n",
      "2304/55000 - mean: 0.0443 - sparse_categorical_accuracy: 0.9883Step 73 / 1718\n",
      "2336/55000 - mean: 0.0456 - sparse_categorical_accuracy: 0.9876Step 74 / 1718\n",
      "2368/55000 - mean: 0.0472 - sparse_categorical_accuracy: 0.9873Step 75 / 1718\n",
      "2400/55000 - mean: 0.0466 - sparse_categorical_accuracy: 0.9875Step 76 / 1718\n",
      "2432/55000 - mean: 0.0460 - sparse_categorical_accuracy: 0.9877Step 77 / 1718\n",
      "2464/55000 - mean: 0.0454 - sparse_categorical_accuracy: 0.9878Step 78 / 1718\n",
      "2496/55000 - mean: 0.0448 - sparse_categorical_accuracy: 0.9880Step 79 / 1718\n",
      "2528/55000 - mean: 0.0445 - sparse_categorical_accuracy: 0.9881Step 80 / 1718\n",
      "2560/55000 - mean: 0.0439 - sparse_categorical_accuracy: 0.9883Step 81 / 1718\n",
      "2592/55000 - mean: 0.0449 - sparse_categorical_accuracy: 0.9880Step 82 / 1718\n",
      "2624/55000 - mean: 0.0448 - sparse_categorical_accuracy: 0.9878Step 83 / 1718\n",
      "2656/55000 - mean: 0.0452 - sparse_categorical_accuracy: 0.9876Step 84 / 1718\n",
      "2688/55000 - mean: 0.0461 - sparse_categorical_accuracy: 0.9874Step 85 / 1718\n",
      "2720/55000 - mean: 0.0469 - sparse_categorical_accuracy: 0.9871Step 86 / 1718\n",
      "2752/55000 - mean: 0.0463 - sparse_categorical_accuracy: 0.9873Step 87 / 1718\n",
      "2784/55000 - mean: 0.0458 - sparse_categorical_accuracy: 0.9874Step 88 / 1718\n",
      "2816/55000 - mean: 0.0453 - sparse_categorical_accuracy: 0.9876Step 89 / 1718\n",
      "2848/55000 - mean: 0.0448 - sparse_categorical_accuracy: 0.9877Step 90 / 1718\n",
      "2880/55000 - mean: 0.0444 - sparse_categorical_accuracy: 0.9878Step 91 / 1718\n",
      "2912/55000 - mean: 0.0450 - sparse_categorical_accuracy: 0.9876Step 92 / 1718\n",
      "2944/55000 - mean: 0.0445 - sparse_categorical_accuracy: 0.9878Step 93 / 1718\n",
      "2976/55000 - mean: 0.0440 - sparse_categorical_accuracy: 0.9879Step 94 / 1718\n",
      "3008/55000 - mean: 0.0436 - sparse_categorical_accuracy: 0.9880Step 95 / 1718\n",
      "3040/55000 - mean: 0.0432 - sparse_categorical_accuracy: 0.9882Step 96 / 1718\n",
      "3072/55000 - mean: 0.0428 - sparse_categorical_accuracy: 0.9883Step 97 / 1718\n",
      "3104/55000 - mean: 0.0429 - sparse_categorical_accuracy: 0.9881Step 98 / 1718\n",
      "3136/55000 - mean: 0.0427 - sparse_categorical_accuracy: 0.9879Step 99 / 1718\n",
      "3168/55000 - mean: 0.0423 - sparse_categorical_accuracy: 0.9880Step 100 / 1718\n",
      "3200/55000 - mean: 0.0419 - sparse_categorical_accuracy: 0.9881Step 101 / 1718\n",
      "3232/55000 - mean: 0.0415 - sparse_categorical_accuracy: 0.9882Step 102 / 1718\n",
      "3264/55000 - mean: 0.0412 - sparse_categorical_accuracy: 0.9884Step 103 / 1718\n",
      "3296/55000 - mean: 0.0411 - sparse_categorical_accuracy: 0.9885Step 104 / 1718\n",
      "3328/55000 - mean: 0.0407 - sparse_categorical_accuracy: 0.9886Step 105 / 1718\n",
      "3360/55000 - mean: 0.0403 - sparse_categorical_accuracy: 0.9887Step 106 / 1718\n",
      "3392/55000 - mean: 0.0400 - sparse_categorical_accuracy: 0.9888Step 107 / 1718\n",
      "3424/55000 - mean: 0.0396 - sparse_categorical_accuracy: 0.9889Step 108 / 1718\n",
      "3456/55000 - mean: 0.0393 - sparse_categorical_accuracy: 0.9890Step 109 / 1718\n",
      "3488/55000 - mean: 0.0400 - sparse_categorical_accuracy: 0.9885Step 110 / 1718\n",
      "3520/55000 - mean: 0.0397 - sparse_categorical_accuracy: 0.9886Step 111 / 1718\n",
      "3552/55000 - mean: 0.0394 - sparse_categorical_accuracy: 0.9887Step 112 / 1718\n",
      "3584/55000 - mean: 0.0391 - sparse_categorical_accuracy: 0.9888Step 113 / 1718\n",
      "3616/55000 - mean: 0.0387 - sparse_categorical_accuracy: 0.9889Step 114 / 1718\n",
      "3648/55000 - mean: 0.0384 - sparse_categorical_accuracy: 0.9890Step 115 / 1718\n",
      "3680/55000 - mean: 0.0387 - sparse_categorical_accuracy: 0.9889Step 116 / 1718\n",
      "3712/55000 - mean: 0.0394 - sparse_categorical_accuracy: 0.9887Step 117 / 1718\n",
      "3744/55000 - mean: 0.0432 - sparse_categorical_accuracy: 0.9885Step 118 / 1718\n",
      "3776/55000 - mean: 0.0429 - sparse_categorical_accuracy: 0.9886Step 119 / 1718\n",
      "3808/55000 - mean: 0.0426 - sparse_categorical_accuracy: 0.9887Step 120 / 1718\n",
      "3840/55000 - mean: 0.0434 - sparse_categorical_accuracy: 0.9885Step 121 / 1718\n",
      "3872/55000 - mean: 0.0438 - sparse_categorical_accuracy: 0.9881Step 122 / 1718\n",
      "3904/55000 - mean: 0.0441 - sparse_categorical_accuracy: 0.9880Step 123 / 1718\n",
      "3936/55000 - mean: 0.0438 - sparse_categorical_accuracy: 0.9881Step 124 / 1718\n",
      "3968/55000 - mean: 0.0435 - sparse_categorical_accuracy: 0.9882Step 125 / 1718\n",
      "4000/55000 - mean: 0.0431 - sparse_categorical_accuracy: 0.9883Step 126 / 1718\n",
      "4032/55000 - mean: 0.0435 - sparse_categorical_accuracy: 0.9881Step 127 / 1718\n",
      "4064/55000 - mean: 0.0432 - sparse_categorical_accuracy: 0.9882Step 128 / 1718\n",
      "4096/55000 - mean: 0.0429 - sparse_categorical_accuracy: 0.9883Step 129 / 1718\n",
      "4128/55000 - mean: 0.0437 - sparse_categorical_accuracy: 0.9879Step 130 / 1718\n",
      "4160/55000 - mean: 0.0434 - sparse_categorical_accuracy: 0.9880Step 131 / 1718\n",
      "4192/55000 - mean: 0.0432 - sparse_categorical_accuracy: 0.9881Step 132 / 1718\n",
      "4224/55000 - mean: 0.0432 - sparse_categorical_accuracy: 0.9879Step 133 / 1718\n",
      "4256/55000 - mean: 0.0429 - sparse_categorical_accuracy: 0.9880Step 134 / 1718\n",
      "4288/55000 - mean: 0.0426 - sparse_categorical_accuracy: 0.9881Step 135 / 1718\n",
      "4320/55000 - mean: 0.0427 - sparse_categorical_accuracy: 0.9880Step 136 / 1718\n",
      "4352/55000 - mean: 0.0425 - sparse_categorical_accuracy: 0.9881Step 137 / 1718\n",
      "4384/55000 - mean: 0.0422 - sparse_categorical_accuracy: 0.9881Step 138 / 1718\n",
      "4416/55000 - mean: 0.0419 - sparse_categorical_accuracy: 0.9882Step 139 / 1718\n",
      "4448/55000 - mean: 0.0427 - sparse_categorical_accuracy: 0.9879Step 140 / 1718\n",
      "4480/55000 - mean: 0.0427 - sparse_categorical_accuracy: 0.9877Step 141 / 1718\n",
      "4512/55000 - mean: 0.0424 - sparse_categorical_accuracy: 0.9878Step 142 / 1718\n",
      "4544/55000 - mean: 0.0421 - sparse_categorical_accuracy: 0.9879Step 143 / 1718\n",
      "4576/55000 - mean: 0.0419 - sparse_categorical_accuracy: 0.9880Step 144 / 1718\n",
      "4608/55000 - mean: 0.0416 - sparse_categorical_accuracy: 0.9881Step 145 / 1718\n",
      "4640/55000 - mean: 0.0413 - sparse_categorical_accuracy: 0.9881Step 146 / 1718\n",
      "4672/55000 - mean: 0.0410 - sparse_categorical_accuracy: 0.9882Step 147 / 1718\n",
      "4704/55000 - mean: 0.0407 - sparse_categorical_accuracy: 0.9883Step 148 / 1718\n",
      "4736/55000 - mean: 0.0404 - sparse_categorical_accuracy: 0.9884Step 149 / 1718\n",
      "4768/55000 - mean: 0.0402 - sparse_categorical_accuracy: 0.9885Step 150 / 1718\n",
      "4800/55000 - mean: 0.0402 - sparse_categorical_accuracy: 0.9883Step 151 / 1718\n",
      "4832/55000 - mean: 0.0400 - sparse_categorical_accuracy: 0.9884Step 152 / 1718\n",
      "4864/55000 - mean: 0.0397 - sparse_categorical_accuracy: 0.9885Step 153 / 1718\n",
      "4896/55000 - mean: 0.0395 - sparse_categorical_accuracy: 0.9886Step 154 / 1718\n",
      "4928/55000 - mean: 0.0392 - sparse_categorical_accuracy: 0.9886Step 155 / 1718\n",
      "4960/55000 - mean: 0.0392 - sparse_categorical_accuracy: 0.9887Step 156 / 1718\n",
      "4992/55000 - mean: 0.0397 - sparse_categorical_accuracy: 0.9886Step 157 / 1718\n",
      "5024/55000 - mean: 0.0395 - sparse_categorical_accuracy: 0.9887Step 158 / 1718\n",
      "5056/55000 - mean: 0.0392 - sparse_categorical_accuracy: 0.9887Step 159 / 1718\n",
      "5088/55000 - mean: 0.0390 - sparse_categorical_accuracy: 0.9888Step 160 / 1718\n",
      "5120/55000 - mean: 0.0401 - sparse_categorical_accuracy: 0.9887Step 161 / 1718\n",
      "5152/55000 - mean: 0.0399 - sparse_categorical_accuracy: 0.9887Step 162 / 1718\n",
      "5184/55000 - mean: 0.0397 - sparse_categorical_accuracy: 0.9888Step 163 / 1718\n",
      "5216/55000 - mean: 0.0401 - sparse_categorical_accuracy: 0.9887Step 164 / 1718\n",
      "5248/55000 - mean: 0.0399 - sparse_categorical_accuracy: 0.9888Step 165 / 1718\n",
      "5280/55000 - mean: 0.0399 - sparse_categorical_accuracy: 0.9886Step 166 / 1718\n",
      "5312/55000 - mean: 0.0397 - sparse_categorical_accuracy: 0.9887Step 167 / 1718\n",
      "5344/55000 - mean: 0.0395 - sparse_categorical_accuracy: 0.9888Step 168 / 1718\n",
      "5376/55000 - mean: 0.0393 - sparse_categorical_accuracy: 0.9888Step 169 / 1718\n",
      "5408/55000 - mean: 0.0390 - sparse_categorical_accuracy: 0.9889Step 170 / 1718\n",
      "5440/55000 - mean: 0.0389 - sparse_categorical_accuracy: 0.9890Step 171 / 1718\n",
      "5472/55000 - mean: 0.0389 - sparse_categorical_accuracy: 0.9889Step 172 / 1718\n",
      "5504/55000 - mean: 0.0387 - sparse_categorical_accuracy: 0.9889Step 173 / 1718\n",
      "5536/55000 - mean: 0.0385 - sparse_categorical_accuracy: 0.9890Step 174 / 1718\n",
      "5568/55000 - mean: 0.0383 - sparse_categorical_accuracy: 0.9890Step 175 / 1718\n",
      "5600/55000 - mean: 0.0381 - sparse_categorical_accuracy: 0.9891Step 176 / 1718\n",
      "5632/55000 - mean: 0.0379 - sparse_categorical_accuracy: 0.9892Step 177 / 1718\n",
      "5664/55000 - mean: 0.0380 - sparse_categorical_accuracy: 0.9891Step 178 / 1718\n",
      "5696/55000 - mean: 0.0378 - sparse_categorical_accuracy: 0.9891Step 179 / 1718\n",
      "5728/55000 - mean: 0.0377 - sparse_categorical_accuracy: 0.9892Step 180 / 1718\n",
      "5760/55000 - mean: 0.0386 - sparse_categorical_accuracy: 0.9891Step 181 / 1718\n",
      "5792/55000 - mean: 0.0384 - sparse_categorical_accuracy: 0.9891Step 182 / 1718\n",
      "5824/55000 - mean: 0.0382 - sparse_categorical_accuracy: 0.9892Step 183 / 1718\n",
      "5856/55000 - mean: 0.0392 - sparse_categorical_accuracy: 0.9891Step 184 / 1718\n",
      "5888/55000 - mean: 0.0390 - sparse_categorical_accuracy: 0.9891Step 185 / 1718\n",
      "5920/55000 - mean: 0.0387 - sparse_categorical_accuracy: 0.9892Step 186 / 1718\n",
      "5952/55000 - mean: 0.0386 - sparse_categorical_accuracy: 0.9892Step 187 / 1718\n",
      "5984/55000 - mean: 0.0384 - sparse_categorical_accuracy: 0.9893Step 188 / 1718\n",
      "6016/55000 - mean: 0.0385 - sparse_categorical_accuracy: 0.9892Step 189 / 1718\n",
      "6048/55000 - mean: 0.0383 - sparse_categorical_accuracy: 0.9893Step 190 / 1718\n",
      "6080/55000 - mean: 0.0381 - sparse_categorical_accuracy: 0.9893Step 191 / 1718\n",
      "6112/55000 - mean: 0.0382 - sparse_categorical_accuracy: 0.9892Step 192 / 1718\n",
      "6144/55000 - mean: 0.0381 - sparse_categorical_accuracy: 0.9893Step 193 / 1718\n",
      "6176/55000 - mean: 0.0380 - sparse_categorical_accuracy: 0.9892Step 194 / 1718\n",
      "6208/55000 - mean: 0.0378 - sparse_categorical_accuracy: 0.9892Step 195 / 1718\n",
      "6240/55000 - mean: 0.0376 - sparse_categorical_accuracy: 0.9893Step 196 / 1718\n",
      "6272/55000 - mean: 0.0375 - sparse_categorical_accuracy: 0.9893Step 197 / 1718\n",
      "6304/55000 - mean: 0.0373 - sparse_categorical_accuracy: 0.9894Step 198 / 1718\n",
      "6336/55000 - mean: 0.0371 - sparse_categorical_accuracy: 0.9894Step 199 / 1718\n",
      "6368/55000 - mean: 0.0369 - sparse_categorical_accuracy: 0.9895Step 200 / 1718\n",
      "6400/55000 - mean: 0.0367 - sparse_categorical_accuracy: 0.9895Step 201 / 1718\n",
      "6432/55000 - mean: 0.0366 - sparse_categorical_accuracy: 0.9896Step 202 / 1718\n",
      "6464/55000 - mean: 0.0364 - sparse_categorical_accuracy: 0.9896Step 203 / 1718\n",
      "6496/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9897Step 204 / 1718\n",
      "6528/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9897Step 205 / 1718\n",
      "6560/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9896Step 206 / 1718\n",
      "6592/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9894Step 207 / 1718\n",
      "6624/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9894Step 208 / 1718\n",
      "6656/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9895Step 209 / 1718\n",
      "6688/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9895Step 210 / 1718\n",
      "6720/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9893Step 211 / 1718\n",
      "6752/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9893Step 212 / 1718\n",
      "6784/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9894Step 213 / 1718\n",
      "6816/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9894Step 214 / 1718\n",
      "6848/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9893Step 215 / 1718\n",
      "6880/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9894Step 216 / 1718\n",
      "6912/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9894Step 217 / 1718\n",
      "6944/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9893Step 218 / 1718\n",
      "6976/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9892Step 219 / 1718\n",
      "7008/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9893Step 220 / 1718\n",
      "7040/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9893Step 221 / 1718\n",
      "7072/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9893Step 222 / 1718\n",
      "7104/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9893Step 223 / 1718\n",
      "7136/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9893Step 224 / 1718\n",
      "7168/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9894Step 225 / 1718\n",
      "7200/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9894Step 226 / 1718\n",
      "7232/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9895Step 227 / 1718\n",
      "7264/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9895Step 228 / 1718\n",
      "7296/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9896Step 229 / 1718\n",
      "7328/55000 - mean: 0.0347 - sparse_categorical_accuracy: 0.9896Step 230 / 1718\n",
      "7360/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9897Step 231 / 1718\n",
      "7392/55000 - mean: 0.0346 - sparse_categorical_accuracy: 0.9896Step 232 / 1718\n",
      "7424/55000 - mean: 0.0344 - sparse_categorical_accuracy: 0.9896Step 233 / 1718\n",
      "7456/55000 - mean: 0.0343 - sparse_categorical_accuracy: 0.9897Step 234 / 1718\n",
      "7488/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9897Step 235 / 1718\n",
      "7520/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9898Step 236 / 1718\n",
      "7552/55000 - mean: 0.0339 - sparse_categorical_accuracy: 0.9898Step 237 / 1718\n",
      "7584/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9898Step 238 / 1718\n",
      "7616/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9899Step 239 / 1718\n",
      "7648/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9899Step 240 / 1718\n",
      "7680/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9896Step 241 / 1718\n",
      "7712/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9896Step 242 / 1718\n",
      "7744/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9897Step 243 / 1718\n",
      "7776/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9897Step 244 / 1718\n",
      "7808/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9896Step 245 / 1718\n",
      "7840/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9895Step 246 / 1718\n",
      "7872/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9895Step 247 / 1718\n",
      "7904/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9895Step 248 / 1718\n",
      "7936/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9895Step 249 / 1718\n",
      "7968/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9895Step 250 / 1718\n",
      "8000/55000 - mean: 0.0372 - sparse_categorical_accuracy: 0.9894Step 251 / 1718\n",
      "8032/55000 - mean: 0.0371 - sparse_categorical_accuracy: 0.9894Step 252 / 1718\n",
      "8064/55000 - mean: 0.0369 - sparse_categorical_accuracy: 0.9895Step 253 / 1718\n",
      "8096/55000 - mean: 0.0375 - sparse_categorical_accuracy: 0.9894Step 254 / 1718\n",
      "8128/55000 - mean: 0.0376 - sparse_categorical_accuracy: 0.9893Step 255 / 1718\n",
      "8160/55000 - mean: 0.0375 - sparse_categorical_accuracy: 0.9893Step 256 / 1718\n",
      "8192/55000 - mean: 0.0374 - sparse_categorical_accuracy: 0.9894Step 257 / 1718\n",
      "8224/55000 - mean: 0.0372 - sparse_categorical_accuracy: 0.9894Step 258 / 1718\n",
      "8256/55000 - mean: 0.0371 - sparse_categorical_accuracy: 0.9895Step 259 / 1718\n",
      "8288/55000 - mean: 0.0370 - sparse_categorical_accuracy: 0.9895Step 260 / 1718\n",
      "8320/55000 - mean: 0.0368 - sparse_categorical_accuracy: 0.9895Step 261 / 1718\n",
      "8352/55000 - mean: 0.0377 - sparse_categorical_accuracy: 0.9895Step 262 / 1718\n",
      "8384/55000 - mean: 0.0377 - sparse_categorical_accuracy: 0.9894Step 263 / 1718\n",
      "8416/55000 - mean: 0.0376 - sparse_categorical_accuracy: 0.9894Step 264 / 1718\n",
      "8448/55000 - mean: 0.0374 - sparse_categorical_accuracy: 0.9895Step 265 / 1718\n",
      "8480/55000 - mean: 0.0373 - sparse_categorical_accuracy: 0.9895Step 266 / 1718\n",
      "8512/55000 - mean: 0.0373 - sparse_categorical_accuracy: 0.9894Step 267 / 1718\n",
      "8544/55000 - mean: 0.0375 - sparse_categorical_accuracy: 0.9893Step 268 / 1718\n",
      "8576/55000 - mean: 0.0374 - sparse_categorical_accuracy: 0.9894Step 269 / 1718\n",
      "8608/55000 - mean: 0.0373 - sparse_categorical_accuracy: 0.9894Step 270 / 1718\n",
      "8640/55000 - mean: 0.0372 - sparse_categorical_accuracy: 0.9895Step 271 / 1718\n",
      "8672/55000 - mean: 0.0371 - sparse_categorical_accuracy: 0.9895Step 272 / 1718\n",
      "8704/55000 - mean: 0.0372 - sparse_categorical_accuracy: 0.9894Step 273 / 1718\n",
      "8736/55000 - mean: 0.0371 - sparse_categorical_accuracy: 0.9895Step 274 / 1718\n",
      "8768/55000 - mean: 0.0369 - sparse_categorical_accuracy: 0.9895Step 275 / 1718\n",
      "8800/55000 - mean: 0.0370 - sparse_categorical_accuracy: 0.9894Step 276 / 1718\n",
      "8832/55000 - mean: 0.0368 - sparse_categorical_accuracy: 0.9895Step 277 / 1718\n",
      "8864/55000 - mean: 0.0372 - sparse_categorical_accuracy: 0.9894Step 278 / 1718\n",
      "8896/55000 - mean: 0.0371 - sparse_categorical_accuracy: 0.9894Step 279 / 1718\n",
      "8928/55000 - mean: 0.0376 - sparse_categorical_accuracy: 0.9894Step 280 / 1718\n",
      "8960/55000 - mean: 0.0375 - sparse_categorical_accuracy: 0.9894Step 281 / 1718\n",
      "8992/55000 - mean: 0.0374 - sparse_categorical_accuracy: 0.9894Step 282 / 1718\n",
      "9024/55000 - mean: 0.0373 - sparse_categorical_accuracy: 0.9895Step 283 / 1718\n",
      "9056/55000 - mean: 0.0372 - sparse_categorical_accuracy: 0.9895Step 284 / 1718\n",
      "9088/55000 - mean: 0.0383 - sparse_categorical_accuracy: 0.9893Step 285 / 1718\n",
      "9120/55000 - mean: 0.0382 - sparse_categorical_accuracy: 0.9894Step 286 / 1718\n",
      "9152/55000 - mean: 0.0380 - sparse_categorical_accuracy: 0.9894Step 287 / 1718\n",
      "9184/55000 - mean: 0.0379 - sparse_categorical_accuracy: 0.9894Step 288 / 1718\n",
      "9216/55000 - mean: 0.0378 - sparse_categorical_accuracy: 0.9895Step 289 / 1718\n",
      "9248/55000 - mean: 0.0380 - sparse_categorical_accuracy: 0.9894Step 290 / 1718\n",
      "9280/55000 - mean: 0.0380 - sparse_categorical_accuracy: 0.9893Step 291 / 1718\n",
      "9312/55000 - mean: 0.0380 - sparse_categorical_accuracy: 0.9894Step 292 / 1718\n",
      "9344/55000 - mean: 0.0379 - sparse_categorical_accuracy: 0.9894Step 293 / 1718\n",
      "9376/55000 - mean: 0.0377 - sparse_categorical_accuracy: 0.9894Step 294 / 1718\n",
      "9408/55000 - mean: 0.0377 - sparse_categorical_accuracy: 0.9894Step 295 / 1718\n",
      "9440/55000 - mean: 0.0376 - sparse_categorical_accuracy: 0.9894Step 296 / 1718\n",
      "9472/55000 - mean: 0.0379 - sparse_categorical_accuracy: 0.9893Step 297 / 1718\n",
      "9504/55000 - mean: 0.0378 - sparse_categorical_accuracy: 0.9894Step 298 / 1718\n",
      "9536/55000 - mean: 0.0379 - sparse_categorical_accuracy: 0.9893Step 299 / 1718\n",
      "9568/55000 - mean: 0.0378 - sparse_categorical_accuracy: 0.9893Step 300 / 1718\n",
      "9600/55000 - mean: 0.0376 - sparse_categorical_accuracy: 0.9894Step 301 / 1718\n",
      "9632/55000 - mean: 0.0375 - sparse_categorical_accuracy: 0.9894Step 302 / 1718\n",
      "9664/55000 - mean: 0.0374 - sparse_categorical_accuracy: 0.9894Step 303 / 1718\n",
      "9696/55000 - mean: 0.0375 - sparse_categorical_accuracy: 0.9894Step 304 / 1718\n",
      "9728/55000 - mean: 0.0382 - sparse_categorical_accuracy: 0.9893Step 305 / 1718\n",
      "9760/55000 - mean: 0.0381 - sparse_categorical_accuracy: 0.9893Step 306 / 1718\n",
      "9792/55000 - mean: 0.0381 - sparse_categorical_accuracy: 0.9893Step 307 / 1718\n",
      "9824/55000 - mean: 0.0380 - sparse_categorical_accuracy: 0.9893Step 308 / 1718\n",
      "9856/55000 - mean: 0.0379 - sparse_categorical_accuracy: 0.9893Step 309 / 1718\n",
      "9888/55000 - mean: 0.0378 - sparse_categorical_accuracy: 0.9894Step 310 / 1718\n",
      "9920/55000 - mean: 0.0376 - sparse_categorical_accuracy: 0.9894Step 311 / 1718\n",
      "9952/55000 - mean: 0.0375 - sparse_categorical_accuracy: 0.9894Step 312 / 1718\n",
      "9984/55000 - mean: 0.0374 - sparse_categorical_accuracy: 0.9895Step 313 / 1718\n",
      "10016/55000 - mean: 0.0373 - sparse_categorical_accuracy: 0.9895Step 314 / 1718\n",
      "10048/55000 - mean: 0.0372 - sparse_categorical_accuracy: 0.9896Step 315 / 1718\n",
      "10080/55000 - mean: 0.0371 - sparse_categorical_accuracy: 0.9896Step 316 / 1718\n",
      "10112/55000 - mean: 0.0369 - sparse_categorical_accuracy: 0.9896Step 317 / 1718\n",
      "10144/55000 - mean: 0.0368 - sparse_categorical_accuracy: 0.9896Step 318 / 1718\n",
      "10176/55000 - mean: 0.0368 - sparse_categorical_accuracy: 0.9897Step 319 / 1718\n",
      "10208/55000 - mean: 0.0367 - sparse_categorical_accuracy: 0.9897Step 320 / 1718\n",
      "10240/55000 - mean: 0.0366 - sparse_categorical_accuracy: 0.9897Step 321 / 1718\n",
      "10272/55000 - mean: 0.0365 - sparse_categorical_accuracy: 0.9898Step 322 / 1718\n",
      "10304/55000 - mean: 0.0370 - sparse_categorical_accuracy: 0.9897Step 323 / 1718\n",
      "10336/55000 - mean: 0.0373 - sparse_categorical_accuracy: 0.9896Step 324 / 1718\n",
      "10368/55000 - mean: 0.0373 - sparse_categorical_accuracy: 0.9896Step 325 / 1718\n",
      "10400/55000 - mean: 0.0373 - sparse_categorical_accuracy: 0.9895Step 326 / 1718\n",
      "10432/55000 - mean: 0.0372 - sparse_categorical_accuracy: 0.9896Step 327 / 1718\n",
      "10464/55000 - mean: 0.0371 - sparse_categorical_accuracy: 0.9895Step 328 / 1718\n",
      "10496/55000 - mean: 0.0370 - sparse_categorical_accuracy: 0.9895Step 329 / 1718\n",
      "10528/55000 - mean: 0.0369 - sparse_categorical_accuracy: 0.9896Step 330 / 1718\n",
      "10560/55000 - mean: 0.0369 - sparse_categorical_accuracy: 0.9895Step 331 / 1718\n",
      "10592/55000 - mean: 0.0368 - sparse_categorical_accuracy: 0.9895Step 332 / 1718\n",
      "10624/55000 - mean: 0.0367 - sparse_categorical_accuracy: 0.9896Step 333 / 1718\n",
      "10656/55000 - mean: 0.0366 - sparse_categorical_accuracy: 0.9896Step 334 / 1718\n",
      "10688/55000 - mean: 0.0365 - sparse_categorical_accuracy: 0.9896Step 335 / 1718\n",
      "10720/55000 - mean: 0.0364 - sparse_categorical_accuracy: 0.9896Step 336 / 1718\n",
      "10752/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9897Step 337 / 1718\n",
      "10784/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9896Step 338 / 1718\n",
      "10816/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9896Step 339 / 1718\n",
      "10848/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9897Step 340 / 1718\n",
      "10880/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9896Step 341 / 1718\n",
      "10912/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9896Step 342 / 1718\n",
      "10944/55000 - mean: 0.0367 - sparse_categorical_accuracy: 0.9896Step 343 / 1718\n",
      "10976/55000 - mean: 0.0375 - sparse_categorical_accuracy: 0.9894Step 344 / 1718\n",
      "11008/55000 - mean: 0.0374 - sparse_categorical_accuracy: 0.9895Step 345 / 1718\n",
      "11040/55000 - mean: 0.0378 - sparse_categorical_accuracy: 0.9894Step 346 / 1718\n",
      "11072/55000 - mean: 0.0377 - sparse_categorical_accuracy: 0.9894Step 347 / 1718\n",
      "11104/55000 - mean: 0.0376 - sparse_categorical_accuracy: 0.9895Step 348 / 1718\n",
      "11136/55000 - mean: 0.0378 - sparse_categorical_accuracy: 0.9894Step 349 / 1718\n",
      "11168/55000 - mean: 0.0377 - sparse_categorical_accuracy: 0.9894Step 350 / 1718\n",
      "11200/55000 - mean: 0.0376 - sparse_categorical_accuracy: 0.9895Step 351 / 1718\n",
      "11232/55000 - mean: 0.0375 - sparse_categorical_accuracy: 0.9895Step 352 / 1718\n",
      "11264/55000 - mean: 0.0374 - sparse_categorical_accuracy: 0.9895Step 353 / 1718\n",
      "11296/55000 - mean: 0.0373 - sparse_categorical_accuracy: 0.9896Step 354 / 1718\n",
      "11328/55000 - mean: 0.0372 - sparse_categorical_accuracy: 0.9896Step 355 / 1718\n",
      "11360/55000 - mean: 0.0371 - sparse_categorical_accuracy: 0.9896Step 356 / 1718\n",
      "11392/55000 - mean: 0.0370 - sparse_categorical_accuracy: 0.9896Step 357 / 1718\n",
      "11424/55000 - mean: 0.0370 - sparse_categorical_accuracy: 0.9896Step 358 / 1718\n",
      "11456/55000 - mean: 0.0371 - sparse_categorical_accuracy: 0.9894Step 359 / 1718\n",
      "11488/55000 - mean: 0.0370 - sparse_categorical_accuracy: 0.9895Step 360 / 1718\n",
      "11520/55000 - mean: 0.0369 - sparse_categorical_accuracy: 0.9895Step 361 / 1718\n",
      "11552/55000 - mean: 0.0368 - sparse_categorical_accuracy: 0.9895Step 362 / 1718\n",
      "11584/55000 - mean: 0.0367 - sparse_categorical_accuracy: 0.9896Step 363 / 1718\n",
      "11616/55000 - mean: 0.0366 - sparse_categorical_accuracy: 0.9896Step 364 / 1718\n",
      "11648/55000 - mean: 0.0366 - sparse_categorical_accuracy: 0.9896Step 365 / 1718\n",
      "11680/55000 - mean: 0.0365 - sparse_categorical_accuracy: 0.9896Step 366 / 1718\n",
      "11712/55000 - mean: 0.0364 - sparse_categorical_accuracy: 0.9897Step 367 / 1718\n",
      "11744/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9897Step 368 / 1718\n",
      "11776/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9897Step 369 / 1718\n",
      "11808/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9898Step 370 / 1718\n",
      "11840/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9898Step 371 / 1718\n",
      "11872/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9898Step 372 / 1718\n",
      "11904/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9898Step 373 / 1718\n",
      "11936/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9899Step 374 / 1718\n",
      "11968/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9898Step 375 / 1718\n",
      "12000/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9898Step 376 / 1718\n",
      "12032/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9899Step 377 / 1718\n",
      "12064/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9899Step 378 / 1718\n",
      "12096/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9899Step 379 / 1718\n",
      "12128/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9899Step 380 / 1718\n",
      "12160/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9899Step 381 / 1718\n",
      "12192/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9899Step 382 / 1718\n",
      "12224/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9899Step 383 / 1718\n",
      "12256/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9900Step 384 / 1718\n",
      "12288/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9900Step 385 / 1718\n",
      "12320/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9900Step 386 / 1718\n",
      "12352/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9900Step 387 / 1718\n",
      "12384/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9901Step 388 / 1718\n",
      "12416/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9900Step 389 / 1718\n",
      "12448/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9900Step 390 / 1718\n",
      "12480/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9899Step 391 / 1718\n",
      "12512/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9899Step 392 / 1718\n",
      "12544/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9900Step 393 / 1718\n",
      "12576/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9899Step 394 / 1718\n",
      "12608/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9899Step 395 / 1718\n",
      "12640/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9900Step 396 / 1718\n",
      "12672/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9900Step 397 / 1718\n",
      "12704/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9900Step 398 / 1718\n",
      "12736/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9900Step 399 / 1718\n",
      "12768/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9901Step 400 / 1718\n",
      "12800/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9901Step 401 / 1718\n",
      "12832/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9901Step 402 / 1718\n",
      "12864/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9901Step 403 / 1718\n",
      "12896/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9902Step 404 / 1718\n",
      "12928/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9900Step 405 / 1718\n",
      "12960/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9900Step 406 / 1718\n",
      "12992/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9900Step 407 / 1718\n",
      "13024/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9900Step 408 / 1718\n",
      "13056/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9900Step 409 / 1718\n",
      "13088/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9901Step 410 / 1718\n",
      "13120/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9901Step 411 / 1718\n",
      "13152/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9901Step 412 / 1718\n",
      "13184/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9901Step 413 / 1718\n",
      "13216/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9901Step 414 / 1718\n",
      "13248/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9901Step 415 / 1718\n",
      "13280/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9901Step 416 / 1718\n",
      "13312/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9902Step 417 / 1718\n",
      "13344/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9901Step 418 / 1718\n",
      "13376/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9901Step 419 / 1718\n",
      "13408/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9900Step 420 / 1718\n",
      "13440/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9900Step 421 / 1718\n",
      "13472/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9901Step 422 / 1718\n",
      "13504/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9900Step 423 / 1718\n",
      "13536/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9900Step 424 / 1718\n",
      "13568/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9901Step 425 / 1718\n",
      "13600/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9901Step 426 / 1718\n",
      "13632/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9901Step 427 / 1718\n",
      "13664/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9901Step 428 / 1718\n",
      "13696/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9901Step 429 / 1718\n",
      "13728/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9901Step 430 / 1718\n",
      "13760/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9900Step 431 / 1718\n",
      "13792/55000 - mean: 0.0347 - sparse_categorical_accuracy: 0.9901Step 432 / 1718\n",
      "13824/55000 - mean: 0.0346 - sparse_categorical_accuracy: 0.9901Step 433 / 1718\n",
      "13856/55000 - mean: 0.0346 - sparse_categorical_accuracy: 0.9901Step 434 / 1718\n",
      "13888/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9901Step 435 / 1718\n",
      "13920/55000 - mean: 0.0344 - sparse_categorical_accuracy: 0.9902Step 436 / 1718\n",
      "13952/55000 - mean: 0.0343 - sparse_categorical_accuracy: 0.9902Step 437 / 1718\n",
      "13984/55000 - mean: 0.0343 - sparse_categorical_accuracy: 0.9902Step 438 / 1718\n",
      "14016/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9902Step 439 / 1718\n",
      "14048/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9902Step 440 / 1718\n",
      "14080/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9901Step 441 / 1718\n",
      "14112/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9900Step 442 / 1718\n",
      "14144/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9900Step 443 / 1718\n",
      "14176/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9901Step 444 / 1718\n",
      "14208/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9901Step 445 / 1718\n",
      "14240/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9900Step 446 / 1718\n",
      "14272/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9901Step 447 / 1718\n",
      "14304/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9901Step 448 / 1718\n",
      "14336/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9901Step 449 / 1718\n",
      "14368/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9901Step 450 / 1718\n",
      "14400/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9901Step 451 / 1718\n",
      "14432/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9901Step 452 / 1718\n",
      "14464/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9901Step 453 / 1718\n",
      "14496/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9901Step 454 / 1718\n",
      "14528/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9901Step 455 / 1718\n",
      "14560/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9901Step 456 / 1718\n",
      "14592/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9901Step 457 / 1718\n",
      "14624/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9900Step 458 / 1718\n",
      "14656/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9900Step 459 / 1718\n",
      "14688/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9900Step 460 / 1718\n",
      "14720/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9900Step 461 / 1718\n",
      "14752/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9900Step 462 / 1718\n",
      "14784/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9901Step 463 / 1718\n",
      "14816/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9901Step 464 / 1718\n",
      "14848/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9901Step 465 / 1718\n",
      "14880/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9901Step 466 / 1718\n",
      "14912/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9901Step 467 / 1718\n",
      "14944/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9900Step 468 / 1718\n",
      "14976/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9900Step 469 / 1718\n",
      "15008/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9900Step 470 / 1718\n",
      "15040/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9900Step 471 / 1718\n",
      "15072/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9900Step 472 / 1718\n",
      "15104/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9901Step 473 / 1718\n",
      "15136/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9900Step 474 / 1718\n",
      "15168/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9900Step 475 / 1718\n",
      "15200/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9901Step 476 / 1718\n",
      "15232/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9900Step 477 / 1718\n",
      "15264/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9900Step 478 / 1718\n",
      "15296/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9900Step 479 / 1718\n",
      "15328/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9900Step 480 / 1718\n",
      "15360/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9900Step 481 / 1718\n",
      "15392/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9901Step 482 / 1718\n",
      "15424/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9901Step 483 / 1718\n",
      "15456/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9901Step 484 / 1718\n",
      "15488/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9901Step 485 / 1718\n",
      "15520/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9901Step 486 / 1718\n",
      "15552/55000 - mean: 0.0347 - sparse_categorical_accuracy: 0.9902Step 487 / 1718\n",
      "15584/55000 - mean: 0.0346 - sparse_categorical_accuracy: 0.9902Step 488 / 1718\n",
      "15616/55000 - mean: 0.0346 - sparse_categorical_accuracy: 0.9902Step 489 / 1718\n",
      "15648/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9902Step 490 / 1718\n",
      "15680/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9902Step 491 / 1718\n",
      "15712/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9901Step 492 / 1718\n",
      "15744/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9902Step 493 / 1718\n",
      "15776/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9900Step 494 / 1718\n",
      "15808/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9901Step 495 / 1718\n",
      "15840/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9901Step 496 / 1718\n",
      "15872/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9901Step 497 / 1718\n",
      "15904/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9901Step 498 / 1718\n",
      "15936/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9901Step 499 / 1718\n",
      "15968/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9902Step 500 / 1718\n",
      "16000/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9902Step 501 / 1718\n",
      "16032/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9902Step 502 / 1718\n",
      "16064/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9902Step 503 / 1718\n",
      "16096/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9902Step 504 / 1718\n",
      "16128/55000 - mean: 0.0347 - sparse_categorical_accuracy: 0.9903Step 505 / 1718\n",
      "16160/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9902Step 506 / 1718\n",
      "16192/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9902Step 507 / 1718\n",
      "16224/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9902Step 508 / 1718\n",
      "16256/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9902Step 509 / 1718\n",
      "16288/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9902Step 510 / 1718\n",
      "16320/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9903Step 511 / 1718\n",
      "16352/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9902Step 512 / 1718\n",
      "16384/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9902Step 513 / 1718\n",
      "16416/55000 - mean: 0.0364 - sparse_categorical_accuracy: 0.9901Step 514 / 1718\n",
      "16448/55000 - mean: 0.0364 - sparse_categorical_accuracy: 0.9901Step 515 / 1718\n",
      "16480/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9901Step 516 / 1718\n",
      "16512/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 517 / 1718\n",
      "16544/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 518 / 1718\n",
      "16576/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9902Step 519 / 1718\n",
      "16608/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9901Step 520 / 1718\n",
      "16640/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9901Step 521 / 1718\n",
      "16672/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9902Step 522 / 1718\n",
      "16704/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9902Step 523 / 1718\n",
      "16736/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9902Step 524 / 1718\n",
      "16768/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9902Step 525 / 1718\n",
      "16800/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 526 / 1718\n",
      "16832/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9903Step 527 / 1718\n",
      "16864/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9903Step 528 / 1718\n",
      "16896/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9903Step 529 / 1718\n",
      "16928/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 530 / 1718\n",
      "16960/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9902Step 531 / 1718\n",
      "16992/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9902Step 532 / 1718\n",
      "17024/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9902Step 533 / 1718\n",
      "17056/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9902Step 534 / 1718\n",
      "17088/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9902Step 535 / 1718\n",
      "17120/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9902Step 536 / 1718\n",
      "17152/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9902Step 537 / 1718\n",
      "17184/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9902Step 538 / 1718\n",
      "17216/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9902Step 539 / 1718\n",
      "17248/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9902Step 540 / 1718\n",
      "17280/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9902Step 541 / 1718\n",
      "17312/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9902Step 542 / 1718\n",
      "17344/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9902Step 543 / 1718\n",
      "17376/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9902Step 544 / 1718\n",
      "17408/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9902Step 545 / 1718\n",
      "17440/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9903Step 546 / 1718\n",
      "17472/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9903Step 547 / 1718\n",
      "17504/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9902Step 548 / 1718\n",
      "17536/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9902Step 549 / 1718\n",
      "17568/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9902Step 550 / 1718\n",
      "17600/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9902Step 551 / 1718\n",
      "17632/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9902Step 552 / 1718\n",
      "17664/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9901Step 553 / 1718\n",
      "17696/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9902Step 554 / 1718\n",
      "17728/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9902Step 555 / 1718\n",
      "17760/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9902Step 556 / 1718\n",
      "17792/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9902Step 557 / 1718\n",
      "17824/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9902Step 558 / 1718\n",
      "17856/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9902Step 559 / 1718\n",
      "17888/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9902Step 560 / 1718\n",
      "17920/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9902Step 561 / 1718\n",
      "17952/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9903Step 562 / 1718\n",
      "17984/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9903Step 563 / 1718\n",
      "18016/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9902Step 564 / 1718\n",
      "18048/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9902Step 565 / 1718\n",
      "18080/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9902Step 566 / 1718\n",
      "18112/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9902Step 567 / 1718\n",
      "18144/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9902Step 568 / 1718\n",
      "18176/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9903Step 569 / 1718\n",
      "18208/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9903Step 570 / 1718\n",
      "18240/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9903Step 571 / 1718\n",
      "18272/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9903Step 572 / 1718\n",
      "18304/55000 - mean: 0.0353 - sparse_categorical_accuracy: 0.9903Step 573 / 1718\n",
      "18336/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9903Step 574 / 1718\n",
      "18368/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9904Step 575 / 1718\n",
      "18400/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9904Step 576 / 1718\n",
      "18432/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9904Step 577 / 1718\n",
      "18464/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9904Step 578 / 1718\n",
      "18496/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9904Step 579 / 1718\n",
      "18528/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9904Step 580 / 1718\n",
      "18560/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9905Step 581 / 1718\n",
      "18592/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9905Step 582 / 1718\n",
      "18624/55000 - mean: 0.0347 - sparse_categorical_accuracy: 0.9905Step 583 / 1718\n",
      "18656/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9905Step 584 / 1718\n",
      "18688/55000 - mean: 0.0347 - sparse_categorical_accuracy: 0.9905Step 585 / 1718\n",
      "18720/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9904Step 586 / 1718\n",
      "18752/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9904Step 587 / 1718\n",
      "18784/55000 - mean: 0.0347 - sparse_categorical_accuracy: 0.9904Step 588 / 1718\n",
      "18816/55000 - mean: 0.0347 - sparse_categorical_accuracy: 0.9904Step 589 / 1718\n",
      "18848/55000 - mean: 0.0346 - sparse_categorical_accuracy: 0.9904Step 590 / 1718\n",
      "18880/55000 - mean: 0.0346 - sparse_categorical_accuracy: 0.9905Step 591 / 1718\n",
      "18912/55000 - mean: 0.0346 - sparse_categorical_accuracy: 0.9905Step 592 / 1718\n",
      "18944/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9905Step 593 / 1718\n",
      "18976/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9905Step 594 / 1718\n",
      "19008/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9905Step 595 / 1718\n",
      "19040/55000 - mean: 0.0344 - sparse_categorical_accuracy: 0.9905Step 596 / 1718\n",
      "19072/55000 - mean: 0.0346 - sparse_categorical_accuracy: 0.9905Step 597 / 1718\n",
      "19104/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9905Step 598 / 1718\n",
      "19136/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9905Step 599 / 1718\n",
      "19168/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9905Step 600 / 1718\n",
      "19200/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9904Step 601 / 1718\n",
      "19232/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9904Step 602 / 1718\n",
      "19264/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9904Step 603 / 1718\n",
      "19296/55000 - mean: 0.0344 - sparse_categorical_accuracy: 0.9904Step 604 / 1718\n",
      "19328/55000 - mean: 0.0344 - sparse_categorical_accuracy: 0.9904Step 605 / 1718\n",
      "19360/55000 - mean: 0.0343 - sparse_categorical_accuracy: 0.9904Step 606 / 1718\n",
      "19392/55000 - mean: 0.0343 - sparse_categorical_accuracy: 0.9905Step 607 / 1718\n",
      "19424/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9905Step 608 / 1718\n",
      "19456/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9904Step 609 / 1718\n",
      "19488/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9905Step 610 / 1718\n",
      "19520/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9905Step 611 / 1718\n",
      "19552/55000 - mean: 0.0343 - sparse_categorical_accuracy: 0.9904Step 612 / 1718\n",
      "19584/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9905Step 613 / 1718\n",
      "19616/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9905Step 614 / 1718\n",
      "19648/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9905Step 615 / 1718\n",
      "19680/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9905Step 616 / 1718\n",
      "19712/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9905Step 617 / 1718\n",
      "19744/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9905Step 618 / 1718\n",
      "19776/55000 - mean: 0.0339 - sparse_categorical_accuracy: 0.9905Step 619 / 1718\n",
      "19808/55000 - mean: 0.0343 - sparse_categorical_accuracy: 0.9905Step 620 / 1718\n",
      "19840/55000 - mean: 0.0343 - sparse_categorical_accuracy: 0.9905Step 621 / 1718\n",
      "19872/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9905Step 622 / 1718\n",
      "19904/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9906Step 623 / 1718\n",
      "19936/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9906Step 624 / 1718\n",
      "19968/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9906Step 625 / 1718\n",
      "20000/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9906Step 626 / 1718\n",
      "20032/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9906Step 627 / 1718\n",
      "20064/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9906Step 628 / 1718\n",
      "20096/55000 - mean: 0.0339 - sparse_categorical_accuracy: 0.9906Step 629 / 1718\n",
      "20128/55000 - mean: 0.0339 - sparse_categorical_accuracy: 0.9907Step 630 / 1718\n",
      "20160/55000 - mean: 0.0338 - sparse_categorical_accuracy: 0.9907Step 631 / 1718\n",
      "20192/55000 - mean: 0.0338 - sparse_categorical_accuracy: 0.9907Step 632 / 1718\n",
      "20224/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9907Step 633 / 1718\n",
      "20256/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9907Step 634 / 1718\n",
      "20288/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9907Step 635 / 1718\n",
      "20320/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9907Step 636 / 1718\n",
      "20352/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9908Step 637 / 1718\n",
      "20384/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9908Step 638 / 1718\n",
      "20416/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9908Step 639 / 1718\n",
      "20448/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9908Step 640 / 1718\n",
      "20480/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9908Step 641 / 1718\n",
      "20512/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9908Step 642 / 1718\n",
      "20544/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9908Step 643 / 1718\n",
      "20576/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9908Step 644 / 1718\n",
      "20608/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9908Step 645 / 1718\n",
      "20640/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9908Step 646 / 1718\n",
      "20672/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9908Step 647 / 1718\n",
      "20704/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9907Step 648 / 1718\n",
      "20736/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9907Step 649 / 1718\n",
      "20768/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9908Step 650 / 1718\n",
      "20800/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9908Step 651 / 1718\n",
      "20832/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9908Step 652 / 1718\n",
      "20864/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9908Step 653 / 1718\n",
      "20896/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9908Step 654 / 1718\n",
      "20928/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9908Step 655 / 1718\n",
      "20960/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9908Step 656 / 1718\n",
      "20992/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9909Step 657 / 1718\n",
      "21024/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9909Step 658 / 1718\n",
      "21056/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9909Step 659 / 1718\n",
      "21088/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9909Step 660 / 1718\n",
      "21120/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9909Step 661 / 1718\n",
      "21152/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9909Step 662 / 1718\n",
      "21184/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9909Step 663 / 1718\n",
      "21216/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9909Step 664 / 1718\n",
      "21248/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9909Step 665 / 1718\n",
      "21280/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9909Step 666 / 1718\n",
      "21312/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9909Step 667 / 1718\n",
      "21344/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9909Step 668 / 1718\n",
      "21376/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9909Step 669 / 1718\n",
      "21408/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9909Step 670 / 1718\n",
      "21440/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9909Step 671 / 1718\n",
      "21472/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9909Step 672 / 1718\n",
      "21504/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9909Step 673 / 1718\n",
      "21536/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9909Step 674 / 1718\n",
      "21568/55000 - mean: 0.0344 - sparse_categorical_accuracy: 0.9908Step 675 / 1718\n",
      "21600/55000 - mean: 0.0345 - sparse_categorical_accuracy: 0.9908Step 676 / 1718\n",
      "21632/55000 - mean: 0.0344 - sparse_categorical_accuracy: 0.9908Step 677 / 1718\n",
      "21664/55000 - mean: 0.0344 - sparse_categorical_accuracy: 0.9908Step 678 / 1718\n",
      "21696/55000 - mean: 0.0343 - sparse_categorical_accuracy: 0.9908Step 679 / 1718\n",
      "21728/55000 - mean: 0.0343 - sparse_categorical_accuracy: 0.9908Step 680 / 1718\n",
      "21760/55000 - mean: 0.0343 - sparse_categorical_accuracy: 0.9909Step 681 / 1718\n",
      "21792/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9909Step 682 / 1718\n",
      "21824/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9909Step 683 / 1718\n",
      "21856/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9908Step 684 / 1718\n",
      "21888/55000 - mean: 0.0343 - sparse_categorical_accuracy: 0.9908Step 685 / 1718\n",
      "21920/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9908Step 686 / 1718\n",
      "21952/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9908Step 687 / 1718\n",
      "21984/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9909Step 688 / 1718\n",
      "22016/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9909Step 689 / 1718\n",
      "22048/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9909Step 690 / 1718\n",
      "22080/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9909Step 691 / 1718\n",
      "22112/55000 - mean: 0.0339 - sparse_categorical_accuracy: 0.9909Step 692 / 1718\n",
      "22144/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9909Step 693 / 1718\n",
      "22176/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9909Step 694 / 1718\n",
      "22208/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9909Step 695 / 1718\n",
      "22240/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9909Step 696 / 1718\n",
      "22272/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9909Step 697 / 1718\n",
      "22304/55000 - mean: 0.0339 - sparse_categorical_accuracy: 0.9909Step 698 / 1718\n",
      "22336/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9909Step 699 / 1718\n",
      "22368/55000 - mean: 0.0339 - sparse_categorical_accuracy: 0.9909Step 700 / 1718\n",
      "22400/55000 - mean: 0.0339 - sparse_categorical_accuracy: 0.9909Step 701 / 1718\n",
      "22432/55000 - mean: 0.0338 - sparse_categorical_accuracy: 0.9909Step 702 / 1718\n",
      "22464/55000 - mean: 0.0338 - sparse_categorical_accuracy: 0.9909Step 703 / 1718\n",
      "22496/55000 - mean: 0.0339 - sparse_categorical_accuracy: 0.9909Step 704 / 1718\n",
      "22528/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9909Step 705 / 1718\n",
      "22560/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9909Step 706 / 1718\n",
      "22592/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9909Step 707 / 1718\n",
      "22624/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9909Step 708 / 1718\n",
      "22656/55000 - mean: 0.0339 - sparse_categorical_accuracy: 0.9909Step 709 / 1718\n",
      "22688/55000 - mean: 0.0339 - sparse_categorical_accuracy: 0.9909Step 710 / 1718\n",
      "22720/55000 - mean: 0.0339 - sparse_categorical_accuracy: 0.9909Step 711 / 1718\n",
      "22752/55000 - mean: 0.0338 - sparse_categorical_accuracy: 0.9909Step 712 / 1718\n",
      "22784/55000 - mean: 0.0338 - sparse_categorical_accuracy: 0.9909Step 713 / 1718\n",
      "22816/55000 - mean: 0.0338 - sparse_categorical_accuracy: 0.9909Step 714 / 1718\n",
      "22848/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9909Step 715 / 1718\n",
      "22880/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9910Step 716 / 1718\n",
      "22912/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9910Step 717 / 1718\n",
      "22944/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9910Step 718 / 1718\n",
      "22976/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9910Step 719 / 1718\n",
      "23008/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9910Step 720 / 1718\n",
      "23040/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9910Step 721 / 1718\n",
      "23072/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9910Step 722 / 1718\n",
      "23104/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9910Step 723 / 1718\n",
      "23136/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9911Step 724 / 1718\n",
      "23168/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9911Step 725 / 1718\n",
      "23200/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9911Step 726 / 1718\n",
      "23232/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9911Step 727 / 1718\n",
      "23264/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9911Step 728 / 1718\n",
      "23296/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9911Step 729 / 1718\n",
      "23328/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9911Step 730 / 1718\n",
      "23360/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9911Step 731 / 1718\n",
      "23392/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9912Step 732 / 1718\n",
      "23424/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9911Step 733 / 1718\n",
      "23456/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9911Step 734 / 1718\n",
      "23488/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9911Step 735 / 1718\n",
      "23520/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9911Step 736 / 1718\n",
      "23552/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9911Step 737 / 1718\n",
      "23584/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9911Step 738 / 1718\n",
      "23616/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9911Step 739 / 1718\n",
      "23648/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9911Step 740 / 1718\n",
      "23680/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9911Step 741 / 1718\n",
      "23712/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9911Step 742 / 1718\n",
      "23744/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9911Step 743 / 1718\n",
      "23776/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9911Step 744 / 1718\n",
      "23808/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9911Step 745 / 1718\n",
      "23840/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9911Step 746 / 1718\n",
      "23872/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9911Step 747 / 1718\n",
      "23904/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9911Step 748 / 1718\n",
      "23936/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9911Step 749 / 1718\n",
      "23968/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9911Step 750 / 1718\n",
      "24000/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9911Step 751 / 1718\n",
      "24032/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9911Step 752 / 1718\n",
      "24064/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9911Step 753 / 1718\n",
      "24096/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9911Step 754 / 1718\n",
      "24128/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9911Step 755 / 1718\n",
      "24160/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9911Step 756 / 1718\n",
      "24192/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9911Step 757 / 1718\n",
      "24224/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9911Step 758 / 1718\n",
      "24256/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9911Step 759 / 1718\n",
      "24288/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9911Step 760 / 1718\n",
      "24320/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9911Step 761 / 1718\n",
      "24352/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9911Step 762 / 1718\n",
      "24384/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9911Step 763 / 1718\n",
      "24416/55000 - mean: 0.0338 - sparse_categorical_accuracy: 0.9911Step 764 / 1718\n",
      "24448/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9911Step 765 / 1718\n",
      "24480/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9911Step 766 / 1718\n",
      "24512/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9911Step 767 / 1718\n",
      "24544/55000 - mean: 0.0338 - sparse_categorical_accuracy: 0.9911Step 768 / 1718\n",
      "24576/55000 - mean: 0.0338 - sparse_categorical_accuracy: 0.9910Step 769 / 1718\n",
      "24608/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9911Step 770 / 1718\n",
      "24640/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9911Step 771 / 1718\n",
      "24672/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9911Step 772 / 1718\n",
      "24704/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9911Step 773 / 1718\n",
      "24736/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9911Step 774 / 1718\n",
      "24768/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9911Step 775 / 1718\n",
      "24800/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9911Step 776 / 1718\n",
      "24832/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9911Step 777 / 1718\n",
      "24864/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9911Step 778 / 1718\n",
      "24896/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9911Step 779 / 1718\n",
      "24928/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9911Step 780 / 1718\n",
      "24960/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9911Step 781 / 1718\n",
      "24992/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9912Step 782 / 1718\n",
      "25024/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9912Step 783 / 1718\n",
      "25056/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9912Step 784 / 1718\n",
      "25088/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9912Step 785 / 1718\n",
      "25120/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9912Step 786 / 1718\n",
      "25152/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9911Step 787 / 1718\n",
      "25184/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9911Step 788 / 1718\n",
      "25216/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9912Step 789 / 1718\n",
      "25248/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9912Step 790 / 1718\n",
      "25280/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9912Step 791 / 1718\n",
      "25312/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9912Step 792 / 1718\n",
      "25344/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9912Step 793 / 1718\n",
      "25376/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9912Step 794 / 1718\n",
      "25408/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9912Step 795 / 1718\n",
      "25440/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9912Step 796 / 1718\n",
      "25472/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9912Step 797 / 1718\n",
      "25504/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9913Step 798 / 1718\n",
      "25536/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9913Step 799 / 1718\n",
      "25568/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9912Step 800 / 1718\n",
      "25600/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9912Step 801 / 1718\n",
      "25632/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9912Step 802 / 1718\n",
      "25664/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9912Step 803 / 1718\n",
      "25696/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9912Step 804 / 1718\n",
      "25728/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9912Step 805 / 1718\n",
      "25760/55000 - mean: 0.0330 - sparse_categorical_accuracy: 0.9912Step 806 / 1718\n",
      "25792/55000 - mean: 0.0329 - sparse_categorical_accuracy: 0.9912Step 807 / 1718\n",
      "25824/55000 - mean: 0.0329 - sparse_categorical_accuracy: 0.9912Step 808 / 1718\n",
      "25856/55000 - mean: 0.0328 - sparse_categorical_accuracy: 0.9913Step 809 / 1718\n",
      "25888/55000 - mean: 0.0328 - sparse_categorical_accuracy: 0.9913Step 810 / 1718\n",
      "25920/55000 - mean: 0.0328 - sparse_categorical_accuracy: 0.9912Step 811 / 1718\n",
      "25952/55000 - mean: 0.0328 - sparse_categorical_accuracy: 0.9913Step 812 / 1718\n",
      "25984/55000 - mean: 0.0328 - sparse_categorical_accuracy: 0.9913Step 813 / 1718\n",
      "26016/55000 - mean: 0.0327 - sparse_categorical_accuracy: 0.9913Step 814 / 1718\n",
      "26048/55000 - mean: 0.0327 - sparse_categorical_accuracy: 0.9913Step 815 / 1718\n",
      "26080/55000 - mean: 0.0326 - sparse_categorical_accuracy: 0.9913Step 816 / 1718\n",
      "26112/55000 - mean: 0.0326 - sparse_categorical_accuracy: 0.9913Step 817 / 1718\n",
      "26144/55000 - mean: 0.0326 - sparse_categorical_accuracy: 0.9913Step 818 / 1718\n",
      "26176/55000 - mean: 0.0325 - sparse_categorical_accuracy: 0.9913Step 819 / 1718\n",
      "26208/55000 - mean: 0.0325 - sparse_categorical_accuracy: 0.9913Step 820 / 1718\n",
      "26240/55000 - mean: 0.0324 - sparse_categorical_accuracy: 0.9913Step 821 / 1718\n",
      "26272/55000 - mean: 0.0324 - sparse_categorical_accuracy: 0.9914Step 822 / 1718\n",
      "26304/55000 - mean: 0.0324 - sparse_categorical_accuracy: 0.9914Step 823 / 1718\n",
      "26336/55000 - mean: 0.0323 - sparse_categorical_accuracy: 0.9914Step 824 / 1718\n",
      "26368/55000 - mean: 0.0323 - sparse_categorical_accuracy: 0.9914Step 825 / 1718\n",
      "26400/55000 - mean: 0.0323 - sparse_categorical_accuracy: 0.9914Step 826 / 1718\n",
      "26432/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9914Step 827 / 1718\n",
      "26464/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9914Step 828 / 1718\n",
      "26496/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9914Step 829 / 1718\n",
      "26528/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9914Step 830 / 1718\n",
      "26560/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9914Step 831 / 1718\n",
      "26592/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9914Step 832 / 1718\n",
      "26624/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9914Step 833 / 1718\n",
      "26656/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9914Step 834 / 1718\n",
      "26688/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9914Step 835 / 1718\n",
      "26720/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9914Step 836 / 1718\n",
      "26752/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9914Step 837 / 1718\n",
      "26784/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9914Step 838 / 1718\n",
      "26816/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9914Step 839 / 1718\n",
      "26848/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9914Step 840 / 1718\n",
      "26880/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9914Step 841 / 1718\n",
      "26912/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9915Step 842 / 1718\n",
      "26944/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9914Step 843 / 1718\n",
      "26976/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9914Step 844 / 1718\n",
      "27008/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9914Step 845 / 1718\n",
      "27040/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9914Step 846 / 1718\n",
      "27072/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9914Step 847 / 1718\n",
      "27104/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9914Step 848 / 1718\n",
      "27136/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9914Step 849 / 1718\n",
      "27168/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9914Step 850 / 1718\n",
      "27200/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9914Step 851 / 1718\n",
      "27232/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9914Step 852 / 1718\n",
      "27264/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9914Step 853 / 1718\n",
      "27296/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9914Step 854 / 1718\n",
      "27328/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9914Step 855 / 1718\n",
      "27360/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9913Step 856 / 1718\n",
      "27392/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9913Step 857 / 1718\n",
      "27424/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9913Step 858 / 1718\n",
      "27456/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9913Step 859 / 1718\n",
      "27488/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9912Step 860 / 1718\n",
      "27520/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9912Step 861 / 1718\n",
      "27552/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9912Step 862 / 1718\n",
      "27584/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9912Step 863 / 1718\n",
      "27616/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9912Step 864 / 1718\n",
      "27648/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9912Step 865 / 1718\n",
      "27680/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9913Step 866 / 1718\n",
      "27712/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9913Step 867 / 1718\n",
      "27744/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9913Step 868 / 1718\n",
      "27776/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9913Step 869 / 1718\n",
      "27808/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9912Step 870 / 1718\n",
      "27840/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9912Step 871 / 1718\n",
      "27872/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9912Step 872 / 1718\n",
      "27904/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9913Step 873 / 1718\n",
      "27936/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9913Step 874 / 1718\n",
      "27968/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9913Step 875 / 1718\n",
      "28000/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9913Step 876 / 1718\n",
      "28032/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9913Step 877 / 1718\n",
      "28064/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9913Step 878 / 1718\n",
      "28096/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9913Step 879 / 1718\n",
      "28128/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9913Step 880 / 1718\n",
      "28160/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9913Step 881 / 1718\n",
      "28192/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9913Step 882 / 1718\n",
      "28224/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9913Step 883 / 1718\n",
      "28256/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9913Step 884 / 1718\n",
      "28288/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9913Step 885 / 1718\n",
      "28320/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9913Step 886 / 1718\n",
      "28352/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9913Step 887 / 1718\n",
      "28384/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9913Step 888 / 1718\n",
      "28416/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9913Step 889 / 1718\n",
      "28448/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9912Step 890 / 1718\n",
      "28480/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9912Step 891 / 1718\n",
      "28512/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9912Step 892 / 1718\n",
      "28544/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9912Step 893 / 1718\n",
      "28576/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9912Step 894 / 1718\n",
      "28608/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9912Step 895 / 1718\n",
      "28640/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9912Step 896 / 1718\n",
      "28672/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9912Step 897 / 1718\n",
      "28704/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9912Step 898 / 1718\n",
      "28736/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9912Step 899 / 1718\n",
      "28768/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9912Step 900 / 1718\n",
      "28800/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9912Step 901 / 1718\n",
      "28832/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9912Step 902 / 1718\n",
      "28864/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9912Step 903 / 1718\n",
      "28896/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9912Step 904 / 1718\n",
      "28928/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9913Step 905 / 1718\n",
      "28960/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9913Step 906 / 1718\n",
      "28992/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9913Step 907 / 1718\n",
      "29024/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9913Step 908 / 1718\n",
      "29056/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9913Step 909 / 1718\n",
      "29088/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9913Step 910 / 1718\n",
      "29120/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9913Step 911 / 1718\n",
      "29152/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9913Step 912 / 1718\n",
      "29184/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9913Step 913 / 1718\n",
      "29216/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9913Step 914 / 1718\n",
      "29248/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9913Step 915 / 1718\n",
      "29280/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9913Step 916 / 1718\n",
      "29312/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9913Step 917 / 1718\n",
      "29344/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9913Step 918 / 1718\n",
      "29376/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9913Step 919 / 1718\n",
      "29408/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9913Step 920 / 1718\n",
      "29440/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9913Step 921 / 1718\n",
      "29472/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9913Step 922 / 1718\n",
      "29504/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9913Step 923 / 1718\n",
      "29536/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9913Step 924 / 1718\n",
      "29568/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9913Step 925 / 1718\n",
      "29600/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9914Step 926 / 1718\n",
      "29632/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9914Step 927 / 1718\n",
      "29664/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9914Step 928 / 1718\n",
      "29696/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9914Step 929 / 1718\n",
      "29728/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9914Step 930 / 1718\n",
      "29760/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9914Step 931 / 1718\n",
      "29792/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9913Step 932 / 1718\n",
      "29824/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9913Step 933 / 1718\n",
      "29856/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9913Step 934 / 1718\n",
      "29888/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9912Step 935 / 1718\n",
      "29920/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9912Step 936 / 1718\n",
      "29952/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9912Step 937 / 1718\n",
      "29984/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9912Step 938 / 1718\n",
      "30016/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9912Step 939 / 1718\n",
      "30048/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9912Step 940 / 1718\n",
      "30080/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9913Step 941 / 1718\n",
      "30112/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9913Step 942 / 1718\n",
      "30144/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9913Step 943 / 1718\n",
      "30176/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9913Step 944 / 1718\n",
      "30208/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9913Step 945 / 1718\n",
      "30240/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9913Step 946 / 1718\n",
      "30272/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9913Step 947 / 1718\n",
      "30304/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9913Step 948 / 1718\n",
      "30336/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9913Step 949 / 1718\n",
      "30368/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9913Step 950 / 1718\n",
      "30400/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9913Step 951 / 1718\n",
      "30432/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9913Step 952 / 1718\n",
      "30464/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9913Step 953 / 1718\n",
      "30496/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9913Step 954 / 1718\n",
      "30528/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9913Step 955 / 1718\n",
      "30560/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9913Step 956 / 1718\n",
      "30592/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9913Step 957 / 1718\n",
      "30624/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9913Step 958 / 1718\n",
      "30656/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9913Step 959 / 1718\n",
      "30688/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9913Step 960 / 1718\n",
      "30720/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9913Step 961 / 1718\n",
      "30752/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9914Step 962 / 1718\n",
      "30784/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9914Step 963 / 1718\n",
      "30816/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9914Step 964 / 1718\n",
      "30848/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9914Step 965 / 1718\n",
      "30880/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9914Step 966 / 1718\n",
      "30912/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9914Step 967 / 1718\n",
      "30944/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9914Step 968 / 1718\n",
      "30976/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9914Step 969 / 1718\n",
      "31008/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9914Step 970 / 1718\n",
      "31040/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9914Step 971 / 1718\n",
      "31072/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9914Step 972 / 1718\n",
      "31104/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9914Step 973 / 1718\n",
      "31136/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9914Step 974 / 1718\n",
      "31168/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9914Step 975 / 1718\n",
      "31200/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9914Step 976 / 1718\n",
      "31232/55000 - mean: 0.0313 - sparse_categorical_accuracy: 0.9914Step 977 / 1718\n",
      "31264/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9914Step 978 / 1718\n",
      "31296/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9914Step 979 / 1718\n",
      "31328/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9914Step 980 / 1718\n",
      "31360/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9914Step 981 / 1718\n",
      "31392/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9914Step 982 / 1718\n",
      "31424/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9914Step 983 / 1718\n",
      "31456/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9914Step 984 / 1718\n",
      "31488/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9914Step 985 / 1718\n",
      "31520/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9914Step 986 / 1718\n",
      "31552/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9914Step 987 / 1718\n",
      "31584/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9914Step 988 / 1718\n",
      "31616/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9914Step 989 / 1718\n",
      "31648/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9914Step 990 / 1718\n",
      "31680/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9914Step 991 / 1718\n",
      "31712/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9914Step 992 / 1718\n",
      "31744/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9914Step 993 / 1718\n",
      "31776/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9914Step 994 / 1718\n",
      "31808/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9914Step 995 / 1718\n",
      "31840/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9914Step 996 / 1718\n",
      "31872/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9914Step 997 / 1718\n",
      "31904/55000 - mean: 0.0311 - sparse_categorical_accuracy: 0.9914Step 998 / 1718\n",
      "31936/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9914Step 999 / 1718\n",
      "31968/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9914Step 1000 / 1718\n",
      "32000/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9914Step 1001 / 1718\n",
      "32032/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9914Step 1002 / 1718\n",
      "32064/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9914Step 1003 / 1718\n",
      "32096/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9914Step 1004 / 1718\n",
      "32128/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9914Step 1005 / 1718\n",
      "32160/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9914Step 1006 / 1718\n",
      "32192/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9914Step 1007 / 1718\n",
      "32224/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9914Step 1008 / 1718\n",
      "32256/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9914Step 1009 / 1718\n",
      "32288/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9915Step 1010 / 1718\n",
      "32320/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9915Step 1011 / 1718\n",
      "32352/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9914Step 1012 / 1718\n",
      "32384/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9914Step 1013 / 1718\n",
      "32416/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9915Step 1014 / 1718\n",
      "32448/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9914Step 1015 / 1718\n",
      "32480/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9914Step 1016 / 1718\n",
      "32512/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9914Step 1017 / 1718\n",
      "32544/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9914Step 1018 / 1718\n",
      "32576/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9914Step 1019 / 1718\n",
      "32608/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9914Step 1020 / 1718\n",
      "32640/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9915Step 1021 / 1718\n",
      "32672/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9914Step 1022 / 1718\n",
      "32704/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9914Step 1023 / 1718\n",
      "32736/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9914Step 1024 / 1718\n",
      "32768/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9915Step 1025 / 1718\n",
      "32800/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9915Step 1026 / 1718\n",
      "32832/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9915Step 1027 / 1718\n",
      "32864/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9914Step 1028 / 1718\n",
      "32896/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9915Step 1029 / 1718\n",
      "32928/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9915Step 1030 / 1718\n",
      "32960/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1031 / 1718\n",
      "32992/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1032 / 1718\n",
      "33024/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1033 / 1718\n",
      "33056/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1034 / 1718\n",
      "33088/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1035 / 1718\n",
      "33120/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9915Step 1036 / 1718\n",
      "33152/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9915Step 1037 / 1718\n",
      "33184/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9915Step 1038 / 1718\n",
      "33216/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9915Step 1039 / 1718\n",
      "33248/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9915Step 1040 / 1718\n",
      "33280/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9915Step 1041 / 1718\n",
      "33312/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9915Step 1042 / 1718\n",
      "33344/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9915Step 1043 / 1718\n",
      "33376/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1044 / 1718\n",
      "33408/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1045 / 1718\n",
      "33440/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9915Step 1046 / 1718\n",
      "33472/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9915Step 1047 / 1718\n",
      "33504/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1048 / 1718\n",
      "33536/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1049 / 1718\n",
      "33568/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9916Step 1050 / 1718\n",
      "33600/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9916Step 1051 / 1718\n",
      "33632/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9916Step 1052 / 1718\n",
      "33664/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9916Step 1053 / 1718\n",
      "33696/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1054 / 1718\n",
      "33728/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1055 / 1718\n",
      "33760/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1056 / 1718\n",
      "33792/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9915Step 1057 / 1718\n",
      "33824/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9915Step 1058 / 1718\n",
      "33856/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1059 / 1718\n",
      "33888/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1060 / 1718\n",
      "33920/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1061 / 1718\n",
      "33952/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9915Step 1062 / 1718\n",
      "33984/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9915Step 1063 / 1718\n",
      "34016/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1064 / 1718\n",
      "34048/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1065 / 1718\n",
      "34080/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1066 / 1718\n",
      "34112/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1067 / 1718\n",
      "34144/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1068 / 1718\n",
      "34176/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1069 / 1718\n",
      "34208/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9915Step 1070 / 1718\n",
      "34240/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9915Step 1071 / 1718\n",
      "34272/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9915Step 1072 / 1718\n",
      "34304/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1073 / 1718\n",
      "34336/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1074 / 1718\n",
      "34368/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1075 / 1718\n",
      "34400/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9915Step 1076 / 1718\n",
      "34432/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9915Step 1077 / 1718\n",
      "34464/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9915Step 1078 / 1718\n",
      "34496/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9914Step 1079 / 1718\n",
      "34528/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1080 / 1718\n",
      "34560/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1081 / 1718\n",
      "34592/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1082 / 1718\n",
      "34624/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1083 / 1718\n",
      "34656/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1084 / 1718\n",
      "34688/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9915Step 1085 / 1718\n",
      "34720/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9915Step 1086 / 1718\n",
      "34752/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9915Step 1087 / 1718\n",
      "34784/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9915Step 1088 / 1718\n",
      "34816/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9915Step 1089 / 1718\n",
      "34848/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9915Step 1090 / 1718\n",
      "34880/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9915Step 1091 / 1718\n",
      "34912/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9915Step 1092 / 1718\n",
      "34944/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9915Step 1093 / 1718\n",
      "34976/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9915Step 1094 / 1718\n",
      "35008/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9915Step 1095 / 1718\n",
      "35040/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1096 / 1718\n",
      "35072/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1097 / 1718\n",
      "35104/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1098 / 1718\n",
      "35136/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9916Step 1099 / 1718\n",
      "35168/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9916Step 1100 / 1718\n",
      "35200/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9916Step 1101 / 1718\n",
      "35232/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9916Step 1102 / 1718\n",
      "35264/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9916Step 1103 / 1718\n",
      "35296/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9916Step 1104 / 1718\n",
      "35328/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9916Step 1105 / 1718\n",
      "35360/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9916Step 1106 / 1718\n",
      "35392/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9916Step 1107 / 1718\n",
      "35424/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9916Step 1108 / 1718\n",
      "35456/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9916Step 1109 / 1718\n",
      "35488/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9916Step 1110 / 1718\n",
      "35520/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9916Step 1111 / 1718\n",
      "35552/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9916Step 1112 / 1718\n",
      "35584/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9916Step 1113 / 1718\n",
      "35616/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9916Step 1114 / 1718\n",
      "35648/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9916Step 1115 / 1718\n",
      "35680/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9916Step 1116 / 1718\n",
      "35712/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9916Step 1117 / 1718\n",
      "35744/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9916Step 1118 / 1718\n",
      "35776/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9916Step 1119 / 1718\n",
      "35808/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1120 / 1718\n",
      "35840/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1121 / 1718\n",
      "35872/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1122 / 1718\n",
      "35904/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1123 / 1718\n",
      "35936/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9917Step 1124 / 1718\n",
      "35968/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1125 / 1718\n",
      "36000/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1126 / 1718\n",
      "36032/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1127 / 1718\n",
      "36064/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1128 / 1718\n",
      "36096/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1129 / 1718\n",
      "36128/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1130 / 1718\n",
      "36160/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1131 / 1718\n",
      "36192/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1132 / 1718\n",
      "36224/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9917Step 1133 / 1718\n",
      "36256/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9917Step 1134 / 1718\n",
      "36288/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9917Step 1135 / 1718\n",
      "36320/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9917Step 1136 / 1718\n",
      "36352/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9916Step 1137 / 1718\n",
      "36384/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9916Step 1138 / 1718\n",
      "36416/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9916Step 1139 / 1718\n",
      "36448/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9916Step 1140 / 1718\n",
      "36480/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9916Step 1141 / 1718\n",
      "36512/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9916Step 1142 / 1718\n",
      "36544/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1143 / 1718\n",
      "36576/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1144 / 1718\n",
      "36608/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1145 / 1718\n",
      "36640/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1146 / 1718\n",
      "36672/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1147 / 1718\n",
      "36704/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1148 / 1718\n",
      "36736/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1149 / 1718\n",
      "36768/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1150 / 1718\n",
      "36800/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1151 / 1718\n",
      "36832/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1152 / 1718\n",
      "36864/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1153 / 1718\n",
      "36896/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1154 / 1718\n",
      "36928/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1155 / 1718\n",
      "36960/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1156 / 1718\n",
      "36992/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1157 / 1718\n",
      "37024/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1158 / 1718\n",
      "37056/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1159 / 1718\n",
      "37088/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1160 / 1718\n",
      "37120/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9917Step 1161 / 1718\n",
      "37152/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9917Step 1162 / 1718\n",
      "37184/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1163 / 1718\n",
      "37216/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1164 / 1718\n",
      "37248/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1165 / 1718\n",
      "37280/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1166 / 1718\n",
      "37312/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1167 / 1718\n",
      "37344/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9918Step 1168 / 1718\n",
      "37376/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1169 / 1718\n",
      "37408/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1170 / 1718\n",
      "37440/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9917Step 1171 / 1718\n",
      "37472/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1172 / 1718\n",
      "37504/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1173 / 1718\n",
      "37536/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1174 / 1718\n",
      "37568/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9917Step 1175 / 1718\n",
      "37600/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9917Step 1176 / 1718\n",
      "37632/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1177 / 1718\n",
      "37664/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1178 / 1718\n",
      "37696/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9916Step 1179 / 1718\n",
      "37728/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1180 / 1718\n",
      "37760/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1181 / 1718\n",
      "37792/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1182 / 1718\n",
      "37824/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9916Step 1183 / 1718\n",
      "37856/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9917Step 1184 / 1718\n",
      "37888/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1185 / 1718\n",
      "37920/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1186 / 1718\n",
      "37952/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9916Step 1187 / 1718\n",
      "37984/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9916Step 1188 / 1718\n",
      "38016/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9916Step 1189 / 1718\n",
      "38048/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9916Step 1190 / 1718\n",
      "38080/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9916Step 1191 / 1718\n",
      "38112/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9916Step 1192 / 1718\n",
      "38144/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9916Step 1193 / 1718\n",
      "38176/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9916Step 1194 / 1718\n",
      "38208/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1195 / 1718\n",
      "38240/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1196 / 1718\n",
      "38272/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9917Step 1197 / 1718\n",
      "38304/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9917Step 1198 / 1718\n",
      "38336/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9917Step 1199 / 1718\n",
      "38368/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9917Step 1200 / 1718\n",
      "38400/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9917Step 1201 / 1718\n",
      "38432/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9917Step 1202 / 1718\n",
      "38464/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9917Step 1203 / 1718\n",
      "38496/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9917Step 1204 / 1718\n",
      "38528/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9917Step 1205 / 1718\n",
      "38560/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9917Step 1206 / 1718\n",
      "38592/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9917Step 1207 / 1718\n",
      "38624/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9917Step 1208 / 1718\n",
      "38656/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9917Step 1209 / 1718\n",
      "38688/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1210 / 1718\n",
      "38720/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9917Step 1211 / 1718\n",
      "38752/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9917Step 1212 / 1718\n",
      "38784/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9917Step 1213 / 1718\n",
      "38816/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9917Step 1214 / 1718\n",
      "38848/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9917Step 1215 / 1718\n",
      "38880/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9917Step 1216 / 1718\n",
      "38912/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9917Step 1217 / 1718\n",
      "38944/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9917Step 1218 / 1718\n",
      "38976/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9917Step 1219 / 1718\n",
      "39008/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9917Step 1220 / 1718\n",
      "39040/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9917Step 1221 / 1718\n",
      "39072/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9917Step 1222 / 1718\n",
      "39104/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9917Step 1223 / 1718\n",
      "39136/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9917Step 1224 / 1718\n",
      "39168/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9917Step 1225 / 1718\n",
      "39200/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9917Step 1226 / 1718\n",
      "39232/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9917Step 1227 / 1718\n",
      "39264/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9917Step 1228 / 1718\n",
      "39296/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9917Step 1229 / 1718\n",
      "39328/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9917Step 1230 / 1718\n",
      "39360/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9917Step 1231 / 1718\n",
      "39392/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9917Step 1232 / 1718\n",
      "39424/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9917Step 1233 / 1718\n",
      "39456/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9917Step 1234 / 1718\n",
      "39488/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9917Step 1235 / 1718\n",
      "39520/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9917Step 1236 / 1718\n",
      "39552/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9917Step 1237 / 1718\n",
      "39584/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9917Step 1238 / 1718\n",
      "39616/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9917Step 1239 / 1718\n",
      "39648/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9918Step 1240 / 1718\n",
      "39680/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9918Step 1241 / 1718\n",
      "39712/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9918Step 1242 / 1718\n",
      "39744/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9918Step 1243 / 1718\n",
      "39776/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9918Step 1244 / 1718\n",
      "39808/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9918Step 1245 / 1718\n",
      "39840/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9918Step 1246 / 1718\n",
      "39872/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9918Step 1247 / 1718\n",
      "39904/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9918Step 1248 / 1718\n",
      "39936/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9918Step 1249 / 1718\n",
      "39968/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9918Step 1250 / 1718\n",
      "40000/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9918Step 1251 / 1718\n",
      "40032/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9918Step 1252 / 1718\n",
      "40064/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9918Step 1253 / 1718\n",
      "40096/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9918Step 1254 / 1718\n",
      "40128/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9918Step 1255 / 1718\n",
      "40160/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9918Step 1256 / 1718\n",
      "40192/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9918Step 1257 / 1718\n",
      "40224/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9918Step 1258 / 1718\n",
      "40256/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9918Step 1259 / 1718\n",
      "40288/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9918Step 1260 / 1718\n",
      "40320/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9918Step 1261 / 1718\n",
      "40352/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9918Step 1262 / 1718\n",
      "40384/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9918Step 1263 / 1718\n",
      "40416/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9918Step 1264 / 1718\n",
      "40448/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9918Step 1265 / 1718\n",
      "40480/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9918Step 1266 / 1718\n",
      "40512/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9919Step 1267 / 1718\n",
      "40544/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9919Step 1268 / 1718\n",
      "40576/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9919Step 1269 / 1718\n",
      "40608/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9919Step 1270 / 1718\n",
      "40640/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1271 / 1718\n",
      "40672/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1272 / 1718\n",
      "40704/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1273 / 1718\n",
      "40736/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1274 / 1718\n",
      "40768/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1275 / 1718\n",
      "40800/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1276 / 1718\n",
      "40832/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1277 / 1718\n",
      "40864/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 1278 / 1718\n",
      "40896/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 1279 / 1718\n",
      "40928/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 1280 / 1718\n",
      "40960/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 1281 / 1718\n",
      "40992/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1282 / 1718\n",
      "41024/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1283 / 1718\n",
      "41056/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1284 / 1718\n",
      "41088/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 1285 / 1718\n",
      "41120/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 1286 / 1718\n",
      "41152/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 1287 / 1718\n",
      "41184/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 1288 / 1718\n",
      "41216/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9919Step 1289 / 1718\n",
      "41248/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1290 / 1718\n",
      "41280/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1291 / 1718\n",
      "41312/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9919Step 1292 / 1718\n",
      "41344/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9919Step 1293 / 1718\n",
      "41376/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 1294 / 1718\n",
      "41408/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 1295 / 1718\n",
      "41440/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 1296 / 1718\n",
      "41472/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9919Step 1297 / 1718\n",
      "41504/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1298 / 1718\n",
      "41536/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1299 / 1718\n",
      "41568/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1300 / 1718\n",
      "41600/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1301 / 1718\n",
      "41632/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1302 / 1718\n",
      "41664/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1303 / 1718\n",
      "41696/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1304 / 1718\n",
      "41728/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1305 / 1718\n",
      "41760/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1306 / 1718\n",
      "41792/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9920Step 1307 / 1718\n",
      "41824/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9920Step 1308 / 1718\n",
      "41856/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9920Step 1309 / 1718\n",
      "41888/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1310 / 1718\n",
      "41920/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1311 / 1718\n",
      "41952/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1312 / 1718\n",
      "41984/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1313 / 1718\n",
      "42016/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1314 / 1718\n",
      "42048/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1315 / 1718\n",
      "42080/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1316 / 1718\n",
      "42112/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1317 / 1718\n",
      "42144/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1318 / 1718\n",
      "42176/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1319 / 1718\n",
      "42208/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1320 / 1718\n",
      "42240/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1321 / 1718\n",
      "42272/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1322 / 1718\n",
      "42304/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1323 / 1718\n",
      "42336/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1324 / 1718\n",
      "42368/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1325 / 1718\n",
      "42400/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1326 / 1718\n",
      "42432/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1327 / 1718\n",
      "42464/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1328 / 1718\n",
      "42496/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1329 / 1718\n",
      "42528/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1330 / 1718\n",
      "42560/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1331 / 1718\n",
      "42592/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1332 / 1718\n",
      "42624/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1333 / 1718\n",
      "42656/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1334 / 1718\n",
      "42688/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9920Step 1335 / 1718\n",
      "42720/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9920Step 1336 / 1718\n",
      "42752/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9920Step 1337 / 1718\n",
      "42784/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9920Step 1338 / 1718\n",
      "42816/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1339 / 1718\n",
      "42848/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1340 / 1718\n",
      "42880/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1341 / 1718\n",
      "42912/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1342 / 1718\n",
      "42944/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1343 / 1718\n",
      "42976/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1344 / 1718\n",
      "43008/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1345 / 1718\n",
      "43040/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1346 / 1718\n",
      "43072/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1347 / 1718\n",
      "43104/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1348 / 1718\n",
      "43136/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1349 / 1718\n",
      "43168/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1350 / 1718\n",
      "43200/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1351 / 1718\n",
      "43232/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1352 / 1718\n",
      "43264/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1353 / 1718\n",
      "43296/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1354 / 1718\n",
      "43328/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1355 / 1718\n",
      "43360/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1356 / 1718\n",
      "43392/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1357 / 1718\n",
      "43424/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9920Step 1358 / 1718\n",
      "43456/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9920Step 1359 / 1718\n",
      "43488/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9920Step 1360 / 1718\n",
      "43520/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9920Step 1361 / 1718\n",
      "43552/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1362 / 1718\n",
      "43584/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1363 / 1718\n",
      "43616/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9920Step 1364 / 1718\n",
      "43648/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9920Step 1365 / 1718\n",
      "43680/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9920Step 1366 / 1718\n",
      "43712/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9920Step 1367 / 1718\n",
      "43744/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9920Step 1368 / 1718\n",
      "43776/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9920Step 1369 / 1718\n",
      "43808/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9920Step 1370 / 1718\n",
      "43840/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9920Step 1371 / 1718\n",
      "43872/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9920Step 1372 / 1718\n",
      "43904/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9920Step 1373 / 1718\n",
      "43936/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9920Step 1374 / 1718\n",
      "43968/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9920Step 1375 / 1718\n",
      "44000/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9920Step 1376 / 1718\n",
      "44032/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9920Step 1377 / 1718\n",
      "44064/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1378 / 1718\n",
      "44096/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1379 / 1718\n",
      "44128/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1380 / 1718\n",
      "44160/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1381 / 1718\n",
      "44192/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1382 / 1718\n",
      "44224/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1383 / 1718\n",
      "44256/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1384 / 1718\n",
      "44288/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1385 / 1718\n",
      "44320/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9920Step 1386 / 1718\n",
      "44352/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9920Step 1387 / 1718\n",
      "44384/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9920Step 1388 / 1718\n",
      "44416/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1389 / 1718\n",
      "44448/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1390 / 1718\n",
      "44480/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1391 / 1718\n",
      "44512/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9920Step 1392 / 1718\n",
      "44544/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9920Step 1393 / 1718\n",
      "44576/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9920Step 1394 / 1718\n",
      "44608/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9920Step 1395 / 1718\n",
      "44640/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9920Step 1396 / 1718\n",
      "44672/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9920Step 1397 / 1718\n",
      "44704/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9920Step 1398 / 1718\n",
      "44736/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9920Step 1399 / 1718\n",
      "44768/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9920Step 1400 / 1718\n",
      "44800/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9920Step 1401 / 1718\n",
      "44832/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 1402 / 1718\n",
      "44864/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 1403 / 1718\n",
      "44896/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1404 / 1718\n",
      "44928/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1405 / 1718\n",
      "44960/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1406 / 1718\n",
      "44992/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1407 / 1718\n",
      "45024/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1408 / 1718\n",
      "45056/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1409 / 1718\n",
      "45088/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1410 / 1718\n",
      "45120/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 1411 / 1718\n",
      "45152/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 1412 / 1718\n",
      "45184/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 1413 / 1718\n",
      "45216/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1414 / 1718\n",
      "45248/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9919Step 1415 / 1718\n",
      "45280/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9919Step 1416 / 1718\n",
      "45312/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9919Step 1417 / 1718\n",
      "45344/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9919Step 1418 / 1718\n",
      "45376/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9919Step 1419 / 1718\n",
      "45408/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9919Step 1420 / 1718\n",
      "45440/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9919Step 1421 / 1718\n",
      "45472/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9919Step 1422 / 1718\n",
      "45504/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9919Step 1423 / 1718\n",
      "45536/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9919Step 1424 / 1718\n",
      "45568/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9919Step 1425 / 1718\n",
      "45600/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9918Step 1426 / 1718\n",
      "45632/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9918Step 1427 / 1718\n",
      "45664/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9919Step 1428 / 1718\n",
      "45696/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9919Step 1429 / 1718\n",
      "45728/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9919Step 1430 / 1718\n",
      "45760/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9919Step 1431 / 1718\n",
      "45792/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9919Step 1432 / 1718\n",
      "45824/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9919Step 1433 / 1718\n",
      "45856/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9919Step 1434 / 1718\n",
      "45888/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9919Step 1435 / 1718\n",
      "45920/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9919Step 1436 / 1718\n",
      "45952/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9918Step 1437 / 1718\n",
      "45984/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9918Step 1438 / 1718\n",
      "46016/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9918Step 1439 / 1718\n",
      "46048/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9918Step 1440 / 1718\n",
      "46080/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9918Step 1441 / 1718\n",
      "46112/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9918Step 1442 / 1718\n",
      "46144/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1443 / 1718\n",
      "46176/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1444 / 1718\n",
      "46208/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9919Step 1445 / 1718\n",
      "46240/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9919Step 1446 / 1718\n",
      "46272/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1447 / 1718\n",
      "46304/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1448 / 1718\n",
      "46336/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1449 / 1718\n",
      "46368/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9919Step 1450 / 1718\n",
      "46400/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1451 / 1718\n",
      "46432/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1452 / 1718\n",
      "46464/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1453 / 1718\n",
      "46496/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1454 / 1718\n",
      "46528/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9919Step 1455 / 1718\n",
      "46560/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1456 / 1718\n",
      "46592/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1457 / 1718\n",
      "46624/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1458 / 1718\n",
      "46656/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1459 / 1718\n",
      "46688/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1460 / 1718\n",
      "46720/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1461 / 1718\n",
      "46752/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1462 / 1718\n",
      "46784/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1463 / 1718\n",
      "46816/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1464 / 1718\n",
      "46848/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1465 / 1718\n",
      "46880/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1466 / 1718\n",
      "46912/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9919Step 1467 / 1718\n",
      "46944/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9919Step 1468 / 1718\n",
      "46976/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9919Step 1469 / 1718\n",
      "47008/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9919Step 1470 / 1718\n",
      "47040/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9919Step 1471 / 1718\n",
      "47072/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9919Step 1472 / 1718\n",
      "47104/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9919Step 1473 / 1718\n",
      "47136/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9919Step 1474 / 1718\n",
      "47168/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9919Step 1475 / 1718\n",
      "47200/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9919Step 1476 / 1718\n",
      "47232/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9920Step 1477 / 1718\n",
      "47264/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9920Step 1478 / 1718\n",
      "47296/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9920Step 1479 / 1718\n",
      "47328/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9919Step 1480 / 1718\n",
      "47360/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9920Step 1481 / 1718\n",
      "47392/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9920Step 1482 / 1718\n",
      "47424/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9920Step 1483 / 1718\n",
      "47456/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9920Step 1484 / 1718\n",
      "47488/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9920Step 1485 / 1718\n",
      "47520/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9920Step 1486 / 1718\n",
      "47552/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9920Step 1487 / 1718\n",
      "47584/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9920Step 1488 / 1718\n",
      "47616/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9920Step 1489 / 1718\n",
      "47648/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9920Step 1490 / 1718\n",
      "47680/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9920Step 1491 / 1718\n",
      "47712/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9920Step 1492 / 1718\n",
      "47744/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9920Step 1493 / 1718\n",
      "47776/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9920Step 1494 / 1718\n",
      "47808/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9920Step 1495 / 1718\n",
      "47840/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9920Step 1496 / 1718\n",
      "47872/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9920Step 1497 / 1718\n",
      "47904/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9920Step 1498 / 1718\n",
      "47936/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9921Step 1499 / 1718\n",
      "47968/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9921Step 1500 / 1718\n",
      "48000/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1501 / 1718\n",
      "48032/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1502 / 1718\n",
      "48064/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1503 / 1718\n",
      "48096/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1504 / 1718\n",
      "48128/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1505 / 1718\n",
      "48160/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1506 / 1718\n",
      "48192/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1507 / 1718\n",
      "48224/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1508 / 1718\n",
      "48256/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1509 / 1718\n",
      "48288/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1510 / 1718\n",
      "48320/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9921Step 1511 / 1718\n",
      "48352/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9921Step 1512 / 1718\n",
      "48384/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9921Step 1513 / 1718\n",
      "48416/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9921Step 1514 / 1718\n",
      "48448/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9921Step 1515 / 1718\n",
      "48480/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9921Step 1516 / 1718\n",
      "48512/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9921Step 1517 / 1718\n",
      "48544/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9921Step 1518 / 1718\n",
      "48576/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9921Step 1519 / 1718\n",
      "48608/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9921Step 1520 / 1718\n",
      "48640/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9921Step 1521 / 1718\n",
      "48672/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9921Step 1522 / 1718\n",
      "48704/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9921Step 1523 / 1718\n",
      "48736/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9921Step 1524 / 1718\n",
      "48768/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9921Step 1525 / 1718\n",
      "48800/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9921Step 1526 / 1718\n",
      "48832/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9921Step 1527 / 1718\n",
      "48864/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9921Step 1528 / 1718\n",
      "48896/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9921Step 1529 / 1718\n",
      "48928/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9922Step 1530 / 1718\n",
      "48960/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9922Step 1531 / 1718\n",
      "48992/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9922Step 1532 / 1718\n",
      "49024/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9921Step 1533 / 1718\n",
      "49056/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9922Step 1534 / 1718\n",
      "49088/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9922Step 1535 / 1718\n",
      "49120/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9922Step 1536 / 1718\n",
      "49152/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9922Step 1537 / 1718\n",
      "49184/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9922Step 1538 / 1718\n",
      "49216/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9922Step 1539 / 1718\n",
      "49248/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9922Step 1540 / 1718\n",
      "49280/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9922Step 1541 / 1718\n",
      "49312/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9922Step 1542 / 1718\n",
      "49344/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9922Step 1543 / 1718\n",
      "49376/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9922Step 1544 / 1718\n",
      "49408/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9922Step 1545 / 1718\n",
      "49440/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9922Step 1546 / 1718\n",
      "49472/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9922Step 1547 / 1718\n",
      "49504/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9921Step 1548 / 1718\n",
      "49536/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9921Step 1549 / 1718\n",
      "49568/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9922Step 1550 / 1718\n",
      "49600/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1551 / 1718\n",
      "49632/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1552 / 1718\n",
      "49664/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1553 / 1718\n",
      "49696/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9921Step 1554 / 1718\n",
      "49728/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1555 / 1718\n",
      "49760/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1556 / 1718\n",
      "49792/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1557 / 1718\n",
      "49824/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1558 / 1718\n",
      "49856/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1559 / 1718\n",
      "49888/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1560 / 1718\n",
      "49920/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1561 / 1718\n",
      "49952/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1562 / 1718\n",
      "49984/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1563 / 1718\n",
      "50016/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1564 / 1718\n",
      "50048/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9920Step 1565 / 1718\n",
      "50080/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1566 / 1718\n",
      "50112/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1567 / 1718\n",
      "50144/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1568 / 1718\n",
      "50176/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1569 / 1718\n",
      "50208/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1570 / 1718\n",
      "50240/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1571 / 1718\n",
      "50272/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1572 / 1718\n",
      "50304/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1573 / 1718\n",
      "50336/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9921Step 1574 / 1718\n",
      "50368/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1575 / 1718\n",
      "50400/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1576 / 1718\n",
      "50432/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1577 / 1718\n",
      "50464/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1578 / 1718\n",
      "50496/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1579 / 1718\n",
      "50528/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1580 / 1718\n",
      "50560/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1581 / 1718\n",
      "50592/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1582 / 1718\n",
      "50624/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1583 / 1718\n",
      "50656/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1584 / 1718\n",
      "50688/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1585 / 1718\n",
      "50720/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1586 / 1718\n",
      "50752/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9920Step 1587 / 1718\n",
      "50784/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1588 / 1718\n",
      "50816/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1589 / 1718\n",
      "50848/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1590 / 1718\n",
      "50880/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1591 / 1718\n",
      "50912/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9920Step 1592 / 1718\n",
      "50944/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1593 / 1718\n",
      "50976/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1594 / 1718\n",
      "51008/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1595 / 1718\n",
      "51040/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1596 / 1718\n",
      "51072/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1597 / 1718\n",
      "51104/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1598 / 1718\n",
      "51136/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1599 / 1718\n",
      "51168/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1600 / 1718\n",
      "51200/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1601 / 1718\n",
      "51232/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9921Step 1602 / 1718\n",
      "51264/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1603 / 1718\n",
      "51296/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1604 / 1718\n",
      "51328/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1605 / 1718\n",
      "51360/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1606 / 1718\n",
      "51392/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1607 / 1718\n",
      "51424/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1608 / 1718\n",
      "51456/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1609 / 1718\n",
      "51488/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1610 / 1718\n",
      "51520/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1611 / 1718\n",
      "51552/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1612 / 1718\n",
      "51584/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1613 / 1718\n",
      "51616/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1614 / 1718\n",
      "51648/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1615 / 1718\n",
      "51680/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9920Step 1616 / 1718\n",
      "51712/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1617 / 1718\n",
      "51744/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1618 / 1718\n",
      "51776/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1619 / 1718\n",
      "51808/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1620 / 1718\n",
      "51840/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1621 / 1718\n",
      "51872/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1622 / 1718\n",
      "51904/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1623 / 1718\n",
      "51936/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1624 / 1718\n",
      "51968/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1625 / 1718\n",
      "52000/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1626 / 1718\n",
      "52032/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1627 / 1718\n",
      "52064/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1628 / 1718\n",
      "52096/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1629 / 1718\n",
      "52128/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1630 / 1718\n",
      "52160/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1631 / 1718\n",
      "52192/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1632 / 1718\n",
      "52224/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1633 / 1718\n",
      "52256/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1634 / 1718\n",
      "52288/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1635 / 1718\n",
      "52320/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1636 / 1718\n",
      "52352/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1637 / 1718\n",
      "52384/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1638 / 1718\n",
      "52416/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1639 / 1718\n",
      "52448/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1640 / 1718\n",
      "52480/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1641 / 1718\n",
      "52512/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1642 / 1718\n",
      "52544/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1643 / 1718\n",
      "52576/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1644 / 1718\n",
      "52608/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1645 / 1718\n",
      "52640/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1646 / 1718\n",
      "52672/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1647 / 1718\n",
      "52704/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1648 / 1718\n",
      "52736/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9921Step 1649 / 1718\n",
      "52768/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9921Step 1650 / 1718\n",
      "52800/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1651 / 1718\n",
      "52832/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1652 / 1718\n",
      "52864/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1653 / 1718\n",
      "52896/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1654 / 1718\n",
      "52928/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1655 / 1718\n",
      "52960/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1656 / 1718\n",
      "52992/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1657 / 1718\n",
      "53024/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1658 / 1718\n",
      "53056/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1659 / 1718\n",
      "53088/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1660 / 1718\n",
      "53120/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1661 / 1718\n",
      "53152/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1662 / 1718\n",
      "53184/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1663 / 1718\n",
      "53216/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1664 / 1718\n",
      "53248/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1665 / 1718\n",
      "53280/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1666 / 1718\n",
      "53312/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1667 / 1718\n",
      "53344/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1668 / 1718\n",
      "53376/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1669 / 1718\n",
      "53408/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9921Step 1670 / 1718\n",
      "53440/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1671 / 1718\n",
      "53472/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1672 / 1718\n",
      "53504/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1673 / 1718\n",
      "53536/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1674 / 1718\n",
      "53568/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1675 / 1718\n",
      "53600/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9921Step 1676 / 1718\n",
      "53632/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1677 / 1718\n",
      "53664/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1678 / 1718\n",
      "53696/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1679 / 1718\n",
      "53728/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1680 / 1718\n",
      "53760/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1681 / 1718\n",
      "53792/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1682 / 1718\n",
      "53824/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1683 / 1718\n",
      "53856/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1684 / 1718\n",
      "53888/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9922Step 1685 / 1718\n",
      "53920/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9922Step 1686 / 1718\n",
      "53952/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9922Step 1687 / 1718\n",
      "53984/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9922Step 1688 / 1718\n",
      "54016/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9922Step 1689 / 1718\n",
      "54048/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9922Step 1690 / 1718\n",
      "54080/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9922Step 1691 / 1718\n",
      "54112/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9922Step 1692 / 1718\n",
      "54144/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9922Step 1693 / 1718\n",
      "54176/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9922Step 1694 / 1718\n",
      "54208/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9922Step 1695 / 1718\n",
      "54240/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9922Step 1696 / 1718\n",
      "54272/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9922Step 1697 / 1718\n",
      "54304/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9922Step 1698 / 1718\n",
      "54336/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9922Step 1699 / 1718\n",
      "54368/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1700 / 1718\n",
      "54400/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1701 / 1718\n",
      "54432/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1702 / 1718\n",
      "54464/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1703 / 1718\n",
      "54496/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1704 / 1718\n",
      "54528/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1705 / 1718\n",
      "54560/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1706 / 1718\n",
      "54592/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1707 / 1718\n",
      "54624/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1708 / 1718\n",
      "54656/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9921Step 1709 / 1718\n",
      "54688/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9921Step 1710 / 1718\n",
      "54720/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9921Step 1711 / 1718\n",
      "54752/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1712 / 1718\n",
      "54784/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9921Step 1713 / 1718\n",
      "54816/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9921Step 1714 / 1718\n",
      "54848/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9921Step 1715 / 1718\n",
      "54880/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9921Step 1716 / 1718\n",
      "54912/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9921Step 1717 / 1718\n",
      "54944/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9921Step 1718 / 1718\n",
      "55000/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9921\n",
      "5000/5000 - mean: 0.1692 - sparse_categorical_accuracy: 0.9738\n",
      "Epoch 5 / 5\n",
      "Step 1 / 1718\n",
      "32/55000 - mean: 0.0001 - sparse_categorical_accuracy: 1.0000Step 2 / 1718\n",
      "64/55000 - mean: 0.0492 - sparse_categorical_accuracy: 0.9688Step 3 / 1718\n",
      "96/55000 - mean: 0.0387 - sparse_categorical_accuracy: 0.9792Step 4 / 1718\n",
      "128/55000 - mean: 0.1199 - sparse_categorical_accuracy: 0.9688Step 5 / 1718\n",
      "160/55000 - mean: 0.1018 - sparse_categorical_accuracy: 0.9688Step 6 / 1718\n",
      "192/55000 - mean: 0.0855 - sparse_categorical_accuracy: 0.9740Step 7 / 1718\n",
      "224/55000 - mean: 0.0899 - sparse_categorical_accuracy: 0.9732Step 8 / 1718\n",
      "256/55000 - mean: 0.0787 - sparse_categorical_accuracy: 0.9766Step 9 / 1718\n",
      "288/55000 - mean: 0.0701 - sparse_categorical_accuracy: 0.9792Step 10 / 1718\n",
      "320/55000 - mean: 0.0654 - sparse_categorical_accuracy: 0.9812Step 11 / 1718\n",
      "352/55000 - mean: 0.0722 - sparse_categorical_accuracy: 0.9801Step 12 / 1718\n",
      "384/55000 - mean: 0.0687 - sparse_categorical_accuracy: 0.9818Step 13 / 1718\n",
      "416/55000 - mean: 0.0704 - sparse_categorical_accuracy: 0.9808Step 14 / 1718\n",
      "448/55000 - mean: 0.0787 - sparse_categorical_accuracy: 0.9777Step 15 / 1718\n",
      "480/55000 - mean: 0.0781 - sparse_categorical_accuracy: 0.9771Step 16 / 1718\n",
      "512/55000 - mean: 0.0802 - sparse_categorical_accuracy: 0.9766Step 17 / 1718\n",
      "544/55000 - mean: 0.0756 - sparse_categorical_accuracy: 0.9779Step 18 / 1718\n",
      "576/55000 - mean: 0.0714 - sparse_categorical_accuracy: 0.9792Step 19 / 1718\n",
      "608/55000 - mean: 0.0677 - sparse_categorical_accuracy: 0.9803Step 20 / 1718\n",
      "640/55000 - mean: 0.0667 - sparse_categorical_accuracy: 0.9797Step 21 / 1718\n",
      "672/55000 - mean: 0.0686 - sparse_categorical_accuracy: 0.9792Step 22 / 1718\n",
      "704/55000 - mean: 0.0662 - sparse_categorical_accuracy: 0.9801Step 23 / 1718\n",
      "736/55000 - mean: 0.0635 - sparse_categorical_accuracy: 0.9810Step 24 / 1718\n",
      "768/55000 - mean: 0.0610 - sparse_categorical_accuracy: 0.9818Step 25 / 1718\n",
      "800/55000 - mean: 0.0586 - sparse_categorical_accuracy: 0.9825Step 26 / 1718\n",
      "832/55000 - mean: 0.0563 - sparse_categorical_accuracy: 0.9832Step 27 / 1718\n",
      "864/55000 - mean: 0.0544 - sparse_categorical_accuracy: 0.9838Step 28 / 1718\n",
      "896/55000 - mean: 0.0525 - sparse_categorical_accuracy: 0.9844Step 29 / 1718\n",
      "928/55000 - mean: 0.0509 - sparse_categorical_accuracy: 0.9849Step 30 / 1718\n",
      "960/55000 - mean: 0.0492 - sparse_categorical_accuracy: 0.9854Step 31 / 1718\n",
      "992/55000 - mean: 0.0476 - sparse_categorical_accuracy: 0.9859Step 32 / 1718\n",
      "1024/55000 - mean: 0.0461 - sparse_categorical_accuracy: 0.9863Step 33 / 1718\n",
      "1056/55000 - mean: 0.0448 - sparse_categorical_accuracy: 0.9867Step 34 / 1718\n",
      "1088/55000 - mean: 0.0435 - sparse_categorical_accuracy: 0.9871Step 35 / 1718\n",
      "1120/55000 - mean: 0.0423 - sparse_categorical_accuracy: 0.9875Step 36 / 1718\n",
      "1152/55000 - mean: 0.0412 - sparse_categorical_accuracy: 0.9878Step 37 / 1718\n",
      "1184/55000 - mean: 0.0438 - sparse_categorical_accuracy: 0.9873Step 38 / 1718\n",
      "1216/55000 - mean: 0.0432 - sparse_categorical_accuracy: 0.9877Step 39 / 1718\n",
      "1248/55000 - mean: 0.0421 - sparse_categorical_accuracy: 0.9880Step 40 / 1718\n",
      "1280/55000 - mean: 0.0493 - sparse_categorical_accuracy: 0.9875Step 41 / 1718\n",
      "1312/55000 - mean: 0.0489 - sparse_categorical_accuracy: 0.9870Step 42 / 1718\n",
      "1344/55000 - mean: 0.0478 - sparse_categorical_accuracy: 0.9874Step 43 / 1718\n",
      "1376/55000 - mean: 0.0493 - sparse_categorical_accuracy: 0.9869Step 44 / 1718\n",
      "1408/55000 - mean: 0.0482 - sparse_categorical_accuracy: 0.9872Step 45 / 1718\n",
      "1440/55000 - mean: 0.0471 - sparse_categorical_accuracy: 0.9875Step 46 / 1718\n",
      "1472/55000 - mean: 0.0465 - sparse_categorical_accuracy: 0.9878Step 47 / 1718\n",
      "1504/55000 - mean: 0.0456 - sparse_categorical_accuracy: 0.9880Step 48 / 1718\n",
      "1536/55000 - mean: 0.0446 - sparse_categorical_accuracy: 0.9883Step 49 / 1718\n",
      "1568/55000 - mean: 0.0437 - sparse_categorical_accuracy: 0.9885Step 50 / 1718\n",
      "1600/55000 - mean: 0.0429 - sparse_categorical_accuracy: 0.9887Step 51 / 1718\n",
      "1632/55000 - mean: 0.0420 - sparse_categorical_accuracy: 0.9890Step 52 / 1718\n",
      "1664/55000 - mean: 0.0417 - sparse_categorical_accuracy: 0.9886Step 53 / 1718\n",
      "1696/55000 - mean: 0.0410 - sparse_categorical_accuracy: 0.9888Step 54 / 1718\n",
      "1728/55000 - mean: 0.0405 - sparse_categorical_accuracy: 0.9890Step 55 / 1718\n",
      "1760/55000 - mean: 0.0398 - sparse_categorical_accuracy: 0.9892Step 56 / 1718\n",
      "1792/55000 - mean: 0.0391 - sparse_categorical_accuracy: 0.9894Step 57 / 1718\n",
      "1824/55000 - mean: 0.0396 - sparse_categorical_accuracy: 0.9890Step 58 / 1718\n",
      "1856/55000 - mean: 0.0390 - sparse_categorical_accuracy: 0.9892Step 59 / 1718\n",
      "1888/55000 - mean: 0.0405 - sparse_categorical_accuracy: 0.9889Step 60 / 1718\n",
      "1920/55000 - mean: 0.0398 - sparse_categorical_accuracy: 0.9891Step 61 / 1718\n",
      "1952/55000 - mean: 0.0392 - sparse_categorical_accuracy: 0.9892Step 62 / 1718\n",
      "1984/55000 - mean: 0.0387 - sparse_categorical_accuracy: 0.9894Step 63 / 1718\n",
      "2016/55000 - mean: 0.0411 - sparse_categorical_accuracy: 0.9886Step 64 / 1718\n",
      "2048/55000 - mean: 0.0405 - sparse_categorical_accuracy: 0.9888Step 65 / 1718\n",
      "2080/55000 - mean: 0.0399 - sparse_categorical_accuracy: 0.9889Step 66 / 1718\n",
      "2112/55000 - mean: 0.0394 - sparse_categorical_accuracy: 0.9891Step 67 / 1718\n",
      "2144/55000 - mean: 0.0399 - sparse_categorical_accuracy: 0.9888Step 68 / 1718\n",
      "2176/55000 - mean: 0.0394 - sparse_categorical_accuracy: 0.9890Step 69 / 1718\n",
      "2208/55000 - mean: 0.0389 - sparse_categorical_accuracy: 0.9891Step 70 / 1718\n",
      "2240/55000 - mean: 0.0408 - sparse_categorical_accuracy: 0.9888Step 71 / 1718\n",
      "2272/55000 - mean: 0.0440 - sparse_categorical_accuracy: 0.9881Step 72 / 1718\n",
      "2304/55000 - mean: 0.0434 - sparse_categorical_accuracy: 0.9883Step 73 / 1718\n",
      "2336/55000 - mean: 0.0431 - sparse_categorical_accuracy: 0.9884Step 74 / 1718\n",
      "2368/55000 - mean: 0.0426 - sparse_categorical_accuracy: 0.9886Step 75 / 1718\n",
      "2400/55000 - mean: 0.0420 - sparse_categorical_accuracy: 0.9887Step 76 / 1718\n",
      "2432/55000 - mean: 0.0415 - sparse_categorical_accuracy: 0.9889Step 77 / 1718\n",
      "2464/55000 - mean: 0.0409 - sparse_categorical_accuracy: 0.9890Step 78 / 1718\n",
      "2496/55000 - mean: 0.0412 - sparse_categorical_accuracy: 0.9888Step 79 / 1718\n",
      "2528/55000 - mean: 0.0426 - sparse_categorical_accuracy: 0.9885Step 80 / 1718\n",
      "2560/55000 - mean: 0.0434 - sparse_categorical_accuracy: 0.9883Step 81 / 1718\n",
      "2592/55000 - mean: 0.0429 - sparse_categorical_accuracy: 0.9884Step 82 / 1718\n",
      "2624/55000 - mean: 0.0427 - sparse_categorical_accuracy: 0.9882Step 83 / 1718\n",
      "2656/55000 - mean: 0.0423 - sparse_categorical_accuracy: 0.9883Step 84 / 1718\n",
      "2688/55000 - mean: 0.0419 - sparse_categorical_accuracy: 0.9885Step 85 / 1718\n",
      "2720/55000 - mean: 0.0414 - sparse_categorical_accuracy: 0.9886Step 86 / 1718\n",
      "2752/55000 - mean: 0.0423 - sparse_categorical_accuracy: 0.9884Step 87 / 1718\n",
      "2784/55000 - mean: 0.0419 - sparse_categorical_accuracy: 0.9885Step 88 / 1718\n",
      "2816/55000 - mean: 0.0421 - sparse_categorical_accuracy: 0.9883Step 89 / 1718\n",
      "2848/55000 - mean: 0.0416 - sparse_categorical_accuracy: 0.9884Step 90 / 1718\n",
      "2880/55000 - mean: 0.0433 - sparse_categorical_accuracy: 0.9878Step 91 / 1718\n",
      "2912/55000 - mean: 0.0428 - sparse_categorical_accuracy: 0.9880Step 92 / 1718\n",
      "2944/55000 - mean: 0.0423 - sparse_categorical_accuracy: 0.9881Step 93 / 1718\n",
      "2976/55000 - mean: 0.0419 - sparse_categorical_accuracy: 0.9882Step 94 / 1718\n",
      "3008/55000 - mean: 0.0428 - sparse_categorical_accuracy: 0.9880Step 95 / 1718\n",
      "3040/55000 - mean: 0.0424 - sparse_categorical_accuracy: 0.9882Step 96 / 1718\n",
      "3072/55000 - mean: 0.0420 - sparse_categorical_accuracy: 0.9883Step 97 / 1718\n",
      "3104/55000 - mean: 0.0416 - sparse_categorical_accuracy: 0.9884Step 98 / 1718\n",
      "3136/55000 - mean: 0.0412 - sparse_categorical_accuracy: 0.9885Step 99 / 1718\n",
      "3168/55000 - mean: 0.0408 - sparse_categorical_accuracy: 0.9886Step 100 / 1718\n",
      "3200/55000 - mean: 0.0404 - sparse_categorical_accuracy: 0.9887Step 101 / 1718\n",
      "3232/55000 - mean: 0.0402 - sparse_categorical_accuracy: 0.9889Step 102 / 1718\n",
      "3264/55000 - mean: 0.0398 - sparse_categorical_accuracy: 0.9890Step 103 / 1718\n",
      "3296/55000 - mean: 0.0400 - sparse_categorical_accuracy: 0.9888Step 104 / 1718\n",
      "3328/55000 - mean: 0.0396 - sparse_categorical_accuracy: 0.9889Step 105 / 1718\n",
      "3360/55000 - mean: 0.0392 - sparse_categorical_accuracy: 0.9890Step 106 / 1718\n",
      "3392/55000 - mean: 0.0389 - sparse_categorical_accuracy: 0.9891Step 107 / 1718\n",
      "3424/55000 - mean: 0.0385 - sparse_categorical_accuracy: 0.9892Step 108 / 1718\n",
      "3456/55000 - mean: 0.0382 - sparse_categorical_accuracy: 0.9893Step 109 / 1718\n",
      "3488/55000 - mean: 0.0379 - sparse_categorical_accuracy: 0.9894Step 110 / 1718\n",
      "3520/55000 - mean: 0.0376 - sparse_categorical_accuracy: 0.9895Step 111 / 1718\n",
      "3552/55000 - mean: 0.0373 - sparse_categorical_accuracy: 0.9896Step 112 / 1718\n",
      "3584/55000 - mean: 0.0372 - sparse_categorical_accuracy: 0.9894Step 113 / 1718\n",
      "3616/55000 - mean: 0.0370 - sparse_categorical_accuracy: 0.9895Step 114 / 1718\n",
      "3648/55000 - mean: 0.0371 - sparse_categorical_accuracy: 0.9893Step 115 / 1718\n",
      "3680/55000 - mean: 0.0368 - sparse_categorical_accuracy: 0.9894Step 116 / 1718\n",
      "3712/55000 - mean: 0.0371 - sparse_categorical_accuracy: 0.9892Step 117 / 1718\n",
      "3744/55000 - mean: 0.0368 - sparse_categorical_accuracy: 0.9893Step 118 / 1718\n",
      "3776/55000 - mean: 0.0365 - sparse_categorical_accuracy: 0.9894Step 119 / 1718\n",
      "3808/55000 - mean: 0.0371 - sparse_categorical_accuracy: 0.9890Step 120 / 1718\n",
      "3840/55000 - mean: 0.0369 - sparse_categorical_accuracy: 0.9891Step 121 / 1718\n",
      "3872/55000 - mean: 0.0369 - sparse_categorical_accuracy: 0.9889Step 122 / 1718\n",
      "3904/55000 - mean: 0.0366 - sparse_categorical_accuracy: 0.9890Step 123 / 1718\n",
      "3936/55000 - mean: 0.0363 - sparse_categorical_accuracy: 0.9891Step 124 / 1718\n",
      "3968/55000 - mean: 0.0361 - sparse_categorical_accuracy: 0.9892Step 125 / 1718\n",
      "4000/55000 - mean: 0.0358 - sparse_categorical_accuracy: 0.9893Step 126 / 1718\n",
      "4032/55000 - mean: 0.0355 - sparse_categorical_accuracy: 0.9893Step 127 / 1718\n",
      "4064/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9894Step 128 / 1718\n",
      "4096/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9895Step 129 / 1718\n",
      "4128/55000 - mean: 0.0347 - sparse_categorical_accuracy: 0.9896Step 130 / 1718\n",
      "4160/55000 - mean: 0.0344 - sparse_categorical_accuracy: 0.9897Step 131 / 1718\n",
      "4192/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9897Step 132 / 1718\n",
      "4224/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9896Step 133 / 1718\n",
      "4256/55000 - mean: 0.0340 - sparse_categorical_accuracy: 0.9897Step 134 / 1718\n",
      "4288/55000 - mean: 0.0342 - sparse_categorical_accuracy: 0.9895Step 135 / 1718\n",
      "4320/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9894Step 136 / 1718\n",
      "4352/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9892Step 137 / 1718\n",
      "4384/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9893Step 138 / 1718\n",
      "4416/55000 - mean: 0.0349 - sparse_categorical_accuracy: 0.9894Step 139 / 1718\n",
      "4448/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9894Step 140 / 1718\n",
      "4480/55000 - mean: 0.0346 - sparse_categorical_accuracy: 0.9895Step 141 / 1718\n",
      "4512/55000 - mean: 0.0343 - sparse_categorical_accuracy: 0.9896Step 142 / 1718\n",
      "4544/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9897Step 143 / 1718\n",
      "4576/55000 - mean: 0.0339 - sparse_categorical_accuracy: 0.9897Step 144 / 1718\n",
      "4608/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9898Step 145 / 1718\n",
      "4640/55000 - mean: 0.0335 - sparse_categorical_accuracy: 0.9899Step 146 / 1718\n",
      "4672/55000 - mean: 0.0332 - sparse_categorical_accuracy: 0.9899Step 147 / 1718\n",
      "4704/55000 - mean: 0.0356 - sparse_categorical_accuracy: 0.9898Step 148 / 1718\n",
      "4736/55000 - mean: 0.0362 - sparse_categorical_accuracy: 0.9897Step 149 / 1718\n",
      "4768/55000 - mean: 0.0360 - sparse_categorical_accuracy: 0.9897Step 150 / 1718\n",
      "4800/55000 - mean: 0.0359 - sparse_categorical_accuracy: 0.9896Step 151 / 1718\n",
      "4832/55000 - mean: 0.0357 - sparse_categorical_accuracy: 0.9897Step 152 / 1718\n",
      "4864/55000 - mean: 0.0354 - sparse_categorical_accuracy: 0.9897Step 153 / 1718\n",
      "4896/55000 - mean: 0.0352 - sparse_categorical_accuracy: 0.9898Step 154 / 1718\n",
      "4928/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9899Step 155 / 1718\n",
      "4960/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9899Step 156 / 1718\n",
      "4992/55000 - mean: 0.0346 - sparse_categorical_accuracy: 0.9900Step 157 / 1718\n",
      "5024/55000 - mean: 0.0344 - sparse_categorical_accuracy: 0.9900Step 158 / 1718\n",
      "5056/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9901Step 159 / 1718\n",
      "5088/55000 - mean: 0.0339 - sparse_categorical_accuracy: 0.9902Step 160 / 1718\n",
      "5120/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9902Step 161 / 1718\n",
      "5152/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9901Step 162 / 1718\n",
      "5184/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9900Step 163 / 1718\n",
      "5216/55000 - mean: 0.0339 - sparse_categorical_accuracy: 0.9900Step 164 / 1718\n",
      "5248/55000 - mean: 0.0351 - sparse_categorical_accuracy: 0.9899Step 165 / 1718\n",
      "5280/55000 - mean: 0.0350 - sparse_categorical_accuracy: 0.9900Step 166 / 1718\n",
      "5312/55000 - mean: 0.0348 - sparse_categorical_accuracy: 0.9900Step 167 / 1718\n",
      "5344/55000 - mean: 0.0346 - sparse_categorical_accuracy: 0.9901Step 168 / 1718\n",
      "5376/55000 - mean: 0.0343 - sparse_categorical_accuracy: 0.9901Step 169 / 1718\n",
      "5408/55000 - mean: 0.0341 - sparse_categorical_accuracy: 0.9902Step 170 / 1718\n",
      "5440/55000 - mean: 0.0339 - sparse_categorical_accuracy: 0.9903Step 171 / 1718\n",
      "5472/55000 - mean: 0.0337 - sparse_categorical_accuracy: 0.9903Step 172 / 1718\n",
      "5504/55000 - mean: 0.0336 - sparse_categorical_accuracy: 0.9904Step 173 / 1718\n",
      "5536/55000 - mean: 0.0334 - sparse_categorical_accuracy: 0.9904Step 174 / 1718\n",
      "5568/55000 - mean: 0.0333 - sparse_categorical_accuracy: 0.9905Step 175 / 1718\n",
      "5600/55000 - mean: 0.0331 - sparse_categorical_accuracy: 0.9905Step 176 / 1718\n",
      "5632/55000 - mean: 0.0329 - sparse_categorical_accuracy: 0.9906Step 177 / 1718\n",
      "5664/55000 - mean: 0.0327 - sparse_categorical_accuracy: 0.9906Step 178 / 1718\n",
      "5696/55000 - mean: 0.0326 - sparse_categorical_accuracy: 0.9907Step 179 / 1718\n",
      "5728/55000 - mean: 0.0324 - sparse_categorical_accuracy: 0.9907Step 180 / 1718\n",
      "5760/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9908Step 181 / 1718\n",
      "5792/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9908Step 182 / 1718\n",
      "5824/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9909Step 183 / 1718\n",
      "5856/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9909Step 184 / 1718\n",
      "5888/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9908Step 185 / 1718\n",
      "5920/55000 - mean: 0.0319 - sparse_categorical_accuracy: 0.9909Step 186 / 1718\n",
      "5952/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9909Step 187 / 1718\n",
      "5984/55000 - mean: 0.0316 - sparse_categorical_accuracy: 0.9910Step 188 / 1718\n",
      "6016/55000 - mean: 0.0327 - sparse_categorical_accuracy: 0.9909Step 189 / 1718\n",
      "6048/55000 - mean: 0.0325 - sparse_categorical_accuracy: 0.9909Step 190 / 1718\n",
      "6080/55000 - mean: 0.0323 - sparse_categorical_accuracy: 0.9910Step 191 / 1718\n",
      "6112/55000 - mean: 0.0322 - sparse_categorical_accuracy: 0.9910Step 192 / 1718\n",
      "6144/55000 - mean: 0.0321 - sparse_categorical_accuracy: 0.9909Step 193 / 1718\n",
      "6176/55000 - mean: 0.0320 - sparse_categorical_accuracy: 0.9909Step 194 / 1718\n",
      "6208/55000 - mean: 0.0318 - sparse_categorical_accuracy: 0.9910Step 195 / 1718\n",
      "6240/55000 - mean: 0.0317 - sparse_categorical_accuracy: 0.9910Step 196 / 1718\n",
      "6272/55000 - mean: 0.0315 - sparse_categorical_accuracy: 0.9911Step 197 / 1718\n",
      "6304/55000 - mean: 0.0314 - sparse_categorical_accuracy: 0.9911Step 198 / 1718\n",
      "6336/55000 - mean: 0.0312 - sparse_categorical_accuracy: 0.9912Step 199 / 1718\n",
      "6368/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9912Step 200 / 1718\n",
      "6400/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9912Step 201 / 1718\n",
      "6432/55000 - mean: 0.0310 - sparse_categorical_accuracy: 0.9911Step 202 / 1718\n",
      "6464/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9912Step 203 / 1718\n",
      "6496/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9912Step 204 / 1718\n",
      "6528/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9913Step 205 / 1718\n",
      "6560/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9912Step 206 / 1718\n",
      "6592/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9912Step 207 / 1718\n",
      "6624/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9912Step 208 / 1718\n",
      "6656/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9913Step 209 / 1718\n",
      "6688/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9913Step 210 / 1718\n",
      "6720/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9914Step 211 / 1718\n",
      "6752/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9914Step 212 / 1718\n",
      "6784/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9913Step 213 / 1718\n",
      "6816/55000 - mean: 0.0306 - sparse_categorical_accuracy: 0.9912Step 214 / 1718\n",
      "6848/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9912Step 215 / 1718\n",
      "6880/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9913Step 216 / 1718\n",
      "6912/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9912Step 217 / 1718\n",
      "6944/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9912Step 218 / 1718\n",
      "6976/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9913Step 219 / 1718\n",
      "7008/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9913Step 220 / 1718\n",
      "7040/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9913Step 221 / 1718\n",
      "7072/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9914Step 222 / 1718\n",
      "7104/55000 - mean: 0.0298 - sparse_categorical_accuracy: 0.9914Step 223 / 1718\n",
      "7136/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9915Step 224 / 1718\n",
      "7168/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9915Step 225 / 1718\n",
      "7200/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9915Step 226 / 1718\n",
      "7232/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9916Step 227 / 1718\n",
      "7264/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9916Step 228 / 1718\n",
      "7296/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9914Step 229 / 1718\n",
      "7328/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9913Step 230 / 1718\n",
      "7360/55000 - mean: 0.0309 - sparse_categorical_accuracy: 0.9912Step 231 / 1718\n",
      "7392/55000 - mean: 0.0308 - sparse_categorical_accuracy: 0.9912Step 232 / 1718\n",
      "7424/55000 - mean: 0.0307 - sparse_categorical_accuracy: 0.9912Step 233 / 1718\n",
      "7456/55000 - mean: 0.0305 - sparse_categorical_accuracy: 0.9913Step 234 / 1718\n",
      "7488/55000 - mean: 0.0304 - sparse_categorical_accuracy: 0.9913Step 235 / 1718\n",
      "7520/55000 - mean: 0.0303 - sparse_categorical_accuracy: 0.9914Step 236 / 1718\n",
      "7552/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9914Step 237 / 1718\n",
      "7584/55000 - mean: 0.0302 - sparse_categorical_accuracy: 0.9913Step 238 / 1718\n",
      "7616/55000 - mean: 0.0301 - sparse_categorical_accuracy: 0.9913Step 239 / 1718\n",
      "7648/55000 - mean: 0.0300 - sparse_categorical_accuracy: 0.9914Step 240 / 1718\n",
      "7680/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9914Step 241 / 1718\n",
      "7712/55000 - mean: 0.0299 - sparse_categorical_accuracy: 0.9913Step 242 / 1718\n",
      "7744/55000 - mean: 0.0297 - sparse_categorical_accuracy: 0.9913Step 243 / 1718\n",
      "7776/55000 - mean: 0.0296 - sparse_categorical_accuracy: 0.9914Step 244 / 1718\n",
      "7808/55000 - mean: 0.0295 - sparse_categorical_accuracy: 0.9914Step 245 / 1718\n",
      "7840/55000 - mean: 0.0294 - sparse_categorical_accuracy: 0.9915Step 246 / 1718\n",
      "7872/55000 - mean: 0.0293 - sparse_categorical_accuracy: 0.9915Step 247 / 1718\n",
      "7904/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9915Step 248 / 1718\n",
      "7936/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9916Step 249 / 1718\n",
      "7968/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9916Step 250 / 1718\n",
      "8000/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9916Step 251 / 1718\n",
      "8032/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 252 / 1718\n",
      "8064/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9914Step 253 / 1718\n",
      "8096/55000 - mean: 0.0292 - sparse_categorical_accuracy: 0.9914Step 254 / 1718\n",
      "8128/55000 - mean: 0.0291 - sparse_categorical_accuracy: 0.9914Step 255 / 1718\n",
      "8160/55000 - mean: 0.0290 - sparse_categorical_accuracy: 0.9914Step 256 / 1718\n",
      "8192/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9913Step 257 / 1718\n",
      "8224/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9914Step 258 / 1718\n",
      "8256/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9914Step 259 / 1718\n",
      "8288/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9914Step 260 / 1718\n",
      "8320/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9915Step 261 / 1718\n",
      "8352/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9915Step 262 / 1718\n",
      "8384/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9914Step 263 / 1718\n",
      "8416/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9913Step 264 / 1718\n",
      "8448/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9914Step 265 / 1718\n",
      "8480/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9914Step 266 / 1718\n",
      "8512/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9913Step 267 / 1718\n",
      "8544/55000 - mean: 0.0289 - sparse_categorical_accuracy: 0.9913Step 268 / 1718\n",
      "8576/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9914Step 269 / 1718\n",
      "8608/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9914Step 270 / 1718\n",
      "8640/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9914Step 271 / 1718\n",
      "8672/55000 - mean: 0.0288 - sparse_categorical_accuracy: 0.9912Step 272 / 1718\n",
      "8704/55000 - mean: 0.0287 - sparse_categorical_accuracy: 0.9913Step 273 / 1718\n",
      "8736/55000 - mean: 0.0286 - sparse_categorical_accuracy: 0.9913Step 274 / 1718\n",
      "8768/55000 - mean: 0.0285 - sparse_categorical_accuracy: 0.9913Step 275 / 1718\n",
      "8800/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9914Step 276 / 1718\n",
      "8832/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9914Step 277 / 1718\n",
      "8864/55000 - mean: 0.0284 - sparse_categorical_accuracy: 0.9913Step 278 / 1718\n",
      "8896/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9913Step 279 / 1718\n",
      "8928/55000 - mean: 0.0283 - sparse_categorical_accuracy: 0.9913Step 280 / 1718\n",
      "8960/55000 - mean: 0.0282 - sparse_categorical_accuracy: 0.9913Step 281 / 1718\n",
      "8992/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9913Step 282 / 1718\n",
      "9024/55000 - mean: 0.0280 - sparse_categorical_accuracy: 0.9914Step 283 / 1718\n",
      "9056/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9914Step 284 / 1718\n",
      "9088/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9914Step 285 / 1718\n",
      "9120/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9914Step 286 / 1718\n",
      "9152/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9915Step 287 / 1718\n",
      "9184/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9915Step 288 / 1718\n",
      "9216/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9915Step 289 / 1718\n",
      "9248/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9916Step 290 / 1718\n",
      "9280/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9916Step 291 / 1718\n",
      "9312/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9916Step 292 / 1718\n",
      "9344/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9917Step 293 / 1718\n",
      "9376/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9916Step 294 / 1718\n",
      "9408/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9916Step 295 / 1718\n",
      "9440/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9916Step 296 / 1718\n",
      "9472/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9917Step 297 / 1718\n",
      "9504/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9917Step 298 / 1718\n",
      "9536/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9917Step 299 / 1718\n",
      "9568/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9917Step 300 / 1718\n",
      "9600/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9917Step 301 / 1718\n",
      "9632/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9917Step 302 / 1718\n",
      "9664/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9917Step 303 / 1718\n",
      "9696/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9917Step 304 / 1718\n",
      "9728/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9918Step 305 / 1718\n",
      "9760/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9917Step 306 / 1718\n",
      "9792/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9916Step 307 / 1718\n",
      "9824/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9917Step 308 / 1718\n",
      "9856/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9917Step 309 / 1718\n",
      "9888/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9917Step 310 / 1718\n",
      "9920/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9917Step 311 / 1718\n",
      "9952/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9918Step 312 / 1718\n",
      "9984/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9917Step 313 / 1718\n",
      "10016/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9917Step 314 / 1718\n",
      "10048/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9917Step 315 / 1718\n",
      "10080/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9918Step 316 / 1718\n",
      "10112/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9918Step 317 / 1718\n",
      "10144/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9918Step 318 / 1718\n",
      "10176/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9918Step 319 / 1718\n",
      "10208/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9919Step 320 / 1718\n",
      "10240/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9919Step 321 / 1718\n",
      "10272/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9919Step 322 / 1718\n",
      "10304/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9919Step 323 / 1718\n",
      "10336/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9920Step 324 / 1718\n",
      "10368/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9920Step 325 / 1718\n",
      "10400/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9919Step 326 / 1718\n",
      "10432/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9919Step 327 / 1718\n",
      "10464/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9920Step 328 / 1718\n",
      "10496/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9920Step 329 / 1718\n",
      "10528/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9920Step 330 / 1718\n",
      "10560/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9920Step 331 / 1718\n",
      "10592/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9921Step 332 / 1718\n",
      "10624/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9921Step 333 / 1718\n",
      "10656/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9921Step 334 / 1718\n",
      "10688/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9921Step 335 / 1718\n",
      "10720/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9922Step 336 / 1718\n",
      "10752/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9921Step 337 / 1718\n",
      "10784/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9921Step 338 / 1718\n",
      "10816/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9921Step 339 / 1718\n",
      "10848/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9922Step 340 / 1718\n",
      "10880/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9922Step 341 / 1718\n",
      "10912/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9922Step 342 / 1718\n",
      "10944/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9922Step 343 / 1718\n",
      "10976/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9923Step 344 / 1718\n",
      "11008/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9923Step 345 / 1718\n",
      "11040/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9923Step 346 / 1718\n",
      "11072/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9923Step 347 / 1718\n",
      "11104/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9923Step 348 / 1718\n",
      "11136/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9924Step 349 / 1718\n",
      "11168/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9924Step 350 / 1718\n",
      "11200/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9924Step 351 / 1718\n",
      "11232/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9924Step 352 / 1718\n",
      "11264/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9925Step 353 / 1718\n",
      "11296/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9925Step 354 / 1718\n",
      "11328/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9925Step 355 / 1718\n",
      "11360/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9925Step 356 / 1718\n",
      "11392/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9925Step 357 / 1718\n",
      "11424/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9926Step 358 / 1718\n",
      "11456/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9926Step 359 / 1718\n",
      "11488/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9926Step 360 / 1718\n",
      "11520/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9926Step 361 / 1718\n",
      "11552/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9926Step 362 / 1718\n",
      "11584/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9926Step 363 / 1718\n",
      "11616/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9926Step 364 / 1718\n",
      "11648/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9926Step 365 / 1718\n",
      "11680/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9926Step 366 / 1718\n",
      "11712/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9926Step 367 / 1718\n",
      "11744/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9926Step 368 / 1718\n",
      "11776/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9926Step 369 / 1718\n",
      "11808/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9926Step 370 / 1718\n",
      "11840/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9927Step 371 / 1718\n",
      "11872/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9927Step 372 / 1718\n",
      "11904/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9926Step 373 / 1718\n",
      "11936/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9926Step 374 / 1718\n",
      "11968/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9926Step 375 / 1718\n",
      "12000/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9927Step 376 / 1718\n",
      "12032/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9926Step 377 / 1718\n",
      "12064/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9926Step 378 / 1718\n",
      "12096/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9926Step 379 / 1718\n",
      "12128/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9927Step 380 / 1718\n",
      "12160/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9927Step 381 / 1718\n",
      "12192/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9927Step 382 / 1718\n",
      "12224/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9927Step 383 / 1718\n",
      "12256/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9927Step 384 / 1718\n",
      "12288/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9928Step 385 / 1718\n",
      "12320/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9928Step 386 / 1718\n",
      "12352/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9928Step 387 / 1718\n",
      "12384/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9928Step 388 / 1718\n",
      "12416/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9928Step 389 / 1718\n",
      "12448/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9929Step 390 / 1718\n",
      "12480/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9929Step 391 / 1718\n",
      "12512/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9929Step 392 / 1718\n",
      "12544/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9929Step 393 / 1718\n",
      "12576/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9929Step 394 / 1718\n",
      "12608/55000 - mean: 0.0236 - sparse_categorical_accuracy: 0.9929Step 395 / 1718\n",
      "12640/55000 - mean: 0.0236 - sparse_categorical_accuracy: 0.9930Step 396 / 1718\n",
      "12672/55000 - mean: 0.0235 - sparse_categorical_accuracy: 0.9930Step 397 / 1718\n",
      "12704/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9928Step 398 / 1718\n",
      "12736/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9929Step 399 / 1718\n",
      "12768/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9929Step 400 / 1718\n",
      "12800/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9927Step 401 / 1718\n",
      "12832/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9928Step 402 / 1718\n",
      "12864/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9928Step 403 / 1718\n",
      "12896/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9927Step 404 / 1718\n",
      "12928/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9927Step 405 / 1718\n",
      "12960/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9927Step 406 / 1718\n",
      "12992/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9925Step 407 / 1718\n",
      "13024/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9926Step 408 / 1718\n",
      "13056/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9926Step 409 / 1718\n",
      "13088/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9926Step 410 / 1718\n",
      "13120/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9926Step 411 / 1718\n",
      "13152/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9926Step 412 / 1718\n",
      "13184/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9926Step 413 / 1718\n",
      "13216/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9927Step 414 / 1718\n",
      "13248/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9927Step 415 / 1718\n",
      "13280/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9927Step 416 / 1718\n",
      "13312/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9926Step 417 / 1718\n",
      "13344/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9927Step 418 / 1718\n",
      "13376/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9927Step 419 / 1718\n",
      "13408/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9927Step 420 / 1718\n",
      "13440/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9927Step 421 / 1718\n",
      "13472/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9927Step 422 / 1718\n",
      "13504/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9927Step 423 / 1718\n",
      "13536/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9927Step 424 / 1718\n",
      "13568/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9927Step 425 / 1718\n",
      "13600/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9926Step 426 / 1718\n",
      "13632/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9927Step 427 / 1718\n",
      "13664/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9927Step 428 / 1718\n",
      "13696/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9927Step 429 / 1718\n",
      "13728/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9926Step 430 / 1718\n",
      "13760/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9927Step 431 / 1718\n",
      "13792/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9926Step 432 / 1718\n",
      "13824/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9926Step 433 / 1718\n",
      "13856/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9926Step 434 / 1718\n",
      "13888/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9927Step 435 / 1718\n",
      "13920/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9927Step 436 / 1718\n",
      "13952/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9926Step 437 / 1718\n",
      "13984/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9926Step 438 / 1718\n",
      "14016/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9927Step 439 / 1718\n",
      "14048/55000 - mean: 0.0236 - sparse_categorical_accuracy: 0.9927Step 440 / 1718\n",
      "14080/55000 - mean: 0.0236 - sparse_categorical_accuracy: 0.9927Step 441 / 1718\n",
      "14112/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9926Step 442 / 1718\n",
      "14144/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9926Step 443 / 1718\n",
      "14176/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9926Step 444 / 1718\n",
      "14208/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9925Step 445 / 1718\n",
      "14240/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9925Step 446 / 1718\n",
      "14272/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9924Step 447 / 1718\n",
      "14304/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9924Step 448 / 1718\n",
      "14336/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9924Step 449 / 1718\n",
      "14368/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9923Step 450 / 1718\n",
      "14400/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9924Step 451 / 1718\n",
      "14432/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9924Step 452 / 1718\n",
      "14464/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9924Step 453 / 1718\n",
      "14496/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9924Step 454 / 1718\n",
      "14528/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9922Step 455 / 1718\n",
      "14560/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9922Step 456 / 1718\n",
      "14592/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9923Step 457 / 1718\n",
      "14624/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9923Step 458 / 1718\n",
      "14656/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9923Step 459 / 1718\n",
      "14688/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9923Step 460 / 1718\n",
      "14720/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9923Step 461 / 1718\n",
      "14752/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9923Step 462 / 1718\n",
      "14784/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9924Step 463 / 1718\n",
      "14816/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9924Step 464 / 1718\n",
      "14848/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9923Step 465 / 1718\n",
      "14880/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9923Step 466 / 1718\n",
      "14912/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9923Step 467 / 1718\n",
      "14944/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9923Step 468 / 1718\n",
      "14976/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9923Step 469 / 1718\n",
      "15008/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9923Step 470 / 1718\n",
      "15040/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9923Step 471 / 1718\n",
      "15072/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9923Step 472 / 1718\n",
      "15104/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9923Step 473 / 1718\n",
      "15136/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9923Step 474 / 1718\n",
      "15168/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9924Step 475 / 1718\n",
      "15200/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9924Step 476 / 1718\n",
      "15232/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9923Step 477 / 1718\n",
      "15264/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9923Step 478 / 1718\n",
      "15296/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9924Step 479 / 1718\n",
      "15328/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9924Step 480 / 1718\n",
      "15360/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9923Step 481 / 1718\n",
      "15392/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9923Step 482 / 1718\n",
      "15424/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9923Step 483 / 1718\n",
      "15456/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9924Step 484 / 1718\n",
      "15488/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9924Step 485 / 1718\n",
      "15520/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9924Step 486 / 1718\n",
      "15552/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9924Step 487 / 1718\n",
      "15584/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9924Step 488 / 1718\n",
      "15616/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9924Step 489 / 1718\n",
      "15648/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9924Step 490 / 1718\n",
      "15680/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9924Step 491 / 1718\n",
      "15712/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9924Step 492 / 1718\n",
      "15744/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9924Step 493 / 1718\n",
      "15776/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9924Step 494 / 1718\n",
      "15808/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9924Step 495 / 1718\n",
      "15840/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9924Step 496 / 1718\n",
      "15872/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9924Step 497 / 1718\n",
      "15904/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 498 / 1718\n",
      "15936/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 499 / 1718\n",
      "15968/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9925Step 500 / 1718\n",
      "16000/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9925Step 501 / 1718\n",
      "16032/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9925Step 502 / 1718\n",
      "16064/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9925Step 503 / 1718\n",
      "16096/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9925Step 504 / 1718\n",
      "16128/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9926Step 505 / 1718\n",
      "16160/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9925Step 506 / 1718\n",
      "16192/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 507 / 1718\n",
      "16224/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9925Step 508 / 1718\n",
      "16256/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9925Step 509 / 1718\n",
      "16288/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 510 / 1718\n",
      "16320/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 511 / 1718\n",
      "16352/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9925Step 512 / 1718\n",
      "16384/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9925Step 513 / 1718\n",
      "16416/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 514 / 1718\n",
      "16448/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 515 / 1718\n",
      "16480/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 516 / 1718\n",
      "16512/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9925Step 517 / 1718\n",
      "16544/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 518 / 1718\n",
      "16576/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 519 / 1718\n",
      "16608/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9924Step 520 / 1718\n",
      "16640/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9924Step 521 / 1718\n",
      "16672/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9924Step 522 / 1718\n",
      "16704/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9924Step 523 / 1718\n",
      "16736/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9924Step 524 / 1718\n",
      "16768/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9924Step 525 / 1718\n",
      "16800/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9924Step 526 / 1718\n",
      "16832/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9925Step 527 / 1718\n",
      "16864/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9925Step 528 / 1718\n",
      "16896/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9925Step 529 / 1718\n",
      "16928/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9924Step 530 / 1718\n",
      "16960/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 531 / 1718\n",
      "16992/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 532 / 1718\n",
      "17024/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9925Step 533 / 1718\n",
      "17056/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9925Step 534 / 1718\n",
      "17088/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 535 / 1718\n",
      "17120/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 536 / 1718\n",
      "17152/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 537 / 1718\n",
      "17184/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9925Step 538 / 1718\n",
      "17216/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9924Step 539 / 1718\n",
      "17248/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9924Step 540 / 1718\n",
      "17280/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9924Step 541 / 1718\n",
      "17312/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9924Step 542 / 1718\n",
      "17344/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9924Step 543 / 1718\n",
      "17376/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9925Step 544 / 1718\n",
      "17408/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9925Step 545 / 1718\n",
      "17440/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9925Step 546 / 1718\n",
      "17472/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9925Step 547 / 1718\n",
      "17504/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9925Step 548 / 1718\n",
      "17536/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9924Step 549 / 1718\n",
      "17568/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9924Step 550 / 1718\n",
      "17600/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9924Step 551 / 1718\n",
      "17632/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9924Step 552 / 1718\n",
      "17664/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9924Step 553 / 1718\n",
      "17696/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9924Step 554 / 1718\n",
      "17728/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9924Step 555 / 1718\n",
      "17760/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9925Step 556 / 1718\n",
      "17792/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9924Step 557 / 1718\n",
      "17824/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9924Step 558 / 1718\n",
      "17856/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9924Step 559 / 1718\n",
      "17888/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9925Step 560 / 1718\n",
      "17920/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9925Step 561 / 1718\n",
      "17952/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9925Step 562 / 1718\n",
      "17984/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9924Step 563 / 1718\n",
      "18016/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9924Step 564 / 1718\n",
      "18048/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9924Step 565 / 1718\n",
      "18080/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9924Step 566 / 1718\n",
      "18112/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9924Step 567 / 1718\n",
      "18144/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9924Step 568 / 1718\n",
      "18176/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9924Step 569 / 1718\n",
      "18208/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9924Step 570 / 1718\n",
      "18240/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9924Step 571 / 1718\n",
      "18272/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9924Step 572 / 1718\n",
      "18304/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9925Step 573 / 1718\n",
      "18336/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9924Step 574 / 1718\n",
      "18368/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9924Step 575 / 1718\n",
      "18400/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9924Step 576 / 1718\n",
      "18432/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9925Step 577 / 1718\n",
      "18464/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9925Step 578 / 1718\n",
      "18496/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9925Step 579 / 1718\n",
      "18528/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9925Step 580 / 1718\n",
      "18560/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9925Step 581 / 1718\n",
      "18592/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9924Step 582 / 1718\n",
      "18624/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9924Step 583 / 1718\n",
      "18656/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9924Step 584 / 1718\n",
      "18688/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9924Step 585 / 1718\n",
      "18720/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9924Step 586 / 1718\n",
      "18752/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9924Step 587 / 1718\n",
      "18784/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9924Step 588 / 1718\n",
      "18816/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9924Step 589 / 1718\n",
      "18848/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9924Step 590 / 1718\n",
      "18880/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9924Step 591 / 1718\n",
      "18912/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9924Step 592 / 1718\n",
      "18944/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9925Step 593 / 1718\n",
      "18976/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9925Step 594 / 1718\n",
      "19008/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9925Step 595 / 1718\n",
      "19040/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9924Step 596 / 1718\n",
      "19072/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9924Step 597 / 1718\n",
      "19104/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9924Step 598 / 1718\n",
      "19136/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9924Step 599 / 1718\n",
      "19168/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9924Step 600 / 1718\n",
      "19200/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9924Step 601 / 1718\n",
      "19232/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9925Step 602 / 1718\n",
      "19264/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9925Step 603 / 1718\n",
      "19296/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9925Step 604 / 1718\n",
      "19328/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9925Step 605 / 1718\n",
      "19360/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9925Step 606 / 1718\n",
      "19392/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9925Step 607 / 1718\n",
      "19424/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9925Step 608 / 1718\n",
      "19456/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9925Step 609 / 1718\n",
      "19488/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9926Step 610 / 1718\n",
      "19520/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9926Step 611 / 1718\n",
      "19552/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9926Step 612 / 1718\n",
      "19584/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9926Step 613 / 1718\n",
      "19616/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9926Step 614 / 1718\n",
      "19648/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9926Step 615 / 1718\n",
      "19680/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9925Step 616 / 1718\n",
      "19712/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9925Step 617 / 1718\n",
      "19744/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9926Step 618 / 1718\n",
      "19776/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9926Step 619 / 1718\n",
      "19808/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9926Step 620 / 1718\n",
      "19840/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9926Step 621 / 1718\n",
      "19872/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9926Step 622 / 1718\n",
      "19904/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9926Step 623 / 1718\n",
      "19936/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9926Step 624 / 1718\n",
      "19968/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9926Step 625 / 1718\n",
      "20000/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9926Step 626 / 1718\n",
      "20032/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9927Step 627 / 1718\n",
      "20064/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9927Step 628 / 1718\n",
      "20096/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9927Step 629 / 1718\n",
      "20128/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9927Step 630 / 1718\n",
      "20160/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9927Step 631 / 1718\n",
      "20192/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9927Step 632 / 1718\n",
      "20224/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9927Step 633 / 1718\n",
      "20256/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9927Step 634 / 1718\n",
      "20288/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9927Step 635 / 1718\n",
      "20320/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9927Step 636 / 1718\n",
      "20352/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9927Step 637 / 1718\n",
      "20384/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9927Step 638 / 1718\n",
      "20416/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9928Step 639 / 1718\n",
      "20448/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9928Step 640 / 1718\n",
      "20480/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9927Step 641 / 1718\n",
      "20512/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9927Step 642 / 1718\n",
      "20544/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9927Step 643 / 1718\n",
      "20576/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9927Step 644 / 1718\n",
      "20608/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9927Step 645 / 1718\n",
      "20640/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9927Step 646 / 1718\n",
      "20672/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9927Step 647 / 1718\n",
      "20704/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9927Step 648 / 1718\n",
      "20736/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9927Step 649 / 1718\n",
      "20768/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9927Step 650 / 1718\n",
      "20800/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9927Step 651 / 1718\n",
      "20832/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9927Step 652 / 1718\n",
      "20864/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9927Step 653 / 1718\n",
      "20896/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9927Step 654 / 1718\n",
      "20928/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9927Step 655 / 1718\n",
      "20960/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9927Step 656 / 1718\n",
      "20992/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9928Step 657 / 1718\n",
      "21024/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9928Step 658 / 1718\n",
      "21056/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9928Step 659 / 1718\n",
      "21088/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9928Step 660 / 1718\n",
      "21120/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9928Step 661 / 1718\n",
      "21152/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9928Step 662 / 1718\n",
      "21184/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9928Step 663 / 1718\n",
      "21216/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9928Step 664 / 1718\n",
      "21248/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9928Step 665 / 1718\n",
      "21280/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9928Step 666 / 1718\n",
      "21312/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9928Step 667 / 1718\n",
      "21344/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9928Step 668 / 1718\n",
      "21376/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9927Step 669 / 1718\n",
      "21408/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9927Step 670 / 1718\n",
      "21440/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9927Step 671 / 1718\n",
      "21472/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9927Step 672 / 1718\n",
      "21504/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9927Step 673 / 1718\n",
      "21536/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9928Step 674 / 1718\n",
      "21568/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9927Step 675 / 1718\n",
      "21600/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9927Step 676 / 1718\n",
      "21632/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9927Step 677 / 1718\n",
      "21664/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9927Step 678 / 1718\n",
      "21696/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9927Step 679 / 1718\n",
      "21728/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9927Step 680 / 1718\n",
      "21760/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9926Step 681 / 1718\n",
      "21792/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9925Step 682 / 1718\n",
      "21824/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9925Step 683 / 1718\n",
      "21856/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9925Step 684 / 1718\n",
      "21888/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9925Step 685 / 1718\n",
      "21920/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9925Step 686 / 1718\n",
      "21952/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9925Step 687 / 1718\n",
      "21984/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9925Step 688 / 1718\n",
      "22016/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9925Step 689 / 1718\n",
      "22048/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9925Step 690 / 1718\n",
      "22080/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9924Step 691 / 1718\n",
      "22112/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9924Step 692 / 1718\n",
      "22144/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9924Step 693 / 1718\n",
      "22176/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9923Step 694 / 1718\n",
      "22208/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9923Step 695 / 1718\n",
      "22240/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9924Step 696 / 1718\n",
      "22272/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9923Step 697 / 1718\n",
      "22304/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9923Step 698 / 1718\n",
      "22336/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9923Step 699 / 1718\n",
      "22368/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9924Step 700 / 1718\n",
      "22400/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9924Step 701 / 1718\n",
      "22432/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9924Step 702 / 1718\n",
      "22464/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9923Step 703 / 1718\n",
      "22496/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9924Step 704 / 1718\n",
      "22528/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9923Step 705 / 1718\n",
      "22560/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9923Step 706 / 1718\n",
      "22592/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9923Step 707 / 1718\n",
      "22624/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9924Step 708 / 1718\n",
      "22656/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9924Step 709 / 1718\n",
      "22688/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9924Step 710 / 1718\n",
      "22720/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9924Step 711 / 1718\n",
      "22752/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9924Step 712 / 1718\n",
      "22784/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9924Step 713 / 1718\n",
      "22816/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9923Step 714 / 1718\n",
      "22848/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9923Step 715 / 1718\n",
      "22880/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9923Step 716 / 1718\n",
      "22912/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9923Step 717 / 1718\n",
      "22944/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9922Step 718 / 1718\n",
      "22976/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9923Step 719 / 1718\n",
      "23008/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9923Step 720 / 1718\n",
      "23040/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9923Step 721 / 1718\n",
      "23072/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9923Step 722 / 1718\n",
      "23104/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9923Step 723 / 1718\n",
      "23136/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9923Step 724 / 1718\n",
      "23168/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9923Step 725 / 1718\n",
      "23200/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9923Step 726 / 1718\n",
      "23232/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9923Step 727 / 1718\n",
      "23264/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9923Step 728 / 1718\n",
      "23296/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9923Step 729 / 1718\n",
      "23328/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9923Step 730 / 1718\n",
      "23360/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9923Step 731 / 1718\n",
      "23392/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9923Step 732 / 1718\n",
      "23424/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9923Step 733 / 1718\n",
      "23456/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9923Step 734 / 1718\n",
      "23488/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9923Step 735 / 1718\n",
      "23520/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9923Step 736 / 1718\n",
      "23552/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9923Step 737 / 1718\n",
      "23584/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9923Step 738 / 1718\n",
      "23616/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9923Step 739 / 1718\n",
      "23648/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9923Step 740 / 1718\n",
      "23680/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9924Step 741 / 1718\n",
      "23712/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9923Step 742 / 1718\n",
      "23744/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9923Step 743 / 1718\n",
      "23776/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9923Step 744 / 1718\n",
      "23808/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9924Step 745 / 1718\n",
      "23840/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9924Step 746 / 1718\n",
      "23872/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9923Step 747 / 1718\n",
      "23904/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9923Step 748 / 1718\n",
      "23936/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9923Step 749 / 1718\n",
      "23968/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9923Step 750 / 1718\n",
      "24000/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9923Step 751 / 1718\n",
      "24032/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9923Step 752 / 1718\n",
      "24064/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9924Step 753 / 1718\n",
      "24096/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9924Step 754 / 1718\n",
      "24128/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9923Step 755 / 1718\n",
      "24160/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9923Step 756 / 1718\n",
      "24192/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9923Step 757 / 1718\n",
      "24224/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9923Step 758 / 1718\n",
      "24256/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9923Step 759 / 1718\n",
      "24288/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9923Step 760 / 1718\n",
      "24320/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9924Step 761 / 1718\n",
      "24352/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9924Step 762 / 1718\n",
      "24384/55000 - mean: 0.0260 - sparse_categorical_accuracy: 0.9924Step 763 / 1718\n",
      "24416/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9924Step 764 / 1718\n",
      "24448/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9924Step 765 / 1718\n",
      "24480/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9924Step 766 / 1718\n",
      "24512/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9924Step 767 / 1718\n",
      "24544/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9924Step 768 / 1718\n",
      "24576/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9924Step 769 / 1718\n",
      "24608/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9924Step 770 / 1718\n",
      "24640/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9924Step 771 / 1718\n",
      "24672/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9924Step 772 / 1718\n",
      "24704/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9924Step 773 / 1718\n",
      "24736/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9924Step 774 / 1718\n",
      "24768/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9924Step 775 / 1718\n",
      "24800/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9925Step 776 / 1718\n",
      "24832/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9925Step 777 / 1718\n",
      "24864/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9925Step 778 / 1718\n",
      "24896/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9925Step 779 / 1718\n",
      "24928/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9925Step 780 / 1718\n",
      "24960/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9925Step 781 / 1718\n",
      "24992/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9925Step 782 / 1718\n",
      "25024/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9925Step 783 / 1718\n",
      "25056/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9925Step 784 / 1718\n",
      "25088/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9925Step 785 / 1718\n",
      "25120/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9925Step 786 / 1718\n",
      "25152/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9925Step 787 / 1718\n",
      "25184/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9925Step 788 / 1718\n",
      "25216/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9925Step 789 / 1718\n",
      "25248/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9925Step 790 / 1718\n",
      "25280/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9925Step 791 / 1718\n",
      "25312/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 792 / 1718\n",
      "25344/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 793 / 1718\n",
      "25376/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9925Step 794 / 1718\n",
      "25408/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9925Step 795 / 1718\n",
      "25440/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 796 / 1718\n",
      "25472/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9925Step 797 / 1718\n",
      "25504/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9926Step 798 / 1718\n",
      "25536/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9926Step 799 / 1718\n",
      "25568/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9926Step 800 / 1718\n",
      "25600/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9926Step 801 / 1718\n",
      "25632/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9926Step 802 / 1718\n",
      "25664/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9926Step 803 / 1718\n",
      "25696/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9926Step 804 / 1718\n",
      "25728/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9926Step 805 / 1718\n",
      "25760/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9926Step 806 / 1718\n",
      "25792/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9926Step 807 / 1718\n",
      "25824/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9926Step 808 / 1718\n",
      "25856/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9926Step 809 / 1718\n",
      "25888/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9926Step 810 / 1718\n",
      "25920/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9926Step 811 / 1718\n",
      "25952/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9926Step 812 / 1718\n",
      "25984/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9926Step 813 / 1718\n",
      "26016/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9927Step 814 / 1718\n",
      "26048/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9927Step 815 / 1718\n",
      "26080/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9927Step 816 / 1718\n",
      "26112/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9927Step 817 / 1718\n",
      "26144/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9927Step 818 / 1718\n",
      "26176/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9927Step 819 / 1718\n",
      "26208/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9927Step 820 / 1718\n",
      "26240/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9927Step 821 / 1718\n",
      "26272/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9927Step 822 / 1718\n",
      "26304/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9927Step 823 / 1718\n",
      "26336/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9927Step 824 / 1718\n",
      "26368/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9927Step 825 / 1718\n",
      "26400/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9927Step 826 / 1718\n",
      "26432/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9927Step 827 / 1718\n",
      "26464/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9927Step 828 / 1718\n",
      "26496/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9927Step 829 / 1718\n",
      "26528/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9927Step 830 / 1718\n",
      "26560/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9927Step 831 / 1718\n",
      "26592/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9927Step 832 / 1718\n",
      "26624/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9927Step 833 / 1718\n",
      "26656/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9927Step 834 / 1718\n",
      "26688/55000 - mean: 0.0247 - sparse_categorical_accuracy: 0.9927Step 835 / 1718\n",
      "26720/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9927Step 836 / 1718\n",
      "26752/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9927Step 837 / 1718\n",
      "26784/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9927Step 838 / 1718\n",
      "26816/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9927Step 839 / 1718\n",
      "26848/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9927Step 840 / 1718\n",
      "26880/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9927Step 841 / 1718\n",
      "26912/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9927Step 842 / 1718\n",
      "26944/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9927Step 843 / 1718\n",
      "26976/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9927Step 844 / 1718\n",
      "27008/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9927Step 845 / 1718\n",
      "27040/55000 - mean: 0.0246 - sparse_categorical_accuracy: 0.9927Step 846 / 1718\n",
      "27072/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9927Step 847 / 1718\n",
      "27104/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9927Step 848 / 1718\n",
      "27136/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9927Step 849 / 1718\n",
      "27168/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9927Step 850 / 1718\n",
      "27200/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9928Step 851 / 1718\n",
      "27232/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9928Step 852 / 1718\n",
      "27264/55000 - mean: 0.0245 - sparse_categorical_accuracy: 0.9927Step 853 / 1718\n",
      "27296/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9927Step 854 / 1718\n",
      "27328/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9927Step 855 / 1718\n",
      "27360/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9927Step 856 / 1718\n",
      "27392/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9927Step 857 / 1718\n",
      "27424/55000 - mean: 0.0244 - sparse_categorical_accuracy: 0.9927Step 858 / 1718\n",
      "27456/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9928Step 859 / 1718\n",
      "27488/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9928Step 860 / 1718\n",
      "27520/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9928Step 861 / 1718\n",
      "27552/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9928Step 862 / 1718\n",
      "27584/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9928Step 863 / 1718\n",
      "27616/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9928Step 864 / 1718\n",
      "27648/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9928Step 865 / 1718\n",
      "27680/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9928Step 866 / 1718\n",
      "27712/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9928Step 867 / 1718\n",
      "27744/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9928Step 868 / 1718\n",
      "27776/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9928Step 869 / 1718\n",
      "27808/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9928Step 870 / 1718\n",
      "27840/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9928Step 871 / 1718\n",
      "27872/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9928Step 872 / 1718\n",
      "27904/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9928Step 873 / 1718\n",
      "27936/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9928Step 874 / 1718\n",
      "27968/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9928Step 875 / 1718\n",
      "28000/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9928Step 876 / 1718\n",
      "28032/55000 - mean: 0.0240 - sparse_categorical_accuracy: 0.9928Step 877 / 1718\n",
      "28064/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9928Step 878 / 1718\n",
      "28096/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9928Step 879 / 1718\n",
      "28128/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9929Step 880 / 1718\n",
      "28160/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9929Step 881 / 1718\n",
      "28192/55000 - mean: 0.0239 - sparse_categorical_accuracy: 0.9929Step 882 / 1718\n",
      "28224/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9929Step 883 / 1718\n",
      "28256/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9929Step 884 / 1718\n",
      "28288/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9929Step 885 / 1718\n",
      "28320/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9929Step 886 / 1718\n",
      "28352/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9929Step 887 / 1718\n",
      "28384/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9929Step 888 / 1718\n",
      "28416/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9929Step 889 / 1718\n",
      "28448/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9929Step 890 / 1718\n",
      "28480/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9929Step 891 / 1718\n",
      "28512/55000 - mean: 0.0238 - sparse_categorical_accuracy: 0.9929Step 892 / 1718\n",
      "28544/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9929Step 893 / 1718\n",
      "28576/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9929Step 894 / 1718\n",
      "28608/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9929Step 895 / 1718\n",
      "28640/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9929Step 896 / 1718\n",
      "28672/55000 - mean: 0.0237 - sparse_categorical_accuracy: 0.9929Step 897 / 1718\n",
      "28704/55000 - mean: 0.0236 - sparse_categorical_accuracy: 0.9929Step 898 / 1718\n",
      "28736/55000 - mean: 0.0236 - sparse_categorical_accuracy: 0.9929Step 899 / 1718\n",
      "28768/55000 - mean: 0.0236 - sparse_categorical_accuracy: 0.9929Step 900 / 1718\n",
      "28800/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9929Step 901 / 1718\n",
      "28832/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9929Step 902 / 1718\n",
      "28864/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9929Step 903 / 1718\n",
      "28896/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9929Step 904 / 1718\n",
      "28928/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9929Step 905 / 1718\n",
      "28960/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9930Step 906 / 1718\n",
      "28992/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9930Step 907 / 1718\n",
      "29024/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9929Step 908 / 1718\n",
      "29056/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9929Step 909 / 1718\n",
      "29088/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9930Step 910 / 1718\n",
      "29120/55000 - mean: 0.0243 - sparse_categorical_accuracy: 0.9930Step 911 / 1718\n",
      "29152/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9930Step 912 / 1718\n",
      "29184/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9930Step 913 / 1718\n",
      "29216/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9930Step 914 / 1718\n",
      "29248/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9930Step 915 / 1718\n",
      "29280/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9930Step 916 / 1718\n",
      "29312/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9930Step 917 / 1718\n",
      "29344/55000 - mean: 0.0241 - sparse_categorical_accuracy: 0.9930Step 918 / 1718\n",
      "29376/55000 - mean: 0.0242 - sparse_categorical_accuracy: 0.9930Step 919 / 1718\n",
      "29408/55000 - mean: 0.0248 - sparse_categorical_accuracy: 0.9930Step 920 / 1718\n",
      "29440/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9929Step 921 / 1718\n",
      "29472/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9929Step 922 / 1718\n",
      "29504/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9930Step 923 / 1718\n",
      "29536/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9930Step 924 / 1718\n",
      "29568/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 925 / 1718\n",
      "29600/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 926 / 1718\n",
      "29632/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9929Step 927 / 1718\n",
      "29664/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9930Step 928 / 1718\n",
      "29696/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 929 / 1718\n",
      "29728/55000 - mean: 0.0249 - sparse_categorical_accuracy: 0.9929Step 930 / 1718\n",
      "29760/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 931 / 1718\n",
      "29792/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 932 / 1718\n",
      "29824/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 933 / 1718\n",
      "29856/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 934 / 1718\n",
      "29888/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 935 / 1718\n",
      "29920/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 936 / 1718\n",
      "29952/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 937 / 1718\n",
      "29984/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 938 / 1718\n",
      "30016/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 939 / 1718\n",
      "30048/55000 - mean: 0.0250 - sparse_categorical_accuracy: 0.9929Step 940 / 1718\n",
      "30080/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 941 / 1718\n",
      "30112/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 942 / 1718\n",
      "30144/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 943 / 1718\n",
      "30176/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 944 / 1718\n",
      "30208/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 945 / 1718\n",
      "30240/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 946 / 1718\n",
      "30272/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 947 / 1718\n",
      "30304/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 948 / 1718\n",
      "30336/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 949 / 1718\n",
      "30368/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 950 / 1718\n",
      "30400/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 951 / 1718\n",
      "30432/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 952 / 1718\n",
      "30464/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 953 / 1718\n",
      "30496/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 954 / 1718\n",
      "30528/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 955 / 1718\n",
      "30560/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 956 / 1718\n",
      "30592/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 957 / 1718\n",
      "30624/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 958 / 1718\n",
      "30656/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 959 / 1718\n",
      "30688/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 960 / 1718\n",
      "30720/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 961 / 1718\n",
      "30752/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9928Step 962 / 1718\n",
      "30784/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9928Step 963 / 1718\n",
      "30816/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9928Step 964 / 1718\n",
      "30848/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 965 / 1718\n",
      "30880/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 966 / 1718\n",
      "30912/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 967 / 1718\n",
      "30944/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 968 / 1718\n",
      "30976/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 969 / 1718\n",
      "31008/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 970 / 1718\n",
      "31040/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 971 / 1718\n",
      "31072/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 972 / 1718\n",
      "31104/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 973 / 1718\n",
      "31136/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9928Step 974 / 1718\n",
      "31168/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9928Step 975 / 1718\n",
      "31200/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9928Step 976 / 1718\n",
      "31232/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 977 / 1718\n",
      "31264/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 978 / 1718\n",
      "31296/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 979 / 1718\n",
      "31328/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 980 / 1718\n",
      "31360/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 981 / 1718\n",
      "31392/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9929Step 982 / 1718\n",
      "31424/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9929Step 983 / 1718\n",
      "31456/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 984 / 1718\n",
      "31488/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9929Step 985 / 1718\n",
      "31520/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9929Step 986 / 1718\n",
      "31552/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 987 / 1718\n",
      "31584/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 988 / 1718\n",
      "31616/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 989 / 1718\n",
      "31648/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 990 / 1718\n",
      "31680/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 991 / 1718\n",
      "31712/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 992 / 1718\n",
      "31744/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 993 / 1718\n",
      "31776/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 994 / 1718\n",
      "31808/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 995 / 1718\n",
      "31840/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 996 / 1718\n",
      "31872/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 997 / 1718\n",
      "31904/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 998 / 1718\n",
      "31936/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 999 / 1718\n",
      "31968/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 1000 / 1718\n",
      "32000/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 1001 / 1718\n",
      "32032/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1002 / 1718\n",
      "32064/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1003 / 1718\n",
      "32096/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1004 / 1718\n",
      "32128/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1005 / 1718\n",
      "32160/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1006 / 1718\n",
      "32192/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1007 / 1718\n",
      "32224/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1008 / 1718\n",
      "32256/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1009 / 1718\n",
      "32288/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1010 / 1718\n",
      "32320/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1011 / 1718\n",
      "32352/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1012 / 1718\n",
      "32384/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1013 / 1718\n",
      "32416/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1014 / 1718\n",
      "32448/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1015 / 1718\n",
      "32480/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1016 / 1718\n",
      "32512/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1017 / 1718\n",
      "32544/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9927Step 1018 / 1718\n",
      "32576/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9928Step 1019 / 1718\n",
      "32608/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9928Step 1020 / 1718\n",
      "32640/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9928Step 1021 / 1718\n",
      "32672/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9928Step 1022 / 1718\n",
      "32704/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9928Step 1023 / 1718\n",
      "32736/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9928Step 1024 / 1718\n",
      "32768/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9928Step 1025 / 1718\n",
      "32800/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9928Step 1026 / 1718\n",
      "32832/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9928Step 1027 / 1718\n",
      "32864/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9927Step 1028 / 1718\n",
      "32896/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9927Step 1029 / 1718\n",
      "32928/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9927Step 1030 / 1718\n",
      "32960/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9927Step 1031 / 1718\n",
      "32992/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9928Step 1032 / 1718\n",
      "33024/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9928Step 1033 / 1718\n",
      "33056/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9927Step 1034 / 1718\n",
      "33088/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9927Step 1035 / 1718\n",
      "33120/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 1036 / 1718\n",
      "33152/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9927Step 1037 / 1718\n",
      "33184/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9927Step 1038 / 1718\n",
      "33216/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9927Step 1039 / 1718\n",
      "33248/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 1040 / 1718\n",
      "33280/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 1041 / 1718\n",
      "33312/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 1042 / 1718\n",
      "33344/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 1043 / 1718\n",
      "33376/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 1044 / 1718\n",
      "33408/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 1045 / 1718\n",
      "33440/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 1046 / 1718\n",
      "33472/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1047 / 1718\n",
      "33504/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1048 / 1718\n",
      "33536/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1049 / 1718\n",
      "33568/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1050 / 1718\n",
      "33600/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 1051 / 1718\n",
      "33632/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 1052 / 1718\n",
      "33664/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1053 / 1718\n",
      "33696/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1054 / 1718\n",
      "33728/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1055 / 1718\n",
      "33760/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1056 / 1718\n",
      "33792/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1057 / 1718\n",
      "33824/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9928Step 1058 / 1718\n",
      "33856/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1059 / 1718\n",
      "33888/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1060 / 1718\n",
      "33920/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1061 / 1718\n",
      "33952/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1062 / 1718\n",
      "33984/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1063 / 1718\n",
      "34016/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1064 / 1718\n",
      "34048/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1065 / 1718\n",
      "34080/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1066 / 1718\n",
      "34112/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1067 / 1718\n",
      "34144/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1068 / 1718\n",
      "34176/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1069 / 1718\n",
      "34208/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1070 / 1718\n",
      "34240/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1071 / 1718\n",
      "34272/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1072 / 1718\n",
      "34304/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1073 / 1718\n",
      "34336/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1074 / 1718\n",
      "34368/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1075 / 1718\n",
      "34400/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 1076 / 1718\n",
      "34432/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9928Step 1077 / 1718\n",
      "34464/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 1078 / 1718\n",
      "34496/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 1079 / 1718\n",
      "34528/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 1080 / 1718\n",
      "34560/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 1081 / 1718\n",
      "34592/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9929Step 1082 / 1718\n",
      "34624/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9929Step 1083 / 1718\n",
      "34656/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9929Step 1084 / 1718\n",
      "34688/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9929Step 1085 / 1718\n",
      "34720/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9929Step 1086 / 1718\n",
      "34752/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1087 / 1718\n",
      "34784/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1088 / 1718\n",
      "34816/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1089 / 1718\n",
      "34848/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1090 / 1718\n",
      "34880/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1091 / 1718\n",
      "34912/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1092 / 1718\n",
      "34944/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1093 / 1718\n",
      "34976/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1094 / 1718\n",
      "35008/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1095 / 1718\n",
      "35040/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9930Step 1096 / 1718\n",
      "35072/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9930Step 1097 / 1718\n",
      "35104/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9930Step 1098 / 1718\n",
      "35136/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1099 / 1718\n",
      "35168/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1100 / 1718\n",
      "35200/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9930Step 1101 / 1718\n",
      "35232/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9930Step 1102 / 1718\n",
      "35264/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9930Step 1103 / 1718\n",
      "35296/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1104 / 1718\n",
      "35328/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9930Step 1105 / 1718\n",
      "35360/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9930Step 1106 / 1718\n",
      "35392/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1107 / 1718\n",
      "35424/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1108 / 1718\n",
      "35456/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1109 / 1718\n",
      "35488/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9930Step 1110 / 1718\n",
      "35520/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9930Step 1111 / 1718\n",
      "35552/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9930Step 1112 / 1718\n",
      "35584/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9930Step 1113 / 1718\n",
      "35616/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9930Step 1114 / 1718\n",
      "35648/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9930Step 1115 / 1718\n",
      "35680/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9930Step 1116 / 1718\n",
      "35712/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9930Step 1117 / 1718\n",
      "35744/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9930Step 1118 / 1718\n",
      "35776/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9930Step 1119 / 1718\n",
      "35808/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9930Step 1120 / 1718\n",
      "35840/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9930Step 1121 / 1718\n",
      "35872/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 1122 / 1718\n",
      "35904/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 1123 / 1718\n",
      "35936/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 1124 / 1718\n",
      "35968/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 1125 / 1718\n",
      "36000/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1126 / 1718\n",
      "36032/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9930Step 1127 / 1718\n",
      "36064/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9930Step 1128 / 1718\n",
      "36096/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9930Step 1129 / 1718\n",
      "36128/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9930Step 1130 / 1718\n",
      "36160/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9930Step 1131 / 1718\n",
      "36192/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 1132 / 1718\n",
      "36224/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 1133 / 1718\n",
      "36256/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 1134 / 1718\n",
      "36288/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9930Step 1135 / 1718\n",
      "36320/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 1136 / 1718\n",
      "36352/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 1137 / 1718\n",
      "36384/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 1138 / 1718\n",
      "36416/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 1139 / 1718\n",
      "36448/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 1140 / 1718\n",
      "36480/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9930Step 1141 / 1718\n",
      "36512/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9930Step 1142 / 1718\n",
      "36544/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9930Step 1143 / 1718\n",
      "36576/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9930Step 1144 / 1718\n",
      "36608/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9930Step 1145 / 1718\n",
      "36640/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1146 / 1718\n",
      "36672/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1147 / 1718\n",
      "36704/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1148 / 1718\n",
      "36736/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1149 / 1718\n",
      "36768/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1150 / 1718\n",
      "36800/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1151 / 1718\n",
      "36832/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1152 / 1718\n",
      "36864/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1153 / 1718\n",
      "36896/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1154 / 1718\n",
      "36928/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1155 / 1718\n",
      "36960/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1156 / 1718\n",
      "36992/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1157 / 1718\n",
      "37024/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1158 / 1718\n",
      "37056/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1159 / 1718\n",
      "37088/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1160 / 1718\n",
      "37120/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1161 / 1718\n",
      "37152/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1162 / 1718\n",
      "37184/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1163 / 1718\n",
      "37216/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1164 / 1718\n",
      "37248/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1165 / 1718\n",
      "37280/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1166 / 1718\n",
      "37312/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1167 / 1718\n",
      "37344/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1168 / 1718\n",
      "37376/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1169 / 1718\n",
      "37408/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1170 / 1718\n",
      "37440/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1171 / 1718\n",
      "37472/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1172 / 1718\n",
      "37504/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1173 / 1718\n",
      "37536/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1174 / 1718\n",
      "37568/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1175 / 1718\n",
      "37600/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1176 / 1718\n",
      "37632/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1177 / 1718\n",
      "37664/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1178 / 1718\n",
      "37696/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1179 / 1718\n",
      "37728/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1180 / 1718\n",
      "37760/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1181 / 1718\n",
      "37792/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1182 / 1718\n",
      "37824/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1183 / 1718\n",
      "37856/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1184 / 1718\n",
      "37888/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1185 / 1718\n",
      "37920/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1186 / 1718\n",
      "37952/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1187 / 1718\n",
      "37984/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1188 / 1718\n",
      "38016/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1189 / 1718\n",
      "38048/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1190 / 1718\n",
      "38080/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1191 / 1718\n",
      "38112/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1192 / 1718\n",
      "38144/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1193 / 1718\n",
      "38176/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1194 / 1718\n",
      "38208/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1195 / 1718\n",
      "38240/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1196 / 1718\n",
      "38272/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1197 / 1718\n",
      "38304/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1198 / 1718\n",
      "38336/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 1199 / 1718\n",
      "38368/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 1200 / 1718\n",
      "38400/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 1201 / 1718\n",
      "38432/55000 - mean: 0.0251 - sparse_categorical_accuracy: 0.9929Step 1202 / 1718\n",
      "38464/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1203 / 1718\n",
      "38496/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1204 / 1718\n",
      "38528/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1205 / 1718\n",
      "38560/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1206 / 1718\n",
      "38592/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1207 / 1718\n",
      "38624/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1208 / 1718\n",
      "38656/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1209 / 1718\n",
      "38688/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9929Step 1210 / 1718\n",
      "38720/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9929Step 1211 / 1718\n",
      "38752/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9929Step 1212 / 1718\n",
      "38784/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1213 / 1718\n",
      "38816/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1214 / 1718\n",
      "38848/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1215 / 1718\n",
      "38880/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1216 / 1718\n",
      "38912/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1217 / 1718\n",
      "38944/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1218 / 1718\n",
      "38976/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1219 / 1718\n",
      "39008/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1220 / 1718\n",
      "39040/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1221 / 1718\n",
      "39072/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9929Step 1222 / 1718\n",
      "39104/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9929Step 1223 / 1718\n",
      "39136/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9929Step 1224 / 1718\n",
      "39168/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9929Step 1225 / 1718\n",
      "39200/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9929Step 1226 / 1718\n",
      "39232/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9929Step 1227 / 1718\n",
      "39264/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9929Step 1228 / 1718\n",
      "39296/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9929Step 1229 / 1718\n",
      "39328/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9929Step 1230 / 1718\n",
      "39360/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9929Step 1231 / 1718\n",
      "39392/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9929Step 1232 / 1718\n",
      "39424/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9929Step 1233 / 1718\n",
      "39456/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9929Step 1234 / 1718\n",
      "39488/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9929Step 1235 / 1718\n",
      "39520/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9929Step 1236 / 1718\n",
      "39552/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9929Step 1237 / 1718\n",
      "39584/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9929Step 1238 / 1718\n",
      "39616/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9929Step 1239 / 1718\n",
      "39648/55000 - mean: 0.0259 - sparse_categorical_accuracy: 0.9929Step 1240 / 1718\n",
      "39680/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9929Step 1241 / 1718\n",
      "39712/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9929Step 1242 / 1718\n",
      "39744/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9929Step 1243 / 1718\n",
      "39776/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9929Step 1244 / 1718\n",
      "39808/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9929Step 1245 / 1718\n",
      "39840/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 1246 / 1718\n",
      "39872/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 1247 / 1718\n",
      "39904/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 1248 / 1718\n",
      "39936/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 1249 / 1718\n",
      "39968/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 1250 / 1718\n",
      "40000/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 1251 / 1718\n",
      "40032/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 1252 / 1718\n",
      "40064/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 1253 / 1718\n",
      "40096/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 1254 / 1718\n",
      "40128/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 1255 / 1718\n",
      "40160/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9930Step 1256 / 1718\n",
      "40192/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9930Step 1257 / 1718\n",
      "40224/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9929Step 1258 / 1718\n",
      "40256/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9929Step 1259 / 1718\n",
      "40288/55000 - mean: 0.0258 - sparse_categorical_accuracy: 0.9929Step 1260 / 1718\n",
      "40320/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 1261 / 1718\n",
      "40352/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 1262 / 1718\n",
      "40384/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 1263 / 1718\n",
      "40416/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 1264 / 1718\n",
      "40448/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9930Step 1265 / 1718\n",
      "40480/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9930Step 1266 / 1718\n",
      "40512/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9930Step 1267 / 1718\n",
      "40544/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 1268 / 1718\n",
      "40576/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9930Step 1269 / 1718\n",
      "40608/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9930Step 1270 / 1718\n",
      "40640/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9929Step 1271 / 1718\n",
      "40672/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9929Step 1272 / 1718\n",
      "40704/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9929Step 1273 / 1718\n",
      "40736/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9930Step 1274 / 1718\n",
      "40768/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9930Step 1275 / 1718\n",
      "40800/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9930Step 1276 / 1718\n",
      "40832/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9930Step 1277 / 1718\n",
      "40864/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9930Step 1278 / 1718\n",
      "40896/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9930Step 1279 / 1718\n",
      "40928/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9930Step 1280 / 1718\n",
      "40960/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9930Step 1281 / 1718\n",
      "40992/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1282 / 1718\n",
      "41024/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1283 / 1718\n",
      "41056/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1284 / 1718\n",
      "41088/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1285 / 1718\n",
      "41120/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1286 / 1718\n",
      "41152/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1287 / 1718\n",
      "41184/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1288 / 1718\n",
      "41216/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1289 / 1718\n",
      "41248/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1290 / 1718\n",
      "41280/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1291 / 1718\n",
      "41312/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1292 / 1718\n",
      "41344/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1293 / 1718\n",
      "41376/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1294 / 1718\n",
      "41408/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1295 / 1718\n",
      "41440/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1296 / 1718\n",
      "41472/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1297 / 1718\n",
      "41504/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1298 / 1718\n",
      "41536/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1299 / 1718\n",
      "41568/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9930Step 1300 / 1718\n",
      "41600/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1301 / 1718\n",
      "41632/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1302 / 1718\n",
      "41664/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1303 / 1718\n",
      "41696/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1304 / 1718\n",
      "41728/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1305 / 1718\n",
      "41760/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1306 / 1718\n",
      "41792/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1307 / 1718\n",
      "41824/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1308 / 1718\n",
      "41856/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1309 / 1718\n",
      "41888/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1310 / 1718\n",
      "41920/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1311 / 1718\n",
      "41952/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1312 / 1718\n",
      "41984/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1313 / 1718\n",
      "42016/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1314 / 1718\n",
      "42048/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1315 / 1718\n",
      "42080/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1316 / 1718\n",
      "42112/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1317 / 1718\n",
      "42144/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1318 / 1718\n",
      "42176/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1319 / 1718\n",
      "42208/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1320 / 1718\n",
      "42240/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1321 / 1718\n",
      "42272/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1322 / 1718\n",
      "42304/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1323 / 1718\n",
      "42336/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1324 / 1718\n",
      "42368/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1325 / 1718\n",
      "42400/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1326 / 1718\n",
      "42432/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1327 / 1718\n",
      "42464/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1328 / 1718\n",
      "42496/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1329 / 1718\n",
      "42528/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1330 / 1718\n",
      "42560/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1331 / 1718\n",
      "42592/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1332 / 1718\n",
      "42624/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1333 / 1718\n",
      "42656/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1334 / 1718\n",
      "42688/55000 - mean: 0.0252 - sparse_categorical_accuracy: 0.9929Step 1335 / 1718\n",
      "42720/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1336 / 1718\n",
      "42752/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1337 / 1718\n",
      "42784/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1338 / 1718\n",
      "42816/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1339 / 1718\n",
      "42848/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1340 / 1718\n",
      "42880/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1341 / 1718\n",
      "42912/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1342 / 1718\n",
      "42944/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1343 / 1718\n",
      "42976/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1344 / 1718\n",
      "43008/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1345 / 1718\n",
      "43040/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1346 / 1718\n",
      "43072/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1347 / 1718\n",
      "43104/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1348 / 1718\n",
      "43136/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1349 / 1718\n",
      "43168/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1350 / 1718\n",
      "43200/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1351 / 1718\n",
      "43232/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9929Step 1352 / 1718\n",
      "43264/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1353 / 1718\n",
      "43296/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1354 / 1718\n",
      "43328/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1355 / 1718\n",
      "43360/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1356 / 1718\n",
      "43392/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1357 / 1718\n",
      "43424/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1358 / 1718\n",
      "43456/55000 - mean: 0.0255 - sparse_categorical_accuracy: 0.9928Step 1359 / 1718\n",
      "43488/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9928Step 1360 / 1718\n",
      "43520/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1361 / 1718\n",
      "43552/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1362 / 1718\n",
      "43584/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9928Step 1363 / 1718\n",
      "43616/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9928Step 1364 / 1718\n",
      "43648/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1365 / 1718\n",
      "43680/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9929Step 1366 / 1718\n",
      "43712/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9928Step 1367 / 1718\n",
      "43744/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9928Step 1368 / 1718\n",
      "43776/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9928Step 1369 / 1718\n",
      "43808/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1370 / 1718\n",
      "43840/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1371 / 1718\n",
      "43872/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1372 / 1718\n",
      "43904/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1373 / 1718\n",
      "43936/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1374 / 1718\n",
      "43968/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9928Step 1375 / 1718\n",
      "44000/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9928Step 1376 / 1718\n",
      "44032/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9928Step 1377 / 1718\n",
      "44064/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9929Step 1378 / 1718\n",
      "44096/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9928Step 1379 / 1718\n",
      "44128/55000 - mean: 0.0254 - sparse_categorical_accuracy: 0.9928Step 1380 / 1718\n",
      "44160/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9928Step 1381 / 1718\n",
      "44192/55000 - mean: 0.0253 - sparse_categorical_accuracy: 0.9928Step 1382 / 1718\n",
      "44224/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 1383 / 1718\n",
      "44256/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 1384 / 1718\n",
      "44288/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 1385 / 1718\n",
      "44320/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 1386 / 1718\n",
      "44352/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 1387 / 1718\n",
      "44384/55000 - mean: 0.0257 - sparse_categorical_accuracy: 0.9928Step 1388 / 1718\n",
      "44416/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 1389 / 1718\n",
      "44448/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 1390 / 1718\n",
      "44480/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 1391 / 1718\n",
      "44512/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 1392 / 1718\n",
      "44544/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 1393 / 1718\n",
      "44576/55000 - mean: 0.0256 - sparse_categorical_accuracy: 0.9928Step 1394 / 1718\n",
      "44608/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9928Step 1395 / 1718\n",
      "44640/55000 - mean: 0.0261 - sparse_categorical_accuracy: 0.9928Step 1396 / 1718\n",
      "44672/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9928Step 1397 / 1718\n",
      "44704/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9928Step 1398 / 1718\n",
      "44736/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9928Step 1399 / 1718\n",
      "44768/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9928Step 1400 / 1718\n",
      "44800/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9928Step 1401 / 1718\n",
      "44832/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9928Step 1402 / 1718\n",
      "44864/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9928Step 1403 / 1718\n",
      "44896/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9927Step 1404 / 1718\n",
      "44928/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9927Step 1405 / 1718\n",
      "44960/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9927Step 1406 / 1718\n",
      "44992/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9928Step 1407 / 1718\n",
      "45024/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9928Step 1408 / 1718\n",
      "45056/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9928Step 1409 / 1718\n",
      "45088/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9928Step 1410 / 1718\n",
      "45120/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9928Step 1411 / 1718\n",
      "45152/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9928Step 1412 / 1718\n",
      "45184/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9928Step 1413 / 1718\n",
      "45216/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9928Step 1414 / 1718\n",
      "45248/55000 - mean: 0.0263 - sparse_categorical_accuracy: 0.9928Step 1415 / 1718\n",
      "45280/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9928Step 1416 / 1718\n",
      "45312/55000 - mean: 0.0262 - sparse_categorical_accuracy: 0.9928Step 1417 / 1718\n",
      "45344/55000 - mean: 0.0264 - sparse_categorical_accuracy: 0.9928Step 1418 / 1718\n",
      "45376/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9928Step 1419 / 1718\n",
      "45408/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9928Step 1420 / 1718\n",
      "45440/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9928Step 1421 / 1718\n",
      "45472/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9927Step 1422 / 1718\n",
      "45504/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9927Step 1423 / 1718\n",
      "45536/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9927Step 1424 / 1718\n",
      "45568/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9927Step 1425 / 1718\n",
      "45600/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9927Step 1426 / 1718\n",
      "45632/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9927Step 1427 / 1718\n",
      "45664/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9928Step 1428 / 1718\n",
      "45696/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9927Step 1429 / 1718\n",
      "45728/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9927Step 1430 / 1718\n",
      "45760/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9927Step 1431 / 1718\n",
      "45792/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9927Step 1432 / 1718\n",
      "45824/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9927Step 1433 / 1718\n",
      "45856/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9927Step 1434 / 1718\n",
      "45888/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9927Step 1435 / 1718\n",
      "45920/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9927Step 1436 / 1718\n",
      "45952/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9927Step 1437 / 1718\n",
      "45984/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9927Step 1438 / 1718\n",
      "46016/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9927Step 1439 / 1718\n",
      "46048/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9927Step 1440 / 1718\n",
      "46080/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9927Step 1441 / 1718\n",
      "46112/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9927Step 1442 / 1718\n",
      "46144/55000 - mean: 0.0265 - sparse_categorical_accuracy: 0.9927Step 1443 / 1718\n",
      "46176/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9927Step 1444 / 1718\n",
      "46208/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9927Step 1445 / 1718\n",
      "46240/55000 - mean: 0.0266 - sparse_categorical_accuracy: 0.9927Step 1446 / 1718\n",
      "46272/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1447 / 1718\n",
      "46304/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1448 / 1718\n",
      "46336/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1449 / 1718\n",
      "46368/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1450 / 1718\n",
      "46400/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1451 / 1718\n",
      "46432/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1452 / 1718\n",
      "46464/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1453 / 1718\n",
      "46496/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1454 / 1718\n",
      "46528/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1455 / 1718\n",
      "46560/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1456 / 1718\n",
      "46592/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1457 / 1718\n",
      "46624/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1458 / 1718\n",
      "46656/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1459 / 1718\n",
      "46688/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1460 / 1718\n",
      "46720/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1461 / 1718\n",
      "46752/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1462 / 1718\n",
      "46784/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1463 / 1718\n",
      "46816/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1464 / 1718\n",
      "46848/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1465 / 1718\n",
      "46880/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1466 / 1718\n",
      "46912/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1467 / 1718\n",
      "46944/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1468 / 1718\n",
      "46976/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1469 / 1718\n",
      "47008/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1470 / 1718\n",
      "47040/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1471 / 1718\n",
      "47072/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1472 / 1718\n",
      "47104/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1473 / 1718\n",
      "47136/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1474 / 1718\n",
      "47168/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1475 / 1718\n",
      "47200/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1476 / 1718\n",
      "47232/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1477 / 1718\n",
      "47264/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1478 / 1718\n",
      "47296/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1479 / 1718\n",
      "47328/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1480 / 1718\n",
      "47360/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1481 / 1718\n",
      "47392/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1482 / 1718\n",
      "47424/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1483 / 1718\n",
      "47456/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1484 / 1718\n",
      "47488/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1485 / 1718\n",
      "47520/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1486 / 1718\n",
      "47552/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1487 / 1718\n",
      "47584/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1488 / 1718\n",
      "47616/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9928Step 1489 / 1718\n",
      "47648/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9928Step 1490 / 1718\n",
      "47680/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1491 / 1718\n",
      "47712/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1492 / 1718\n",
      "47744/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1493 / 1718\n",
      "47776/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1494 / 1718\n",
      "47808/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1495 / 1718\n",
      "47840/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1496 / 1718\n",
      "47872/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1497 / 1718\n",
      "47904/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1498 / 1718\n",
      "47936/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1499 / 1718\n",
      "47968/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1500 / 1718\n",
      "48000/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1501 / 1718\n",
      "48032/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1502 / 1718\n",
      "48064/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1503 / 1718\n",
      "48096/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1504 / 1718\n",
      "48128/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1505 / 1718\n",
      "48160/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1506 / 1718\n",
      "48192/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1507 / 1718\n",
      "48224/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1508 / 1718\n",
      "48256/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1509 / 1718\n",
      "48288/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1510 / 1718\n",
      "48320/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1511 / 1718\n",
      "48352/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1512 / 1718\n",
      "48384/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1513 / 1718\n",
      "48416/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1514 / 1718\n",
      "48448/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1515 / 1718\n",
      "48480/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1516 / 1718\n",
      "48512/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1517 / 1718\n",
      "48544/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1518 / 1718\n",
      "48576/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1519 / 1718\n",
      "48608/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1520 / 1718\n",
      "48640/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1521 / 1718\n",
      "48672/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1522 / 1718\n",
      "48704/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1523 / 1718\n",
      "48736/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1524 / 1718\n",
      "48768/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1525 / 1718\n",
      "48800/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1526 / 1718\n",
      "48832/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1527 / 1718\n",
      "48864/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1528 / 1718\n",
      "48896/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1529 / 1718\n",
      "48928/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1530 / 1718\n",
      "48960/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1531 / 1718\n",
      "48992/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1532 / 1718\n",
      "49024/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1533 / 1718\n",
      "49056/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1534 / 1718\n",
      "49088/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1535 / 1718\n",
      "49120/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1536 / 1718\n",
      "49152/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1537 / 1718\n",
      "49184/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1538 / 1718\n",
      "49216/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1539 / 1718\n",
      "49248/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1540 / 1718\n",
      "49280/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1541 / 1718\n",
      "49312/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1542 / 1718\n",
      "49344/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1543 / 1718\n",
      "49376/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1544 / 1718\n",
      "49408/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1545 / 1718\n",
      "49440/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1546 / 1718\n",
      "49472/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1547 / 1718\n",
      "49504/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1548 / 1718\n",
      "49536/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1549 / 1718\n",
      "49568/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1550 / 1718\n",
      "49600/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1551 / 1718\n",
      "49632/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1552 / 1718\n",
      "49664/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1553 / 1718\n",
      "49696/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1554 / 1718\n",
      "49728/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1555 / 1718\n",
      "49760/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9926Step 1556 / 1718\n",
      "49792/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9926Step 1557 / 1718\n",
      "49824/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1558 / 1718\n",
      "49856/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1559 / 1718\n",
      "49888/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1560 / 1718\n",
      "49920/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1561 / 1718\n",
      "49952/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1562 / 1718\n",
      "49984/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1563 / 1718\n",
      "50016/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1564 / 1718\n",
      "50048/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1565 / 1718\n",
      "50080/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1566 / 1718\n",
      "50112/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1567 / 1718\n",
      "50144/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1568 / 1718\n",
      "50176/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1569 / 1718\n",
      "50208/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1570 / 1718\n",
      "50240/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1571 / 1718\n",
      "50272/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1572 / 1718\n",
      "50304/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1573 / 1718\n",
      "50336/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1574 / 1718\n",
      "50368/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1575 / 1718\n",
      "50400/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1576 / 1718\n",
      "50432/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1577 / 1718\n",
      "50464/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1578 / 1718\n",
      "50496/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1579 / 1718\n",
      "50528/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1580 / 1718\n",
      "50560/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1581 / 1718\n",
      "50592/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1582 / 1718\n",
      "50624/55000 - mean: 0.0270 - sparse_categorical_accuracy: 0.9927Step 1583 / 1718\n",
      "50656/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1584 / 1718\n",
      "50688/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1585 / 1718\n",
      "50720/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1586 / 1718\n",
      "50752/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1587 / 1718\n",
      "50784/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1588 / 1718\n",
      "50816/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1589 / 1718\n",
      "50848/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1590 / 1718\n",
      "50880/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1591 / 1718\n",
      "50912/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1592 / 1718\n",
      "50944/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1593 / 1718\n",
      "50976/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1594 / 1718\n",
      "51008/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1595 / 1718\n",
      "51040/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9928Step 1596 / 1718\n",
      "51072/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9928Step 1597 / 1718\n",
      "51104/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9928Step 1598 / 1718\n",
      "51136/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9928Step 1599 / 1718\n",
      "51168/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9928Step 1600 / 1718\n",
      "51200/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9928Step 1601 / 1718\n",
      "51232/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1602 / 1718\n",
      "51264/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1603 / 1718\n",
      "51296/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1604 / 1718\n",
      "51328/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9928Step 1605 / 1718\n",
      "51360/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1606 / 1718\n",
      "51392/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9927Step 1607 / 1718\n",
      "51424/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9927Step 1608 / 1718\n",
      "51456/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9928Step 1609 / 1718\n",
      "51488/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9928Step 1610 / 1718\n",
      "51520/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9927Step 1611 / 1718\n",
      "51552/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9927Step 1612 / 1718\n",
      "51584/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9927Step 1613 / 1718\n",
      "51616/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9927Step 1614 / 1718\n",
      "51648/55000 - mean: 0.0267 - sparse_categorical_accuracy: 0.9927Step 1615 / 1718\n",
      "51680/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1616 / 1718\n",
      "51712/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1617 / 1718\n",
      "51744/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1618 / 1718\n",
      "51776/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1619 / 1718\n",
      "51808/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1620 / 1718\n",
      "51840/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1621 / 1718\n",
      "51872/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1622 / 1718\n",
      "51904/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1623 / 1718\n",
      "51936/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1624 / 1718\n",
      "51968/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1625 / 1718\n",
      "52000/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1626 / 1718\n",
      "52032/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1627 / 1718\n",
      "52064/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1628 / 1718\n",
      "52096/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1629 / 1718\n",
      "52128/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1630 / 1718\n",
      "52160/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1631 / 1718\n",
      "52192/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1632 / 1718\n",
      "52224/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1633 / 1718\n",
      "52256/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1634 / 1718\n",
      "52288/55000 - mean: 0.0268 - sparse_categorical_accuracy: 0.9927Step 1635 / 1718\n",
      "52320/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1636 / 1718\n",
      "52352/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1637 / 1718\n",
      "52384/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1638 / 1718\n",
      "52416/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1639 / 1718\n",
      "52448/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1640 / 1718\n",
      "52480/55000 - mean: 0.0269 - sparse_categorical_accuracy: 0.9927Step 1641 / 1718\n",
      "52512/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1642 / 1718\n",
      "52544/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1643 / 1718\n",
      "52576/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1644 / 1718\n",
      "52608/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1645 / 1718\n",
      "52640/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1646 / 1718\n",
      "52672/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1647 / 1718\n",
      "52704/55000 - mean: 0.0271 - sparse_categorical_accuracy: 0.9927Step 1648 / 1718\n",
      "52736/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1649 / 1718\n",
      "52768/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9926Step 1650 / 1718\n",
      "52800/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1651 / 1718\n",
      "52832/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1652 / 1718\n",
      "52864/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9926Step 1653 / 1718\n",
      "52896/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9926Step 1654 / 1718\n",
      "52928/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1655 / 1718\n",
      "52960/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1656 / 1718\n",
      "52992/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1657 / 1718\n",
      "53024/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1658 / 1718\n",
      "53056/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1659 / 1718\n",
      "53088/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1660 / 1718\n",
      "53120/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1661 / 1718\n",
      "53152/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1662 / 1718\n",
      "53184/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1663 / 1718\n",
      "53216/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1664 / 1718\n",
      "53248/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1665 / 1718\n",
      "53280/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1666 / 1718\n",
      "53312/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1667 / 1718\n",
      "53344/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1668 / 1718\n",
      "53376/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1669 / 1718\n",
      "53408/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1670 / 1718\n",
      "53440/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9926Step 1671 / 1718\n",
      "53472/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1672 / 1718\n",
      "53504/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1673 / 1718\n",
      "53536/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1674 / 1718\n",
      "53568/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9926Step 1675 / 1718\n",
      "53600/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9926Step 1676 / 1718\n",
      "53632/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1677 / 1718\n",
      "53664/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1678 / 1718\n",
      "53696/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1679 / 1718\n",
      "53728/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1680 / 1718\n",
      "53760/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1681 / 1718\n",
      "53792/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1682 / 1718\n",
      "53824/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9926Step 1683 / 1718\n",
      "53856/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9926Step 1684 / 1718\n",
      "53888/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9926Step 1685 / 1718\n",
      "53920/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9926Step 1686 / 1718\n",
      "53952/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9926Step 1687 / 1718\n",
      "53984/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9926Step 1688 / 1718\n",
      "54016/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1689 / 1718\n",
      "54048/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1690 / 1718\n",
      "54080/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1691 / 1718\n",
      "54112/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1692 / 1718\n",
      "54144/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1693 / 1718\n",
      "54176/55000 - mean: 0.0273 - sparse_categorical_accuracy: 0.9927Step 1694 / 1718\n",
      "54208/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1695 / 1718\n",
      "54240/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1696 / 1718\n",
      "54272/55000 - mean: 0.0272 - sparse_categorical_accuracy: 0.9927Step 1697 / 1718\n",
      "54304/55000 - mean: 0.0275 - sparse_categorical_accuracy: 0.9927Step 1698 / 1718\n",
      "54336/55000 - mean: 0.0274 - sparse_categorical_accuracy: 0.9927Step 1699 / 1718\n",
      "54368/55000 - mean: 0.0276 - sparse_categorical_accuracy: 0.9926Step 1700 / 1718\n",
      "54400/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9926Step 1701 / 1718\n",
      "54432/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9926Step 1702 / 1718\n",
      "54464/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9926Step 1703 / 1718\n",
      "54496/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9926Step 1704 / 1718\n",
      "54528/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9926Step 1705 / 1718\n",
      "54560/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9926Step 1706 / 1718\n",
      "54592/55000 - mean: 0.0278 - sparse_categorical_accuracy: 0.9926Step 1707 / 1718\n",
      "54624/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9926Step 1708 / 1718\n",
      "54656/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9926Step 1709 / 1718\n",
      "54688/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9926Step 1710 / 1718\n",
      "54720/55000 - mean: 0.0277 - sparse_categorical_accuracy: 0.9926Step 1711 / 1718\n",
      "54752/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9926Step 1712 / 1718\n",
      "54784/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9926Step 1713 / 1718\n",
      "54816/55000 - mean: 0.0279 - sparse_categorical_accuracy: 0.9926Step 1714 / 1718\n",
      "54848/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9926Step 1715 / 1718\n",
      "54880/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9926Step 1716 / 1718\n",
      "54912/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9926Step 1717 / 1718\n",
      "54944/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9926Step 1718 / 1718\n",
      "55000/55000 - mean: 0.0281 - sparse_categorical_accuracy: 0.9926\n",
      "5000/5000 - mean: 0.2220 - sparse_categorical_accuracy: 0.9734\n"
     ]
    }
   ],
   "source": [
    "for epoch in range(1, n_epochs + 1):\n",
    "    print(f\"Epoch {epoch} / {n_epochs}\")\n",
    "    for step in range(1, n_steps + 1):\n",
    "        print(f\"Step {step} / {n_steps}\")\n",
    "        X_batch = X_train[(step-1)*batch_size:(step*batch_size)]\n",
    "        y_batch = y_train[(step-1)*batch_size:(step*batch_size)]\n",
    "        with tf.GradientTape(persistent=True) as tape:\n",
    "            y_pred = model(X_batch, training=True)\n",
    "            loss = tf.reduce_mean(loss_fn(y_batch, y_pred))\n",
    "        gradients_lower = tape.gradient(loss, model.trainable_variables[:2])\n",
    "        gradients_upper = tape.gradient(loss, model.trainable_variables[2:])\n",
    "        del tape\n",
    "        optimizer_lower.apply_gradients(zip(gradients_lower, model.trainable_variables[:2]))\n",
    "        optimizer_upper.apply_gradients(zip(gradients_upper, model.trainable_variables[2:]))\n",
    "        mean_loss(loss)\n",
    "\n",
    "        \n",
    "        for metric in metrics:\n",
    "            metric(y_batch, y_pred)\n",
    "\n",
    "            \n",
    "        print_status_bar(step * batch_size, len(y_train), mean_loss, metrics)\n",
    "\n",
    "    ### Validation  at the end of every epoch###\n",
    "    y_pred_valid = model(X_valid)\n",
    "    validation_loss = tf.reduce_mean(loss_fn(y_valid,y_pred_valid))\n",
    "    mean_validation_loss(validation_loss)\n",
    "    for metric in validation_metrics:\n",
    "        metric(y_valid, y_pred_valid)\n",
    "    \n",
    "    print_status_bar(len(y_train), len(y_train), mean_loss, metrics)\n",
    "    print_status_bar(len(y_valid), len(y_valid), mean_validation_loss, validation_metrics)\n",
    "\n",
    "    for metric in [mean_loss] + metrics:\n",
    "        metric.reset_states()\n",
    "        \n",
    "    for metric in [mean_validation_loss] + validation_metrics:\n",
    "        metric.reset_states()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.9.12 ('ml')",
   "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": "3.9.12"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "976019bd7028fad645a45aaa2b51f9e68e9bcad9b4e9047d93bf7b882332665b"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}