good_review_lstm_classifier.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Classifying Good Product Reviews: Deep Learning Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Organize imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "application/javascript": [
       "IPython.notebook.set_autosave_interval(5000)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Autosaving every 5 seconds\n"
     ]
    }
   ],
   "source": [
    "from gensim.models import Word2Vec\n",
    "from lib.amazon_model import *\n",
    "from lib.data_mgmt import *\n",
    "from nltk.data import find\n",
    "from sklearn.metrics import roc_curve, auc\n",
    "\n",
    "import gensim\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "import tensorflow as tf\n",
    "\n",
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "%autosave 5"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### add check in order to see if TensorFlow with GPU is used"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[name: \"/device:CPU:0\"\n",
       " device_type: \"CPU\"\n",
       " memory_limit: 268435456\n",
       " locality {\n",
       " }\n",
       " incarnation: 9855489813237813305, name: \"/device:GPU:0\"\n",
       " device_type: \"GPU\"\n",
       " memory_limit: 15770877952\n",
       " locality {\n",
       "   bus_id: 1\n",
       " }\n",
       " incarnation: 5623224763006627716\n",
       " physical_device_desc: \"device: 0, name: Quadro P5000, pci bus id: 0000:02:00.0, compute capability: 6.1\"]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tensorflow.python.client import device_lib\n",
    "device_lib.list_local_devices()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Set network parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Dataset/Featurization Config\n",
    "path = 'data/full_cols_good_50k.csv'\n",
    "label = 'good'\n",
    "max_size = 200\n",
    "max_tfidf = 12000\n",
    "n_classes = 2\n",
    "n_features = 300\n",
    "test_size = 0.25\n",
    "verbose = True\n",
    "\n",
    "# LSTM Config\n",
    "lstm_size = 50\n",
    "num_lstm_layers = 2\n",
    "lstm_dropout = 0.15\n",
    "\n",
    "# FC Config\n",
    "fc_layer_sizes = [75, 50, 30]\n",
    "fc_dropout = 0.10\n",
    "\n",
    "# Training config\n",
    "run_name = label + '_run_' + path.split('.')[0].split('_')[-1]\n",
    "learning_rate = 1e-3\n",
    "reg_weight = 1e-5\n",
    "training_iters = 15000 * 10\n",
    "batch_size = 15\n",
    "display_step = 10\n",
    "save_every = 5000\n",
    "save_fn = 'chkpts/model_' + run_name + '.ckpt'\n",
    "should_log_metrics = True"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "training_loss_iter_fn = 'log/training_loss_iter_' + run_name + '.csv'\n",
    "training_acc_iter_fn = 'log/training_acc_iter_' + run_name + '.csv'\n",
    "testing_loss_iter_fn = 'log/testing_loss_iter_' + run_name + '.csv'\n",
    "testing_acc_iter_fn = 'log/testing_acc_iter_' + run_name + '.csv'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Set up word embedding model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 8.38 s, sys: 152 ms, total: 8.53 s\n",
      "Wall time: 8.54 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "word2vec_sample = str(find('models/word2vec_sample/pruned.word2vec.txt'))\n",
    "we_model = gensim.models.KeyedVectors.load_word2vec_format(word2vec_sample, binary=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Configure dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Initializing dataset at data/full_cols_good_50k.csv.\n",
      "Splitting training and test sets.\n",
      "Building reviewer/ASIN maps.\n",
      "Fitting tf-idf featurizer.\n",
      "doc list created: 50000\n",
      "Tuning tf-idf classifier.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/u/44/prollm1/unix/anaconda3/envs/tensorflow/lib/python3.6/site-packages/sklearn/linear_model/logistic.py:1228: UserWarning: 'n_jobs' > 1 does not have any effect when 'solver' is set to 'liblinear'. Got 'n_jobs' = -1.\n",
      "  \" = {}.\".format(self.n_jobs))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Featurizing training set.\n",
      "Featurizing test set.\n",
      "Finished configuring dataset.\n",
      "CPU times: user 36min 19s, sys: 1min 37s, total: 37min 56s\n",
      "Wall time: 7min 28s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "dataset = Dataset(path, we_model, {\n",
    "    'label'      : label,\n",
    "    'max_size'   : max_size,\n",
    "    'max_tfidf'  : max_tfidf,\n",
    "    'n_features' : n_features,\n",
    "    'test_size'  : test_size,\n",
    "    'verbose'    : verbose\n",
    "})\n",
    "\n",
    "n_cols = dataset.get_n_cols() # Number of additional DNN features (including tf-idf features)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Prepare model configurations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class ModelConfig(object):\n",
    "    def __init__(self):\n",
    "        self.lstm_size = lstm_size\n",
    "        self.num_lstm_layers = num_lstm_layers\n",
    "        self.lstm_keep_prob = 1.0 - lstm_dropout\n",
    "        self.fc_keep_prob = 1.0 - fc_dropout\n",
    "        self.learning_rate = learning_rate\n",
    "        self.reg_weight = reg_weight\n",
    "        self.fc_layer_sizes = fc_layer_sizes\n",
    "        self.n_classes = n_classes\n",
    "        self.n_cols = n_cols\n",
    "        self.n_fc_layers = len(fc_layer_sizes) + 1\n",
    "\n",
    "config = ModelConfig()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Set up Tensorflow Graph"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "tf.reset_default_graph()\n",
    "\n",
    "amazon_model = AmazonModel(config)\n",
    "pred = amazon_model.build_graph()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Input placeholders\n",
    "x = tf.placeholder(\"float\", [None, max_size, n_features])\n",
    "x2 = tf.placeholder(\"float\", [None, n_cols])\n",
    "seqlen = tf.placeholder(tf.int32, [None])\n",
    "lstm_keep_prob = tf.placeholder(tf.float32)\n",
    "fc_keep_prob = tf.placeholder(tf.float32)\n",
    "\n",
    "# Output placeholder\n",
    "y = tf.placeholder(\"float\", [None, n_classes])\n",
    "\n",
    "# Checkpointing\n",
    "saver = tf.train.Saver()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/u/44/prollm1/unix/anaconda3/envs/tensorflow/lib/python3.6/site-packages/tensorflow/python/ops/gradients_impl.py:96: UserWarning: Converting sparse IndexedSlices to a dense Tensor of unknown shape. This may consume a large amount of memory.\n",
      "  \"Converting sparse IndexedSlices to a dense Tensor of unknown shape. \"\n"
     ]
    }
   ],
   "source": [
    "# Run forward pass\n",
    "logits, weights = pred(x, x2, seqlen, lstm_keep_prob, fc_keep_prob)\n",
    "\n",
    "# Define loss and optimizer\n",
    "cost = tf.add(tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=logits, labels=y)),\n",
    "              amazon_model.regularization_penalty())\n",
    "\n",
    "optimizer = tf.train.AdamOptimizer(learning_rate=config.learning_rate).minimize(cost)\n",
    "\n",
    "# Evaluate model\n",
    "correct_pred = tf.equal(tf.argmax(logits,1), tf.argmax(y,1))\n",
    "accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Initializing the variables\n",
    "init = tf.global_variables_initializer()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "if should_log_metrics:\n",
    "    train_loss_log = open(training_loss_iter_fn, 'a')\n",
    "    test_loss_log = open(testing_loss_iter_fn, 'a')\n",
    "    train_acc_log = open(training_acc_iter_fn, 'a')\n",
    "    test_acc_log = open(testing_acc_iter_fn, 'a')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 150, Minibatch Loss= 0.695298, Training Accuracy= 0.53333\n",
      "Iter 300, Minibatch Loss= 0.641469, Training Accuracy= 0.73333\n",
      "Iter 450, Minibatch Loss= 0.699551, Training Accuracy= 0.33333\n",
      "Iter 600, Minibatch Loss= 0.639029, Training Accuracy= 0.60000\n",
      "Iter 750, Minibatch Loss= 0.681501, Training Accuracy= 0.53333\n",
      "Validation Accuracy: 0.58792\n",
      "Validation Loss: 0.6795162\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e8713be80>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 900, Minibatch Loss= 0.673586, Training Accuracy= 0.60000\n",
      "Iter 1050, Minibatch Loss= 0.625853, Training Accuracy= 0.80000\n",
      "Iter 1200, Minibatch Loss= 0.690564, Training Accuracy= 0.53333\n",
      "Iter 1350, Minibatch Loss= 0.677733, Training Accuracy= 0.60000\n",
      "Iter 1500, Minibatch Loss= 0.687299, Training Accuracy= 0.40000\n",
      "Validation Accuracy: 0.6304\n",
      "Validation Loss: 0.67217153\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e84a9e7f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 1650, Minibatch Loss= 0.656828, Training Accuracy= 0.60000\n",
      "Iter 1800, Minibatch Loss= 0.555429, Training Accuracy= 0.80000\n",
      "Iter 1950, Minibatch Loss= 0.564421, Training Accuracy= 0.66667\n",
      "Iter 2100, Minibatch Loss= 0.673784, Training Accuracy= 0.53333\n",
      "Iter 2250, Minibatch Loss= 0.593735, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.62384\n",
      "Validation Loss: 0.65544045\n",
      "Iter 2400, Minibatch Loss= 0.641976, Training Accuracy= 0.60000\n",
      "Iter 2550, Minibatch Loss= 0.698457, Training Accuracy= 0.53333\n",
      "Iter 2700, Minibatch Loss= 0.561642, Training Accuracy= 0.60000\n",
      "Iter 2850, Minibatch Loss= 0.687600, Training Accuracy= 0.60000\n",
      "Iter 3000, Minibatch Loss= 0.563226, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.63184\n",
      "Validation Loss: 0.64095694\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e79703198>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 3150, Minibatch Loss= 0.539343, Training Accuracy= 0.60000\n",
      "Iter 3300, Minibatch Loss= 0.551157, Training Accuracy= 0.73333\n",
      "Iter 3450, Minibatch Loss= 0.681271, Training Accuracy= 0.60000\n",
      "Iter 3600, Minibatch Loss= 0.671206, Training Accuracy= 0.53333\n",
      "Iter 3750, Minibatch Loss= 0.599751, Training Accuracy= 0.60000\n",
      "Validation Accuracy: 0.64536\n",
      "Validation Loss: 0.6375112\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e6399c5c0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 3900, Minibatch Loss= 0.549958, Training Accuracy= 0.66667\n",
      "Iter 4050, Minibatch Loss= 0.658181, Training Accuracy= 0.53333\n",
      "Iter 4200, Minibatch Loss= 0.671059, Training Accuracy= 0.60000\n",
      "Iter 4350, Minibatch Loss= 0.760393, Training Accuracy= 0.46667\n",
      "Iter 4500, Minibatch Loss= 0.638184, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.65168\n",
      "Validation Loss: 0.6374605\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYoAAAEWCAYAAAB42tAoAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvNQv5yAAAIABJREFUeJzs3Xd4FNX6wPHvm55AKAnIRZCO9CYRQZAqHRFEfxSFK6KICKgogoqiCBZEEKSriF6ucm0ISgcRLLQgTYqAgBBEOkmAJCTZ8/tjQjZAygayO0n2/TxPHubMnpl9MyT75sw5c44YY1BKKaUy4mN3AEoppXI3TRRKKaUypYlCKaVUpjRRKKWUypQmCqWUUpnSRKGUUipTmiiUUkplShOFyldE5JCIxInIeRH5R0TmiEjBq+rcKSI/iEisiESLyHciUv2qOoVE5D0ROZxyrv0p5WIZvK+IyBAR+V1ELohIlIh8KSK13Pn9KuUJmihUfnSPMaYgUBeoB7xw+QURaQQsBxYANwPlgW3ALyJSIaVOALAKqAG0AwoBdwKngQYZvOck4ClgCBAG3Ap8C3TMbvAi4pfdY5RyJ9Ens1V+IiKHgEeNMStTyuOAGsaYjinln4AdxpiBVx23BDhpjOkjIo8CY4GKxpjzLrxnZWAP0MgYszGDOj8Cc40xH6aUH06Js0lK2QCDgKcBP2AZcN4Y81yacywA1hhjJojIzcD7QFPgPDDRGDPZhUukVLZpi0LlWyJSGmgP7E8ph2C1DL5Mp/oXQOuU7buBpa4kiRStgKiMkkQ2dAHuAKoDnwHdRUQARKQo0AaYJyI+wHdYLaFSKe//tIi0vcH3VypdmihUfvStiMQCR4ATwKiU/WFYP/PH0jnmGHC5/yE8gzoZyW79jLxpjDljjIkDfgIMcFfKa/cD64wxfwO3A8WNMaONMZeMMQeAD4AeORCDUtfQRKHyoy7GmFCgOVAVZwI4CziAkukcUxI4lbJ9OoM6Gclu/YwcubxhrHvC84CeKbt6Af9N2S4L3Cwi5y5/AS8CJXIgBqWuoYlC5VvGmDXAHGB8SvkCsA54IJ3q/4fVgQ2wEmgrIgVcfKtVQGkRicikzgUgJE35X+mFfFX5c+B+ESmLdUvq65T9R4CDxpgiab5CjTEdXIxXqWzRRKHyu/eA1iJSN6U8Avh3ylDWUBEpKiJjgEbAayl1/oP1Yfy1iFQVER8RCReRF0Xkmg9jY8w+YBrwuYg0F5EAEQkSkR4iMiKl2lbgPhEJEZFKQL+sAjfGbAFOAh8Cy4wx51Je2gjEiMhwEQkWEV8RqSkit1/PBVIqK5ooVL5mjDkJfAq8nFL+GWgL3IfVr/AX1hDaJikf+BhjErA6tPcAK4AYrA/nYsCGDN5qCDAFmAqcA/4EumJ1OgNMBC4Bx4FPcN5GysrnKbF8luZ7SgbuwRr+exDrltmHQGEXz6lUtujwWKWUUpnSFoVSSqlMaaJQSimVKU0USimlMqWJQimlVKby3ORjxYoVM+XKlbM7DKWUylM2b958yhhT/HqOzXOJoly5ckRGRtodhlJK5Ski8tf1Hqu3npRSSmVKE4VSSqlMaaJQSimVKU0USimlMqWJQimlVKY0USillMqU2xKFiMwWkRMi8nsGr4uITBaR/SKyXURuc1csSimlrp87n6OYgzXt8qcZvN4eqJzydQcwPeVfpZRSlyVehEsxYBxpvpKvKqfZH38WxBdMEjiSwSRzKSHxhkJwW6IwxqwVkXKZVLkX+DRlycf1IlJEREoaY3Ji7WGllLJf8iWIOQyJFyA5AS4cA8TaPn8UkuLh5DYIKgrH1kNwMYhaA4XKwdm9ORLCsO9as+XvG1up184ns0uRZo1gICpl3zWJQkT6A/0BypQp45HglFIqS3Fn4Ng6OPwDnNltJYSkOLhwHGIPX/9500sSBUqC+KT58r2qnPIVd9r6t0hFEF9q1ijO5F/KXX8s2JsoJJ196a6iZIyZBcwCiIiI0JWWlFKelXwJ9n0Dp36HHR9AyE0QGwUJ57I+FsAvGIrVBJ8AiDkEJRuCXxDEnYLCFcC/AITXAMclqzURXByCw8A/1GptSHofl+nbteskv/12jIe61wagzwOGZsOiKV9+dPa/78vhX/eRNy4KuCVNuTTwt02xKKWUdSso5jCc3gXHI63y5nevrXfxhPWvjx84kqB8eyhU3koG4dUhpAQUKAH+BcHX3yOhX7yYyJgxa3nnnV/x9RUaNixNpUphiAjlyhW5oXPbmSgWAoNEZB5WJ3a09k8opdzOkQQnt1t9AxdPwO+zs9cfcNtTULwOlG0DBW/O1l/77rJkyT6efHIxBw9aLZx+/eoTHh6cY+d3W6IQkc+B5kAxEYkCRgH+AMaYGcBioAOwH7gI9HVXLEopL2MMHN9s3SqKPWJ1IiddhEPLICHa6kfIjPhA2dYQWBRubmTdDqrYydqfixw9GsPTTy/jq692AVC7dglmzOhIo0a3ZHFk9rhz1FPPLF43wJPuen+llBdJToTTO63Wwe7/QvyZzOsHFYXwmtZtofIdISAUyt4NoWXAx9czMeeAJ59czIIFfxAS4s/o0c156qmG+PnlfDLLc+tRKKW8nCMJdnwIf620RhrFn4EL/6RfN6AQ1HoMQktBQGGrUzm4GJRtletaB65KSnKkJoO3374bf39f3n23DWXKFHbbe2qiUErlbsZYzxwcWGQ9a7BzTsZ1y7WDcm2g2oPWyKFc0H+QU6Kj4xk58gf27j3D0qUPIiJUqVKML798wO3vrYlCKZW7JF6AQ8th/7fwxzxraGp6CpWDGv+2+hBKNoRA9/1FbSdjDF9+uYunn17KsWPn8fUVtm79h3r1buwhuuzQRKGUsldSPPy1AjaMhWMbMq7nFwLFa0HjsXBL8zzVl3C9/vzzDIMGLWHp0v0ANGpUmhkzOlG7dgmPxqGJQinlWY4kOLIGNk+Aw6us6SzSU/BmqNgZIp6zHkrLR7eRXDF+/K+8/PJq4uOTKFIkiLffvptHH70NHx/PXwdNFEop90pOtKa5OLENtk23OqDTU6gs1H8Gqv8bgm7sAbH84OLFROLjk+jduzbjx7fhppsK2BaLJgqlVM478wesfR4unrSSRHpqPQqlm1lfhXJ23H9edPLkBf744zRNmljz2Q0f3pjmzcvRtGlZmyPTRKGUygnGAVE/wfrXrdtJ6SlUFm59ACp0tPoYFAAOh2H27C08//wK/Px82LNnEGFhwQQG+uWKJAGaKJRS1yMxznrA7cRW2PUJHP352johJaBmX+tWUnhVz8eYB/z++wkGDPieX36xJtJu3boCFy8mEhaWc9Nv5ARNFEqprF1+luHvdbB9FhxemX692v2hZCOo1EX7GTJx4cIlRo9ew4QJ60lKclCiRAHee68d3bvXQHJhp70mCqVUxs7shcUPWjOppqd8e+sZhrCqUL4DBBT0bHx51P33f8nSpfsRgYEDIxg7thVFigTZHVaGNFEopZyMgRNb4JeRVod09IFr65RrB7cPgzItPR9fPjF8eGOOHz/P9OkdueOO0naHkyVNFEp5M0ey1fl8aLm1IM+lmGvrFKtljVCq9Rj4565753lBUpKD99/fwKFD55g0qT0AzZuXIzKyvy3PRFwPTRRKeaMT22DTONjzWcZ1mk+Aqr2sBXjUddm48SiPP/49W7dakxb271+fGjVuAsgzSQI0USjlHZLirem393wOZ/ddu55zsVrQcCSEVYOwKuAbYE+c+cS5c/G8+OIqZsyIxBgoW7YwU6Z0SE0SeY0mCqXyq79WwW8T4eBSMMnXvl7tQag3GEre4fnY8rF5837n6aeXcvz4Bfz8fHj22Ua8/HJTChTIu8lXE4VS+UnMYauvYf2Ya18LLGxNvd3gBaj+kLYa3GT58j85fvwCjRvfwvTpHalVK+/futNEoVReF3cGtk2DXf+5du3nIpWgyVi4+U4Izf2ja/KihIQkjh6NpUKFogCMG9eau+4qw7//XTdP9UNkRhOFUnmRMdYQ1g1vpP96mVbQfKI1Lbdymx9+OMgTTyzCx0fYtm0AAQG+FCsWQt++9ewOLUdpolAqL7l4Cn59xZqFNa2walCshnVbqcRt9sTmRY4fP89zz61g7tztAFStWoyoqJjUVkV+o4lCqbwg6mdrNtarZ2K9fTg0egX8Q+yJy8s4HIYPPtjMiBGrOHcunqAgP0aOvIthwxoTEJB/F1LSRKFUbnZ6F3zREi4ev3J/k7FQdxAEFrInLi/Vtev/WLjwDwDatq3I1KkdqFgxzOao3E8ThVK50bk/4cu7IeaQc1/xunDXm1Curdet9pZb3HdfVTZuPMqkSe144IHquXICP3fQRKFUbnL+GKx4HA5859xXsDS0fB8qd7EvLi+1cOEfREXFMHDg7QD06VOH++6rRmhooM2ReZYmCqVyg6O/wrzG1+5/7JC14I/yqMOHoxkyZAkLFvxBYKAv7dpVokKFooiI1yUJ0EShlL2SE+HrNnDkxyv31x0ELSfrLSYPS0xMZvLkDYwa9SMXLiQSGhrAmDEtKVu2sN2h2UoThVJ22f2ZtdZDWl0WQsV77InHy61fH8Xjj3/P9u3WwIEHHqjOxIltKVVKBwxoolDK0878AT+/BPu+du4rUR96rgNff/vi8nIvv7ya7duPU758EaZM6UCHDpXtDinX0EShlKdcOg/ftL9yfWm/EBh0VuddsoExhtjYSxQqZPU5TJnSnk8/3cZLLzUlJEQTdlqaKJTyhH3fwsKuznJ4DWgxCcq2si8mL/bHH6cYOHAxIrBiRW9EhCpVijF2rP5/pEcThVLuZBzWA3NRa5z7unwHFTvZF5MXi49P4s03f+Ktt37h0qVkwsODOXToHOXL58+pN3KKJgql3CH5EkS+C5snQNwp5/5B0fo0tU1WrPiTgQMXs3//GQAeeaQu48a1Jjxcpz/JilsThYi0AyYBvsCHxpi3rnq9DPAJUCSlzghjzGJ3xqSUWzmSYetUWP2Uc59/QajWC+6eocNdbWCMoV+/hXz88VYAqlcvzowZHbnrLn0+xVVuSxQi4gtMBVoDUcAmEVlojNmVptpI4AtjzHQRqQ4sBsq5Kyal3GrDG9ZoprQajLDmZRIfe2JSiAjlyhUhONiPV15pxtChjfL1BH7u4M4WRQNgvzHmAICIzAPuBdImCgNcbocXBv52YzxKucfBpVZHdVK8c1/dJ6HJG3qbySZbt/7DsWOxtG9vDXEdPrwxvXvX1r6I6+TORFEKOJKmHAVcvTjvq8ByERkMFADuTu9EItIf6A9QpkyZHA9UqesSfQg+qmh1WF9Wvj3c+60Od7VJbGwCo0b9yKRJGwgPD2bPnkGEhQUTGOinSeIGuDNRpHcz1lxV7gnMMca8KyKNgP+ISE1j0v7mgTFmFjALICIi4upzKOVZJ3fAj0Ph8ErnvoBC0HsLFKlgX1xezBjDt9/uYciQpURFxeDjI/TqVQt/f73llxPcmSiigFvSlEtz7a2lfkA7AGPMOhEJAooBJ9wYl1LXJ/kSrB8L60c795VqAs0nwL9uty8uL/fXX+cYNGgJ339vrRceEXEzM2d24rbbStocWf7hzkSxCagsIuWBo0APoNdVdQ4DrYA5IlINCAJOujEmpbIv8QJ80/HKZyH8guGeL6FCR/viUhhj6NbtCzZvPkahQoG88UZLBgyIwNdXWxI5yW2JwhiTJCKDgGVYQ19nG2N2ishoINIYsxB4FvhARJ7Bui31sDFGby2p3GPvV/DdA1fua/MR1OyrQ11t5HAYfHwEEWH8+DbMmBHJxIltKVky1O7Q8iXJa5/LERERJjIy0u4wVH53YDH8MQ92/ce57+Y74YFV4BdkX1xe7vTpi4wYYfUNffBBZ5ujyVtEZLMxJuJ6jtUns5VKK/6c1YJI21EdWgYeWAlFdTZRuxhj+PTTbTz33ApOnbpIQIAvo0Y1p3RpHX7sCZoolLrs949h2SNX7uv5K9zcyJ54FAC7d5/kiScWsWbNXwA0b16O6dM7apLwIE0USm3/EDa+CdEHrHJYVajaExq9Ym9cXs4YwyuvrObtt38hMdFBsWIhvPtuG3r3ro1o/5BHaaJQ3ssYWNIHds917ms0ykoQOuWG7USEo0djSUx08Nhjt/HWW3cTFhZsd1heSROF8k6nd8Hc+s5pN4KLQb/9EOjdayPb7e+/Yzl16iK1a5cAYNy41vTrV4/GjXVGBjvpn03K+2wcB3NqOJNEmVbwxAlNEjZKTnYwZcpGqlWbSo8eX3HpUjIAxYqFaJLIBbRFobzLxnHw03BrO7Aw9NoAYVXsjcnL/fbbMR5//HsiI62JG5o2LUtMTALFiuk6EbmFS4lCRAKAMsaY/W6ORyn3cCTB543hn41WOaQEPPYX+AXaG5cXi4lJ4OWXf2DKlE04HIbSpQsxeXI7unSpqp3VuUyWiUJEOgITgACgvIjUBUYZY7pmfqRSucS5P+GjSs5yxc7QZYF98SiMMTRt+jHbth3H11cYOrQhr77anNBQTdy5kSt9FKOxpgc/B2CM2QpUyvQIpXKLlQOvTBK3PmBNA65sJSI880xDGjQoRWRkf959t60miVzMlVtPicaYc1c1BfPWvB/KO0X9BNumO8vd10DppvbF48UuXUpmwoR1+PoKw4Y1BqBPnzo89FBtncAvD3AlUewWkf8DfFJmgn0KWO/esJS6QYdWwNdtnOWhyfpshE1++ukvBgxYxK5dJwkM9KVPnzqUKFEQEcHXV/si8gJXfnMGAfUBB/ANEI+VLJTKfWKjYFrxK5PEwzs1Sdjg1KmLPPLIApo2ncOuXSepXDmM77/vRYkSBe0OTWWTKy2KtsaY4cDwyztE5D6spKFU7rFqEGyd6ixX7AxtPoSQ4vbF5IWMMcyZs5Vhw1Zw+nQcAQG+vPBCE0aMaEJQkI7Iz4tc+V8bybVJ4aV09illD2NgalFIiLbKAYWsDusyLeyNy4vNnbuD06fjaNmyPNOmdaBKlWJ2h6RuQIaJQkTaYi1TWkpEJqR5qRDWbSil7GcMzLvLmSQCi8DAk+Cjf7l60sWLiURHx1OyZCgiwrRpHdi06W8efLCWPhORD2T223QC+B2rT2Jnmv2xwAh3BqVUlowDDq+Gr+527itUFh47ZFtI3mrJkn08+eRiKlQoyooVvRERqlQppq2IfCTDRGGM2QJsEZH/GmPiPRiTUpnbv8BaXMiR6NxX/1lo9o59MXmho0djePrpZXz11S4AQkMDOX06TqfeyIdcaZ+XEpGxQHUgdQ1IY8ytbotKqfQYAz+/CBvfcu6r8wTcORpC9K9XT0lOdjB16iZGjvyB2NhLFCjgz+jRLRgy5A78/HR0WX7kSqKYA4wBxgPtgb5oH4XytFM74ZOaznLBUtaEfqGl7IvJCzkchmbN5vDLL0cA6NKlKpMmtaNMGZ15Nz9zJVGEGGOWich4Y8yfwEgR+cndgSkFwOndsKwvHNvg3FeiPvRarx3WNvDxEdq0qcjhw9FMmdKBzp115l1v4MpvWoJYwxb+FJEBwFHgJveGpRQQtRb+18xZrnAP3PEi3NzQvpi8jDGGL77YiZ+fD926VQdg+PDGDB3aiIIFA2yOTnmKK4niGaAgMAQYCxQGHsn0CKVu1Jap8MMgZ/nRg1C4nG3heKM//zzDwIGLWb78T4oXD6Fly/IULRpMYKAfgTp/n1fJMlEYYy63+WOB3gAiUtqdQSkvFn8WPqkN56Oc+3r+qknCgxISknjnnV8ZO/Yn4uOTKFo0iLFjW1K4cFDWB6t8KdNEISK3A6WAn40xp0SkBtZUHi0BTRYq5ziSrRFNm8Y599XsB3dPB19/++LyMj/+eIgnnljEnj2nAOjduzbjx7fhppsK2ByZslNmT2a/CXQDtmF1YM/HmgzwbWCAZ8JTXuHYBvi2M1w84dzX/lOo3tu+mLxQcrKDgQOtJFGlSjjTp3ekRYvydoelcoHMWhT3AnWMMXEiEgb8nVL+wzOhqXwv+qDVWR1rDbXEvyBU6gJtZ2srwkMcDkN8fBIhIf74+vowfXpH1q79i+efb0xgoI4qU5bMfhLijTFxAMaYMyKyR5OEyjE7PoIVj4NJtsohN0G//RAQam9cXmTHjuMMGLCIqlXD+eijewFo1qwczZqVszcwletkligqiMjlGWIFKJemjDHmPrdGpvKnxDjYPgN+HGqVC5aCdnOg7N2ZHqZyzoULlxg9eg0TJqwnKcnBwYNnOXs2jqJFg+0OTeVSmSWKbleVp7gzEOUFTu+CT+uAI8kq+/hB3z0QoAvZeMp33/3BoEFLOHw4GhEYODCCsWNbUaSIjmhSGctsUsBVngxE5XPRB2FODWe5eF245wtNEh6SlOSge/ev+Oab3QDUrfsvZs7sRIMGOgWKypr2Vin3Mg5YORC2z3Tue2QvFK1sX0xeyM/Ph8KFAylYMIDXX2/BoEENdAI/5TIxxrjv5CLtgEmAL/ChMeatdOr8H/AqYIBtxphemZ0zIiLCREZGuiFaleOSE+G9QKz/2hQ9foZSjW0LyZts2GA9tHjHHdYjT6dPXyQuLonSpQvZGZayiYhsNsZEXM+xLrcoRCTQGJOQjfq+wFSgNRAFbBKRhcaYXWnqVAZeABobY86KiM4hlV8YAwu7kpokitWCrt9Ziwsptzp3Lp4XXljJzJmbqVq1GFu3DiAgwJfwcF0nQl2fLNueItJARHYA+1LKdUTkfRfO3QDYb4w5YIy5BMzDejYjrceAqcaYswDGmBOovC/+LEwrBgcWWeUafeHf2zVJuJkxhs8+20HVqlOYMWMzvr4+dO5cheRkXRVA3RhXWhSTgU7AtwDGmG0i4sqq9aWAI2nKUcAdV9W5FUBEfsG6PfWqMWapC+dWudWJrfD5nZAUZ5XveBGajLU3Ji+wb99pBg5czMqVBwBo3PgWZszoRM2a2khXN86VROFjjPnrqgXSk104Lr0V1a/uEPEDKgPNseaO+klEahpjzl1xIpH+QH+AMmXKuPDWyhYb34af0iyn3m0ZlGtjXzxeIjExmZYtPyUqKoawsGDGjbubvn3r4eOT3q+gUtnnSqI4IiINAJPS7zAY2OvCcVHALWnKpbGmAbm6znpjTCJwUET+wEocm9JWMsbMAmaB1ZntwnsrT7p69TmAx/6CQprU3ckYg4jg7+/L2LEtWb36EOPG3U3x4jqBn8pZroyPewIYCpQBjgMNU/ZlZRNQWUTKi0gA0ANYeFWdb4EWACJSDOtW1AHXQle5QmzUlUmiwj0w1KFJwo2OHz9P797zGTNmbeq+Pn3q8PHH92qSUG7hSosiyRjTI7snNsYkicggYBlW/8NsY8xOERkNRBpjFqa81kZEdmHdzhpmjDmd3fdSNok7A7PSNBq7r4XSd9kXTz7ncBg++GAzI0as4ty5eIoUCeLppxsSGqqrCCn3ciVRbEq5JfQ/4BtjTKyrJzfGLAYWX7XvlTTbBqu1MtTVc6pcwpEM/2vqLHeYq0nCjbZt+4cBAxaxfr31bES7dpWYOrWDJgnlEa6scFdRRO7EunX0mohsBeYZY+a5PTqVO8Wfhc8awdmUyYS7LYdyre2NKZ9KTEzmhRdW8d5760lONpQsWZBJk9px//3VuWqAiVJu49Iz/MaYX40xQ4DbgBjgv26NSuVuC7o6k0TnrzVJuJGfnw9btvyDw2EYPLgBu3c/yQMP1NAkoTwqyxaFiBTEelCuB1ANWADc6ea4VG4Ufxb+Uw9i/rLKXRdBhQ72xpQPHT4cTXKyg/LliyIizJjRkejoBCIibrY7NOWlXOmj+B34DhhnjPnJzfGo3Mg4YO0IiHzHua/245okclhiYjKTJm1g1KgfadSoNCtW9EZEqFw53O7QlJdzJVFUMMboHADeKuYwfNMBTu907msxGW4bbF9M+dC6dUcYMGAR27cfByAsLJiLFxMpUCDA5siUyiRRiMi7xphnga9F5JqH3HSFOy/w62uwfrTVogCo2Q9azwQfX3vjykfOno1jxIiVzJr1GwDlyxdh6tQOtG+v07Cr3COzFsX/Uv7Vle280e8fw7pXnWVdQyLHJSQkUbfuTA4fjsbf34dhw+7kpZeaEhLib3doSl0hsxXuNqZsVjPGXJEsUh6k0xXw8qvId2HNc87yU/Hgp+P1c1pgoB/9+tVj1aqDTJ/ekerVi9sdklLpynLhIhH5zRhz21X7thhj6rk1sgzowkVutmM2LO/nLD99CXz1L9ycEB+fxJtv/kSVKsXo1asWYC1R6usrOtxVuZ1bFi4Ske5YQ2LLi8g3aV4KBc6lf5TK0/bNvzJJDE0G0eUyc8KKFX8ycOBi9u8/w003FaBr16oEB/vrcqQqT8isj2IjcBpr1tepafbHAlvcGZSywaHlsDDN+IQnz2qSyAH//HOeoUOX8fnnvwNQo0ZxZszoRHCwttJU3pFZH8VB4CCw0nPhKFskRMPXbVMKAoOjISDU1pDyuuRkBzNnbubFF1cRHZ1AcLAfo0Y145lnGhEQoKPGVN6S2a2nNcaYZiJylisXHBKs+fzC3B6dcr/kSzCliLP86J+aJHJAcrLh/fc3Eh2dQIcOlZkypT3lyxe1Oyylrktmt54uL3dazBOBKJvMqe7c7rYcCpe3L5Y8LjY2geRkQ5EiQQQE+PLBB/dw/Ph57ruvmnZWqzwtw5vQaZ7GvgXwNcYkA42AxwFdHSU/WDkQzv1pbXecp5P7XSdjDN98s5tq1aby7LPLUvc3aVKGbt10lleV97nSW/kt1jKoFYFPsSYG/MytUSn3W/YobJtubYdVhard7Y0njzp06BydO8+jW7cvOHo0lt9/P0l8fJLdYSmVo1yZ68lhjEkUkfuA94wxk0VERz3lVRf+sdaSiDlklYvXhp7rbQ0pL0pMTGbChHW89toa4uKSKFQokDfeaMmAARH4+upoMZW/uLQUqog8APQGuqTs07F9eVH8OZhR0lmu3d+au0lly8WLiTRs+CE7dpwAoEePmkyY0IaSJXUQgMqfXEkUjwADsaYZPyAi5YHP3RuWcosvmju3O34OVbO9FLoCQkL8iYi4mYsXE5k2rSNt2lS0OySl3CrLKTwARMQPqJRS3G+Mse0mrE7hcZ2Tt81WAAAgAElEQVS+7wl/pKxe2+I9uO0pe+PJQ4wxfPrpNipWDKNJkzIAREfHExDgqw/OqTzDLVN4pDn5XcB/gKNYz1D8S0R6G2N+uZ43VDZY+rAzSdzxkiaJbNi9+yRPPLGINWv+olq1YmzdOoCAAF8KFw6yOzSlPMaVW08TgQ7GmF0AIlINK3FcV2ZSHrZlCuz8xNq+40VoMsbeePKIuLhExo79iXHjfiEx0UHx4iG88EIT/P21o1p5H1cSRcDlJAFgjNktIrrsVl6wbQb8kLISXWgZaKxJwhVLl+7nyScXc+DAWQAee+w23nrrbsLCgm2OTCl7uJIofhORmVitCIAH0UkBc7fkRFj/uvUF4BcMjx0CffArS+fPX6J37/mcOnWRmjVvYsaMjjRuXMbusJSylSuJYgAwBHgeq49iLfC+O4NSN+CfSPjv7c5y0Srw0CZNEplITnbgcBj8/X0pWDCASZPaERUVwzPPNMTfXyfwUyrTRCEitYCKwHxjzDjPhKSuW/KlK5OEjm7K0ubNf/P4499z771VePnlZgCpiwoppSwZ9syJyItY03c8CKwQkUc8FpXKvjN74YOyzvIjezVJZCImJoGnnlpCgwYfsnnzMf7zn+0kJibbHZZSuVJmLYoHgdrGmAsiUhxYDMz2TFgqWw4uhW/aW9viC+0/haKV7Y0plzLG8NVXu3jqqaUcO3YeX19h6NCGvPZaC73NpFQGMksUCcaYCwDGmJMiutxZrhN9CJb0gaM/Off126dThWcgNjaB7t2/YsmS/QDccUcpZszoRN26/7I5MqVyt8wSRYU0a2ULUDHt2tnGmPvSP0x5RNwZ+DBNQqj/LDQYASG6fEhGChYMICEhmcKFA3nrrbvp378+Pj7aya9UVjJLFN2uKk9xZyAqG84dgI/SzC/U+Ruo3NW+eHKxtWv/omTJglSuHI6IMHt2Z4KC/ChRoqDdoSmVZ2S2ZvYqTwaiXHRqJ3xS01nu9D9NEuk4deoizz+/go8/3kqrVuVZsaI3IkLZskWyPlgpdQVXnqNQuUVi3JVJos92KK5DOdNyOAxz5mxl2LAVnDkTR0CAL3fdVYbkZIOfn95mUup6uLWDWkTaicgfIrJfREZkUu9+ETEiovNHZST5EkxOswJt3z80SVxl584TNG8+h379FnLmTBytWpVnx44nGDWqOX5+OhZDqevlcotCRAKNMQnZqO8LTAVaA1HAJhFZmHbeqJR6oVhPfm9w9dxeJzkRPmsIpEwJX38ohN1qa0i5TXR0PA0bfsT585e46aYCTJjQhl69aul61UrlgCz/zBKRBiKyA9iXUq4jIq5M4dEAa+2KA8aYS8A84N506r0OjAPiXQ/by8ytDydSpte6/Xlo/q698eQil9dTKVw4iOHDGzNgQH327HmSBx+srUlCqRziSnt8MtAJOA1gjNkGtHDhuFLAkTTlqJR9qUSkHnCLMeb7zE4kIv1FJFJEIk+ePOnCW+cjsUfh1A5r+44Xoenb9saTSxw9GsP993/B3LnbU/e99NJdTJ/eiaJFdZZXpXKSK4nCxxjz11X7XJnrIL0/51KX00t5gG8i8GxWJzLGzDLGRBhjIooXL+7CW+cTxzfDrNLOcpOx9sWSSyQlOZg0aT1Vq07l6693M2rUjyQnOwC0BaGUm7jSR3FERBoAJqXfYTCw14XjooBb0pRLA3+nKYcCNYEfU37B/wUsFJHOxhhd6zQpHuam6dt/xJVLnr9t2nSUAQMW8dtvxwDo0qUqkye3w9dXO6qVcidXEsUTWLefygDHgZUp+7KyCagsIuWxllHtAfS6/KIxJhpIfYxYRH4EntMkAez+Lyx+yFnuusir5266cOESw4evZNq0TRgDZcoU5v3329O5cxW7Q1PKK2SZKIwxJ7A+5LPFGJMkIoOAZYAvMNsYs1NERgORxpiF2Y7WG6wYANtnOsvdlkK5tvbFkwv4+fmwcuUBfHyEoUMbMWpUMwoU0EUWlfIUuTxqJMMKIh+Qpm/hMmNMf3cFlZmIiAgTGZlPGx07ZsPyftZ2gX/Bg5EQWirzY/KpP/88Q5EiQYSHhwDWbaegID9q1Sphc2RK5U0istkYc13Pqrlyc3clsCrl6xfgJsDl5ymUi3560ZkkitWC/ke8MkkkJCQxZsxaataczvDhK1P33357KU0SStnElVtP/0tbFpH/ACvcFpE3OrENNr5pbYeUgD7bvHLp0h9/PMQTTyxiz55TgDXCKTnZoZ3VStnseuZ6Kg+UzbKWcs3Vy5c+etDrksSJExcYNmwFn366DYAqVcKZPr0jLVrouhpK5QZZJgoROYuzj8IHOANkOG+TyoakBJhWDByJVvnhneDvXQ+LnTp1kWrVpnLmTByBgb689NJdPP98YwIDdb5KpXKLTH8bxXrAoQ7W8FYAh8mq91u57pv2kHje2r7nSwivbm88NihWLIR7761CVFQM06Z1pFKlMLtDUkpdJdNEYYwxIjLfGFPfUwF5jbUj4Mhqa7vZeLj1fnvj8ZALFy4xevQaOna8laZNrTuY06Z1JDDQV5+sViqXcqWXcKOI3Ob2SLzJr6/CppQ5m8KqQkSWs5jkC9999wfVq09j3LhfGThwEQ6H1TgNCvLTJKFULpZhi0JE/IwxSUAT4DER+RO4gDWHkzHGaPK4Hmf3wbrXnOXeW+yLxUOOHInmqaeWMn/+HgDq1fsXM2d20vWqlcojMrv1tBG4DejioVjyP0cyzE6zjsSTZ8AvyL543CwpycHkyRt45ZXVXLiQSMGCAYwZ04Inn2ygCwkplYdkligEwBjzp4diyd+MAyamudx9/4CgovbF4wExMQm8+ebPXLiQSLdu1XjvvXaULl3I7rCUUtmUWaIoLiJDM3rRGDPBDfHkX1+2cm7XG5xvV6g7dy6e4GA/AgP9CAsLZubMTgQG+tKxY/78fpXyBpm1/32BgljTgaf3pVxhDHzeGI78aJVvaQEtJ9sakjsYY/jssx1UqTKFceN+Sd1/333VNEkolcdl1qI4ZowZ7bFI8qtt0+HvX63tGn2h3Wx743GDvXtPM3DgIlatOgjA2rWHMcboSCal8oks+yjUDYiNglVPWts1Hs53SSI+Pom33/6ZN974mUuXkgkLC+add1rz8MN1NUkolY9klihaZfKayoojGWalWeCvxST7YnGDf/45T9OmH7Nv3xkAHn64Lu+805pixUJsjkwpldMyTBTGmDOeDCTf+byRc/uBHyAwf432KVGiALfcUhg/Px+mT+9Is2bl7A5JKeUmOvOaO3zbBf7ZZG2XagJlWtgbTw5wOAwffLCZFi3Kc+ut4YgIn312H0WLBhMQ4Gt3eEopN9KnnnLa8c3w5wJr+5YW0OMne+PJAdu2/UPjxrMZMGARAwcu4vK8kCVKFNQkoZQX0BZFTjq0HL5OWd86KAweWGVvPDfo/PlLvPrqj7z33nqSkw033xzKgAHXtZKiUioP00SRky4nCYD/W52nFyD69ts9DB68hKioGHx8hMGDGzBmTEsKFQq0OzSllIdposgpG8c5t584DiE32RfLDTp6NIYePb4iISGZ+vVLMmNGJyIibrY7LKWUTTRR5IR98+Gn4dZ2pa55MkkkJibj5+eDiFCqVCHGjm1JQIAvAwfermtWK+Xl9BPgRiXEwML7rG0fP2ulujzm11+PUL/+LObO3Z6679ln72Tw4Ds0SSilNFHcsCmFnduP/w0+eWcU0JkzcTz++Hc0bjybHTtOMG1aJLrSrVLqanrr6Ubs/MS5fe8CCCluXyzZYIxh7tztPPvsck6evIi/vw/PP9+Yl166S6feUEpdQxPF9Tq7H5Y+7CxX6mxbKNlx/Ph5evb8mtWrDwHQrFlZpk/vSLVqeSPJKaU8TxPF9Yg/B7MrW9viC3132xtPNhQpEsSxY+cpViyE8eNb06dPHW1FKKUypYnieqRtSTx6AAqVsS0UV6xY8Se33VaS8PAQAgP9+PLLByhZsiDh4TqBn1Iqa9qZnV0xR5xTdDR8JVcniWPHYunZ82vatJnL8OErU/fXrHmTJgmllMu0RZEdxsCH5Z3lhi/ZF0smkpMdzJy5mRdeWEVMTALBwX5UqRKuiwkppa6LJors+KYDmGRru8928A2wN550/PbbMQYM+J5Nm/4GoGPHykyZ0oFy5YrYHJlSKq/SROGqNcPg0FJru84TULyWvfGk49ChczRo8AHJyYZSpUKZPLk9XbtW1VaEUuqGuDVRiEg7YBLgC3xojHnrqteHAo8CScBJ4BFjzF/ujOm6bJ0GkeOt7aAwaDXV3ngyUK5cEfr2rUtoaCCvvdac0FCdwE8pdePc1pktIr7AVKA9UB3oKSLVr6q2BYgwxtQGvgLGkdv8MMS57nWRSvDk6VwzK+yhQ+e4557PWbPmUOq+WbPuYcKEtpoklFI5xp0tigbAfmPMAQARmQfcC+y6XMEYszpN/fXAQ26MJ/u2TIEt7zvLD+/KuK4HJSYmM2HCOl57bQ1xcUmcOnWRdev6AehtJqVUjnNnoigFHElTjgLuyKR+P2BJei+ISH+gP0CZMh4ajupIhh8GX44AnnV45n2z8PPPhxkw4Ht27jwJQI8eNZkwoY3NUSml8jN3Jor0/rRNd8Y5EXkIiACapfe6MWYWMAsgIiLCM7PW/Zxm6OvAUx55y8ycPRvHsGEr+OijLQBUrFiUadM60qZNRZsjU0rld+5MFFHALWnKpYG/r64kIncDLwHNjDEJbozHdYkXYNPb1nbFzhAcZm88gMNhWLDgD/z9fRgxogkvvNCE4GB/u8NSSnkBdyaKTUBlESkPHAV6AL3SVhCResBMoJ0x5oQbY8meBV2d252/ti2MPXtOUb58EQID/QgPD+G//72PMmUKU7VqMdtiUkp5H7eNejLGJAGDgGXAbuALY8xOERktIpenWn0HKAh8KSJbRWShu+Jx2Zd3w18rrO3a/a3FiDzs4sVEXnppFbVrT2fcuF9S97dpU1GThFLK49z6KWiMWQwsvmrfK2m273bn+2fbqsFweJW1Xbs/tJ7p8RCWLt3PwIGLOHjwHACnTl30eAxKKZWWPpl92ZEfYesUZ/nuGR59+7//juXpp5fy5ZfWENxatW5ixoxO3HnnLVkcqZRS7qWJ4rIvWji3n4rz6EN1e/eeJiJiFrGxlwgJ8efVV5vx9NMN8ffPO8uqKqXyL00UAIt7O7d7bQC/II++feXKYdx+eykKFPDn/ffbU7asTuCnlMo9NFF83d452V/dJ6FkA7e/ZUxMAq+8spqBA2/n1lvDEREWLuxBgQK5bzZapZTy7kRxeLUzSfgFQ4tJbn07YwxffbWLp55ayrFj59mz5xRLl1qzlmiSUErlVt6dKL7/P+f2U+4dXXTgwFkGDVrMkiX7AWjYsDRvv527Bn0ppVR6vDdRHP8N4lKm5rh/hdve5tKlZMaP/5XXX19LfHwSRYoE8dZbrXjssfr4+OgEfkqp3M97E8U3HZzbZd33l/2RI9GMHr2GhIRkHnywFu++24YSJQq67f2UUiqneWeiiI2Ci8et7Y7zcvz0Z8/GUaRIECJCxYphTJrUjkqVwmjVqkKOv5dSSrmb26bwyNW+bmf9W/BmqNo9x07rcBhmz95CpUrvM3fu9tT9jz8eoUlCKZVneV+i2L8QTu+0tusMzLHT7tx5gubN59Cv30LOnIlL7bRWSqm8zrtuPV08BQvudZbvePHGT3kxkddfX8P48etISnJw000FmDixLT171rzhcyulVG7gXYliXhPndv8jNzxNx969p2nbdi6HDp1DBAYMqM8bb7SiaNHgGwxUKaVyD+9JFKufhrN/WNstJkFo6Rs+ZdmyhQkK8qNOnRLMmNGJhg1v/Jwq/0hMTCQqKor4+Hi7Q1FeJCgoiNKlS+Pvn3MLm3lHotjwBvyW8tR1w5Fw25DrOk1SkoMZMyLp2bMm4eEhBAb6sXTpg5QqVQg/P+/r7lGZi4qKIjQ0lHLlyiEenGRSeS9jDKdPnyYqKory5cvn2Hnz/6db5LvO9a/LtYPGr1/XaTZuPEqDBh8wePAShg9fmbq/bNkimiRUuuLj4wkPD9ckoTxGRAgPD8/xVmz+blEc+RHWPGdt+wZClwXZPkV0dDwvvfQD06ZtwhgoU6Yw995bJWfjVPmWJgnlae74mcu/iSLmyJVrTAw5n61lTY0x/O9/O3nmmWX88895/Px8GDq0Ia+80kwn8FNKeZX8ec/EGPi4qrP87x3ZXvt627bj9Oz5Nf/8c54777yF337rz9tvt9YkofIUX19f6tatS82aNbnnnns4d+5c6ms7d+6kZcuW3HrrrVSuXJnXX38dY0zq60uWLCEiIoJq1apRtWpVnnvuOTu+hUxt2bKFRx991O4wMvXmm29SqVIlqlSpwrJly9Ktc9ddd1G3bl3q1q3LzTffTJcuXQDrD9YhQ4ZQqVIlateuzW+//QbAyZMnadeunce+B4wxeeqrfv36Jks/vWjMeKyvqJ+zrp8iKSn5ivIzzyw1H3yw2SQnO1w+h1KX7dq1y+4QTIECBVK3+/TpY8aMGWOMMebixYumQoUKZtmyZcYYYy5cuGDatWtnpkyZYowxZseOHaZChQpm9+7dxhhjEhMTzdSpU3M0tsTExBs+x/3332+2bt3q0ffMjp07d5ratWub+Ph4c+DAAVOhQgWTlJSU6TH33Xef+eSTT4wxxixatMi0a9fOOBwOs27dOtOgQYPUeg8//LD5+ef0P9/S+9kDIs11fu7mv1tPyYmwbaa1Xb0PlGrs0mGrVx9k4MDFzJzZiaZNywIwYUJbd0WpvM27buqreNZkXSdFo0aN2L7dmlrms88+o3HjxrRp0waAkJAQpkyZQvPmzXnyyScZN24cL730ElWrWi1zPz8/Bg68diaD8+fPM3jwYCIjIxERRo0aRbdu3ShYsCDnz58H4KuvvuL7779nzpw5PPzww4SFhbFlyxbq1q3L/Pnz2bp1K0WKWKs6VqpUiV9++QUfHx8GDBjA4cOHAXjvvfdo3PjK3+XY2Fi2b99OnTp1ANi4cSNPP/00cXFxBAcH8/HHH1OlShXmzJnDokWLiI+P58KFC/zwww+88847fPHFFyQkJNC1a1dee+01ALp06cKRI0eIj4/nqaeeon///i5f3/QsWLCAHj16EBgYSPny5alUqRIbN26kUaNG6daPjY3lhx9+4OOPP049vk+fPogIDRs25Ny5cxw7doySJUvSpUsX/vvf/15zXdwh/yWKj6tA/GkIKAStZ2VZ/cSJCwwbtoJPP90GwIQJ61IThVL5RXJyMqtWraJfv36Addupfv36V9SpWLEi58+fJyYmht9//51nn302y/O+/vrrFC5cmB07dgBw9uzZLI/Zu3cvK1euxNfXF4fDwfz58+nbty8bNmygXLlylChRgl69evHMM8/QpEkTDh8+TNu2bdm9e/cV54mMjKRmTecMCFWrVmXt2rX4+fmxcuVKXnzxRb7++msA1q1bx/bt2wkLC2P58uXs27ePjRs3Yoyhc+fOrF27lqZNmzJ79mzCwsKIi4vj9ttvp1u3boSHh1/xvs888wyrV6++5vvq0aMHI0aMuGLf0aNHadiwYWq5dOnSHD16NMNrM3/+fFq1akWhQoVSj7/llluuOb5kyZJEREQwcuTIrC53jshfiWLHbIg+aG03HgN+gRlWdTgMH330G8OHr+Ts2XgCA30ZObIpw4bd6aFglVfJxl/+OSkuLo66dety6NAh6tevT+vWrQHrlnNGo2OyM2pm5cqVzJvnnIG5aNGiWR7zwAMP4OvrC0D37t0ZPXo0ffv2Zd68eXTv3j31vLt27Uo9JiYmhtjYWEJDQ1P3HTt2jOLFi6eWo6Oj+fe//82+ffsQERITE1Nfa926NWFhYQAsX76c5cuXU69ePcBqFe3bt4+mTZsyefJk5s+fD8CRI0fYt2/fNYli4sSJrl0cuKLP57LMru/nn39+RZ9LZsffdNNN/P333y7HciPyT6IwBpb3c5ZvG5xh1YMHz/LQQ/P59dcjALRpU5GpUztQqVKYu6NUyqOCg4PZunUr0dHRdOrUialTpzJkyBBq1KjB2rVrr6h74MABChYsSGhoKDVq1GDz5s2pt3UyklHCSbvv6jH9BQoUSN1u1KgR+/fv5+TJk3z77bepfyE7HA7WrVtHcHDG0+EEBwdfce6XX36ZFi1aMH/+fA4dOkTz5s3TfU9jDC+88AKPP/74Fef78ccfWblyJevWrSMkJITmzZun+zxCdloUpUuX5siRI6nlqKgobr755nS/n9OnT7Nx48bURJXV8fHx8Zlen5yUf0Y9fdXauT3wVKZVCxUKZO/e0/zrXwWZN68bS5c+qElC5WuFCxdm8uTJjB8/nsTERB588EF+/vlnVq60Hh6Ni4tjyJAhPP/88wAMGzaMN954g7179wLWB/eECROuOW+bNm2YMmVKavnyracSJUqwe/fu1FtLGRERunbtytChQ6lWrVrqX+9Xn3fr1q3XHFutWjX273fO0hwdHU2pUqUAmDNnTobv2bZtW2bPnp3ah3L06FFOnDhBdHQ0RYsWJSQkhD179rB+/fp0j584cSJbt2695uvqJAHQuXNn5s2bR0JCAgcPHmTfvn00aNAg3fN++eWXdOrUiaCgoCuO//TTTzHGsH79egoXLkzJkiUB6xZe2ltv7pQ/EsXvc+DwKmu71TQIDr+myrJl+0lISAIgPDyEhQt7sGfPk3TvXlMfilJeoV69etSpU4d58+YRHBzMggULGDNmDFWqVKFWrVrcfvvtDBo0CIDatWvz3nvv0bNnT6pVq0bNmjU5duzYNeccOXIkZ8+epWbNmtSpUyf1L+233nqLTp060bJly9QPtox0796duXPnpt52Apg8eTKRkZHUrl2b6tWrM2PGjGuOq1q1KtHR0cTGxgLw/PPP88ILL9C4cWOSk5MzfL82bdrQq1cvGjVqRK1atbj//vuJjY2lXbt2JCUlUbt2bV5++eUr+hauV40aNfi///s/qlevTrt27Zg6dWrqbbcOHTpcceto3rx59OzZ84rjO3ToQIUKFahUqRKPPfYY06ZNS31t9erVdOzY8YZjdIWkdw8sN4uIiDCRkZHOHbFHYVbKZHylmkCPn66of+RINEOGLOXbb/fw+ustGDmyqQejVd5s9+7dVKtWze4w8rWJEycSGhqa65+lcIemTZuyYMGCdPuF0vvZE5HNxpiI63mvvN+iSLu+RLflqZtJSQ4mTFhHtWpT+fbbPRQsGEBYmE7/rVR+8sQTTxAYmPGglfzq5MmTDB061KXBAzkhb3dmn9gGxzdb2w1fAX8rEaxfH8WAAd+zbZu1Lna3btWYNKkdpUoVsitSpZQbBAUF0bt3b7vD8LjixYunPr3tCXk7UaRtTdz5KgAbNkRx550fYQyUK1eEKVPa07HjrfbEp7xeZsNQlXIHd3Qn5N1EcWQNxPxlbXf6InW1ugYNStG2bSXq1fsXI0c2JSQk5xbvUCo7goKCOH36tE41rjzGpKxHkXbkVE7Iu4niB2t0xr6EujzzbCITJpzm1lutX8hFi3rh46O/mMpepUuXJioqipMnT9odivIil1e4y0l5M1Es6UPCP7t564dmvPljSxIu7SMoyI+vvvo/AE0SKlfw9/fP0VXGlLKLW0c9iUg7EflDRPaLyDVPo4hIoIj8L+X1DSJSLsuTxp9l1YKfqf3uE7y6vAUJlwx9+9ZlxoxObvgOlFJKua1FISK+wFSgNRAFbBKRhcaYXWmq9QPOGmMqiUgP4G2g+7Vnczp48Ax373wagGrVijFjRiedxE8ppdzInS2KBsB+Y8wBY8wlYB5w71V17gU+Sdn+CmglWfT6nb0YTJBfIm+8XJetWwdoklBKKTdz25PZInI/0M4Y82hKuTdwhzFmUJo6v6fUiUop/5lS59RV5+oPXJ4Yvibwu1uCznuKAZlPbOU99Fo46bVw0mvhVMUYE5p1tWu5szM7vZbB1VnJlToYY2YBswBEJPJ6H0PPb/RaOOm1cNJr4aTXwklEIrOulT533nqKAm5JUy4NXD15emodEfEDCgNn3BiTUkqpbHJnotgEVBaR8iISAPQAFl5VZyHw75Tt+4EfTF6bpVAppfI5t916MsYkicggYBngC8w2xuwUkdFYi3wvBD4C/iMi+7FaEj1cOHXW65t6D70WTnotnPRaOOm1cLrua5HnphlXSinlWXl/mnGllFJupYlCKaVUpnJtonDL9B95lAvXYqiI7BKR7SKySkTy7VOIWV2LNPXuFxEjIvl2aKQr10JE/i/lZ2OniHzm6Rg9xYXfkTIislpEtqT8nnSwI053E5HZInIi5Rm19F4XEZmccp22i8htLp3YGJPrvrA6v/8EKgABwDag+lV1BgIzUrZ7AP+zO24br0ULICRl+wlvvhYp9UKBtcB6IMLuuG38uagMbAGKppRvsjtuG6/FLOCJlO3qwCG743bTtWgK3Ab8nsHrHYAlWM+wNQQ2uHLe3NqicMv0H3lUltfCGLPaGHMxpbge65mV/MiVnwuA14FxQLwng/MwV67FY8BUY8xZAGPMCQ/H6CmuXAsDXF7isjDXPtOVLxhj1pL5s2j3Ap8ay3qgiIiUzOq8uTVRlAKOpClHpexLt44xJgmIBsI9Ep1nuXIt0uqH9RdDfpTltRCResAtxpjvPRmYDVz5ubgVuFVEfhGR9SLSzmPReZYr1+JV4CERiQIWA4M9E1quk93PEyD3rkeRY9N/5AMuf58i8hAQATRza0T2yfRaiIgPMBF42FMB2ciVnws/rNtPzbFamT+JSE1jzDk3x+ZprlyLnsAcY8y7ItII6/mtmsYYh/vDy1Wu63Mzt7YodPoPJ1euBSJyN/AS0NkYk+Ch2Dwtq2sRijVp5I8icgjrHuzCfNqh7ervyAJjTKIx5iD8f3v3GyJVFRo3NfIAAATqSURBVMZx/PsjLDVJECmSoC0MK0mXspB8EWZJf0hKxC1M20hCKULLXoRBBb2QrBeZ2VoSGpiYoiGVmIRZyJpK+KcWyzCRQEpCJGyD2H69OGdz2mZn7m7+md15PjCwc+7ce84c2PvMfe7lOXxHChz9TZG5eAz4AMB2KzCQVDCw3hQ6n3RVq4Eiyn+cVnUucrplOSlI9Nc8NFSZC9snbQ+33WC7gXS/ZortXhdDq2FF/kc+JD3ogKThpFTU4XM6ynOjyFwcBSYBSLqOFCjqcY3aTcCs/PTTeOCk7WPVdqrJ1JPPXvmPPqfgXCwGhgDr8v38o7annLdBnyUF56IuFJyLLcBkSW1AB/Cs7V/P36jPjoJz8QzwjqT5pFRLc3/8YSlpDSnVODzfj3kBGABgu4V0f+Ye4Afgd+DRQsfth3MVQgjhDKrV1FMIIYQaEYEihBBCRREoQgghVBSBIoQQQkURKEIIIVQUgSLUHEkdkvaWvBoqfLahu0qZPezz81x9dF8ueTGqF8eYI2lW/rtZ0oiSbSskXX+Gx7lbUmOBfeZJGvx/+w71KwJFqEXtthtLXkfOUb8zbI8lFZtc3NOdbbfYfi+/bQZGlGybbbvtjIzy9DiXUWyc84AIFKHXIlCEPiFfOXwp6ev8urXMZ0ZL2pWvQvZLuia3P1zSvlzSBVW6+wIYmfedlNcwOJBr/V+U2xfp9Bogr+a2FyUtkDSNVHNrde5zUL4SGCdprqRXSsbcLOmNXo6zlZKCbpLekrRHae2Jl3LbU6SAtU3Sttw2WVJrnsd1koZU6SfUuQgUoRYNKkk7bcxtvwB32r4RaAKWlNlvDvC67UbSifqnXK6hCZiQ2zuAGVX6vw84IGkgsBJosn0DqZLBXEnDgAeA0bbHAC+X7mx7PbCH9Mu/0XZ7yeb1wNSS903A2l6O8y5SmY5OC22PA8YAt0kaY3sJqZbPRNsTcymP54E78lzuAZ6u0k+oczVZwiPUvfZ8siw1AFiac/IdpLpFXbUCCyVdAWywfUjSJOAmYHcubzKIFHTKWS2pHThCKkM9CvjR9vd5+yrgCWApaa2LFZI+BgqXNLd9XNLhXGfnUO5jRz5uT8Z5MalcRekKZdMlPU76v76ctEDP/i77js/tO3I/F5LmLYRuRaAIfcV84GdgLOlK+D+LEtl+X9JXwL3AFkmzSWWVV9l+rkAfM0oLCEoqu75Jri10C6nI3IPAk8DtPfgua4HpwEFgo20rnbULj5O0itsi4E1gqqSrgAXAzbZPSFpJKnzXlYCtth/qwXhDnYvUU+grhgLH8voBM0m/pv9F0tXA4Zxu2URKwXwGTJN0af7MMBVfU/wg0CBpZH4/E9iec/pDbX9CulFc7smj30hlz8vZANxPWiNhbW7r0Tht/0lKIY3PaatLgFPASUmXAXd3M5adwITO7yRpsKRyV2ch/CMCRegrlgGPSNpJSjudKvOZJuAbSXuBa0lLPraRTqifStoPbCWlZaqy/QepuuY6SQeAv4AW0kn3o3y87aSrna5WAi2dN7O7HPcE0AZcaXtXbuvxOPO9j9eABbb3kdbH/hZ4l5TO6vQ2sFnSNtvHSU9krcn97CTNVQjdiuqxIYQQKoorihBCCBVFoAghhFBRBIoQQggVRaAIIYRQUQSKEEIIFUWgCCGEUFEEihBCCBX9DT6v3SkwJ31oAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e6188c128>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 4650, Minibatch Loss= 0.566554, Training Accuracy= 0.66667\n",
      "Iter 4800, Minibatch Loss= 0.761132, Training Accuracy= 0.40000\n",
      "Iter 4950, Minibatch Loss= 0.609254, Training Accuracy= 0.66667\n",
      "Iter 5100, Minibatch Loss= 0.712238, Training Accuracy= 0.46667\n",
      "Iter 5250, Minibatch Loss= 0.616506, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.64216\n",
      "Validation Loss: 0.6398711\n",
      "Iter 5400, Minibatch Loss= 0.743520, Training Accuracy= 0.53333\n",
      "Iter 5550, Minibatch Loss= 0.524739, Training Accuracy= 0.80000\n",
      "Iter 5700, Minibatch Loss= 0.680425, Training Accuracy= 0.46667\n",
      "Iter 5850, Minibatch Loss= 0.604846, Training Accuracy= 0.73333\n",
      "Iter 6000, Minibatch Loss= 0.725271, Training Accuracy= 0.46667\n",
      "Validation Accuracy: 0.6492\n",
      "Validation Loss: 0.6278761\n",
      "Iter 6150, Minibatch Loss= 0.502995, Training Accuracy= 0.73333\n",
      "Iter 6300, Minibatch Loss= 0.800564, Training Accuracy= 0.46667\n",
      "Iter 6450, Minibatch Loss= 0.689566, Training Accuracy= 0.46667\n",
      "Iter 6600, Minibatch Loss= 0.608621, Training Accuracy= 0.73333\n",
      "Iter 6750, Minibatch Loss= 0.694111, Training Accuracy= 0.46667\n",
      "Validation Accuracy: 0.60904\n",
      "Validation Loss: 0.6481385\n",
      "Iter 6900, Minibatch Loss= 0.657918, Training Accuracy= 0.73333\n",
      "Iter 7050, Minibatch Loss= 0.605038, Training Accuracy= 0.60000\n",
      "Iter 7200, Minibatch Loss= 0.419353, Training Accuracy= 0.93333\n",
      "Iter 7350, Minibatch Loss= 0.550545, Training Accuracy= 0.66667\n",
      "Iter 7500, Minibatch Loss= 0.669654, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.67232\n",
      "Validation Loss: 0.6144397\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e8b7dca20>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 7650, Minibatch Loss= 0.526754, Training Accuracy= 0.86667\n",
      "Iter 7800, Minibatch Loss= 0.576889, Training Accuracy= 0.60000\n",
      "Iter 7950, Minibatch Loss= 0.728425, Training Accuracy= 0.66667\n",
      "Iter 8100, Minibatch Loss= 0.482154, Training Accuracy= 0.86667\n",
      "Iter 8250, Minibatch Loss= 0.587959, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.68352\n",
      "Validation Loss: 0.605211\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e7ea7d5f8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 8400, Minibatch Loss= 0.526586, Training Accuracy= 0.73333\n",
      "Iter 8550, Minibatch Loss= 0.573156, Training Accuracy= 0.66667\n",
      "Iter 8700, Minibatch Loss= 0.617248, Training Accuracy= 0.60000\n",
      "Iter 8850, Minibatch Loss= 0.607214, Training Accuracy= 0.66667\n",
      "Iter 9000, Minibatch Loss= 0.591978, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.7088\n",
      "Validation Loss: 0.58450794\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e7883e630>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 9150, Minibatch Loss= 0.525191, Training Accuracy= 0.80000\n",
      "Iter 9300, Minibatch Loss= 0.527337, Training Accuracy= 0.80000\n",
      "Iter 9450, Minibatch Loss= 0.736057, Training Accuracy= 0.40000\n",
      "Iter 9600, Minibatch Loss= 0.717461, Training Accuracy= 0.46667\n",
      "Iter 9750, Minibatch Loss= 0.512423, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.58608\n",
      "Validation Loss: 0.6514029\n",
      "Iter 9900, Minibatch Loss= 0.771592, Training Accuracy= 0.60000\n",
      "Iter 10050, Minibatch Loss= 0.531081, Training Accuracy= 0.80000\n",
      "Iter 10200, Minibatch Loss= 0.725332, Training Accuracy= 0.60000\n",
      "Iter 10350, Minibatch Loss= 0.847340, Training Accuracy= 0.66667\n",
      "Iter 10500, Minibatch Loss= 0.458600, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.68752\n",
      "Validation Loss: 0.60545194\n",
      "Iter 10650, Minibatch Loss= 0.615282, Training Accuracy= 0.66667\n",
      "Iter 10800, Minibatch Loss= 0.590871, Training Accuracy= 0.66667\n",
      "Iter 10950, Minibatch Loss= 0.705090, Training Accuracy= 0.60000\n",
      "Iter 11100, Minibatch Loss= 0.605701, Training Accuracy= 0.60000\n",
      "Iter 11250, Minibatch Loss= 0.527114, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.68904\n",
      "Validation Loss: 0.5928098\n",
      "Iter 11400, Minibatch Loss= 0.708284, Training Accuracy= 0.40000\n",
      "Iter 11550, Minibatch Loss= 0.680018, Training Accuracy= 0.46667\n",
      "Iter 11700, Minibatch Loss= 0.699345, Training Accuracy= 0.73333\n",
      "Iter 11850, Minibatch Loss= 0.522543, Training Accuracy= 0.80000\n",
      "Iter 12000, Minibatch Loss= 0.575883, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.66312\n",
      "Validation Loss: 0.6075193\n",
      "Iter 12150, Minibatch Loss= 0.609695, Training Accuracy= 0.66667\n",
      "Iter 12300, Minibatch Loss= 0.461812, Training Accuracy= 0.86667\n",
      "Iter 12450, Minibatch Loss= 0.410615, Training Accuracy= 0.86667\n",
      "Iter 12600, Minibatch Loss= 0.608866, Training Accuracy= 0.73333\n",
      "Iter 12750, Minibatch Loss= 0.516651, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.72856\n",
      "Validation Loss: 0.5563914\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e61c49390>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 12900, Minibatch Loss= 0.586507, Training Accuracy= 0.73333\n",
      "Iter 13050, Minibatch Loss= 0.805206, Training Accuracy= 0.40000\n",
      "Iter 13200, Minibatch Loss= 0.562430, Training Accuracy= 0.80000\n",
      "Iter 13350, Minibatch Loss= 0.531074, Training Accuracy= 0.73333\n",
      "Iter 13500, Minibatch Loss= 0.455489, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.7144\n",
      "Validation Loss: 0.5823389\n",
      "Iter 13650, Minibatch Loss= 0.639921, Training Accuracy= 0.73333\n",
      "Iter 13800, Minibatch Loss= 0.547538, Training Accuracy= 0.66667\n",
      "Iter 13950, Minibatch Loss= 0.627845, Training Accuracy= 0.60000\n",
      "Iter 14100, Minibatch Loss= 0.555217, Training Accuracy= 0.86667\n",
      "Iter 14250, Minibatch Loss= 0.454626, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.68856\n",
      "Validation Loss: 0.59199214\n",
      "Iter 14400, Minibatch Loss= 0.640948, Training Accuracy= 0.66667\n",
      "Iter 14550, Minibatch Loss= 0.461343, Training Accuracy= 0.80000\n",
      "Iter 14700, Minibatch Loss= 0.432105, Training Accuracy= 0.86667\n",
      "Iter 14850, Minibatch Loss= 0.540536, Training Accuracy= 0.80000\n",
      "Iter 15000, Minibatch Loss= 0.523123, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.7028\n",
      "Validation Loss: 0.57047874\n",
      "Iter 15150, Minibatch Loss= 0.570924, Training Accuracy= 0.66667\n",
      "Iter 15300, Minibatch Loss= 0.555267, Training Accuracy= 0.66667\n",
      "Iter 15450, Minibatch Loss= 0.300946, Training Accuracy= 0.93333\n",
      "Iter 15600, Minibatch Loss= 0.517552, Training Accuracy= 0.66667\n",
      "Iter 15750, Minibatch Loss= 0.494536, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.75152\n",
      "Validation Loss: 0.5330639\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e66390898>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 15900, Minibatch Loss= 0.538043, Training Accuracy= 0.73333\n",
      "Iter 16050, Minibatch Loss= 0.619295, Training Accuracy= 0.80000\n",
      "Iter 16200, Minibatch Loss= 0.578587, Training Accuracy= 0.73333\n",
      "Iter 16350, Minibatch Loss= 0.414356, Training Accuracy= 0.86667\n",
      "Iter 16500, Minibatch Loss= 0.528505, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.71544\n",
      "Validation Loss: 0.56705153\n",
      "Iter 16650, Minibatch Loss= 0.502665, Training Accuracy= 0.86667\n",
      "Iter 16800, Minibatch Loss= 0.332368, Training Accuracy= 1.00000\n",
      "Iter 16950, Minibatch Loss= 0.502578, Training Accuracy= 0.73333\n",
      "Iter 17100, Minibatch Loss= 0.627657, Training Accuracy= 0.53333\n",
      "Iter 17250, Minibatch Loss= 0.391550, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.71968\n",
      "Validation Loss: 0.56105345\n",
      "Iter 17400, Minibatch Loss= 0.482002, Training Accuracy= 0.86667\n",
      "Iter 17550, Minibatch Loss= 0.771154, Training Accuracy= 0.40000\n",
      "Iter 17700, Minibatch Loss= 0.554778, Training Accuracy= 0.66667\n",
      "Iter 17850, Minibatch Loss= 0.734788, Training Accuracy= 0.53333\n",
      "Iter 18000, Minibatch Loss= 0.428068, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.66704\n",
      "Validation Loss: 0.60161847\n",
      "Iter 18150, Minibatch Loss= 0.400886, Training Accuracy= 0.86667\n",
      "Iter 18300, Minibatch Loss= 0.358799, Training Accuracy= 0.86667\n",
      "Iter 18450, Minibatch Loss= 0.725933, Training Accuracy= 0.53333\n",
      "Iter 18600, Minibatch Loss= 0.786200, Training Accuracy= 0.33333\n",
      "Iter 18750, Minibatch Loss= 0.520839, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.72008\n",
      "Validation Loss: 0.57676524\n",
      "Iter 18900, Minibatch Loss= 0.509391, Training Accuracy= 0.73333\n",
      "Iter 19050, Minibatch Loss= 0.487565, Training Accuracy= 0.80000\n",
      "Iter 19200, Minibatch Loss= 0.488207, Training Accuracy= 0.73333\n",
      "Iter 19350, Minibatch Loss= 0.488369, Training Accuracy= 0.86667\n",
      "Iter 19500, Minibatch Loss= 0.355960, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.72736\n",
      "Validation Loss: 0.55254626\n",
      "Iter 19650, Minibatch Loss= 0.464411, Training Accuracy= 0.73333\n",
      "Iter 19800, Minibatch Loss= 0.511663, Training Accuracy= 0.80000\n",
      "Iter 19950, Minibatch Loss= 0.560352, Training Accuracy= 0.66667\n",
      "Iter 20100, Minibatch Loss= 0.543639, Training Accuracy= 0.86667\n",
      "Iter 20250, Minibatch Loss= 0.417185, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76192\n",
      "Validation Loss: 0.5210202\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e776e41d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 20400, Minibatch Loss= 0.445393, Training Accuracy= 0.86667\n",
      "Iter 20550, Minibatch Loss= 0.706696, Training Accuracy= 0.53333\n",
      "Iter 20700, Minibatch Loss= 0.244871, Training Accuracy= 1.00000\n",
      "Iter 20850, Minibatch Loss= 0.291554, Training Accuracy= 0.93333\n",
      "Iter 21000, Minibatch Loss= 0.550551, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.59432\n",
      "Validation Loss: 0.7583842\n",
      "Iter 21150, Minibatch Loss= 0.784255, Training Accuracy= 0.46667\n",
      "Iter 21300, Minibatch Loss= 0.688535, Training Accuracy= 0.73333\n",
      "Iter 21450, Minibatch Loss= 0.442061, Training Accuracy= 0.80000\n",
      "Iter 21600, Minibatch Loss= 0.383069, Training Accuracy= 0.86667\n",
      "Iter 21750, Minibatch Loss= 0.436592, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.73976\n",
      "Validation Loss: 0.5581036\n",
      "Iter 21900, Minibatch Loss= 0.627996, Training Accuracy= 0.46667\n",
      "Iter 22050, Minibatch Loss= 0.422507, Training Accuracy= 0.80000\n",
      "Iter 22200, Minibatch Loss= 0.262090, Training Accuracy= 1.00000\n",
      "Iter 22350, Minibatch Loss= 0.643039, Training Accuracy= 0.53333\n",
      "Iter 22500, Minibatch Loss= 0.396334, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.7252\n",
      "Validation Loss: 0.5708092\n",
      "Iter 22650, Minibatch Loss= 0.462498, Training Accuracy= 0.80000\n",
      "Iter 22800, Minibatch Loss= 0.537822, Training Accuracy= 0.73333\n",
      "Iter 22950, Minibatch Loss= 0.444811, Training Accuracy= 0.80000\n",
      "Iter 23100, Minibatch Loss= 0.489678, Training Accuracy= 0.66667\n",
      "Iter 23250, Minibatch Loss= 0.393069, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.75776\n",
      "Validation Loss: 0.51496947\n",
      "Iter 23400, Minibatch Loss= 0.475635, Training Accuracy= 0.80000\n",
      "Iter 23550, Minibatch Loss= 0.438887, Training Accuracy= 0.93333\n",
      "Iter 23700, Minibatch Loss= 0.502470, Training Accuracy= 0.73333\n",
      "Iter 23850, Minibatch Loss= 0.579688, Training Accuracy= 0.66667\n",
      "Iter 24000, Minibatch Loss= 0.440733, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.68184\n",
      "Validation Loss: 0.5829029\n",
      "Iter 24150, Minibatch Loss= 0.440224, Training Accuracy= 0.80000\n",
      "Iter 24300, Minibatch Loss= 0.427896, Training Accuracy= 0.80000\n",
      "Iter 24450, Minibatch Loss= 0.362300, Training Accuracy= 0.93333\n",
      "Iter 24600, Minibatch Loss= 0.531466, Training Accuracy= 0.80000\n",
      "Iter 24750, Minibatch Loss= 0.357002, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76168\n",
      "Validation Loss: 0.51107585\n",
      "Iter 24900, Minibatch Loss= 0.368506, Training Accuracy= 0.86667\n",
      "Iter 25050, Minibatch Loss= 0.651684, Training Accuracy= 0.60000\n",
      "Iter 25200, Minibatch Loss= 0.359822, Training Accuracy= 0.93333\n",
      "Iter 25350, Minibatch Loss= 0.510627, Training Accuracy= 0.73333\n",
      "Iter 25500, Minibatch Loss= 0.586596, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.74112\n",
      "Validation Loss: 0.5416284\n",
      "Iter 25650, Minibatch Loss= 0.367744, Training Accuracy= 0.93333\n",
      "Iter 25800, Minibatch Loss= 0.283300, Training Accuracy= 0.93333\n",
      "Iter 25950, Minibatch Loss= 0.574507, Training Accuracy= 0.66667\n",
      "Iter 26100, Minibatch Loss= 0.629280, Training Accuracy= 0.73333\n",
      "Iter 26250, Minibatch Loss= 0.512230, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.76768\n",
      "Validation Loss: 0.51544017\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e79bf3cc0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 26400, Minibatch Loss= 0.567983, Training Accuracy= 0.66667\n",
      "Iter 26550, Minibatch Loss= 0.516913, Training Accuracy= 0.73333\n",
      "Iter 26700, Minibatch Loss= 0.672645, Training Accuracy= 0.53333\n",
      "Iter 26850, Minibatch Loss= 0.423805, Training Accuracy= 0.86667\n",
      "Iter 27000, Minibatch Loss= 0.752752, Training Accuracy= 0.60000\n",
      "Validation Accuracy: 0.7564\n",
      "Validation Loss: 0.5136117\n",
      "Iter 27150, Minibatch Loss= 0.514760, Training Accuracy= 0.73333\n",
      "Iter 27300, Minibatch Loss= 0.402155, Training Accuracy= 0.80000\n",
      "Iter 27450, Minibatch Loss= 0.593757, Training Accuracy= 0.73333\n",
      "Iter 27600, Minibatch Loss= 0.618011, Training Accuracy= 0.73333\n",
      "Iter 27750, Minibatch Loss= 0.727405, Training Accuracy= 0.53333\n",
      "Validation Accuracy: 0.69392\n",
      "Validation Loss: 0.5817039\n",
      "Iter 27900, Minibatch Loss= 0.346569, Training Accuracy= 0.93333\n",
      "Iter 28050, Minibatch Loss= 0.576039, Training Accuracy= 0.73333\n",
      "Iter 28200, Minibatch Loss= 0.555133, Training Accuracy= 0.73333\n",
      "Iter 28350, Minibatch Loss= 0.304876, Training Accuracy= 1.00000\n",
      "Iter 28500, Minibatch Loss= 0.536943, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.71592\n",
      "Validation Loss: 0.5607152\n",
      "Iter 28650, Minibatch Loss= 0.563680, Training Accuracy= 0.66667\n",
      "Iter 28800, Minibatch Loss= 0.664132, Training Accuracy= 0.66667\n",
      "Iter 28950, Minibatch Loss= 0.472782, Training Accuracy= 0.80000\n",
      "Iter 29100, Minibatch Loss= 0.584578, Training Accuracy= 0.66667\n",
      "Iter 29250, Minibatch Loss= 0.570885, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.7428\n",
      "Validation Loss: 0.5241993\n",
      "Iter 29400, Minibatch Loss= 0.501528, Training Accuracy= 0.73333\n",
      "Iter 29550, Minibatch Loss= 0.506056, Training Accuracy= 0.73333\n",
      "Iter 29700, Minibatch Loss= 0.385689, Training Accuracy= 0.86667\n",
      "Iter 29850, Minibatch Loss= 0.418433, Training Accuracy= 0.86667\n",
      "Iter 30000, Minibatch Loss= 0.646110, Training Accuracy= 0.53333\n",
      "Validation Accuracy: 0.72664\n",
      "Validation Loss: 0.539591\n",
      "Iter 30150, Minibatch Loss= 0.374712, Training Accuracy= 0.93333\n",
      "Iter 30300, Minibatch Loss= 0.527616, Training Accuracy= 0.66667\n",
      "Iter 30450, Minibatch Loss= 0.775405, Training Accuracy= 0.80000\n",
      "Iter 30600, Minibatch Loss= 0.361397, Training Accuracy= 0.86667\n",
      "Iter 30750, Minibatch Loss= 0.587631, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.71784\n",
      "Validation Loss: 0.5575264\n",
      "Iter 30900, Minibatch Loss= 0.456791, Training Accuracy= 0.86667\n",
      "Iter 31050, Minibatch Loss= 0.572207, Training Accuracy= 0.73333\n",
      "Iter 31200, Minibatch Loss= 0.491763, Training Accuracy= 0.80000\n",
      "Iter 31350, Minibatch Loss= 0.765367, Training Accuracy= 0.53333\n",
      "Iter 31500, Minibatch Loss= 0.531124, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.75624\n",
      "Validation Loss: 0.51699144\n",
      "Iter 31650, Minibatch Loss= 0.625361, Training Accuracy= 0.66667\n",
      "Iter 31800, Minibatch Loss= 0.607046, Training Accuracy= 0.60000\n",
      "Iter 31950, Minibatch Loss= 0.632996, Training Accuracy= 0.60000\n",
      "Iter 32100, Minibatch Loss= 0.613468, Training Accuracy= 0.60000\n",
      "Iter 32250, Minibatch Loss= 0.421053, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.7344\n",
      "Validation Loss: 0.53484404\n",
      "Iter 32400, Minibatch Loss= 0.455067, Training Accuracy= 0.86667\n",
      "Iter 32550, Minibatch Loss= 0.684659, Training Accuracy= 0.60000\n",
      "Iter 32700, Minibatch Loss= 0.659079, Training Accuracy= 0.60000\n",
      "Iter 32850, Minibatch Loss= 0.521178, Training Accuracy= 0.73333\n",
      "Iter 33000, Minibatch Loss= 0.587387, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.73568\n",
      "Validation Loss: 0.5293205\n",
      "Iter 33150, Minibatch Loss= 0.412231, Training Accuracy= 0.86667\n",
      "Iter 33300, Minibatch Loss= 0.508685, Training Accuracy= 0.73333\n",
      "Iter 33450, Minibatch Loss= 0.342460, Training Accuracy= 0.86667\n",
      "Iter 33600, Minibatch Loss= 0.524627, Training Accuracy= 0.80000\n",
      "Iter 33750, Minibatch Loss= 0.576743, Training Accuracy= 0.60000\n",
      "Validation Accuracy: 0.74992\n",
      "Validation Loss: 0.535965\n",
      "Iter 33900, Minibatch Loss= 0.539648, Training Accuracy= 0.66667\n",
      "Iter 34050, Minibatch Loss= 0.399199, Training Accuracy= 0.93333\n",
      "Iter 34200, Minibatch Loss= 0.365270, Training Accuracy= 0.86667\n",
      "Iter 34350, Minibatch Loss= 0.420191, Training Accuracy= 0.80000\n",
      "Iter 34500, Minibatch Loss= 0.421204, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76344\n",
      "Validation Loss: 0.4976983\n",
      "Iter 34650, Minibatch Loss= 0.337776, Training Accuracy= 0.86667\n",
      "Iter 34800, Minibatch Loss= 0.616970, Training Accuracy= 0.73333\n",
      "Iter 34950, Minibatch Loss= 0.396781, Training Accuracy= 0.86667\n",
      "Iter 35100, Minibatch Loss= 0.724380, Training Accuracy= 0.53333\n",
      "Iter 35250, Minibatch Loss= 0.486436, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.70064\n",
      "Validation Loss: 0.5827911\n",
      "Iter 35400, Minibatch Loss= 0.345053, Training Accuracy= 0.93333\n",
      "Iter 35550, Minibatch Loss= 0.611048, Training Accuracy= 0.73333\n",
      "Iter 35700, Minibatch Loss= 0.398746, Training Accuracy= 0.93333\n",
      "Iter 35850, Minibatch Loss= 0.579529, Training Accuracy= 0.66667\n",
      "Iter 36000, Minibatch Loss= 0.521388, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.762\n",
      "Validation Loss: 0.5075733\n",
      "Iter 36150, Minibatch Loss= 0.266596, Training Accuracy= 0.93333\n",
      "Iter 36300, Minibatch Loss= 0.536506, Training Accuracy= 0.80000\n",
      "Iter 36450, Minibatch Loss= 0.470700, Training Accuracy= 0.80000\n",
      "Iter 36600, Minibatch Loss= 0.551532, Training Accuracy= 0.80000\n",
      "Iter 36750, Minibatch Loss= 0.704177, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.69504\n",
      "Validation Loss: 0.56818813\n",
      "Iter 36900, Minibatch Loss= 0.511472, Training Accuracy= 0.60000\n",
      "Iter 37050, Minibatch Loss= 0.864661, Training Accuracy= 0.46667\n",
      "Iter 37200, Minibatch Loss= 0.501380, Training Accuracy= 0.73333\n",
      "Iter 37350, Minibatch Loss= 0.362650, Training Accuracy= 0.86667\n",
      "Iter 37500, Minibatch Loss= 0.318393, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.7552\n",
      "Validation Loss: 0.5169872\n",
      "Iter 37650, Minibatch Loss= 0.396347, Training Accuracy= 0.86667\n",
      "Iter 37800, Minibatch Loss= 0.415410, Training Accuracy= 0.86667\n",
      "Iter 37950, Minibatch Loss= 0.591191, Training Accuracy= 0.66667\n",
      "Iter 38100, Minibatch Loss= 0.268526, Training Accuracy= 0.93333\n",
      "Iter 38250, Minibatch Loss= 0.602177, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.74296\n",
      "Validation Loss: 0.53628963\n",
      "Iter 38400, Minibatch Loss= 0.555037, Training Accuracy= 0.86667\n",
      "Iter 38550, Minibatch Loss= 0.510491, Training Accuracy= 0.73333\n",
      "Iter 38700, Minibatch Loss= 0.498258, Training Accuracy= 0.80000\n",
      "Iter 38850, Minibatch Loss= 0.451290, Training Accuracy= 0.86667\n",
      "Iter 39000, Minibatch Loss= 0.625549, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.72296\n",
      "Validation Loss: 0.53560567\n",
      "Iter 39150, Minibatch Loss= 0.938708, Training Accuracy= 0.80000\n",
      "Iter 39300, Minibatch Loss= 0.277113, Training Accuracy= 0.93333\n",
      "Iter 39450, Minibatch Loss= 0.459422, Training Accuracy= 0.73333\n",
      "Iter 39600, Minibatch Loss= 0.402713, Training Accuracy= 0.80000\n",
      "Iter 39750, Minibatch Loss= 0.455063, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.77064\n",
      "Validation Loss: 0.487641\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e7ea65748>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 39900, Minibatch Loss= 0.660705, Training Accuracy= 0.73333\n",
      "Iter 40050, Minibatch Loss= 0.445777, Training Accuracy= 0.80000\n",
      "Iter 40200, Minibatch Loss= 0.336384, Training Accuracy= 0.86667\n",
      "Iter 40350, Minibatch Loss= 0.430892, Training Accuracy= 0.80000\n",
      "Iter 40500, Minibatch Loss= 0.570805, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.72848\n",
      "Validation Loss: 0.54206204\n",
      "Iter 40650, Minibatch Loss= 0.522117, Training Accuracy= 0.73333\n",
      "Iter 40800, Minibatch Loss= 0.377517, Training Accuracy= 0.86667\n",
      "Iter 40950, Minibatch Loss= 0.419448, Training Accuracy= 0.80000\n",
      "Iter 41100, Minibatch Loss= 0.602261, Training Accuracy= 0.66667\n",
      "Iter 41250, Minibatch Loss= 0.618349, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.76448\n",
      "Validation Loss: 0.5104968\n",
      "Iter 41400, Minibatch Loss= 0.470328, Training Accuracy= 0.66667\n",
      "Iter 41550, Minibatch Loss= 0.552013, Training Accuracy= 0.86667\n",
      "Iter 41700, Minibatch Loss= 0.464598, Training Accuracy= 0.86667\n",
      "Iter 41850, Minibatch Loss= 0.572474, Training Accuracy= 0.73333\n",
      "Iter 42000, Minibatch Loss= 0.368977, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.77368\n",
      "Validation Loss: 0.49607688\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e65085a20>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 42150, Minibatch Loss= 0.373230, Training Accuracy= 0.80000\n",
      "Iter 42300, Minibatch Loss= 0.424648, Training Accuracy= 0.73333\n",
      "Iter 42450, Minibatch Loss= 0.534525, Training Accuracy= 0.80000\n",
      "Iter 42600, Minibatch Loss= 0.498540, Training Accuracy= 0.73333\n",
      "Iter 42750, Minibatch Loss= 0.371230, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.758\n",
      "Validation Loss: 0.5131673\n",
      "Iter 42900, Minibatch Loss= 0.540290, Training Accuracy= 0.73333\n",
      "Iter 43050, Minibatch Loss= 0.379885, Training Accuracy= 0.93333\n",
      "Iter 43200, Minibatch Loss= 0.709623, Training Accuracy= 0.60000\n",
      "Iter 43350, Minibatch Loss= 0.389069, Training Accuracy= 0.80000\n",
      "Iter 43500, Minibatch Loss= 0.620034, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.7648\n",
      "Validation Loss: 0.5093962\n",
      "Iter 43650, Minibatch Loss= 0.515487, Training Accuracy= 0.66667\n",
      "Iter 43800, Minibatch Loss= 0.840265, Training Accuracy= 0.60000\n",
      "Iter 43950, Minibatch Loss= 0.573592, Training Accuracy= 0.73333\n",
      "Iter 44100, Minibatch Loss= 0.575791, Training Accuracy= 0.66667\n",
      "Iter 44250, Minibatch Loss= 0.436397, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.76064\n",
      "Validation Loss: 0.50382423\n",
      "Iter 44400, Minibatch Loss= 0.620643, Training Accuracy= 0.60000\n",
      "Iter 44550, Minibatch Loss= 0.393671, Training Accuracy= 0.80000\n",
      "Iter 44700, Minibatch Loss= 0.206934, Training Accuracy= 1.00000\n",
      "Iter 44850, Minibatch Loss= 0.454479, Training Accuracy= 0.86667\n",
      "Iter 45000, Minibatch Loss= 0.485208, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.76072\n",
      "Validation Loss: 0.5023482\n",
      "Iter 45150, Minibatch Loss= 0.289223, Training Accuracy= 1.00000\n",
      "Iter 45300, Minibatch Loss= 0.262933, Training Accuracy= 0.93333\n",
      "Iter 45450, Minibatch Loss= 0.729010, Training Accuracy= 0.66667\n",
      "Iter 45600, Minibatch Loss= 0.271122, Training Accuracy= 1.00000\n",
      "Iter 45750, Minibatch Loss= 0.271937, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.76624\n",
      "Validation Loss: 0.50642246\n",
      "Iter 45900, Minibatch Loss= 0.317351, Training Accuracy= 0.86667\n",
      "Iter 46050, Minibatch Loss= 0.447654, Training Accuracy= 0.66667\n",
      "Iter 46200, Minibatch Loss= 0.409376, Training Accuracy= 0.93333\n",
      "Iter 46350, Minibatch Loss= 0.441971, Training Accuracy= 0.80000\n",
      "Iter 46500, Minibatch Loss= 0.503375, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.75392\n",
      "Validation Loss: 0.51523066\n",
      "Iter 46650, Minibatch Loss= 0.481643, Training Accuracy= 0.80000\n",
      "Iter 46800, Minibatch Loss= 0.443290, Training Accuracy= 0.80000\n",
      "Iter 46950, Minibatch Loss= 0.537189, Training Accuracy= 0.73333\n",
      "Iter 47100, Minibatch Loss= 0.530342, Training Accuracy= 0.66667\n",
      "Iter 47250, Minibatch Loss= 0.323396, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.72328\n",
      "Validation Loss: 0.5656364\n",
      "Iter 47400, Minibatch Loss= 0.530525, Training Accuracy= 0.73333\n",
      "Iter 47550, Minibatch Loss= 0.355627, Training Accuracy= 0.93333\n",
      "Iter 47700, Minibatch Loss= 0.524764, Training Accuracy= 0.73333\n",
      "Iter 47850, Minibatch Loss= 0.601909, Training Accuracy= 0.80000\n",
      "Iter 48000, Minibatch Loss= 0.484094, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.74592\n",
      "Validation Loss: 0.5393069\n",
      "Iter 48150, Minibatch Loss= 0.565982, Training Accuracy= 0.66667\n",
      "Iter 48300, Minibatch Loss= 0.496244, Training Accuracy= 0.73333\n",
      "Iter 48450, Minibatch Loss= 0.599344, Training Accuracy= 0.73333\n",
      "Iter 48600, Minibatch Loss= 0.401214, Training Accuracy= 0.73333\n",
      "Iter 48750, Minibatch Loss= 0.350731, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76808\n",
      "Validation Loss: 0.49512222\n",
      "Iter 48900, Minibatch Loss= 0.418433, Training Accuracy= 0.86667\n",
      "Iter 49050, Minibatch Loss= 0.548167, Training Accuracy= 0.66667\n",
      "Iter 49200, Minibatch Loss= 0.299556, Training Accuracy= 0.93333\n",
      "Iter 49350, Minibatch Loss= 0.193769, Training Accuracy= 1.00000\n",
      "Iter 49500, Minibatch Loss= 0.360798, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.7632\n",
      "Validation Loss: 0.5140666\n",
      "Iter 49650, Minibatch Loss= 0.535657, Training Accuracy= 0.66667\n",
      "Iter 49800, Minibatch Loss= 0.290433, Training Accuracy= 0.93333\n",
      "Iter 49950, Minibatch Loss= 0.385770, Training Accuracy= 0.93333\n",
      "Iter 50100, Minibatch Loss= 0.511119, Training Accuracy= 0.66667\n",
      "Iter 50250, Minibatch Loss= 0.488624, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.76096\n",
      "Validation Loss: 0.50851136\n",
      "Iter 50400, Minibatch Loss= 0.550283, Training Accuracy= 0.80000\n",
      "Iter 50550, Minibatch Loss= 0.675345, Training Accuracy= 0.66667\n",
      "Iter 50700, Minibatch Loss= 0.475525, Training Accuracy= 0.73333\n",
      "Iter 50850, Minibatch Loss= 0.436976, Training Accuracy= 0.80000\n",
      "Iter 51000, Minibatch Loss= 0.407544, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76904\n",
      "Validation Loss: 0.49476355\n",
      "Iter 51150, Minibatch Loss= 0.637823, Training Accuracy= 0.66667\n",
      "Iter 51300, Minibatch Loss= 0.521940, Training Accuracy= 0.80000\n",
      "Iter 51450, Minibatch Loss= 0.605493, Training Accuracy= 0.73333\n",
      "Iter 51600, Minibatch Loss= 0.383470, Training Accuracy= 0.80000\n",
      "Iter 51750, Minibatch Loss= 0.298530, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.75976\n",
      "Validation Loss: 0.50795364\n",
      "Iter 51900, Minibatch Loss= 0.560746, Training Accuracy= 0.73333\n",
      "Iter 52050, Minibatch Loss= 0.350186, Training Accuracy= 0.93333\n",
      "Iter 52200, Minibatch Loss= 0.169832, Training Accuracy= 1.00000\n",
      "Iter 52350, Minibatch Loss= 0.328840, Training Accuracy= 0.86667\n",
      "Iter 52500, Minibatch Loss= 0.410999, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.7688\n",
      "Validation Loss: 0.5108928\n",
      "Iter 52650, Minibatch Loss= 0.514961, Training Accuracy= 0.66667\n",
      "Iter 52800, Minibatch Loss= 0.388379, Training Accuracy= 0.73333\n",
      "Iter 52950, Minibatch Loss= 0.283514, Training Accuracy= 0.86667\n",
      "Iter 53100, Minibatch Loss= 0.515644, Training Accuracy= 0.80000\n",
      "Iter 53250, Minibatch Loss= 0.382730, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.75088\n",
      "Validation Loss: 0.52303696\n",
      "Iter 53400, Minibatch Loss= 0.304829, Training Accuracy= 0.93333\n",
      "Iter 53550, Minibatch Loss= 0.417290, Training Accuracy= 0.73333\n",
      "Iter 53700, Minibatch Loss= 0.421585, Training Accuracy= 0.86667\n",
      "Iter 53850, Minibatch Loss= 0.313610, Training Accuracy= 0.86667\n",
      "Iter 54000, Minibatch Loss= 0.258233, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.75592\n",
      "Validation Loss: 0.53101826\n",
      "Iter 54150, Minibatch Loss= 0.456399, Training Accuracy= 0.86667\n",
      "Iter 54300, Minibatch Loss= 0.349374, Training Accuracy= 0.93333\n",
      "Iter 54450, Minibatch Loss= 0.625811, Training Accuracy= 0.60000\n",
      "Iter 54600, Minibatch Loss= 0.566613, Training Accuracy= 0.60000\n",
      "Iter 54750, Minibatch Loss= 0.381670, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.7528\n",
      "Validation Loss: 0.51685095\n",
      "Iter 54900, Minibatch Loss= 0.414349, Training Accuracy= 0.86667\n",
      "Iter 55050, Minibatch Loss= 0.555700, Training Accuracy= 0.60000\n",
      "Iter 55200, Minibatch Loss= 0.227881, Training Accuracy= 0.93333\n",
      "Iter 55350, Minibatch Loss= 0.421877, Training Accuracy= 0.66667\n",
      "Iter 55500, Minibatch Loss= 0.400537, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.75664\n",
      "Validation Loss: 0.54782933\n",
      "Iter 55650, Minibatch Loss= 0.256357, Training Accuracy= 0.93333\n",
      "Iter 55800, Minibatch Loss= 0.207219, Training Accuracy= 0.93333\n",
      "Iter 55950, Minibatch Loss= 0.379096, Training Accuracy= 0.86667\n",
      "Iter 56100, Minibatch Loss= 0.660633, Training Accuracy= 0.66667\n",
      "Iter 56250, Minibatch Loss= 0.488823, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.732\n",
      "Validation Loss: 0.5244397\n",
      "Iter 56400, Minibatch Loss= 0.322450, Training Accuracy= 0.93333\n",
      "Iter 56550, Minibatch Loss= 0.375088, Training Accuracy= 0.80000\n",
      "Iter 56700, Minibatch Loss= 0.315781, Training Accuracy= 0.86667\n",
      "Iter 56850, Minibatch Loss= 0.397964, Training Accuracy= 0.80000\n",
      "Iter 57000, Minibatch Loss= 0.251320, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.76592\n",
      "Validation Loss: 0.4980591\n",
      "Iter 57150, Minibatch Loss= 0.278234, Training Accuracy= 0.93333\n",
      "Iter 57300, Minibatch Loss= 0.583451, Training Accuracy= 0.66667\n",
      "Iter 57450, Minibatch Loss= 0.522963, Training Accuracy= 0.73333\n",
      "Iter 57600, Minibatch Loss= 0.479689, Training Accuracy= 0.86667\n",
      "Iter 57750, Minibatch Loss= 0.409044, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.77216\n",
      "Validation Loss: 0.4925609\n",
      "Iter 57900, Minibatch Loss= 0.378962, Training Accuracy= 0.86667\n",
      "Iter 58050, Minibatch Loss= 0.645952, Training Accuracy= 0.66667\n",
      "Iter 58200, Minibatch Loss= 0.128236, Training Accuracy= 1.00000\n",
      "Iter 58350, Minibatch Loss= 0.230858, Training Accuracy= 0.93333\n",
      "Iter 58500, Minibatch Loss= 0.466460, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.73664\n",
      "Validation Loss: 0.57216847\n",
      "Iter 58650, Minibatch Loss= 0.368164, Training Accuracy= 0.86667\n",
      "Iter 58800, Minibatch Loss= 0.600428, Training Accuracy= 0.66667\n",
      "Iter 58950, Minibatch Loss= 0.275067, Training Accuracy= 0.93333\n",
      "Iter 59100, Minibatch Loss= 0.477867, Training Accuracy= 0.80000\n",
      "Iter 59250, Minibatch Loss= 0.309465, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.7708\n",
      "Validation Loss: 0.5027933\n",
      "Iter 59400, Minibatch Loss= 0.744008, Training Accuracy= 0.53333\n",
      "Iter 59550, Minibatch Loss= 0.426810, Training Accuracy= 0.86667\n",
      "Iter 59700, Minibatch Loss= 0.287393, Training Accuracy= 0.86667\n",
      "Iter 59850, Minibatch Loss= 0.550395, Training Accuracy= 0.73333\n",
      "Iter 60000, Minibatch Loss= 0.381527, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.76552\n",
      "Validation Loss: 0.5125536\n",
      "Iter 60150, Minibatch Loss= 0.304062, Training Accuracy= 0.86667\n",
      "Iter 60300, Minibatch Loss= 0.330212, Training Accuracy= 0.86667\n",
      "Iter 60450, Minibatch Loss= 0.421836, Training Accuracy= 0.80000\n",
      "Iter 60600, Minibatch Loss= 0.434386, Training Accuracy= 0.80000\n",
      "Iter 60750, Minibatch Loss= 0.367111, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.77392\n",
      "Validation Loss: 0.5192365\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e7a8a3f60>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 60900, Minibatch Loss= 0.467088, Training Accuracy= 0.80000\n",
      "Iter 61050, Minibatch Loss= 0.440074, Training Accuracy= 0.80000\n",
      "Iter 61200, Minibatch Loss= 0.322225, Training Accuracy= 0.86667\n",
      "Iter 61350, Minibatch Loss= 0.410736, Training Accuracy= 0.73333\n",
      "Iter 61500, Minibatch Loss= 0.352685, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.766\n",
      "Validation Loss: 0.49900016\n",
      "Iter 61650, Minibatch Loss= 0.473701, Training Accuracy= 0.80000\n",
      "Iter 61800, Minibatch Loss= 0.318356, Training Accuracy= 0.93333\n",
      "Iter 61950, Minibatch Loss= 0.246818, Training Accuracy= 0.93333\n",
      "Iter 62100, Minibatch Loss= 0.618087, Training Accuracy= 0.66667\n",
      "Iter 62250, Minibatch Loss= 0.250690, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.772\n",
      "Validation Loss: 0.4943119\n",
      "Iter 62400, Minibatch Loss= 0.334990, Training Accuracy= 0.80000\n",
      "Iter 62550, Minibatch Loss= 0.380498, Training Accuracy= 0.80000\n",
      "Iter 62700, Minibatch Loss= 0.212520, Training Accuracy= 1.00000\n",
      "Iter 62850, Minibatch Loss= 0.430334, Training Accuracy= 0.80000\n",
      "Iter 63000, Minibatch Loss= 0.484410, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.76416\n",
      "Validation Loss: 0.49271414\n",
      "Iter 63150, Minibatch Loss= 0.326008, Training Accuracy= 0.93333\n",
      "Iter 63300, Minibatch Loss= 0.218899, Training Accuracy= 0.93333\n",
      "Iter 63450, Minibatch Loss= 0.541205, Training Accuracy= 0.73333\n",
      "Iter 63600, Minibatch Loss= 0.544032, Training Accuracy= 0.73333\n",
      "Iter 63750, Minibatch Loss= 0.459698, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.75952\n",
      "Validation Loss: 0.4952641\n",
      "Iter 63900, Minibatch Loss= 0.440509, Training Accuracy= 0.66667\n",
      "Iter 64050, Minibatch Loss= 0.604958, Training Accuracy= 0.73333\n",
      "Iter 64200, Minibatch Loss= 0.787476, Training Accuracy= 0.60000\n",
      "Iter 64350, Minibatch Loss= 0.397878, Training Accuracy= 0.86667\n",
      "Iter 64500, Minibatch Loss= 0.760087, Training Accuracy= 0.60000\n",
      "Validation Accuracy: 0.77104\n",
      "Validation Loss: 0.49497217\n",
      "Iter 64650, Minibatch Loss= 0.483039, Training Accuracy= 0.86667\n",
      "Iter 64800, Minibatch Loss= 0.382729, Training Accuracy= 0.73333\n",
      "Iter 64950, Minibatch Loss= 0.421070, Training Accuracy= 0.86667\n",
      "Iter 65100, Minibatch Loss= 0.531140, Training Accuracy= 0.73333\n",
      "Iter 65250, Minibatch Loss= 0.618988, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.73552\n",
      "Validation Loss: 0.5575602\n",
      "Iter 65400, Minibatch Loss= 0.455845, Training Accuracy= 0.86667\n",
      "Iter 65550, Minibatch Loss= 0.534236, Training Accuracy= 0.86667\n",
      "Iter 65700, Minibatch Loss= 0.505728, Training Accuracy= 0.80000\n",
      "Iter 65850, Minibatch Loss= 0.215533, Training Accuracy= 1.00000\n",
      "Iter 66000, Minibatch Loss= 0.670542, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.7592\n",
      "Validation Loss: 0.5163216\n",
      "Iter 66150, Minibatch Loss= 0.490198, Training Accuracy= 0.80000\n",
      "Iter 66300, Minibatch Loss= 0.655365, Training Accuracy= 0.73333\n",
      "Iter 66450, Minibatch Loss= 0.504342, Training Accuracy= 0.86667\n",
      "Iter 66600, Minibatch Loss= 0.804385, Training Accuracy= 0.60000\n",
      "Iter 66750, Minibatch Loss= 0.557049, Training Accuracy= 0.60000\n",
      "Validation Accuracy: 0.7676\n",
      "Validation Loss: 0.51573694\n",
      "Iter 66900, Minibatch Loss= 0.440741, Training Accuracy= 0.80000\n",
      "Iter 67050, Minibatch Loss= 0.414507, Training Accuracy= 0.80000\n",
      "Iter 67200, Minibatch Loss= 0.388894, Training Accuracy= 0.86667\n",
      "Iter 67350, Minibatch Loss= 0.431060, Training Accuracy= 0.86667\n",
      "Iter 67500, Minibatch Loss= 0.748722, Training Accuracy= 0.46667\n",
      "Validation Accuracy: 0.74808\n",
      "Validation Loss: 0.5184787\n",
      "Iter 67650, Minibatch Loss= 0.362235, Training Accuracy= 0.86667\n",
      "Iter 67800, Minibatch Loss= 0.375719, Training Accuracy= 0.80000\n",
      "Iter 67950, Minibatch Loss= 0.639410, Training Accuracy= 0.80000\n",
      "Iter 68100, Minibatch Loss= 0.434680, Training Accuracy= 0.86667\n",
      "Iter 68250, Minibatch Loss= 0.352358, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76648\n",
      "Validation Loss: 0.5142784\n",
      "Iter 68400, Minibatch Loss= 0.371924, Training Accuracy= 0.86667\n",
      "Iter 68550, Minibatch Loss= 0.471965, Training Accuracy= 0.73333\n",
      "Iter 68700, Minibatch Loss= 0.422806, Training Accuracy= 0.86667\n",
      "Iter 68850, Minibatch Loss= 0.675175, Training Accuracy= 0.66667\n",
      "Iter 69000, Minibatch Loss= 0.489712, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.74888\n",
      "Validation Loss: 0.53168696\n",
      "Iter 69150, Minibatch Loss= 0.304124, Training Accuracy= 0.93333\n",
      "Iter 69300, Minibatch Loss= 0.505635, Training Accuracy= 0.66667\n",
      "Iter 69450, Minibatch Loss= 0.472401, Training Accuracy= 0.86667\n",
      "Iter 69600, Minibatch Loss= 0.612679, Training Accuracy= 0.66667\n",
      "Iter 69750, Minibatch Loss= 0.322515, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76712\n",
      "Validation Loss: 0.48980233\n",
      "Iter 69900, Minibatch Loss= 0.402299, Training Accuracy= 0.86667\n",
      "Iter 70050, Minibatch Loss= 0.563933, Training Accuracy= 0.73333\n",
      "Iter 70200, Minibatch Loss= 0.664060, Training Accuracy= 0.66667\n",
      "Iter 70350, Minibatch Loss= 0.447326, Training Accuracy= 0.73333\n",
      "Iter 70500, Minibatch Loss= 0.485767, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.77056\n",
      "Validation Loss: 0.49175742\n",
      "Iter 70650, Minibatch Loss= 0.340459, Training Accuracy= 0.93333\n",
      "Iter 70800, Minibatch Loss= 0.492149, Training Accuracy= 0.73333\n",
      "Iter 70950, Minibatch Loss= 0.316045, Training Accuracy= 0.86667\n",
      "Iter 71100, Minibatch Loss= 0.558799, Training Accuracy= 0.86667\n",
      "Iter 71250, Minibatch Loss= 0.600380, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.77344\n",
      "Validation Loss: 0.4967271\n",
      "Iter 71400, Minibatch Loss= 0.386758, Training Accuracy= 0.80000\n",
      "Iter 71550, Minibatch Loss= 0.260370, Training Accuracy= 0.93333\n",
      "Iter 71700, Minibatch Loss= 0.293730, Training Accuracy= 0.86667\n",
      "Iter 71850, Minibatch Loss= 0.450698, Training Accuracy= 0.80000\n",
      "Iter 72000, Minibatch Loss= 0.428389, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.75392\n",
      "Validation Loss: 0.53406477\n",
      "Iter 72150, Minibatch Loss= 0.277215, Training Accuracy= 0.93333\n",
      "Iter 72300, Minibatch Loss= 0.612264, Training Accuracy= 0.73333\n",
      "Iter 72450, Minibatch Loss= 0.368829, Training Accuracy= 0.86667\n",
      "Iter 72600, Minibatch Loss= 0.733379, Training Accuracy= 0.53333\n",
      "Iter 72750, Minibatch Loss= 0.325885, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.77424\n",
      "Validation Loss: 0.48946962\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e8499b7b8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 72900, Minibatch Loss= 0.274955, Training Accuracy= 0.93333\n",
      "Iter 73050, Minibatch Loss= 0.623022, Training Accuracy= 0.73333\n",
      "Iter 73200, Minibatch Loss= 0.340834, Training Accuracy= 0.93333\n",
      "Iter 73350, Minibatch Loss= 0.532532, Training Accuracy= 0.80000\n",
      "Iter 73500, Minibatch Loss= 0.467064, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.77032\n",
      "Validation Loss: 0.49309608\n",
      "Iter 73650, Minibatch Loss= 0.186618, Training Accuracy= 1.00000\n",
      "Iter 73800, Minibatch Loss= 0.482780, Training Accuracy= 0.80000\n",
      "Iter 73950, Minibatch Loss= 0.494233, Training Accuracy= 0.73333\n",
      "Iter 74100, Minibatch Loss= 0.491249, Training Accuracy= 0.73333\n",
      "Iter 74250, Minibatch Loss= 0.499695, Training Accuracy= 0.60000\n",
      "Validation Accuracy: 0.77272\n",
      "Validation Loss: 0.49392524\n",
      "Iter 74400, Minibatch Loss= 0.300373, Training Accuracy= 0.80000\n",
      "Iter 74550, Minibatch Loss= 0.966772, Training Accuracy= 0.40000\n",
      "Iter 74700, Minibatch Loss= 0.500121, Training Accuracy= 0.73333\n",
      "Iter 74850, Minibatch Loss= 0.508374, Training Accuracy= 0.66667\n",
      "Iter 75000, Minibatch Loss= 0.250869, Training Accuracy= 1.00000\n",
      "Validation Accuracy: 0.748\n",
      "Validation Loss: 0.52838385\n",
      "Checkpoint saved at chkpts/model_good_run_50k.ckpt\n",
      "Iter 75150, Minibatch Loss= 0.377739, Training Accuracy= 0.86667\n",
      "Iter 75300, Minibatch Loss= 0.400533, Training Accuracy= 0.86667\n",
      "Iter 75450, Minibatch Loss= 0.512362, Training Accuracy= 0.73333\n",
      "Iter 75600, Minibatch Loss= 0.292532, Training Accuracy= 0.93333\n",
      "Iter 75750, Minibatch Loss= 0.511014, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.77368\n",
      "Validation Loss: 0.49221304\n",
      "Iter 75900, Minibatch Loss= 0.424049, Training Accuracy= 0.80000\n",
      "Iter 76050, Minibatch Loss= 0.474719, Training Accuracy= 0.73333\n",
      "Iter 76200, Minibatch Loss= 0.542901, Training Accuracy= 0.66667\n",
      "Iter 76350, Minibatch Loss= 0.426565, Training Accuracy= 0.73333\n",
      "Iter 76500, Minibatch Loss= 0.527142, Training Accuracy= 0.60000\n",
      "Validation Accuracy: 0.77112\n",
      "Validation Loss: 0.48931533\n",
      "Iter 76650, Minibatch Loss= 0.644859, Training Accuracy= 0.73333\n",
      "Iter 76800, Minibatch Loss= 0.276836, Training Accuracy= 0.86667\n",
      "Iter 76950, Minibatch Loss= 0.510561, Training Accuracy= 0.73333\n",
      "Iter 77100, Minibatch Loss= 0.358832, Training Accuracy= 0.80000\n",
      "Iter 77250, Minibatch Loss= 0.446978, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.77216\n",
      "Validation Loss: 0.48889893\n",
      "Iter 77400, Minibatch Loss= 0.667843, Training Accuracy= 0.73333\n",
      "Iter 77550, Minibatch Loss= 0.415463, Training Accuracy= 0.80000\n",
      "Iter 77700, Minibatch Loss= 0.310001, Training Accuracy= 0.86667\n",
      "Iter 77850, Minibatch Loss= 0.378925, Training Accuracy= 0.80000\n",
      "Iter 78000, Minibatch Loss= 0.643118, Training Accuracy= 0.46667\n",
      "Validation Accuracy: 0.64368\n",
      "Validation Loss: 0.62032247\n",
      "Iter 78150, Minibatch Loss= 0.513366, Training Accuracy= 0.66667\n",
      "Iter 78300, Minibatch Loss= 0.367914, Training Accuracy= 0.86667\n",
      "Iter 78450, Minibatch Loss= 0.357259, Training Accuracy= 0.86667\n",
      "Iter 78600, Minibatch Loss= 0.538878, Training Accuracy= 0.53333\n",
      "Iter 78750, Minibatch Loss= 0.584014, Training Accuracy= 0.60000\n",
      "Validation Accuracy: 0.7656\n",
      "Validation Loss: 0.4982377\n",
      "Iter 78900, Minibatch Loss= 0.513156, Training Accuracy= 0.73333\n",
      "Iter 79050, Minibatch Loss= 0.531467, Training Accuracy= 0.80000\n",
      "Iter 79200, Minibatch Loss= 0.435687, Training Accuracy= 0.86667\n",
      "Iter 79350, Minibatch Loss= 0.591479, Training Accuracy= 0.73333\n",
      "Iter 79500, Minibatch Loss= 0.354649, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.7668\n",
      "Validation Loss: 0.49796593\n",
      "Iter 79650, Minibatch Loss= 0.344095, Training Accuracy= 0.86667\n",
      "Iter 79800, Minibatch Loss= 0.449004, Training Accuracy= 0.86667\n",
      "Iter 79950, Minibatch Loss= 0.525308, Training Accuracy= 0.80000\n",
      "Iter 80100, Minibatch Loss= 0.404104, Training Accuracy= 0.80000\n",
      "Iter 80250, Minibatch Loss= 0.288070, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.75752\n",
      "Validation Loss: 0.5057954\n",
      "Iter 80400, Minibatch Loss= 0.612307, Training Accuracy= 0.73333\n",
      "Iter 80550, Minibatch Loss= 0.292155, Training Accuracy= 0.93333\n",
      "Iter 80700, Minibatch Loss= 0.725999, Training Accuracy= 0.53333\n",
      "Iter 80850, Minibatch Loss= 0.364180, Training Accuracy= 0.93333\n",
      "Iter 81000, Minibatch Loss= 0.645175, Training Accuracy= 0.53333\n",
      "Validation Accuracy: 0.7624\n",
      "Validation Loss: 0.5124744\n",
      "Iter 81150, Minibatch Loss= 0.483579, Training Accuracy= 0.73333\n",
      "Iter 81300, Minibatch Loss= 0.801556, Training Accuracy= 0.53333\n",
      "Iter 81450, Minibatch Loss= 0.587286, Training Accuracy= 0.73333\n",
      "Iter 81600, Minibatch Loss= 0.572504, Training Accuracy= 0.66667\n",
      "Iter 81750, Minibatch Loss= 0.457328, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.7484\n",
      "Validation Loss: 0.5201996\n",
      "Iter 81900, Minibatch Loss= 0.631074, Training Accuracy= 0.60000\n",
      "Iter 82050, Minibatch Loss= 0.427409, Training Accuracy= 0.80000\n",
      "Iter 82200, Minibatch Loss= 0.162469, Training Accuracy= 1.00000\n",
      "Iter 82350, Minibatch Loss= 0.429899, Training Accuracy= 0.86667\n",
      "Iter 82500, Minibatch Loss= 0.353943, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.7716\n",
      "Validation Loss: 0.49046817\n",
      "Iter 82650, Minibatch Loss= 0.242550, Training Accuracy= 1.00000\n",
      "Iter 82800, Minibatch Loss= 0.277133, Training Accuracy= 0.93333\n",
      "Iter 82950, Minibatch Loss= 0.699142, Training Accuracy= 0.66667\n",
      "Iter 83100, Minibatch Loss= 0.239458, Training Accuracy= 1.00000\n",
      "Iter 83250, Minibatch Loss= 0.214229, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.76832\n",
      "Validation Loss: 0.5007671\n",
      "Iter 83400, Minibatch Loss= 0.322116, Training Accuracy= 0.80000\n",
      "Iter 83550, Minibatch Loss= 0.496228, Training Accuracy= 0.73333\n",
      "Iter 83700, Minibatch Loss= 0.458261, Training Accuracy= 0.73333\n",
      "Iter 83850, Minibatch Loss= 0.489526, Training Accuracy= 0.80000\n",
      "Iter 84000, Minibatch Loss= 0.463162, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.71936\n",
      "Validation Loss: 0.5394371\n",
      "Iter 84150, Minibatch Loss= 0.450805, Training Accuracy= 0.66667\n",
      "Iter 84300, Minibatch Loss= 0.337370, Training Accuracy= 0.93333\n",
      "Iter 84450, Minibatch Loss= 0.408138, Training Accuracy= 0.80000\n",
      "Iter 84600, Minibatch Loss= 0.491743, Training Accuracy= 0.73333\n",
      "Iter 84750, Minibatch Loss= 0.272358, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.73184\n",
      "Validation Loss: 0.5611978\n",
      "Iter 84900, Minibatch Loss= 0.456526, Training Accuracy= 0.73333\n",
      "Iter 85050, Minibatch Loss= 0.287409, Training Accuracy= 0.93333\n",
      "Iter 85200, Minibatch Loss= 0.513933, Training Accuracy= 0.73333\n",
      "Iter 85350, Minibatch Loss= 0.543616, Training Accuracy= 0.80000\n",
      "Iter 85500, Minibatch Loss= 0.524478, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.7492\n",
      "Validation Loss: 0.5277393\n",
      "Iter 85650, Minibatch Loss= 0.432276, Training Accuracy= 0.80000\n",
      "Iter 85800, Minibatch Loss= 0.462366, Training Accuracy= 0.66667\n",
      "Iter 85950, Minibatch Loss= 0.588777, Training Accuracy= 0.73333\n",
      "Iter 86100, Minibatch Loss= 0.330384, Training Accuracy= 0.86667\n",
      "Iter 86250, Minibatch Loss= 0.314082, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76616\n",
      "Validation Loss: 0.49457702\n",
      "Iter 86400, Minibatch Loss= 0.428231, Training Accuracy= 0.86667\n",
      "Iter 86550, Minibatch Loss= 0.492753, Training Accuracy= 0.73333\n",
      "Iter 86700, Minibatch Loss= 0.294649, Training Accuracy= 0.93333\n",
      "Iter 86850, Minibatch Loss= 0.186945, Training Accuracy= 1.00000\n",
      "Iter 87000, Minibatch Loss= 0.404728, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.77488\n",
      "Validation Loss: 0.4948583\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2e611fc748>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iter 87150, Minibatch Loss= 0.487908, Training Accuracy= 0.80000\n",
      "Iter 87300, Minibatch Loss= 0.240780, Training Accuracy= 0.86667\n",
      "Iter 87450, Minibatch Loss= 0.327747, Training Accuracy= 0.86667\n",
      "Iter 87600, Minibatch Loss= 0.449505, Training Accuracy= 0.73333\n",
      "Iter 87750, Minibatch Loss= 0.395255, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.76328\n",
      "Validation Loss: 0.496542\n",
      "Iter 87900, Minibatch Loss= 0.589134, Training Accuracy= 0.80000\n",
      "Iter 88050, Minibatch Loss= 0.681714, Training Accuracy= 0.73333\n",
      "Iter 88200, Minibatch Loss= 0.455474, Training Accuracy= 0.73333\n",
      "Iter 88350, Minibatch Loss= 0.351046, Training Accuracy= 0.80000\n",
      "Iter 88500, Minibatch Loss= 0.443427, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76664\n",
      "Validation Loss: 0.50349677\n",
      "Iter 88650, Minibatch Loss= 0.679516, Training Accuracy= 0.66667\n",
      "Iter 88800, Minibatch Loss= 0.538070, Training Accuracy= 0.80000\n",
      "Iter 88950, Minibatch Loss= 0.596127, Training Accuracy= 0.73333\n",
      "Iter 89100, Minibatch Loss= 0.464914, Training Accuracy= 0.86667\n",
      "Iter 89250, Minibatch Loss= 0.274011, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.7504\n",
      "Validation Loss: 0.5161352\n",
      "Iter 89400, Minibatch Loss= 0.566159, Training Accuracy= 0.73333\n",
      "Iter 89550, Minibatch Loss= 0.336311, Training Accuracy= 0.93333\n",
      "Iter 89700, Minibatch Loss= 0.224517, Training Accuracy= 1.00000\n",
      "Iter 89850, Minibatch Loss= 0.314433, Training Accuracy= 0.86667\n",
      "Iter 90000, Minibatch Loss= 0.421020, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.77344\n",
      "Validation Loss: 0.5018263\n",
      "Iter 90150, Minibatch Loss= 0.487702, Training Accuracy= 0.80000\n",
      "Iter 90300, Minibatch Loss= 0.365735, Training Accuracy= 0.73333\n",
      "Iter 90450, Minibatch Loss= 0.194338, Training Accuracy= 1.00000\n",
      "Iter 90600, Minibatch Loss= 0.496571, Training Accuracy= 0.80000\n",
      "Iter 90750, Minibatch Loss= 0.355641, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.7692\n",
      "Validation Loss: 0.5073977\n",
      "Iter 90900, Minibatch Loss= 0.327874, Training Accuracy= 0.86667\n",
      "Iter 91050, Minibatch Loss= 0.469101, Training Accuracy= 0.66667\n",
      "Iter 91200, Minibatch Loss= 0.432211, Training Accuracy= 0.80000\n",
      "Iter 91350, Minibatch Loss= 0.454892, Training Accuracy= 0.73333\n",
      "Iter 91500, Minibatch Loss= 0.346213, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.7476\n",
      "Validation Loss: 0.52254844\n",
      "Iter 91650, Minibatch Loss= 0.411938, Training Accuracy= 0.86667\n",
      "Iter 91800, Minibatch Loss= 0.311271, Training Accuracy= 0.93333\n",
      "Iter 91950, Minibatch Loss= 0.737441, Training Accuracy= 0.46667\n",
      "Iter 92100, Minibatch Loss= 0.563293, Training Accuracy= 0.73333\n",
      "Iter 92250, Minibatch Loss= 0.382673, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76792\n",
      "Validation Loss: 0.49850133\n",
      "Iter 92400, Minibatch Loss= 0.437227, Training Accuracy= 0.86667\n",
      "Iter 92550, Minibatch Loss= 0.480108, Training Accuracy= 0.66667\n",
      "Iter 92700, Minibatch Loss= 0.251012, Training Accuracy= 1.00000\n",
      "Iter 92850, Minibatch Loss= 0.521523, Training Accuracy= 0.73333\n",
      "Iter 93000, Minibatch Loss= 0.381294, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76808\n",
      "Validation Loss: 0.49807447\n",
      "Iter 93150, Minibatch Loss= 0.228166, Training Accuracy= 0.93333\n",
      "Iter 93300, Minibatch Loss= 0.203185, Training Accuracy= 0.93333\n",
      "Iter 93450, Minibatch Loss= 0.380973, Training Accuracy= 0.73333\n",
      "Iter 93600, Minibatch Loss= 0.666031, Training Accuracy= 0.66667\n",
      "Iter 93750, Minibatch Loss= 0.421571, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.77112\n",
      "Validation Loss: 0.49226782\n",
      "Iter 93900, Minibatch Loss= 0.366146, Training Accuracy= 0.86667\n",
      "Iter 94050, Minibatch Loss= 0.418876, Training Accuracy= 0.80000\n",
      "Iter 94200, Minibatch Loss= 0.322821, Training Accuracy= 0.86667\n",
      "Iter 94350, Minibatch Loss= 0.359515, Training Accuracy= 0.86667\n",
      "Iter 94500, Minibatch Loss= 0.220026, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.7536\n",
      "Validation Loss: 0.5103371\n",
      "Iter 94650, Minibatch Loss= 0.306223, Training Accuracy= 0.93333\n",
      "Iter 94800, Minibatch Loss= 0.436976, Training Accuracy= 0.73333\n",
      "Iter 94950, Minibatch Loss= 0.533683, Training Accuracy= 0.66667\n",
      "Iter 95100, Minibatch Loss= 0.350552, Training Accuracy= 0.86667\n",
      "Iter 95250, Minibatch Loss= 0.467082, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.75736\n",
      "Validation Loss: 0.51327944\n",
      "Iter 95400, Minibatch Loss= 0.365058, Training Accuracy= 0.86667\n",
      "Iter 95550, Minibatch Loss= 0.673926, Training Accuracy= 0.66667\n",
      "Iter 95700, Minibatch Loss= 0.125607, Training Accuracy= 1.00000\n",
      "Iter 95850, Minibatch Loss= 0.237660, Training Accuracy= 0.93333\n",
      "Iter 96000, Minibatch Loss= 0.447645, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.75544\n",
      "Validation Loss: 0.5349875\n",
      "Iter 96150, Minibatch Loss= 0.387842, Training Accuracy= 0.80000\n",
      "Iter 96300, Minibatch Loss= 0.525328, Training Accuracy= 0.73333\n",
      "Iter 96450, Minibatch Loss= 0.282009, Training Accuracy= 0.93333\n",
      "Iter 96600, Minibatch Loss= 0.343591, Training Accuracy= 0.86667\n",
      "Iter 96750, Minibatch Loss= 0.318480, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.7524\n",
      "Validation Loss: 0.5158715\n",
      "Iter 96900, Minibatch Loss= 0.640903, Training Accuracy= 0.60000\n",
      "Iter 97050, Minibatch Loss= 0.409272, Training Accuracy= 0.80000\n",
      "Iter 97200, Minibatch Loss= 0.273347, Training Accuracy= 0.86667\n",
      "Iter 97350, Minibatch Loss= 0.539347, Training Accuracy= 0.73333\n",
      "Iter 97500, Minibatch Loss= 0.374137, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.76632\n",
      "Validation Loss: 0.508729\n",
      "Iter 97650, Minibatch Loss= 0.286927, Training Accuracy= 0.86667\n",
      "Iter 97800, Minibatch Loss= 0.349396, Training Accuracy= 0.86667\n",
      "Iter 97950, Minibatch Loss= 0.357774, Training Accuracy= 0.80000\n",
      "Iter 98100, Minibatch Loss= 0.441979, Training Accuracy= 0.86667\n",
      "Iter 98250, Minibatch Loss= 0.362004, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76224\n",
      "Validation Loss: 0.5074039\n",
      "Iter 98400, Minibatch Loss= 0.358527, Training Accuracy= 0.86667\n",
      "Iter 98550, Minibatch Loss= 0.396576, Training Accuracy= 0.80000\n",
      "Iter 98700, Minibatch Loss= 0.217507, Training Accuracy= 0.93333\n",
      "Iter 98850, Minibatch Loss= 0.436879, Training Accuracy= 0.73333\n",
      "Iter 99000, Minibatch Loss= 0.418025, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.74848\n",
      "Validation Loss: 0.51430774\n",
      "Iter 99150, Minibatch Loss= 0.471920, Training Accuracy= 0.80000\n",
      "Iter 99300, Minibatch Loss= 0.303687, Training Accuracy= 0.86667\n",
      "Iter 99450, Minibatch Loss= 0.187398, Training Accuracy= 0.93333\n",
      "Iter 99600, Minibatch Loss= 0.641780, Training Accuracy= 0.66667\n",
      "Iter 99750, Minibatch Loss= 0.243190, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.77056\n",
      "Validation Loss: 0.48935857\n",
      "Iter 99900, Minibatch Loss= 0.268985, Training Accuracy= 0.93333\n",
      "Iter 100050, Minibatch Loss= 0.340994, Training Accuracy= 0.86667\n",
      "Iter 100200, Minibatch Loss= 0.252369, Training Accuracy= 1.00000\n",
      "Iter 100350, Minibatch Loss= 0.421639, Training Accuracy= 0.80000\n",
      "Iter 100500, Minibatch Loss= 0.475886, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.75648\n",
      "Validation Loss: 0.5129036\n",
      "Iter 100650, Minibatch Loss= 0.333601, Training Accuracy= 0.93333\n",
      "Iter 100800, Minibatch Loss= 0.195522, Training Accuracy= 0.93333\n",
      "Iter 100950, Minibatch Loss= 0.552792, Training Accuracy= 0.73333\n",
      "Iter 101100, Minibatch Loss= 0.553642, Training Accuracy= 0.73333\n",
      "Iter 101250, Minibatch Loss= 0.483060, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.77224\n",
      "Validation Loss: 0.495644\n",
      "Iter 101400, Minibatch Loss= 0.588155, Training Accuracy= 0.66667\n",
      "Iter 101550, Minibatch Loss= 0.616382, Training Accuracy= 0.73333\n",
      "Iter 101700, Minibatch Loss= 0.794196, Training Accuracy= 0.60000\n",
      "Iter 101850, Minibatch Loss= 0.355306, Training Accuracy= 0.86667\n",
      "Iter 102000, Minibatch Loss= 0.707106, Training Accuracy= 0.60000\n",
      "Validation Accuracy: 0.77136\n",
      "Validation Loss: 0.49451986\n",
      "Iter 102150, Minibatch Loss= 0.490361, Training Accuracy= 0.80000\n",
      "Iter 102300, Minibatch Loss= 0.375653, Training Accuracy= 0.80000\n",
      "Iter 102450, Minibatch Loss= 0.450925, Training Accuracy= 0.86667\n",
      "Iter 102600, Minibatch Loss= 0.817946, Training Accuracy= 0.73333\n",
      "Iter 102750, Minibatch Loss= 0.455842, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.76552\n",
      "Validation Loss: 0.49924046\n",
      "Iter 102900, Minibatch Loss= 0.416983, Training Accuracy= 0.93333\n",
      "Iter 103050, Minibatch Loss= 0.479298, Training Accuracy= 0.86667\n",
      "Iter 103200, Minibatch Loss= 0.512765, Training Accuracy= 0.80000\n",
      "Iter 103350, Minibatch Loss= 0.200552, Training Accuracy= 1.00000\n",
      "Iter 103500, Minibatch Loss= 0.666475, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.75424\n",
      "Validation Loss: 0.53116053\n",
      "Iter 103650, Minibatch Loss= 0.490611, Training Accuracy= 0.80000\n",
      "Iter 103800, Minibatch Loss= 0.654436, Training Accuracy= 0.73333\n",
      "Iter 103950, Minibatch Loss= 0.540619, Training Accuracy= 0.86667\n",
      "Iter 104100, Minibatch Loss= 0.761443, Training Accuracy= 0.66667\n",
      "Iter 104250, Minibatch Loss= 0.532057, Training Accuracy= 0.60000\n",
      "Validation Accuracy: 0.76912\n",
      "Validation Loss: 0.5149357\n",
      "Iter 104400, Minibatch Loss= 0.412566, Training Accuracy= 0.80000\n",
      "Iter 104550, Minibatch Loss= 0.431107, Training Accuracy= 0.80000\n",
      "Iter 104700, Minibatch Loss= 0.426311, Training Accuracy= 0.80000\n",
      "Iter 104850, Minibatch Loss= 0.486544, Training Accuracy= 0.80000\n",
      "Iter 105000, Minibatch Loss= 0.735920, Training Accuracy= 0.53333\n",
      "Validation Accuracy: 0.742\n",
      "Validation Loss: 0.5204136\n",
      "Iter 105150, Minibatch Loss= 0.419980, Training Accuracy= 0.86667\n",
      "Iter 105300, Minibatch Loss= 0.371910, Training Accuracy= 0.80000\n",
      "Iter 105450, Minibatch Loss= 0.637148, Training Accuracy= 0.80000\n",
      "Iter 105600, Minibatch Loss= 0.409684, Training Accuracy= 0.86667\n",
      "Iter 105750, Minibatch Loss= 0.351546, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.77328\n",
      "Validation Loss: 0.49547276\n",
      "Iter 105900, Minibatch Loss= 0.393368, Training Accuracy= 0.86667\n",
      "Iter 106050, Minibatch Loss= 0.465134, Training Accuracy= 0.73333\n",
      "Iter 106200, Minibatch Loss= 0.397656, Training Accuracy= 0.86667\n",
      "Iter 106350, Minibatch Loss= 0.690001, Training Accuracy= 0.66667\n",
      "Iter 106500, Minibatch Loss= 0.458967, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.75872\n",
      "Validation Loss: 0.513372\n",
      "Iter 106650, Minibatch Loss= 0.282641, Training Accuracy= 0.93333\n",
      "Iter 106800, Minibatch Loss= 0.503850, Training Accuracy= 0.80000\n",
      "Iter 106950, Minibatch Loss= 0.493224, Training Accuracy= 0.80000\n",
      "Iter 107100, Minibatch Loss= 0.700531, Training Accuracy= 0.60000\n",
      "Iter 107250, Minibatch Loss= 0.333311, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.77008\n",
      "Validation Loss: 0.48733687\n",
      "Iter 107400, Minibatch Loss= 0.368081, Training Accuracy= 0.93333\n",
      "Iter 107550, Minibatch Loss= 0.541835, Training Accuracy= 0.80000\n",
      "Iter 107700, Minibatch Loss= 0.619812, Training Accuracy= 0.73333\n",
      "Iter 107850, Minibatch Loss= 0.405121, Training Accuracy= 0.86667\n",
      "Iter 108000, Minibatch Loss= 0.483270, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.76992\n",
      "Validation Loss: 0.5032693\n",
      "Iter 108150, Minibatch Loss= 0.289610, Training Accuracy= 0.93333\n",
      "Iter 108300, Minibatch Loss= 0.577750, Training Accuracy= 0.73333\n",
      "Iter 108450, Minibatch Loss= 0.300446, Training Accuracy= 0.93333\n",
      "Iter 108600, Minibatch Loss= 0.516448, Training Accuracy= 0.80000\n",
      "Iter 108750, Minibatch Loss= 0.620640, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.76584\n",
      "Validation Loss: 0.5172431\n",
      "Iter 108900, Minibatch Loss= 0.445825, Training Accuracy= 0.80000\n",
      "Iter 109050, Minibatch Loss= 0.286104, Training Accuracy= 0.93333\n",
      "Iter 109200, Minibatch Loss= 0.300281, Training Accuracy= 0.86667\n",
      "Iter 109350, Minibatch Loss= 0.372306, Training Accuracy= 0.80000\n",
      "Iter 109500, Minibatch Loss= 0.387656, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.73592\n",
      "Validation Loss: 0.5418914\n",
      "Iter 109650, Minibatch Loss= 0.292581, Training Accuracy= 0.93333\n",
      "Iter 109800, Minibatch Loss= 0.611972, Training Accuracy= 0.73333\n",
      "Iter 109950, Minibatch Loss= 0.351443, Training Accuracy= 0.86667\n",
      "Iter 110100, Minibatch Loss= 0.710590, Training Accuracy= 0.60000\n",
      "Iter 110250, Minibatch Loss= 0.308798, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.77088\n",
      "Validation Loss: 0.49323785\n",
      "Iter 110400, Minibatch Loss= 0.269694, Training Accuracy= 0.93333\n",
      "Iter 110550, Minibatch Loss= 0.711083, Training Accuracy= 0.60000\n",
      "Iter 110700, Minibatch Loss= 0.341550, Training Accuracy= 0.93333\n",
      "Iter 110850, Minibatch Loss= 0.512970, Training Accuracy= 0.80000\n",
      "Iter 111000, Minibatch Loss= 0.465066, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.76936\n",
      "Validation Loss: 0.49456584\n",
      "Iter 111150, Minibatch Loss= 0.178710, Training Accuracy= 1.00000\n",
      "Iter 111300, Minibatch Loss= 0.552762, Training Accuracy= 0.80000\n",
      "Iter 111450, Minibatch Loss= 0.518101, Training Accuracy= 0.73333\n",
      "Iter 111600, Minibatch Loss= 0.457072, Training Accuracy= 0.73333\n",
      "Iter 111750, Minibatch Loss= 0.589112, Training Accuracy= 0.60000\n",
      "Validation Accuracy: 0.74576\n",
      "Validation Loss: 0.51877004\n",
      "Iter 111900, Minibatch Loss= 0.280889, Training Accuracy= 0.80000\n",
      "Iter 112050, Minibatch Loss= 0.836820, Training Accuracy= 0.53333\n",
      "Iter 112200, Minibatch Loss= 0.493964, Training Accuracy= 0.66667\n",
      "Iter 112350, Minibatch Loss= 0.419852, Training Accuracy= 0.80000\n",
      "Iter 112500, Minibatch Loss= 0.229388, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.77312\n",
      "Validation Loss: 0.4900122\n",
      "Iter 112650, Minibatch Loss= 0.374216, Training Accuracy= 0.86667\n",
      "Iter 112800, Minibatch Loss= 0.392043, Training Accuracy= 0.86667\n",
      "Iter 112950, Minibatch Loss= 0.434026, Training Accuracy= 0.80000\n",
      "Iter 113100, Minibatch Loss= 0.238574, Training Accuracy= 0.93333\n",
      "Iter 113250, Minibatch Loss= 0.531304, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.76552\n",
      "Validation Loss: 0.50446737\n",
      "Iter 113400, Minibatch Loss= 0.437117, Training Accuracy= 0.86667\n",
      "Iter 113550, Minibatch Loss= 0.501049, Training Accuracy= 0.73333\n",
      "Iter 113700, Minibatch Loss= 0.500353, Training Accuracy= 0.73333\n",
      "Iter 113850, Minibatch Loss= 0.453534, Training Accuracy= 0.66667\n",
      "Iter 114000, Minibatch Loss= 0.543378, Training Accuracy= 0.60000\n",
      "Validation Accuracy: 0.77464\n",
      "Validation Loss: 0.488276\n",
      "Iter 114150, Minibatch Loss= 0.553519, Training Accuracy= 0.80000\n",
      "Iter 114300, Minibatch Loss= 0.323016, Training Accuracy= 0.86667\n",
      "Iter 114450, Minibatch Loss= 0.422673, Training Accuracy= 0.73333\n",
      "Iter 114600, Minibatch Loss= 0.375362, Training Accuracy= 0.80000\n",
      "Iter 114750, Minibatch Loss= 0.453512, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76704\n",
      "Validation Loss: 0.49651647\n",
      "Iter 114900, Minibatch Loss= 0.671436, Training Accuracy= 0.73333\n",
      "Iter 115050, Minibatch Loss= 0.410055, Training Accuracy= 0.80000\n",
      "Iter 115200, Minibatch Loss= 0.317035, Training Accuracy= 0.86667\n",
      "Iter 115350, Minibatch Loss= 0.410225, Training Accuracy= 0.80000\n",
      "Iter 115500, Minibatch Loss= 0.580214, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.71888\n",
      "Validation Loss: 0.55046064\n",
      "Iter 115650, Minibatch Loss= 0.457898, Training Accuracy= 0.66667\n",
      "Iter 115800, Minibatch Loss= 0.387255, Training Accuracy= 0.86667\n",
      "Iter 115950, Minibatch Loss= 0.317561, Training Accuracy= 0.86667\n",
      "Iter 116100, Minibatch Loss= 0.627509, Training Accuracy= 0.66667\n",
      "Iter 116250, Minibatch Loss= 0.621737, Training Accuracy= 0.60000\n",
      "Validation Accuracy: 0.76256\n",
      "Validation Loss: 0.49730054\n",
      "Iter 116400, Minibatch Loss= 0.490976, Training Accuracy= 0.66667\n",
      "Iter 116550, Minibatch Loss= 0.507928, Training Accuracy= 0.80000\n",
      "Iter 116700, Minibatch Loss= 0.421533, Training Accuracy= 0.86667\n",
      "Iter 116850, Minibatch Loss= 0.556381, Training Accuracy= 0.73333\n",
      "Iter 117000, Minibatch Loss= 0.340440, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76672\n",
      "Validation Loss: 0.49923593\n",
      "Iter 117150, Minibatch Loss= 0.346977, Training Accuracy= 0.86667\n",
      "Iter 117300, Minibatch Loss= 0.429254, Training Accuracy= 0.80000\n",
      "Iter 117450, Minibatch Loss= 0.445053, Training Accuracy= 0.80000\n",
      "Iter 117600, Minibatch Loss= 0.376413, Training Accuracy= 0.80000\n",
      "Iter 117750, Minibatch Loss= 0.282769, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76336\n",
      "Validation Loss: 0.5018176\n",
      "Iter 117900, Minibatch Loss= 0.552342, Training Accuracy= 0.73333\n",
      "Iter 118050, Minibatch Loss= 0.289401, Training Accuracy= 0.93333\n",
      "Iter 118200, Minibatch Loss= 0.703592, Training Accuracy= 0.53333\n",
      "Iter 118350, Minibatch Loss= 0.399179, Training Accuracy= 0.93333\n",
      "Iter 118500, Minibatch Loss= 0.623676, Training Accuracy= 0.60000\n",
      "Validation Accuracy: 0.76544\n",
      "Validation Loss: 0.50629145\n",
      "Iter 118650, Minibatch Loss= 0.493419, Training Accuracy= 0.73333\n",
      "Iter 118800, Minibatch Loss= 0.791172, Training Accuracy= 0.66667\n",
      "Iter 118950, Minibatch Loss= 0.609125, Training Accuracy= 0.73333\n",
      "Iter 119100, Minibatch Loss= 0.533987, Training Accuracy= 0.80000\n",
      "Iter 119250, Minibatch Loss= 0.431639, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.74848\n",
      "Validation Loss: 0.5256196\n",
      "Iter 119400, Minibatch Loss= 0.595464, Training Accuracy= 0.60000\n",
      "Iter 119550, Minibatch Loss= 0.431272, Training Accuracy= 0.80000\n",
      "Iter 119700, Minibatch Loss= 0.148618, Training Accuracy= 1.00000\n",
      "Iter 119850, Minibatch Loss= 0.421218, Training Accuracy= 0.86667\n",
      "Iter 120000, Minibatch Loss= 0.348282, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.77176\n",
      "Validation Loss: 0.48439032\n",
      "Iter 120150, Minibatch Loss= 0.246959, Training Accuracy= 1.00000\n",
      "Iter 120300, Minibatch Loss= 0.297207, Training Accuracy= 0.93333\n",
      "Iter 120450, Minibatch Loss= 0.713584, Training Accuracy= 0.66667\n",
      "Iter 120600, Minibatch Loss= 0.223546, Training Accuracy= 0.93333\n",
      "Iter 120750, Minibatch Loss= 0.228438, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.76976\n",
      "Validation Loss: 0.4929744\n",
      "Iter 120900, Minibatch Loss= 0.330495, Training Accuracy= 0.80000\n",
      "Iter 121050, Minibatch Loss= 0.471074, Training Accuracy= 0.80000\n",
      "Iter 121200, Minibatch Loss= 0.457285, Training Accuracy= 0.86667\n",
      "Iter 121350, Minibatch Loss= 0.451577, Training Accuracy= 0.80000\n",
      "Iter 121500, Minibatch Loss= 0.375302, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.7716\n",
      "Validation Loss: 0.48649922\n",
      "Iter 121650, Minibatch Loss= 0.451732, Training Accuracy= 0.80000\n",
      "Iter 121800, Minibatch Loss= 0.332991, Training Accuracy= 0.86667\n",
      "Iter 121950, Minibatch Loss= 0.391228, Training Accuracy= 0.86667\n",
      "Iter 122100, Minibatch Loss= 0.438610, Training Accuracy= 0.73333\n",
      "Iter 122250, Minibatch Loss= 0.385054, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.75464\n",
      "Validation Loss: 0.50067276\n",
      "Iter 122400, Minibatch Loss= 0.607200, Training Accuracy= 0.66667\n",
      "Iter 122550, Minibatch Loss= 0.346232, Training Accuracy= 0.93333\n",
      "Iter 122700, Minibatch Loss= 0.469449, Training Accuracy= 0.80000\n",
      "Iter 122850, Minibatch Loss= 0.498471, Training Accuracy= 0.80000\n",
      "Iter 123000, Minibatch Loss= 0.520867, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.76808\n",
      "Validation Loss: 0.5113861\n",
      "Iter 123150, Minibatch Loss= 0.465039, Training Accuracy= 0.86667\n",
      "Iter 123300, Minibatch Loss= 0.475943, Training Accuracy= 0.73333\n",
      "Iter 123450, Minibatch Loss= 0.574831, Training Accuracy= 0.73333\n",
      "Iter 123600, Minibatch Loss= 0.385091, Training Accuracy= 0.86667\n",
      "Iter 123750, Minibatch Loss= 0.293583, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.75128\n",
      "Validation Loss: 0.51153463\n",
      "Iter 123900, Minibatch Loss= 0.344066, Training Accuracy= 0.86667\n",
      "Iter 124050, Minibatch Loss= 0.375402, Training Accuracy= 0.80000\n",
      "Iter 124200, Minibatch Loss= 0.260330, Training Accuracy= 0.93333\n",
      "Iter 124350, Minibatch Loss= 0.167427, Training Accuracy= 1.00000\n",
      "Iter 124500, Minibatch Loss= 0.405593, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.76056\n",
      "Validation Loss: 0.5079813\n",
      "Iter 124650, Minibatch Loss= 0.535059, Training Accuracy= 0.73333\n",
      "Iter 124800, Minibatch Loss= 0.239956, Training Accuracy= 0.86667\n",
      "Iter 124950, Minibatch Loss= 0.344734, Training Accuracy= 0.86667\n",
      "Iter 125100, Minibatch Loss= 0.406168, Training Accuracy= 0.80000\n",
      "Iter 125250, Minibatch Loss= 0.395533, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.77\n",
      "Validation Loss: 0.49236566\n",
      "Iter 125400, Minibatch Loss= 0.587949, Training Accuracy= 0.80000\n",
      "Iter 125550, Minibatch Loss= 0.691334, Training Accuracy= 0.73333\n",
      "Iter 125700, Minibatch Loss= 0.464043, Training Accuracy= 0.73333\n",
      "Iter 125850, Minibatch Loss= 0.350831, Training Accuracy= 0.80000\n",
      "Iter 126000, Minibatch Loss= 0.454049, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76736\n",
      "Validation Loss: 0.50433326\n",
      "Iter 126150, Minibatch Loss= 0.638107, Training Accuracy= 0.60000\n",
      "Iter 126300, Minibatch Loss= 0.483425, Training Accuracy= 0.80000\n",
      "Iter 126450, Minibatch Loss= 0.579728, Training Accuracy= 0.66667\n",
      "Iter 126600, Minibatch Loss= 0.386356, Training Accuracy= 0.86667\n",
      "Iter 126750, Minibatch Loss= 0.267899, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76624\n",
      "Validation Loss: 0.5276335\n",
      "Iter 126900, Minibatch Loss= 0.572284, Training Accuracy= 0.73333\n",
      "Iter 127050, Minibatch Loss= 0.330977, Training Accuracy= 0.93333\n",
      "Iter 127200, Minibatch Loss= 0.139666, Training Accuracy= 1.00000\n",
      "Iter 127350, Minibatch Loss= 0.240093, Training Accuracy= 0.86667\n",
      "Iter 127500, Minibatch Loss= 0.370253, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.75032\n",
      "Validation Loss: 0.5180975\n",
      "Iter 127650, Minibatch Loss= 0.547971, Training Accuracy= 0.66667\n",
      "Iter 127800, Minibatch Loss= 0.361329, Training Accuracy= 0.80000\n",
      "Iter 127950, Minibatch Loss= 0.201249, Training Accuracy= 1.00000\n",
      "Iter 128100, Minibatch Loss= 0.506418, Training Accuracy= 0.80000\n",
      "Iter 128250, Minibatch Loss= 0.386171, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.7548\n",
      "Validation Loss: 0.53326297\n",
      "Iter 128400, Minibatch Loss= 0.330442, Training Accuracy= 0.86667\n",
      "Iter 128550, Minibatch Loss= 0.431957, Training Accuracy= 0.73333\n",
      "Iter 128700, Minibatch Loss= 0.444217, Training Accuracy= 0.80000\n",
      "Iter 128850, Minibatch Loss= 0.491783, Training Accuracy= 0.73333\n",
      "Iter 129000, Minibatch Loss= 0.317792, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.75264\n",
      "Validation Loss: 0.52555287\n",
      "Iter 129150, Minibatch Loss= 0.383236, Training Accuracy= 0.86667\n",
      "Iter 129300, Minibatch Loss= 0.251598, Training Accuracy= 0.93333\n",
      "Iter 129450, Minibatch Loss= 0.469214, Training Accuracy= 0.80000\n",
      "Iter 129600, Minibatch Loss= 0.501704, Training Accuracy= 0.73333\n",
      "Iter 129750, Minibatch Loss= 0.351354, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76352\n",
      "Validation Loss: 0.49334028\n",
      "Iter 129900, Minibatch Loss= 0.407113, Training Accuracy= 0.86667\n",
      "Iter 130050, Minibatch Loss= 0.454526, Training Accuracy= 0.66667\n",
      "Iter 130200, Minibatch Loss= 0.248464, Training Accuracy= 0.93333\n",
      "Iter 130350, Minibatch Loss= 0.422139, Training Accuracy= 0.66667\n",
      "Iter 130500, Minibatch Loss= 0.459238, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.73384\n",
      "Validation Loss: 0.62873405\n",
      "Iter 130650, Minibatch Loss= 0.252153, Training Accuracy= 0.93333\n",
      "Iter 130800, Minibatch Loss= 0.191208, Training Accuracy= 0.93333\n",
      "Iter 130950, Minibatch Loss= 0.423775, Training Accuracy= 0.73333\n",
      "Iter 131100, Minibatch Loss= 0.682755, Training Accuracy= 0.66667\n",
      "Iter 131250, Minibatch Loss= 0.498519, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.75\n",
      "Validation Loss: 0.53503567\n",
      "Iter 131400, Minibatch Loss= 0.344900, Training Accuracy= 0.86667\n",
      "Iter 131550, Minibatch Loss= 0.413845, Training Accuracy= 0.86667\n",
      "Iter 131700, Minibatch Loss= 0.305991, Training Accuracy= 1.00000\n",
      "Iter 131850, Minibatch Loss= 0.344994, Training Accuracy= 0.86667\n",
      "Iter 132000, Minibatch Loss= 0.223926, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.75944\n",
      "Validation Loss: 0.5057603\n",
      "Iter 132150, Minibatch Loss= 0.386668, Training Accuracy= 0.80000\n",
      "Iter 132300, Minibatch Loss= 0.434895, Training Accuracy= 0.73333\n",
      "Iter 132450, Minibatch Loss= 0.566903, Training Accuracy= 0.73333\n",
      "Iter 132600, Minibatch Loss= 0.374373, Training Accuracy= 0.80000\n",
      "Iter 132750, Minibatch Loss= 0.429530, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.76984\n",
      "Validation Loss: 0.5031053\n",
      "Iter 132900, Minibatch Loss= 0.403985, Training Accuracy= 0.93333\n",
      "Iter 133050, Minibatch Loss= 0.694064, Training Accuracy= 0.73333\n",
      "Iter 133200, Minibatch Loss= 0.147197, Training Accuracy= 1.00000\n",
      "Iter 133350, Minibatch Loss= 0.253502, Training Accuracy= 0.93333\n",
      "Iter 133500, Minibatch Loss= 0.479687, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.72296\n",
      "Validation Loss: 0.60759133\n",
      "Iter 133650, Minibatch Loss= 0.407466, Training Accuracy= 0.73333\n",
      "Iter 133800, Minibatch Loss= 0.511580, Training Accuracy= 0.73333\n",
      "Iter 133950, Minibatch Loss= 0.319753, Training Accuracy= 0.86667\n",
      "Iter 134100, Minibatch Loss= 0.523859, Training Accuracy= 0.80000\n",
      "Iter 134250, Minibatch Loss= 0.329774, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.77288\n",
      "Validation Loss: 0.49669197\n",
      "Iter 134400, Minibatch Loss= 0.458789, Training Accuracy= 0.80000\n",
      "Iter 134550, Minibatch Loss= 0.398070, Training Accuracy= 0.80000\n",
      "Iter 134700, Minibatch Loss= 0.243434, Training Accuracy= 0.93333\n",
      "Iter 134850, Minibatch Loss= 0.531963, Training Accuracy= 0.73333\n",
      "Iter 135000, Minibatch Loss= 0.364318, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.75368\n",
      "Validation Loss: 0.5179706\n",
      "Iter 135150, Minibatch Loss= 0.257468, Training Accuracy= 0.93333\n",
      "Iter 135300, Minibatch Loss= 0.275853, Training Accuracy= 0.86667\n",
      "Iter 135450, Minibatch Loss= 0.351478, Training Accuracy= 0.80000\n",
      "Iter 135600, Minibatch Loss= 0.430952, Training Accuracy= 0.80000\n",
      "Iter 135750, Minibatch Loss= 0.325719, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76096\n",
      "Validation Loss: 0.51173246\n",
      "Iter 135900, Minibatch Loss= 0.359419, Training Accuracy= 0.86667\n",
      "Iter 136050, Minibatch Loss= 0.377277, Training Accuracy= 0.80000\n",
      "Iter 136200, Minibatch Loss= 0.220373, Training Accuracy= 0.93333\n",
      "Iter 136350, Minibatch Loss= 0.406871, Training Accuracy= 0.73333\n",
      "Iter 136500, Minibatch Loss= 0.367770, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.75952\n",
      "Validation Loss: 0.5179454\n",
      "Iter 136650, Minibatch Loss= 0.404190, Training Accuracy= 0.80000\n",
      "Iter 136800, Minibatch Loss= 0.297815, Training Accuracy= 0.93333\n",
      "Iter 136950, Minibatch Loss= 0.230657, Training Accuracy= 0.93333\n",
      "Iter 137100, Minibatch Loss= 0.591196, Training Accuracy= 0.80000\n",
      "Iter 137250, Minibatch Loss= 0.228827, Training Accuracy= 0.93333\n",
      "Validation Accuracy: 0.77\n",
      "Validation Loss: 0.49852672\n",
      "Iter 137400, Minibatch Loss= 0.421529, Training Accuracy= 0.80000\n",
      "Iter 137550, Minibatch Loss= 0.529452, Training Accuracy= 0.73333\n",
      "Iter 137700, Minibatch Loss= 0.297447, Training Accuracy= 0.93333\n",
      "Iter 137850, Minibatch Loss= 0.423579, Training Accuracy= 0.80000\n",
      "Iter 138000, Minibatch Loss= 0.455638, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.7688\n",
      "Validation Loss: 0.50376946\n",
      "Iter 138150, Minibatch Loss= 0.289440, Training Accuracy= 0.86667\n",
      "Iter 138300, Minibatch Loss= 0.185060, Training Accuracy= 1.00000\n",
      "Iter 138450, Minibatch Loss= 0.560387, Training Accuracy= 0.73333\n",
      "Iter 138600, Minibatch Loss= 0.524261, Training Accuracy= 0.73333\n",
      "Iter 138750, Minibatch Loss= 0.465633, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.77136\n",
      "Validation Loss: 0.4892868\n",
      "Iter 138900, Minibatch Loss= 0.429971, Training Accuracy= 0.66667\n",
      "Iter 139050, Minibatch Loss= 0.560713, Training Accuracy= 0.73333\n",
      "Iter 139200, Minibatch Loss= 0.774501, Training Accuracy= 0.60000\n",
      "Iter 139350, Minibatch Loss= 0.370657, Training Accuracy= 0.86667\n",
      "Iter 139500, Minibatch Loss= 0.610428, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.77\n",
      "Validation Loss: 0.4974825\n",
      "Iter 139650, Minibatch Loss= 0.447575, Training Accuracy= 0.80000\n",
      "Iter 139800, Minibatch Loss= 0.343538, Training Accuracy= 0.80000\n",
      "Iter 139950, Minibatch Loss= 0.366251, Training Accuracy= 0.86667\n",
      "Iter 140100, Minibatch Loss= 0.368384, Training Accuracy= 0.93333\n",
      "Iter 140250, Minibatch Loss= 0.569580, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.73536\n",
      "Validation Loss: 0.5327226\n",
      "Iter 140400, Minibatch Loss= 0.488443, Training Accuracy= 0.73333\n",
      "Iter 140550, Minibatch Loss= 0.534326, Training Accuracy= 0.86667\n",
      "Iter 140700, Minibatch Loss= 0.530001, Training Accuracy= 0.80000\n",
      "Iter 140850, Minibatch Loss= 0.174591, Training Accuracy= 1.00000\n",
      "Iter 141000, Minibatch Loss= 0.680143, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.7592\n",
      "Validation Loss: 0.5307113\n",
      "Iter 141150, Minibatch Loss= 0.463006, Training Accuracy= 0.80000\n",
      "Iter 141300, Minibatch Loss= 0.623234, Training Accuracy= 0.73333\n",
      "Iter 141450, Minibatch Loss= 0.520438, Training Accuracy= 0.80000\n",
      "Iter 141600, Minibatch Loss= 0.695960, Training Accuracy= 0.66667\n",
      "Iter 141750, Minibatch Loss= 0.502235, Training Accuracy= 0.66667\n",
      "Validation Accuracy: 0.76768\n",
      "Validation Loss: 0.5044555\n",
      "Iter 141900, Minibatch Loss= 0.429822, Training Accuracy= 0.80000\n",
      "Iter 142050, Minibatch Loss= 0.403617, Training Accuracy= 0.80000\n",
      "Iter 142200, Minibatch Loss= 0.439483, Training Accuracy= 0.80000\n",
      "Iter 142350, Minibatch Loss= 0.472347, Training Accuracy= 0.73333\n",
      "Iter 142500, Minibatch Loss= 0.710252, Training Accuracy= 0.53333\n",
      "Validation Accuracy: 0.76376\n",
      "Validation Loss: 0.49981344\n",
      "Iter 142650, Minibatch Loss= 0.405841, Training Accuracy= 0.80000\n",
      "Iter 142800, Minibatch Loss= 0.317610, Training Accuracy= 0.86667\n",
      "Iter 142950, Minibatch Loss= 0.596794, Training Accuracy= 0.80000\n",
      "Iter 143100, Minibatch Loss= 0.448350, Training Accuracy= 0.80000\n",
      "Iter 143250, Minibatch Loss= 0.277156, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.75512\n",
      "Validation Loss: 0.5371488\n",
      "Iter 143400, Minibatch Loss= 0.332935, Training Accuracy= 0.86667\n",
      "Iter 143550, Minibatch Loss= 0.458672, Training Accuracy= 0.73333\n",
      "Iter 143700, Minibatch Loss= 0.387109, Training Accuracy= 0.93333\n",
      "Iter 143850, Minibatch Loss= 0.715665, Training Accuracy= 0.60000\n",
      "Iter 144000, Minibatch Loss= 0.461531, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.74768\n",
      "Validation Loss: 0.52947664\n",
      "Iter 144150, Minibatch Loss= 0.281779, Training Accuracy= 1.00000\n",
      "Iter 144300, Minibatch Loss= 0.497784, Training Accuracy= 0.73333\n",
      "Iter 144450, Minibatch Loss= 0.494667, Training Accuracy= 0.66667\n",
      "Iter 144600, Minibatch Loss= 0.564377, Training Accuracy= 0.60000\n",
      "Iter 144750, Minibatch Loss= 0.340755, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.7724\n",
      "Validation Loss: 0.48990503\n",
      "Iter 144900, Minibatch Loss= 0.372493, Training Accuracy= 0.93333\n",
      "Iter 145050, Minibatch Loss= 0.534251, Training Accuracy= 0.73333\n",
      "Iter 145200, Minibatch Loss= 0.586090, Training Accuracy= 0.73333\n",
      "Iter 145350, Minibatch Loss= 0.443722, Training Accuracy= 0.73333\n",
      "Iter 145500, Minibatch Loss= 0.445480, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.7712\n",
      "Validation Loss: 0.49580452\n",
      "Iter 145650, Minibatch Loss= 0.280569, Training Accuracy= 0.93333\n",
      "Iter 145800, Minibatch Loss= 0.554071, Training Accuracy= 0.73333\n",
      "Iter 145950, Minibatch Loss= 0.288123, Training Accuracy= 0.86667\n",
      "Iter 146100, Minibatch Loss= 0.433033, Training Accuracy= 0.86667\n",
      "Iter 146250, Minibatch Loss= 0.568276, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.74784\n",
      "Validation Loss: 0.5173216\n",
      "Iter 146400, Minibatch Loss= 0.357157, Training Accuracy= 0.86667\n",
      "Iter 146550, Minibatch Loss= 0.299988, Training Accuracy= 0.93333\n",
      "Iter 146700, Minibatch Loss= 0.299853, Training Accuracy= 0.86667\n",
      "Iter 146850, Minibatch Loss= 0.411233, Training Accuracy= 0.86667\n",
      "Iter 147000, Minibatch Loss= 0.405028, Training Accuracy= 0.80000\n",
      "Validation Accuracy: 0.76928\n",
      "Validation Loss: 0.4922613\n",
      "Iter 147150, Minibatch Loss= 0.251897, Training Accuracy= 0.86667\n",
      "Iter 147300, Minibatch Loss= 0.609320, Training Accuracy= 0.73333\n",
      "Iter 147450, Minibatch Loss= 0.375133, Training Accuracy= 0.86667\n",
      "Iter 147600, Minibatch Loss= 0.770888, Training Accuracy= 0.53333\n",
      "Iter 147750, Minibatch Loss= 0.320773, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.76984\n",
      "Validation Loss: 0.50131553\n",
      "Iter 147900, Minibatch Loss= 0.206202, Training Accuracy= 0.93333\n",
      "Iter 148050, Minibatch Loss= 0.660118, Training Accuracy= 0.66667\n",
      "Iter 148200, Minibatch Loss= 0.326462, Training Accuracy= 0.93333\n",
      "Iter 148350, Minibatch Loss= 0.510225, Training Accuracy= 0.73333\n",
      "Iter 148500, Minibatch Loss= 0.437527, Training Accuracy= 0.86667\n",
      "Validation Accuracy: 0.7696\n",
      "Validation Loss: 0.49408296\n",
      "Iter 148650, Minibatch Loss= 0.155857, Training Accuracy= 1.00000\n",
      "Iter 148800, Minibatch Loss= 0.542397, Training Accuracy= 0.73333\n",
      "Iter 148950, Minibatch Loss= 0.462871, Training Accuracy= 0.73333\n",
      "Iter 149100, Minibatch Loss= 0.418542, Training Accuracy= 0.66667\n",
      "Iter 149250, Minibatch Loss= 0.466522, Training Accuracy= 0.73333\n",
      "Validation Accuracy: 0.77072\n",
      "Validation Loss: 0.49915138\n",
      "Iter 149400, Minibatch Loss= 0.284123, Training Accuracy= 0.86667\n",
      "Iter 149550, Minibatch Loss= 0.875517, Training Accuracy= 0.40000\n",
      "Iter 149700, Minibatch Loss= 0.472929, Training Accuracy= 0.73333\n",
      "Iter 149850, Minibatch Loss= 0.383144, Training Accuracy= 0.80000\n",
      "Optimization Finished!\n",
      "Checkpoint saved at chkpts/model_good_run_50k.ckpt\n",
      "Final Testing Accuracy: 0.76344\n",
      "CPU times: user 1h 28min 1s, sys: 1min 13s, total: 1h 29min 15s\n",
      "Wall time: 1h 25min\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "best_test_acc = -1.0\n",
    "sess = tf.Session()\n",
    "\n",
    "# Launch the graph\n",
    "if True:\n",
    "    sess.run(init)\n",
    "    test_data, test_label, test_seqlen, test_dnn = dataset.get_test_batch()\n",
    "    test_dict = {\n",
    "        x: test_data, \n",
    "        y: test_label,\n",
    "        seqlen: test_seqlen, \n",
    "        x2: test_dnn,\n",
    "        lstm_keep_prob: 1.00,\n",
    "        fc_keep_prob: 1.00\n",
    "    }\n",
    "    step = 1\n",
    "    \n",
    "    # Keep training until reach max iterations\n",
    "    while step * batch_size < training_iters:\n",
    "        batch_x, batch_y, batch_seqlen, batch_dnn = dataset.next(batch_size)\n",
    "        # Run optimization op (backprop)\n",
    "        sess.run(optimizer, feed_dict={\n",
    "            x: batch_x, \n",
    "            y: batch_y,\n",
    "            seqlen: batch_seqlen, \n",
    "            x2: batch_dnn,\n",
    "            lstm_keep_prob: config.lstm_keep_prob,\n",
    "            fc_keep_prob: config.fc_keep_prob\n",
    "        })\n",
    "        \n",
    "        eval_dict = {\n",
    "            x:      batch_x, \n",
    "            y:      batch_y,\n",
    "            seqlen: batch_seqlen, \n",
    "            x2:     batch_dnn,\n",
    "            lstm_keep_prob: 1.00,\n",
    "            fc_keep_prob:   1.00\n",
    "        }\n",
    "        \n",
    "        if should_log_metrics:\n",
    "            acc = sess.run(accuracy, feed_dict=eval_dict)\n",
    "            loss = sess.run(cost, feed_dict=eval_dict)\n",
    "            line_start = str(step*batch_size) + \",\"\n",
    "            train_loss_log.write(line_start + str(loss) + '\\n')\n",
    "            train_loss_log.flush()\n",
    "            train_acc_log.write(line_start + str(acc) + '\\n')\n",
    "            train_acc_log.flush()\n",
    "\n",
    "        if step % display_step == 0:\n",
    "            if not should_log_metrics:\n",
    "                acc = sess.run(accuracy, feed_dict=eval_dict)\n",
    "                loss = sess.run(cost, feed_dict=eval_dict)\n",
    "            print(\"Iter \" + str(step*batch_size) + \", Minibatch Loss= \" + \\\n",
    "                  \"{:.6f}\".format(loss) + \", Training Accuracy= \" + \\\n",
    "                  \"{:.5f}\".format(acc))\n",
    "            \n",
    "        if step % (display_step) == 0:\n",
    "            if should_log_metrics:\n",
    "                test_acc = sess.run(accuracy, feed_dict=test_dict)\n",
    "                test_loss = sess.run(cost, feed_dict=test_dict)\n",
    "                test_loss_log.write(line_start + str(test_loss) + '\\n')\n",
    "                test_loss_log.flush()\n",
    "                test_acc_log.write(line_start + str(test_acc) + '\\n')\n",
    "                test_acc_log.flush()        \n",
    "            if step % (5 * display_step) == 0:\n",
    "                if not should_log_metrics:\n",
    "                    test_acc = sess.run(accuracy, feed_dict=test_dict)\n",
    "                    test_loss = sess.run(cost, feed_dict=test_dict)\n",
    "                print(\"Validation Accuracy:\", test_acc)\n",
    "                print(\"Validation Loss:\", test_loss)\n",
    "                \n",
    "                if test_acc > best_test_acc:\n",
    "                    best_test_acc = test_acc\n",
    "                    # Plot ROC\n",
    "                    answers = sess.run(tf.nn.softmax(logits), feed_dict=test_dict)\n",
    "                    if label == 'good':\n",
    "                        fpr, tpr, _ = roc_curve(1 - np.argmax(test_label, axis=1), answers[:, 0])\n",
    "                    else:\n",
    "                        fpr, tpr, _ = roc_curve(np.argmax(test_label, axis=1), answers[:, 1])\n",
    "                    roc_auc = auc(fpr, tpr)\n",
    "                    plt.figure()\n",
    "                    lw = 2\n",
    "                    plt.plot(fpr, tpr, color='darkorange',\n",
    "                             lw=lw, label='ROC curve (area = %0.2f)' % roc_auc)\n",
    "                    plt.plot([0, 1], [0, 1], color='navy', lw=lw, linestyle='--')\n",
    "                    plt.xlim([0.0, 1.0])\n",
    "                    plt.ylim([0.0, 1.05])\n",
    "                    plt.xlabel('False Positive Rate')\n",
    "                    plt.ylabel('True Positive Rate')\n",
    "                    plt.title('ROC Curve')\n",
    "                    plt.legend(loc=\"lower right\")\n",
    "                    plt.savefig('plots/' + run_name + '_step_' + str(step) + '.png', dpi=300, bbox_inches='tight')\n",
    "                    plt.show()\n",
    "        \n",
    "        if step % save_every == 0:\n",
    "            save_path = saver.save(sess, save_fn)\n",
    "            print(\"Checkpoint saved at %s\" % save_path)\n",
    "        step += 1\n",
    "    print(\"Optimization Finished!\")\n",
    "    \n",
    "    save_path = saver.save(sess, save_fn)\n",
    "    print(\"Checkpoint saved at %s\" % save_path)\n",
    "\n",
    "    # Calculate accuracy\n",
    "    print(\"Final Testing Accuracy:\", sess.run(accuracy, feed_dict=test_dict))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Read out the final weight matrix and save it to disk"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "w = sess.run(weights)\n",
    "n_fc = len(fc_layer_sizes) + 1\n",
    "\n",
    "for i in range(n_fc):\n",
    "    f_name = 'weights/' + label + '_dim_' + str(i) + '.csv'\n",
    "    np.savetxt(f_name, w[i], delimiter=',')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}