associative_memory.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import numpy as np\n",
    "import random\n",
    "import matplotlib.pyplot as plt\n",
    "import importlib\n",
    "import matplotlib\n",
    "from torch.utils.data import Dataset, DataLoader\n",
    "from torchvision import datasets\n",
    "import sys\n",
    "import os\n",
    "import snn_lib.utilities as utilities\n",
    "from snn_lib.snn_layers import *"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Configurations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset_path = './dataset/associative_target.npy'\n",
    "\n",
    "checkpoint_base_name = \"associative_memory_checkpoint_\"\n",
    "checkpoint_base_path = './associative_memory_checkpoint/'\n",
    "\n",
    "torch.manual_seed(2)\n",
    "np.random.seed(0)\n",
    "\n",
    "length = 300\n",
    "input_size = 300\n",
    "batch_size = 32\n",
    "synapse_type = 'dual_exp'\n",
    "epoch = 100\n",
    "tau_m = 8\n",
    "tau_s = 2\n",
    "filter_tau_m = tau_m\n",
    "filter_tau_s = tau_s\n",
    "dropout_rate = 0.3\n",
    "\n",
    "modify_input_prob = 0.6\n",
    "remove_prob = 0.5\n",
    "remove_row_prob = 0.2\n",
    "remove_col_prob = 0.2\n",
    "remove_block_prob = 0.2\n",
    "mutate_prob = 0.7\n",
    "noise_prob = 0.005\n",
    "\n",
    "remove_block_h = 30\n",
    "remove_block_w = 40\n",
    "remove_block_h = 40\n",
    "remove_col_w = 30\n",
    "remove_row_h = 30\n",
    "\n",
    "optimizer_choice = 1\n",
    "scheduler_choice = 1\n",
    "\n",
    "optimizer_config = {0: (torch.optim.Adam,       0.001),\n",
    "                    1: (torch.optim.AdamW,      0.001),\n",
    "                    2: (torch.optim.SGD,        0.0001)}\n",
    "\n",
    "scheduler_config = {0: None,\n",
    "                    1: (torch.optim.lr_scheduler.MultiStepLR, [50,100, 150], 0.1),\n",
    "                    # order: milestones, gamma=0.1\n",
    "                    2: (torch.optim.lr_scheduler.CosineAnnealingWarmRestarts, 1000),\n",
    "                    # T_0\n",
    "                    3: (torch.optim.lr_scheduler.CyclicLR, 0.001, 0.01, 200)\n",
    "                    }"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Prepare dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# %% utility functions\n",
    "def add_noise_spike(spike_array,probability = 0.003):\n",
    "    '''\n",
    "    :param spike_array: 2d array [spike train num, length]\n",
    "    :param probability:\n",
    "    :return:\n",
    "    '''\n",
    "    noise_mat = np.random.rand(*spike_array.shape).astype(np.float32)\n",
    "\n",
    "    noise_mat[np.where(noise_mat > 1-probability)] = 1\n",
    "    noise_mat[np.where(noise_mat <= 1 - probability)] = 0\n",
    "\n",
    "    new_arr = spike_array + noise_mat\n",
    "\n",
    "    new_arr[np.where(new_arr > 1)] = 1\n",
    "\n",
    "    return  new_arr\n",
    "\n",
    "def remove_row(spike_array,remove_width, position = None):\n",
    "    '''\n",
    "    remove a few rows in the spike mat (set rows to 0)\n",
    "    :param spike_array: 2d array [spike train num, length]\n",
    "    :param remove_width: How many rows to remove\n",
    "    :param position: spike will be removed from row position to position+remove_width\n",
    "    :return:\n",
    "    '''\n",
    "\n",
    "    h,w = spike_array.shape\n",
    "    upper_row = np.random.randint(0,h-remove_width)\n",
    "\n",
    "    if position != None:\n",
    "        upper_row = position\n",
    "\n",
    "    new_arr = spike_array\n",
    "    new_arr[upper_row:upper_row+remove_width,:] = 0\n",
    "\n",
    "    return new_arr\n",
    "\n",
    "def remove_col(spike_array,remove_width):\n",
    "    '''\n",
    "    remove a few columns in spike mat (set columns to 0)\n",
    "    :param spike_array: 2d array [spike train num, length]\n",
    "    :param width:\n",
    "    :return:\n",
    "    '''\n",
    "\n",
    "    h,w = spike_array.shape\n",
    "    left_col = np.random.randint(0,w-remove_width)\n",
    "\n",
    "    new_arr = spike_array\n",
    "    new_arr[:,left_col:left_col+remove_width] = 0\n",
    "\n",
    "    return new_arr\n",
    "\n",
    "def remove_block(spike_array,remove_hight, remove_width):\n",
    "    '''\n",
    "    set a block region in spike mat to 0\n",
    "    :param spike_array: 2d array [spike train num, length]\n",
    "    :param width:\n",
    "    :return:\n",
    "    '''\n",
    "\n",
    "    h,w = spike_array.shape\n",
    "    top_left_row = np.random.randint(0,w-remove_hight)\n",
    "    top_left_col = np.random.randint(0,w-remove_width)\n",
    "\n",
    "    new_arr = spike_array\n",
    "    new_arr[top_left_row:top_left_row+remove_hight,top_left_col:top_left_col+remove_width] = 0\n",
    "\n",
    "    return new_arr\n",
    "\n",
    "class PatternDataset(Dataset):\n",
    "    \"\"\"random pattern dataset\"\"\"\n",
    "\n",
    "    def __init__(self, input_pattern, target, filtered_target,length):\n",
    "        '''\n",
    "\n",
    "        :param input_pattern: [number, hight, width/time]\n",
    "        :param label_cat:\n",
    "        :param target:\n",
    "        :param length:\n",
    "        '''\n",
    "        self.input_pattern = input_pattern\n",
    "        self.target = target\n",
    "        self.filtered_target = filtered_target\n",
    "        self.length = length\n",
    "\n",
    "\n",
    "        self.modify_input_prob = modify_input_prob\n",
    "        self.remove_prob = remove_prob\n",
    "        self.remove_row_prob = remove_row_prob\n",
    "        self.remove_col_prob = remove_col_prob\n",
    "        self.noise_prob = noise_prob\n",
    "        self.mutate_prob = mutate_prob\n",
    "\n",
    "        self.remove_block_h = remove_block_h\n",
    "        self.remove_block_w = remove_block_w\n",
    "        self.remove_block_h =remove_block_h\n",
    "        self.remove_col_w = remove_col_w\n",
    "        self.remove_row_h = remove_row_h\n",
    "\n",
    "    def __len__(self):\n",
    "        #input only has 10 samples, so increase length by 100\n",
    "        #each sample is a class\n",
    "        return len(self.input_pattern)*100\n",
    "\n",
    "    def __getitem__(self, idx):\n",
    "        if torch.is_tensor(idx):\n",
    "            idx = idx.tolist()\n",
    "\n",
    "        #input data only has 10 sample\n",
    "        #by mod get the actual idx, which is also the class\n",
    "        real_idx = idx % 10\n",
    "\n",
    "        label = real_idx\n",
    "        input_pattern = self.input_pattern[real_idx].copy()\n",
    "        output_target = self.target[real_idx]\n",
    "        filtered_target = self.filtered_target[real_idx]\n",
    "\n",
    "        if np.random.rand() > self.modify_input_prob:\n",
    "            #modify input data, remove, delete, add noise, mutate\n",
    "            if np.random.rand() < self.mutate_prob:\n",
    "                input_pattern = utilities.mutate_spike_pattern(input_pattern, 0, 1)\n",
    "\n",
    "            #random select remove row, col, or block\n",
    "            choice = np.random.randint(2)\n",
    "\n",
    "            if np.random.rand() < self.remove_prob:\n",
    "                if choice == 0:\n",
    "                    input_pattern = remove_block(input_pattern,remove_block_w, remove_block_h)\n",
    "                elif choice == 1:\n",
    "                    input_pattern = remove_row(input_pattern, remove_block_h)\n",
    "\n",
    "        input_pattern = add_noise_spike(input_pattern, self.noise_prob)\n",
    "\n",
    "        return input_pattern, label, output_target, filtered_target"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Build the network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "class mysnn(torch.nn.Module):\n",
    "    def __init__(self):\n",
    "        super().__init__()\n",
    "\n",
    "        self.length = length\n",
    "        self.batch_size = batch_size\n",
    "\n",
    "        self.train_coefficients = True\n",
    "        self.train_decay_v = False\n",
    "        self.train_v0 = False\n",
    "        self.train_reset_v = False\n",
    "        self.train_reset_decay = False\n",
    "        self.train_threshold = False\n",
    "        self.train_bias = True\n",
    "        self.membrane_filter = False\n",
    "\n",
    "        self.axon1 = dual_exp_iir_layer((input_size,), self.length, \n",
    "        self.batch_size, tau_m, tau_s, self.train_coefficients)\n",
    "\n",
    "        self.dense1 = neuron_layer(input_size, 500, self.length, self.batch_size, \n",
    "        tau_m, self.train_bias, self.membrane_filter)\n",
    "\n",
    "        self.axon2 = dual_exp_iir_layer((500,), self.length, self.batch_size, tau_m, tau_s, \n",
    "        self.train_coefficients)\n",
    "        self.dense2 = neuron_layer(500, 200, self.length, self.batch_size, tau_m,  \n",
    "        self.train_bias, self.membrane_filter)\n",
    "                    \n",
    "        self.axon3 = dual_exp_iir_layer((200,), self.length, self.batch_size, tau_m, tau_s, \n",
    "        self.train_coefficients)\n",
    "\n",
    "        self.dense3 = neuron_layer(200, 500, self.length, self.batch_size, tau_m, \n",
    "        self.train_bias, self.membrane_filter)\n",
    "        \n",
    "        self.axon4 = dual_exp_iir_layer((500,), self.length, self.batch_size, tau_m, tau_s, \n",
    "        self.train_coefficients)\n",
    "\n",
    "        self.dense4 = neuron_layer(500, 300, self.length, self.batch_size, tau_m, \n",
    "        self.train_bias, self.membrane_filter)\n",
    "\n",
    "        self.output_filter = filter_layer(300, self.length, self.batch_size, filter_tau_m, filter_tau_s)                                \n",
    "    def forward(self, inputs):\n",
    "        \"\"\"\n",
    "        :param inputs: [batch, input_size, t]\n",
    "        :return:\n",
    "        \"\"\"\n",
    "        axon1_states = self.axon1.create_init_states()\n",
    "        dense1_states = self.dense1.create_init_states()\n",
    "\n",
    "        axon2_states = self.axon2.create_init_states()\n",
    "        dense2_states = self.dense2.create_init_states()\n",
    "\n",
    "        axon3_states = self.axon3.create_init_states()\n",
    "        dense3_states = self.dense3.create_init_states()\n",
    "\n",
    "        axon4_states = self.axon4.create_init_states()\n",
    "        dense4_states = self.dense4.create_init_states()\n",
    "\n",
    "        axon1_out, axon1_states = self.axon1(inputs, axon1_states)\n",
    "        spike_dense1, dense1_states = self.dense1(axon1_out, dense1_states)\n",
    "\n",
    "        axon2_out, axon2_states = self.axon2(spike_dense1, axon2_states)\n",
    "        spike_dense2, dense2_states = self.dense2(axon2_out, dense2_states)\n",
    "\n",
    "        axon3_out, axon3_states = self.axon3(spike_dense2, axon3_states)\n",
    "        spike_dense3, dense3_states = self.dense3(axon3_out, dense3_states)\n",
    "\n",
    "        axon4_out, axon4_states = self.axon4(spike_dense3, axon4_states)\n",
    "        spike_dense4, dense4_states = self.dense4(axon4_out, dense4_states)\n",
    "\n",
    "        filtered_output = self.output_filter(spike_dense4)\n",
    "\n",
    "        return spike_dense4, filtered_output"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Train and test function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train(model, optimizer, scheduler, data_loader):\n",
    "    '''\n",
    "\n",
    "    '''\n",
    "\n",
    "    model.train()\n",
    "\n",
    "    eval_image_number = 0\n",
    "    correct_total = 0\n",
    "    wrong_total = 0\n",
    "\n",
    "    criterion = torch.nn.MSELoss()\n",
    "\n",
    "    for i_batch, sample_batched in enumerate(data_loader):\n",
    "\n",
    "        x_train = sample_batched[0].to(device)\n",
    "        label = sample_batched[1].to(device)\n",
    "        target_pattern = sample_batched[2].to(device)\n",
    "        filtered_target_pattern = sample_batched[3].to(device)\n",
    "\n",
    "        out_spike, filtered_out_spike = model(x_train)\n",
    "\n",
    "        model.zero_grad()\n",
    "\n",
    "        loss = criterion(filtered_out_spike, filtered_target_pattern)\n",
    "\n",
    "        print('train loss: {}'.format(loss))\n",
    "\n",
    "        loss.backward()\n",
    "\n",
    "        optimizer.step()\n",
    "\n",
    "        if isinstance(scheduler, torch.optim.lr_scheduler.CosineAnnealingWarmRestarts) or \\\n",
    "                isinstance(scheduler, torch.optim.lr_scheduler.CyclicLR):\n",
    "            # isinstance(scheduler, torch.optim.lr_scheduler.OneCycleLR):\n",
    "            scheduler.step()\n",
    "\n",
    "    if isinstance(scheduler, torch.optim.lr_scheduler.MultiStepLR):\n",
    "        scheduler.step()\n",
    "\n",
    "    return loss\n",
    "\n",
    "def test(model, data_loader):\n",
    "\n",
    "    model.eval()\n",
    "\n",
    "    loss_list = []\n",
    "\n",
    "    criterion = torch.nn.MSELoss()\n",
    "\n",
    "    for i_batch, sample_batched in enumerate(data_loader):\n",
    "\n",
    "        criterion = torch.nn.MSELoss()\n",
    "        x_test = sample_batched[0].to(device)\n",
    "        label = sample_batched[1].to(device)\n",
    "        target_pattern = sample_batched[2].to(device)\n",
    "        filtered_target_pattern = sample_batched[3].to(device)\n",
    "\n",
    "        out_spike, filtered_out_spike = model(x_test)\n",
    "\n",
    "        loss = criterion(filtered_out_spike, filtered_target_pattern)\n",
    "\n",
    "        print('Test loss: {}'.format(loss))\n",
    "\n",
    "        loss_list.append(loss.cpu().detach().numpy())\n",
    "\n",
    "    #calculate loss of this epoch\n",
    "    loss_array = np.stack(loss_list)\n",
    "    average_loss = loss_array.mean()\n",
    "\n",
    "    return average_loss"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Train the network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "if torch.cuda.is_available():\n",
    "    device = torch.device('cuda:2')\n",
    "else:\n",
    "    device = torch.device('cpu')\n",
    "\n",
    "original_pattern = np.load(dataset_path).astype(np.float32)\n",
    "\n",
    "delayed_target = np.zeros(original_pattern.shape, dtype=np.float32)\n",
    "\n",
    "delayed_target[:, :, 4:] = original_pattern[:, :, :-4]\n",
    "\n",
    "filtered_target = []\n",
    "for target_idx, delayed_target_pattern in enumerate(delayed_target):\n",
    "    filtered_target.append(utilities.filter_spike_multiple(delayed_target_pattern, filter_type='dual_exp',\n",
    "                                                            tau_m=filter_tau_m, tau_s=filter_tau_s))\n",
    "# shape [pattern idx, spike train idx, time]\n",
    "filtered_target = np.array(filtered_target)\n",
    "\n",
    "snn = mysnn().to(device)\n",
    "\n",
    "params = list(snn.parameters())\n",
    "\n",
    "optimizer_class = optimizer_config[optimizer_choice][0]\n",
    "learning_rate = optimizer_config[optimizer_choice][1]\n",
    "optimizer = optimizer_class(params, learning_rate)\n",
    "\n",
    "scheduler = None\n",
    "if scheduler_choice == 0:\n",
    "    scheduler = None\n",
    "elif scheduler_choice == 1:\n",
    "    scheduler = torch.optim.lr_scheduler.MultiStepLR(optimizer, \n",
    "    scheduler_config[scheduler_choice][1], \n",
    "    scheduler_config[scheduler_choice][2])\n",
    "elif scheduler_choice == 2:\n",
    "    scheduler = torch.optim.lr_scheduler.CosineAnnealingWarmRestarts(optimizer, scheduler_config[scheduler_choice][1])\n",
    "elif scheduler_choice == 3:\n",
    "    scheduler = torch.optim.lr_scheduler.CyclicLR(optimizer, \n",
    "    scheduler_config[scheduler_choice][1], \n",
    "    scheduler_config[scheduler_choice][2], \n",
    "    scheduler_config[scheduler_choice][3])\n",
    "elif scheduler_choice == 4:\n",
    "    scheduler = torch.optim.lr_scheduler.OneCycleLR(optimizer, \n",
    "    scheduler_config[scheduler_choice][1], \n",
    "    scheduler_config[scheduler_choice][2], \n",
    "    scheduler_config[scheduler_choice][3])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train and test dataloader"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_data = PatternDataset(original_pattern, delayed_target, filtered_target, length)\n",
    "train_dataloader = DataLoader(train_data, batch_size=batch_size, shuffle=True, drop_last=True)\n",
    "\n",
    "dev_data = PatternDataset(original_pattern, delayed_target, filtered_target, length)\n",
    "dev_dataloader = DataLoader(dev_data, batch_size=batch_size, shuffle=True, drop_last=True)\n",
    "\n",
    "test_data = PatternDataset(original_pattern, delayed_target, filtered_target, length)\n",
    "test_dataloader = DataLoader(test_data, batch_size=batch_size, shuffle=True, drop_last=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1000, 1000, 1000)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(train_data), len(dev_data), len(test_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "scrolled": true,
    "tags": [
     "outputPrepend"
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train epoch: 0\n",
      "train loss: 0.1468135267496109\n",
      "train loss: 0.145709827542305\n",
      "train loss: 0.1534099280834198\n",
      "train loss: 0.14330919086933136\n",
      "train loss: 0.13321346044540405\n",
      "train loss: 0.1428721398115158\n",
      "train loss: 0.1363387405872345\n",
      "train loss: 0.136715367436409\n",
      "train loss: 0.1453908234834671\n",
      "train loss: 0.13600625097751617\n",
      "train loss: 0.13528677821159363\n",
      "train loss: 0.12958715856075287\n",
      "train loss: 0.1491219699382782\n",
      "train loss: 0.14317646622657776\n",
      "train loss: 0.14920742809772491\n",
      "train loss: 0.14152570068836212\n",
      "train loss: 0.13404697179794312\n",
      "train loss: 0.14115890860557556\n",
      "train loss: 0.14329023659229279\n",
      "train loss: 0.15496914088726044\n",
      "train loss: 0.30767083168029785\n",
      "train loss: 0.16881853342056274\n",
      "train loss: 0.14469237625598907\n",
      "train loss: 0.14680810272693634\n",
      "train loss: 0.146453857421875\n",
      "train loss: 0.16706258058547974\n",
      "train loss: 0.19828636944293976\n",
      "train loss: 0.1432013213634491\n",
      "train loss: 0.16793495416641235\n",
      "train loss: 0.16112378239631653\n",
      "train loss: 0.1478549689054489\n",
      "Test loss: 0.13794079422950745\n",
      "Test loss: 0.14674162864685059\n",
      "Test loss: 0.1310742050409317\n",
      "Test loss: 0.1431283801794052\n",
      "Test loss: 0.1325708031654358\n",
      "Test loss: 0.1467253565788269\n",
      "Test loss: 0.13427571952342987\n",
      "Test loss: 0.13810378313064575\n",
      "Test loss: 0.15153390169143677\n",
      "Test loss: 0.15147870779037476\n",
      "Test loss: 0.13390564918518066\n",
      "Test loss: 0.14046384394168854\n",
      "Test loss: 0.1467360407114029\n",
      "Test loss: 0.14990867674350739\n",
      "Test loss: 0.141195148229599\n",
      "Test loss: 0.14170509576797485\n",
      "Test loss: 0.14731375873088837\n",
      "Test loss: 0.14202135801315308\n",
      "Test loss: 0.12783075869083405\n",
      "Test loss: 0.13822324573993683\n",
      "Test loss: 0.1341988444328308\n",
      "Test loss: 0.1298876702785492\n",
      "Test loss: 0.1392899602651596\n",
      "Test loss: 0.13932153582572937\n",
      "Test loss: 0.13996657729148865\n",
      "Test loss: 0.14472036063671112\n",
      "Test loss: 0.1416633278131485\n",
      "Test loss: 0.1456354409456253\n",
      "Test loss: 0.14131158590316772\n",
      "Test loss: 0.14450867474079132\n",
      "Test loss: 0.14318051934242249\n",
      "Test: 0, loss: 0.1408568024635315\n",
      "Train epoch: 1\n",
      "train loss: 0.14042507112026215\n",
      "train loss: 0.14489448070526123\n",
      "train loss: 0.19407689571380615\n",
      "train loss: 0.18243637681007385\n",
      "train loss: 0.1564803123474121\n",
      "train loss: 0.1394781470298767\n",
      "train loss: 0.14113059639930725\n",
      "train loss: 0.17090968787670135\n",
      "train loss: 0.18955673277378082\n",
      "train loss: 0.1438138633966446\n",
      "train loss: 0.1420627236366272\n",
      "train loss: 0.1392066478729248\n",
      "train loss: 0.1686839759349823\n",
      "train loss: 0.1815432757139206\n",
      "train loss: 0.14354783296585083\n",
      "train loss: 0.12907761335372925\n",
      "train loss: 0.13960419595241547\n",
      "train loss: 0.16438589990139008\n",
      "train loss: 0.17551277577877045\n",
      "train loss: 0.13606306910514832\n",
      "train loss: 0.13269656896591187\n",
      "train loss: 0.1376642882823944\n",
      "train loss: 0.16395796835422516\n",
      "train loss: 0.1470007449388504\n",
      "train loss: 0.13177590072155\n",
      "train loss: 0.1372096985578537\n",
      "train loss: 0.16576436161994934\n",
      "train loss: 0.153227299451828\n",
      "train loss: 0.12967248260974884\n",
      "train loss: 0.13273459672927856\n",
      "train loss: 0.16171474754810333\n",
      "Test loss: 0.15235567092895508\n",
      "Test loss: 0.15098188817501068\n",
      "Test loss: 0.15910671651363373\n",
      "Test loss: 0.15946635603904724\n",
      "Test loss: 0.15648239850997925\n",
      "Test loss: 0.15304242074489594\n",
      "Test loss: 0.15279817581176758\n",
      "Test loss: 0.15484245121479034\n",
      "Test loss: 0.15624748170375824\n",
      "Test loss: 0.15787242352962494\n",
      "Test loss: 0.15546797215938568\n",
      "Test loss: 0.15843738615512848\n",
      "Test loss: 0.1555764228105545\n",
      "Test loss: 0.15515895187854767\n",
      "Test loss: 0.15701977908611298\n",
      "Test loss: 0.15942059457302094\n",
      "Test loss: 0.15804243087768555\n",
      "Test loss: 0.16350765526294708\n",
      "Test loss: 0.16043414175510406\n",
      "Test loss: 0.15536139905452728\n",
      "Test loss: 0.15139555931091309\n",
      "Test loss: 0.15455521643161774\n",
      "Test loss: 0.14928731322288513\n",
      "Test loss: 0.15849921107292175\n",
      "Test loss: 0.1497231423854828\n",
      "Test loss: 0.151272252202034\n",
      "Test loss: 0.15306293964385986\n",
      "Test loss: 0.16075386106967926\n",
      "Test loss: 0.15772300958633423\n",
      "Test loss: 0.15797865390777588\n",
      "Test loss: 0.15568721294403076\n",
      "Test: 1, loss: 0.15585678815841675\n",
      "Train epoch: 2\n",
      "train loss: 0.15959393978118896\n",
      "train loss: 0.12916214764118195\n",
      "train loss: 0.134116992354393\n",
      "train loss: 0.16321876645088196\n",
      "train loss: 0.1480838507413864\n",
      "train loss: 0.1245204359292984\n",
      "train loss: 0.12266534566879272\n",
      "train loss: 0.15309210121631622\n",
      "train loss: 0.14618629217147827\n",
      "train loss: 0.12298420071601868\n",
      "train loss: 0.12363710254430771\n",
      "train loss: 0.13570648431777954\n",
      "train loss: 0.14091455936431885\n",
      "train loss: 0.12619881331920624\n",
      "train loss: 0.1146613284945488\n",
      "train loss: 0.12631581723690033\n",
      "train loss: 0.12633150815963745\n",
      "train loss: 0.12201200425624847\n",
      "train loss: 0.11724915355443954\n",
      "train loss: 0.1202549934387207\n",
      "train loss: 0.12933699786663055\n",
      "train loss: 0.11778686940670013\n",
      "train loss: 0.11779893934726715\n",
      "train loss: 0.1148504763841629\n",
      "train loss: 0.12219534814357758\n",
      "train loss: 0.12319870293140411\n",
      "train loss: 0.11365855485200882\n",
      "train loss: 0.11866150051355362\n",
      "train loss: 0.11712054163217545\n",
      "train loss: 0.11352239549160004\n",
      "train loss: 0.11104618012905121\n",
      "Test loss: 0.11533299088478088\n",
      "Test loss: 0.11575805395841599\n",
      "Test loss: 0.1114724650979042\n",
      "Test loss: 0.1153465285897255\n",
      "Test loss: 0.11483938992023468\n",
      "Test loss: 0.1162417083978653\n",
      "Test loss: 0.10855819284915924\n",
      "Test loss: 0.11936358362436295\n",
      "Test loss: 0.11094976216554642\n",
      "Test loss: 0.11620401591062546\n",
      "Test loss: 0.11499509960412979\n",
      "Test loss: 0.11595892161130905\n",
      "Test loss: 0.11072390526533127\n",
      "Test loss: 0.11059609800577164\n",
      "Test loss: 0.10916657745838165\n",
      "Test loss: 0.11443229019641876\n",
      "Test loss: 0.10792821645736694\n",
      "Test loss: 0.11121489107608795\n",
      "Test loss: 0.11384472995996475\n",
      "Test loss: 0.1187763437628746\n",
      "Test loss: 0.10383396595716476\n",
      "Test loss: 0.11550022661685944\n",
      "Test loss: 0.11307428777217865\n",
      "Test loss: 0.11124620586633682\n",
      "Test loss: 0.11565818637609482\n",
      "Test loss: 0.11216136068105698\n",
      "Test loss: 0.11427854001522064\n",
      "Test loss: 0.11248796433210373\n",
      "Test loss: 0.11170988529920578\n",
      "Test loss: 0.11391066014766693\n",
      "Test loss: 0.11363708972930908\n",
      "Test: 2, loss: 0.11320008337497711\n",
      "Train epoch: 3\n",
      "train loss: 0.11430232226848602\n",
      "train loss: 0.11290209740400314\n",
      "train loss: 0.10926873236894608\n",
      "train loss: 0.10323307663202286\n",
      "train loss: 0.11915429681539536\n",
      "train loss: 0.10159174352884293\n",
      "train loss: 0.10379751771688461\n",
      "train loss: 0.11161202192306519\n",
      "train loss: 0.10806795954704285\n",
      "train loss: 0.1031109094619751\n",
      "train loss: 0.10706117004156113\n",
      "train loss: 0.10650855302810669\n",
      "train loss: 0.10067109763622284\n",
      "train loss: 0.10173844546079636\n",
      "train loss: 0.09798508882522583\n",
      "train loss: 0.10121341049671173\n",
      "train loss: 0.09802017360925674\n",
      "train loss: 0.09855736047029495\n",
      "train loss: 0.10025855153799057\n",
      "train loss: 0.09145382046699524\n",
      "train loss: 0.09971518069505692\n",
      "train loss: 0.09854088723659515\n",
      "train loss: 0.08905395865440369\n",
      "train loss: 0.09042087942361832\n",
      "train loss: 0.09680898487567902\n",
      "train loss: 0.09099855273962021\n",
      "train loss: 0.0897250697016716\n",
      "train loss: 0.09038862586021423\n",
      "train loss: 0.08606182038784027\n",
      "train loss: 0.08921223878860474\n",
      "train loss: 0.08963355422019958\n",
      "Test loss: 0.09034591168165207\n",
      "Test loss: 0.08812026679515839\n",
      "Test loss: 0.08802276849746704\n",
      "Test loss: 0.08352317661046982\n",
      "Test loss: 0.08821728825569153\n",
      "Test loss: 0.08383269608020782\n",
      "Test loss: 0.08228322863578796\n",
      "Test loss: 0.08554908633232117\n",
      "Test loss: 0.0828002467751503\n",
      "Test loss: 0.086248978972435\n",
      "Test loss: 0.08562435209751129\n",
      "Test loss: 0.08493601530790329\n",
      "Test loss: 0.08351762592792511\n",
      "Test loss: 0.08095888793468475\n",
      "Test loss: 0.0865304172039032\n",
      "Test loss: 0.08658313006162643\n",
      "Test loss: 0.08376666903495789\n",
      "Test loss: 0.08677087724208832\n",
      "Test loss: 0.08632315695285797\n",
      "Test loss: 0.0798327699303627\n",
      "Test loss: 0.08611372113227844\n",
      "Test loss: 0.08498803526163101\n",
      "Test loss: 0.08504767715930939\n",
      "Test loss: 0.08581119030714035\n",
      "Test loss: 0.08438744395971298\n",
      "Test loss: 0.08534839749336243\n",
      "Test loss: 0.08532139658927917\n",
      "Test loss: 0.0839843675494194\n",
      "Test loss: 0.08744878321886063\n",
      "Test loss: 0.08605712652206421\n",
      "Test loss: 0.08475429564714432\n",
      "Test: 3, loss: 0.08525968343019485\n",
      "Train epoch: 4\n",
      "train loss: 0.08291464298963547\n",
      "train loss: 0.08271461725234985\n",
      "train loss: 0.08445174992084503\n",
      "train loss: 0.08604929596185684\n",
      "train loss: 0.07944045960903168\n",
      "train loss: 0.08728509396314621\n",
      "train loss: 0.0855163186788559\n",
      "train loss: 0.07964307814836502\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train loss: 0.08257048577070236\n",
      "train loss: 0.08441125601530075\n",
      "train loss: 0.07798291742801666\n",
      "train loss: 0.08386795967817307\n",
      "train loss: 0.0874362662434578\n",
      "train loss: 0.07892649620771408\n",
      "train loss: 0.0797116681933403\n",
      "train loss: 0.09007661789655685\n",
      "train loss: 0.07313978672027588\n",
      "train loss: 0.08107603341341019\n",
      "train loss: 0.08264320343732834\n",
      "train loss: 0.0813702791929245\n",
      "train loss: 0.07487271726131439\n",
      "train loss: 0.08103638142347336\n",
      "train loss: 0.0787733867764473\n",
      "train loss: 0.0807037353515625\n",
      "train loss: 0.08071622252464294\n",
      "train loss: 0.07819138467311859\n",
      "train loss: 0.07786504179239273\n",
      "train loss: 0.07740273326635361\n",
      "train loss: 0.07670041918754578\n",
      "train loss: 0.07637523114681244\n",
      "train loss: 0.07819584757089615\n",
      "Test loss: 0.07446569204330444\n",
      "Test loss: 0.08023999631404877\n",
      "Test loss: 0.0765635073184967\n",
      "Test loss: 0.07246656715869904\n",
      "Test loss: 0.07396260648965836\n",
      "Test loss: 0.0730392336845398\n",
      "Test loss: 0.07247509807348251\n",
      "Test loss: 0.07205464690923691\n",
      "Test loss: 0.07418690621852875\n",
      "Test loss: 0.07076946645975113\n",
      "Test loss: 0.0767989382147789\n",
      "Test loss: 0.0747438594698906\n",
      "Test loss: 0.07515010982751846\n",
      "Test loss: 0.07514140754938126\n",
      "Test loss: 0.0724887028336525\n",
      "Test loss: 0.07513846457004547\n",
      "Test loss: 0.07311555743217468\n",
      "Test loss: 0.07444464415311813\n",
      "Test loss: 0.07511360943317413\n",
      "Test loss: 0.0735054463148117\n",
      "Test loss: 0.07622282952070236\n",
      "Test loss: 0.07392988353967667\n",
      "Test loss: 0.07882821559906006\n",
      "Test loss: 0.07645244896411896\n",
      "Test loss: 0.07572287321090698\n",
      "Test loss: 0.0744422972202301\n",
      "Test loss: 0.0706382542848587\n",
      "Test loss: 0.07608947902917862\n",
      "Test loss: 0.07310539484024048\n",
      "Test loss: 0.07168871909379959\n",
      "Test loss: 0.07228228449821472\n",
      "Test: 4, loss: 0.07436345517635345\n",
      "Train epoch: 5\n",
      "train loss: 0.07763019949197769\n",
      "train loss: 0.07774481177330017\n",
      "train loss: 0.07254061102867126\n",
      "train loss: 0.07420936226844788\n",
      "train loss: 0.07970980554819107\n",
      "train loss: 0.07683239877223969\n",
      "train loss: 0.07447531074285507\n",
      "train loss: 0.0735790953040123\n",
      "train loss: 0.07446711510419846\n",
      "train loss: 0.07142426818609238\n",
      "train loss: 0.07100456207990646\n",
      "train loss: 0.06996911764144897\n",
      "train loss: 0.07087239623069763\n",
      "train loss: 0.07205182313919067\n",
      "train loss: 0.07121291011571884\n",
      "train loss: 0.06948182731866837\n",
      "train loss: 0.0717005506157875\n",
      "train loss: 0.06850448995828629\n",
      "train loss: 0.06866936385631561\n",
      "train loss: 0.07257549464702606\n",
      "train loss: 0.07192539423704147\n",
      "train loss: 0.0663793534040451\n",
      "train loss: 0.07183623313903809\n",
      "train loss: 0.0691160187125206\n",
      "train loss: 0.06816338002681732\n",
      "train loss: 0.06902347505092621\n",
      "train loss: 0.06806302070617676\n",
      "train loss: 0.06736926734447479\n",
      "train loss: 0.06694431602954865\n",
      "train loss: 0.06692081689834595\n",
      "train loss: 0.06694917380809784\n",
      "Test loss: 0.06393155455589294\n",
      "Test loss: 0.06635057926177979\n",
      "Test loss: 0.0673099160194397\n",
      "Test loss: 0.06630712747573853\n",
      "Test loss: 0.06548768281936646\n",
      "Test loss: 0.0671251118183136\n",
      "Test loss: 0.06516966968774796\n",
      "Test loss: 0.06536715477705002\n",
      "Test loss: 0.06729278713464737\n",
      "Test loss: 0.06821007281541824\n",
      "Test loss: 0.06702920794487\n",
      "Test loss: 0.06609941273927689\n",
      "Test loss: 0.06246884912252426\n",
      "Test loss: 0.0665631964802742\n",
      "Test loss: 0.06597047299146652\n",
      "Test loss: 0.06698759645223618\n",
      "Test loss: 0.06573105603456497\n",
      "Test loss: 0.06443488597869873\n",
      "Test loss: 0.06380125135183334\n",
      "Test loss: 0.06497782468795776\n",
      "Test loss: 0.06655468046665192\n",
      "Test loss: 0.06454543769359589\n",
      "Test loss: 0.06628235429525375\n",
      "Test loss: 0.06701318919658661\n",
      "Test loss: 0.06552314013242722\n",
      "Test loss: 0.06804478913545609\n",
      "Test loss: 0.06617531180381775\n",
      "Test loss: 0.06676461547613144\n",
      "Test loss: 0.06659263372421265\n",
      "Test loss: 0.06647446751594543\n",
      "Test loss: 0.06428076326847076\n",
      "Test: 5, loss: 0.06596344709396362\n",
      "Train epoch: 6\n",
      "train loss: 0.06575864553451538\n",
      "train loss: 0.0707879438996315\n",
      "train loss: 0.06736431270837784\n",
      "train loss: 0.06360305100679398\n",
      "train loss: 0.06824181973934174\n",
      "train loss: 0.0674077495932579\n",
      "train loss: 0.06461580842733383\n",
      "train loss: 0.06381942331790924\n",
      "train loss: 0.06573419272899628\n",
      "train loss: 0.0660032108426094\n",
      "train loss: 0.06490261852741241\n",
      "train loss: 0.06209065392613411\n",
      "train loss: 0.06331125646829605\n",
      "train loss: 0.06831251084804535\n",
      "train loss: 0.060505349189043045\n",
      "train loss: 0.06301087886095047\n",
      "train loss: 0.06712145358324051\n",
      "train loss: 0.06469197571277618\n",
      "train loss: 0.0609976127743721\n",
      "train loss: 0.06216121092438698\n",
      "train loss: 0.06494349241256714\n",
      "train loss: 0.0648021250963211\n",
      "train loss: 0.06137626990675926\n",
      "train loss: 0.06157795339822769\n",
      "train loss: 0.06549812108278275\n",
      "train loss: 0.061768293380737305\n",
      "train loss: 0.06091979891061783\n",
      "train loss: 0.06016894429922104\n",
      "train loss: 0.062442079186439514\n",
      "train loss: 0.06109887734055519\n",
      "train loss: 0.06109721213579178\n",
      "Test loss: 0.06112951040267944\n",
      "Test loss: 0.060318946838378906\n",
      "Test loss: 0.05637334659695625\n",
      "Test loss: 0.06203083693981171\n",
      "Test loss: 0.05910573899745941\n",
      "Test loss: 0.0592653751373291\n",
      "Test loss: 0.06009278446435928\n",
      "Test loss: 0.06074298173189163\n",
      "Test loss: 0.06450372189283371\n",
      "Test loss: 0.06139110401272774\n",
      "Test loss: 0.059522416442632675\n",
      "Test loss: 0.05740896239876747\n",
      "Test loss: 0.05930095538496971\n",
      "Test loss: 0.06225655600428581\n",
      "Test loss: 0.06335365772247314\n",
      "Test loss: 0.05991083011031151\n",
      "Test loss: 0.06116204336285591\n",
      "Test loss: 0.06217912212014198\n",
      "Test loss: 0.06150764971971512\n",
      "Test loss: 0.05862889438867569\n",
      "Test loss: 0.060294076800346375\n",
      "Test loss: 0.05703297629952431\n",
      "Test loss: 0.06088857352733612\n",
      "Test loss: 0.05934910476207733\n",
      "Test loss: 0.06153617799282074\n",
      "Test loss: 0.05886567011475563\n",
      "Test loss: 0.06182502210140228\n",
      "Test loss: 0.0630522146821022\n",
      "Test loss: 0.06304311007261276\n",
      "Test loss: 0.06050226464867592\n",
      "Test loss: 0.060003612190485\n",
      "Test: 6, loss: 0.06053478643298149\n",
      "Train epoch: 7\n",
      "train loss: 0.05876724421977997\n",
      "train loss: 0.0634678304195404\n",
      "train loss: 0.05793677270412445\n",
      "train loss: 0.061578910797834396\n",
      "train loss: 0.058479417115449905\n",
      "train loss: 0.059583041816949844\n",
      "train loss: 0.060894791036844254\n",
      "train loss: 0.06085248664021492\n",
      "train loss: 0.05821602791547775\n",
      "train loss: 0.060029760003089905\n",
      "train loss: 0.060436658561229706\n",
      "train loss: 0.0573900081217289\n",
      "train loss: 0.06021267920732498\n",
      "train loss: 0.062177084386348724\n",
      "train loss: 0.05862472951412201\n",
      "train loss: 0.05870125815272331\n",
      "train loss: 0.05761793628334999\n",
      "train loss: 0.06082254275679588\n",
      "train loss: 0.05980346351861954\n",
      "train loss: 0.05785573646426201\n",
      "train loss: 0.055954113602638245\n",
      "train loss: 0.05728675797581673\n",
      "train loss: 0.059091489762067795\n",
      "train loss: 0.0603603757917881\n",
      "train loss: 0.058487895876169205\n",
      "train loss: 0.05751765891909599\n",
      "train loss: 0.05517324432730675\n",
      "train loss: 0.05908464267849922\n",
      "train loss: 0.057060644030570984\n",
      "train loss: 0.05853935703635216\n",
      "train loss: 0.0563228540122509\n",
      "Test loss: 0.055744823068380356\n",
      "Test loss: 0.05587344989180565\n",
      "Test loss: 0.05412033572793007\n",
      "Test loss: 0.05465487390756607\n",
      "Test loss: 0.05867006257176399\n",
      "Test loss: 0.056275613605976105\n",
      "Test loss: 0.05610799789428711\n",
      "Test loss: 0.05766282603144646\n",
      "Test loss: 0.056038178503513336\n",
      "Test loss: 0.056861307471990585\n",
      "Test loss: 0.05571181699633598\n",
      "Test loss: 0.056462451815605164\n",
      "Test loss: 0.05469899624586105\n",
      "Test loss: 0.05573105067014694\n",
      "Test loss: 0.05561824142932892\n",
      "Test loss: 0.05755668506026268\n",
      "Test loss: 0.05542889982461929\n",
      "Test loss: 0.05530066415667534\n",
      "Test loss: 0.05679168179631233\n",
      "Test loss: 0.056663867086172104\n",
      "Test loss: 0.05619947984814644\n",
      "Test loss: 0.057587914168834686\n",
      "Test loss: 0.05559735745191574\n",
      "Test loss: 0.05686064437031746\n",
      "Test loss: 0.05837061628699303\n",
      "Test loss: 0.0554521270096302\n",
      "Test loss: 0.05894983187317848\n",
      "Test loss: 0.056342851370573044\n",
      "Test loss: 0.05685614049434662\n",
      "Test loss: 0.056868404150009155\n",
      "Test loss: 0.0572405606508255\n",
      "Test: 7, loss: 0.056396763771772385\n",
      "Train epoch: 8\n",
      "train loss: 0.05774420499801636\n",
      "train loss: 0.05714142695069313\n",
      "train loss: 0.05588700249791145\n",
      "train loss: 0.054562803357839584\n",
      "train loss: 0.0565778985619545\n",
      "train loss: 0.054548341780900955\n",
      "train loss: 0.05776706710457802\n",
      "train loss: 0.05519787222146988\n",
      "train loss: 0.055652543902397156\n",
      "train loss: 0.05388059467077255\n",
      "train loss: 0.05583348125219345\n",
      "train loss: 0.05643745884299278\n",
      "train loss: 0.05490684136748314\n",
      "train loss: 0.05159910395741463\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train loss: 0.05674370750784874\n",
      "train loss: 0.05526326969265938\n",
      "train loss: 0.053789764642715454\n",
      "train loss: 0.0537978857755661\n",
      "train loss: 0.05492881312966347\n",
      "train loss: 0.05521578714251518\n",
      "train loss: 0.05112311244010925\n",
      "train loss: 0.05712663754820824\n",
      "train loss: 0.053413428366184235\n",
      "train loss: 0.05311840400099754\n",
      "train loss: 0.05398185923695564\n",
      "train loss: 0.05743062496185303\n",
      "train loss: 0.05432690307497978\n",
      "train loss: 0.05176763981580734\n",
      "train loss: 0.048920638859272\n",
      "train loss: 0.05235946178436279\n",
      "train loss: 0.05572611838579178\n",
      "Test loss: 0.04925691336393356\n",
      "Test loss: 0.05211321637034416\n",
      "Test loss: 0.05277649313211441\n",
      "Test loss: 0.052855685353279114\n",
      "Test loss: 0.05159030109643936\n",
      "Test loss: 0.05158325284719467\n",
      "Test loss: 0.05353561416268349\n",
      "Test loss: 0.05073981359601021\n",
      "Test loss: 0.05074818432331085\n",
      "Test loss: 0.051615260541439056\n",
      "Test loss: 0.05157962441444397\n",
      "Test loss: 0.052082255482673645\n",
      "Test loss: 0.05408969148993492\n",
      "Test loss: 0.05053514987230301\n",
      "Test loss: 0.05164514109492302\n",
      "Test loss: 0.05397335812449455\n",
      "Test loss: 0.05050628259778023\n",
      "Test loss: 0.05297031253576279\n",
      "Test loss: 0.049971215426921844\n",
      "Test loss: 0.049445539712905884\n",
      "Test loss: 0.053469207137823105\n",
      "Test loss: 0.05245493724942207\n",
      "Test loss: 0.05492658540606499\n",
      "Test loss: 0.051394086331129074\n",
      "Test loss: 0.05427024886012077\n",
      "Test loss: 0.051254939287900925\n",
      "Test loss: 0.05275886133313179\n",
      "Test loss: 0.05119651183485985\n",
      "Test loss: 0.05055459216237068\n",
      "Test loss: 0.05041472986340523\n",
      "Test loss: 0.05038494989275932\n",
      "Test: 8, loss: 0.05182880908250809\n",
      "Train epoch: 9\n",
      "train loss: 0.05146830528974533\n",
      "train loss: 0.05332230031490326\n",
      "train loss: 0.05472578853368759\n",
      "train loss: 0.05373547226190567\n",
      "train loss: 0.05082722380757332\n",
      "train loss: 0.05093255266547203\n",
      "train loss: 0.052530739456415176\n",
      "train loss: 0.05184516683220863\n",
      "train loss: 0.04802118241786957\n",
      "train loss: 0.05093734338879585\n",
      "train loss: 0.05098751559853554\n",
      "train loss: 0.04924666881561279\n",
      "train loss: 0.04869167134165764\n",
      "train loss: 0.04985126107931137\n",
      "train loss: 0.049606531858444214\n",
      "train loss: 0.05126876011490822\n",
      "train loss: 0.04951968416571617\n",
      "train loss: 0.04730149358510971\n",
      "train loss: 0.04992807283997536\n",
      "train loss: 0.05012257397174835\n",
      "train loss: 0.05399167165160179\n",
      "train loss: 0.0461234487593174\n",
      "train loss: 0.050269726663827896\n",
      "train loss: 0.051178473979234695\n",
      "train loss: 0.05250387266278267\n",
      "train loss: 0.05028736963868141\n",
      "train loss: 0.048598580062389374\n",
      "train loss: 0.04777269437909126\n",
      "train loss: 0.05804550275206566\n",
      "train loss: 0.05647817999124527\n",
      "train loss: 0.05388180539011955\n",
      "Test loss: 0.051196977496147156\n",
      "Test loss: 0.05097386613488197\n",
      "Test loss: 0.05168604478240013\n",
      "Test loss: 0.052390165627002716\n",
      "Test loss: 0.05108889937400818\n",
      "Test loss: 0.05148143321275711\n",
      "Test loss: 0.05133422836661339\n",
      "Test loss: 0.051353972405195236\n",
      "Test loss: 0.05105731263756752\n",
      "Test loss: 0.051097460091114044\n",
      "Test loss: 0.05212371051311493\n",
      "Test loss: 0.05175380781292915\n",
      "Test loss: 0.0481276661157608\n",
      "Test loss: 0.049162957817316055\n",
      "Test loss: 0.05195656418800354\n",
      "Test loss: 0.05114932358264923\n",
      "Test loss: 0.051841169595718384\n",
      "Test loss: 0.050864774733781815\n",
      "Test loss: 0.05048108100891113\n",
      "Test loss: 0.049971334636211395\n",
      "Test loss: 0.05077245086431503\n",
      "Test loss: 0.05294279754161835\n",
      "Test loss: 0.05398080497980118\n",
      "Test loss: 0.050695497542619705\n",
      "Test loss: 0.050695475190877914\n",
      "Test loss: 0.052976708859205246\n",
      "Test loss: 0.04780694469809532\n",
      "Test loss: 0.05018557980656624\n",
      "Test loss: 0.053584832698106766\n",
      "Test loss: 0.048576150089502335\n",
      "Test loss: 0.04921608045697212\n",
      "Test: 9, loss: 0.05104922503232956\n",
      "Train epoch: 10\n",
      "train loss: 0.05067053437232971\n",
      "train loss: 0.0547410286962986\n",
      "train loss: 0.057969868183135986\n",
      "train loss: 0.062160491943359375\n",
      "train loss: 0.08489284664392471\n",
      "train loss: 0.07340487092733383\n",
      "train loss: 0.07927315682172775\n",
      "train loss: 0.0923818051815033\n",
      "train loss: 0.07916221767663956\n",
      "train loss: 0.07702254503965378\n",
      "train loss: 0.09680375456809998\n",
      "train loss: 0.0703040361404419\n",
      "train loss: 0.07203693687915802\n",
      "train loss: 0.1026422306895256\n",
      "train loss: 0.07156534492969513\n",
      "train loss: 0.07017339766025543\n",
      "train loss: 0.08854691684246063\n",
      "train loss: 0.07912489026784897\n",
      "train loss: 0.06630656868219376\n",
      "train loss: 0.06562130153179169\n",
      "train loss: 0.07556179910898209\n",
      "train loss: 0.07525750994682312\n",
      "train loss: 0.06284822523593903\n",
      "train loss: 0.06026574596762657\n",
      "train loss: 0.06692703813314438\n",
      "train loss: 0.07238832116127014\n",
      "train loss: 0.06425617635250092\n",
      "train loss: 0.06293648481369019\n",
      "train loss: 0.06093486398458481\n",
      "train loss: 0.06338018923997879\n",
      "train loss: 0.06297466158866882\n",
      "Test loss: 0.059409160166978836\n",
      "Test loss: 0.059583887457847595\n",
      "Test loss: 0.060691412538290024\n",
      "Test loss: 0.05987636372447014\n",
      "Test loss: 0.059405338019132614\n",
      "Test loss: 0.05955228954553604\n",
      "Test loss: 0.05948525294661522\n",
      "Test loss: 0.06065314635634422\n",
      "Test loss: 0.05669279396533966\n",
      "Test loss: 0.05771204084157944\n",
      "Test loss: 0.05846765637397766\n",
      "Test loss: 0.058550138026475906\n",
      "Test loss: 0.05545087531208992\n",
      "Test loss: 0.05748799070715904\n",
      "Test loss: 0.05837630853056908\n",
      "Test loss: 0.05959390103816986\n",
      "Test loss: 0.058262892067432404\n",
      "Test loss: 0.05648820102214813\n",
      "Test loss: 0.056466519832611084\n",
      "Test loss: 0.060565680265426636\n",
      "Test loss: 0.05762312188744545\n",
      "Test loss: 0.058069635182619095\n",
      "Test loss: 0.05803433060646057\n",
      "Test loss: 0.05767788738012314\n",
      "Test loss: 0.05816679075360298\n",
      "Test loss: 0.058029912412166595\n",
      "Test loss: 0.057353515177965164\n",
      "Test loss: 0.060244303196668625\n",
      "Test loss: 0.058809295296669006\n",
      "Test loss: 0.06036743149161339\n",
      "Test loss: 0.058722879737615585\n",
      "Test: 10, loss: 0.05857647955417633\n",
      "Train epoch: 11\n",
      "train loss: 0.056639283895492554\n",
      "train loss: 0.054891470819711685\n",
      "train loss: 0.05798133835196495\n",
      "train loss: 0.05933109298348427\n",
      "train loss: 0.060855139046907425\n",
      "train loss: 0.05710276588797569\n",
      "train loss: 0.05818016454577446\n",
      "train loss: 0.052897948771715164\n",
      "train loss: 0.05798136070370674\n",
      "train loss: 0.05591031536459923\n",
      "train loss: 0.055953167378902435\n",
      "train loss: 0.0544663742184639\n",
      "train loss: 0.05271086096763611\n",
      "train loss: 0.054694272577762604\n",
      "train loss: 0.054633431136608124\n",
      "train loss: 0.052852533757686615\n",
      "train loss: 0.05409998446702957\n",
      "train loss: 0.05197201669216156\n",
      "train loss: 0.0529198944568634\n",
      "train loss: 0.053398504853248596\n",
      "train loss: 0.05243769660592079\n",
      "train loss: 0.04979975149035454\n",
      "train loss: 0.049731843173503876\n",
      "train loss: 0.051330361515283585\n",
      "train loss: 0.05176043510437012\n",
      "train loss: 0.05341614782810211\n",
      "train loss: 0.05416778847575188\n",
      "train loss: 0.05048427730798721\n",
      "train loss: 0.047925349324941635\n",
      "train loss: 0.04791676625609398\n",
      "train loss: 0.04905374348163605\n",
      "Test loss: 0.05025925487279892\n",
      "Test loss: 0.0498519092798233\n",
      "Test loss: 0.04841003566980362\n",
      "Test loss: 0.047256309539079666\n",
      "Test loss: 0.04940064996480942\n",
      "Test loss: 0.050505381077528\n",
      "Test loss: 0.0487862154841423\n",
      "Test loss: 0.04871397465467453\n",
      "Test loss: 0.05007246136665344\n",
      "Test loss: 0.0495697483420372\n",
      "Test loss: 0.05089394003152847\n",
      "Test loss: 0.05206915736198425\n",
      "Test loss: 0.04913637042045593\n",
      "Test loss: 0.05105895921587944\n",
      "Test loss: 0.04885350540280342\n",
      "Test loss: 0.04707813635468483\n",
      "Test loss: 0.04934467375278473\n",
      "Test loss: 0.04823162034153938\n",
      "Test loss: 0.05000550299882889\n",
      "Test loss: 0.051417168229818344\n",
      "Test loss: 0.04858475551009178\n",
      "Test loss: 0.05126070976257324\n",
      "Test loss: 0.04886860027909279\n",
      "Test loss: 0.04725678637623787\n",
      "Test loss: 0.04839154705405235\n",
      "Test loss: 0.04794388636946678\n",
      "Test loss: 0.04897027462720871\n",
      "Test loss: 0.049282606691122055\n",
      "Test loss: 0.05091933161020279\n",
      "Test loss: 0.048158951103687286\n",
      "Test loss: 0.04781200736761093\n",
      "Test: 11, loss: 0.04930207505822182\n",
      "Train epoch: 12\n",
      "train loss: 0.04906710609793663\n",
      "train loss: 0.04731418937444687\n",
      "train loss: 0.04888341575860977\n",
      "train loss: 0.049016162753105164\n",
      "train loss: 0.04782123118638992\n",
      "train loss: 0.04725636914372444\n",
      "train loss: 0.0501420684158802\n",
      "train loss: 0.04765654355287552\n",
      "train loss: 0.046639833599328995\n",
      "train loss: 0.05078524723649025\n",
      "train loss: 0.049857884645462036\n",
      "train loss: 0.04536030814051628\n",
      "train loss: 0.04924429953098297\n",
      "train loss: 0.046683311462402344\n",
      "train loss: 0.04639093577861786\n",
      "train loss: 0.04622701182961464\n",
      "train loss: 0.050931911915540695\n",
      "train loss: 0.0456865131855011\n",
      "train loss: 0.04852864518761635\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train loss: 0.04723537340760231\n",
      "train loss: 0.04691815376281738\n",
      "train loss: 0.0451740026473999\n",
      "train loss: 0.042714934796094894\n",
      "train loss: 0.045435041189193726\n",
      "train loss: 0.043069325387477875\n",
      "train loss: 0.04628120735287666\n",
      "train loss: 0.04882761090993881\n",
      "train loss: 0.04483155533671379\n",
      "train loss: 0.04413498938083649\n",
      "train loss: 0.04281507804989815\n",
      "train loss: 0.04181588068604469\n",
      "Test loss: 0.044728927314281464\n",
      "Test loss: 0.04776658117771149\n",
      "Test loss: 0.04724062606692314\n",
      "Test loss: 0.0433284193277359\n",
      "Test loss: 0.04738272726535797\n",
      "Test loss: 0.047114137560129166\n",
      "Test loss: 0.04639645293354988\n",
      "Test loss: 0.04602007940411568\n",
      "Test loss: 0.04497884213924408\n",
      "Test loss: 0.04557567462325096\n",
      "Test loss: 0.047515034675598145\n",
      "Test loss: 0.04717816784977913\n",
      "Test loss: 0.04661756381392479\n",
      "Test loss: 0.044822756201028824\n",
      "Test loss: 0.04571494832634926\n",
      "Test loss: 0.04399888589978218\n",
      "Test loss: 0.04486989602446556\n",
      "Test loss: 0.04496695101261139\n",
      "Test loss: 0.04580926150083542\n",
      "Test loss: 0.045520078390836716\n",
      "Test loss: 0.04731205478310585\n",
      "Test loss: 0.044983990490436554\n",
      "Test loss: 0.04619620367884636\n",
      "Test loss: 0.04641813039779663\n",
      "Test loss: 0.04700227826833725\n",
      "Test loss: 0.04570745676755905\n",
      "Test loss: 0.04597904905676842\n",
      "Test loss: 0.046755220741033554\n",
      "Test loss: 0.046701692044734955\n",
      "Test loss: 0.04534628987312317\n",
      "Test loss: 0.0453062541782856\n",
      "Test: 12, loss: 0.0459759496152401\n",
      "Train epoch: 13\n",
      "train loss: 0.046047601848840714\n",
      "train loss: 0.043767355382442474\n",
      "train loss: 0.04835408180952072\n",
      "train loss: 0.04369775578379631\n",
      "train loss: 0.04427538439631462\n",
      "train loss: 0.045379720628261566\n",
      "train loss: 0.043507374823093414\n",
      "train loss: 0.042754411697387695\n",
      "train loss: 0.04273238033056259\n",
      "train loss: 0.040865208953619\n",
      "train loss: 0.04240557551383972\n",
      "train loss: 0.041257865726947784\n",
      "train loss: 0.041869454085826874\n",
      "train loss: 0.04337313398718834\n",
      "train loss: 0.04248548299074173\n",
      "train loss: 0.04074200615286827\n",
      "train loss: 0.043002787977457047\n",
      "train loss: 0.03864729776978493\n",
      "train loss: 0.04431113973259926\n",
      "train loss: 0.04384298250079155\n",
      "train loss: 0.046194009482860565\n",
      "train loss: 0.04542340710759163\n",
      "train loss: 0.0438237302005291\n",
      "train loss: 0.041331756860017776\n",
      "train loss: 0.04006142541766167\n",
      "train loss: 0.03985261917114258\n",
      "train loss: 0.03967839851975441\n",
      "train loss: 0.04275231063365936\n",
      "train loss: 0.04294494912028313\n",
      "train loss: 0.042164940387010574\n",
      "train loss: 0.03968946263194084\n",
      "Test loss: 0.03870897367596626\n",
      "Test loss: 0.0408838614821434\n",
      "Test loss: 0.03901996463537216\n",
      "Test loss: 0.04019182547926903\n",
      "Test loss: 0.03842921182513237\n",
      "Test loss: 0.04008358716964722\n",
      "Test loss: 0.04440465196967125\n",
      "Test loss: 0.03946562483906746\n",
      "Test loss: 0.04027055576443672\n",
      "Test loss: 0.04131048172712326\n",
      "Test loss: 0.03964662179350853\n",
      "Test loss: 0.04341454431414604\n",
      "Test loss: 0.03839747607707977\n",
      "Test loss: 0.04083748906850815\n",
      "Test loss: 0.03982333093881607\n",
      "Test loss: 0.03824913501739502\n",
      "Test loss: 0.039205074310302734\n",
      "Test loss: 0.04205767810344696\n",
      "Test loss: 0.03945358842611313\n",
      "Test loss: 0.03968026489019394\n",
      "Test loss: 0.038397166877985\n",
      "Test loss: 0.041107963770627975\n",
      "Test loss: 0.04328940063714981\n",
      "Test loss: 0.0399838462471962\n",
      "Test loss: 0.04088876023888588\n",
      "Test loss: 0.04154415801167488\n",
      "Test loss: 0.03863050788640976\n",
      "Test loss: 0.039926689118146896\n",
      "Test loss: 0.03925640881061554\n",
      "Test loss: 0.03705526143312454\n",
      "Test loss: 0.04271251708269119\n",
      "Test: 13, loss: 0.040204085409641266\n",
      "Train epoch: 14\n",
      "train loss: 0.04128936305642128\n",
      "train loss: 0.04216208681464195\n",
      "train loss: 0.04052823409438133\n",
      "train loss: 0.04377433657646179\n",
      "train loss: 0.038386259227991104\n",
      "train loss: 0.037909261882305145\n",
      "train loss: 0.04015135020017624\n",
      "train loss: 0.04016256704926491\n",
      "train loss: 0.03914833068847656\n",
      "train loss: 0.0369923897087574\n",
      "train loss: 0.03768819570541382\n",
      "train loss: 0.038314081728458405\n",
      "train loss: 0.03707912564277649\n",
      "train loss: 0.03935634344816208\n",
      "train loss: 0.03928302600979805\n",
      "train loss: 0.03691111505031586\n",
      "train loss: 0.03578108921647072\n",
      "train loss: 0.03583855181932449\n",
      "train loss: 0.03884867578744888\n",
      "train loss: 0.038380902260541916\n",
      "train loss: 0.03764275461435318\n",
      "train loss: 0.036665353924036026\n",
      "train loss: 0.03613249585032463\n",
      "train loss: 0.034759774804115295\n",
      "train loss: 0.037667084485292435\n",
      "train loss: 0.036590151488780975\n",
      "train loss: 0.03522370010614395\n",
      "train loss: 0.03530915081501007\n",
      "train loss: 0.03385534510016441\n",
      "train loss: 0.03218806907534599\n",
      "train loss: 0.03309601917862892\n",
      "Test loss: 0.03533158823847771\n",
      "Test loss: 0.03105299174785614\n",
      "Test loss: 0.031644295901060104\n",
      "Test loss: 0.03371065482497215\n",
      "Test loss: 0.034189652651548386\n",
      "Test loss: 0.03468737378716469\n",
      "Test loss: 0.03361840918660164\n",
      "Test loss: 0.03592706844210625\n",
      "Test loss: 0.03394056856632233\n",
      "Test loss: 0.03499077633023262\n",
      "Test loss: 0.035756103694438934\n",
      "Test loss: 0.03307582437992096\n",
      "Test loss: 0.03343057632446289\n",
      "Test loss: 0.034212686121463776\n",
      "Test loss: 0.03138258680701256\n",
      "Test loss: 0.03440186008810997\n",
      "Test loss: 0.03324286639690399\n",
      "Test loss: 0.03420468047261238\n",
      "Test loss: 0.035364191979169846\n",
      "Test loss: 0.03450002148747444\n",
      "Test loss: 0.03282823786139488\n",
      "Test loss: 0.03444981202483177\n",
      "Test loss: 0.03210991621017456\n",
      "Test loss: 0.03243064135313034\n",
      "Test loss: 0.03119795024394989\n",
      "Test loss: 0.0356503464281559\n",
      "Test loss: 0.03409339115023613\n",
      "Test loss: 0.032715119421482086\n",
      "Test loss: 0.03423195704817772\n",
      "Test loss: 0.03519730269908905\n",
      "Test loss: 0.03348400071263313\n",
      "Test: 14, loss: 0.03377591818571091\n",
      "Train epoch: 15\n",
      "train loss: 0.032108839601278305\n",
      "train loss: 0.03344639763236046\n",
      "train loss: 0.03323774039745331\n",
      "train loss: 0.034977201372385025\n",
      "train loss: 0.03380359336733818\n",
      "train loss: 0.03265722841024399\n",
      "train loss: 0.030528025701642036\n",
      "train loss: 0.030031586065888405\n",
      "train loss: 0.028501752763986588\n",
      "train loss: 0.030129028484225273\n",
      "train loss: 0.030278820544481277\n",
      "train loss: 0.029436517506837845\n",
      "train loss: 0.030390502884984016\n",
      "train loss: 0.029790673404932022\n",
      "train loss: 0.02842002734541893\n",
      "train loss: 0.029318593442440033\n",
      "train loss: 0.03203025460243225\n",
      "train loss: 0.02792229875922203\n",
      "train loss: 0.0281558595597744\n",
      "train loss: 0.027490125969052315\n",
      "train loss: 0.025066785514354706\n",
      "train loss: 0.024269336834549904\n",
      "train loss: 0.025244951248168945\n",
      "train loss: 0.022482577711343765\n",
      "train loss: 0.025434982031583786\n",
      "train loss: 0.02356499619781971\n",
      "train loss: 0.023588702082633972\n",
      "train loss: 0.02336846850812435\n",
      "train loss: 0.02156086638569832\n",
      "train loss: 0.022564547136425972\n",
      "train loss: 0.023760346695780754\n",
      "Test loss: 0.023566430434584618\n",
      "Test loss: 0.026271680369973183\n",
      "Test loss: 0.02332480438053608\n",
      "Test loss: 0.022284239530563354\n",
      "Test loss: 0.02347661182284355\n",
      "Test loss: 0.022165685892105103\n",
      "Test loss: 0.02331480383872986\n",
      "Test loss: 0.023555992171168327\n",
      "Test loss: 0.024993127211928368\n",
      "Test loss: 0.025489676743745804\n",
      "Test loss: 0.024725129827857018\n",
      "Test loss: 0.02582252025604248\n",
      "Test loss: 0.02176886610686779\n",
      "Test loss: 0.02591080032289028\n",
      "Test loss: 0.02768387645483017\n",
      "Test loss: 0.02246548980474472\n",
      "Test loss: 0.02825762890279293\n",
      "Test loss: 0.024891192093491554\n",
      "Test loss: 0.023488514125347137\n",
      "Test loss: 0.0250743106007576\n",
      "Test loss: 0.02429577335715294\n",
      "Test loss: 0.02631337381899357\n",
      "Test loss: 0.025300806388258934\n",
      "Test loss: 0.02383929118514061\n",
      "Test loss: 0.023386353626847267\n",
      "Test loss: 0.023642757907509804\n",
      "Test loss: 0.024685312062501907\n",
      "Test loss: 0.023412266746163368\n",
      "Test loss: 0.02561314031481743\n",
      "Test loss: 0.02453550696372986\n",
      "Test loss: 0.02653731033205986\n",
      "Test: 15, loss: 0.024519139900803566\n",
      "Train epoch: 16\n",
      "train loss: 0.023607753217220306\n",
      "train loss: 0.023071864619851112\n",
      "train loss: 0.022082170471549034\n",
      "train loss: 0.02144964039325714\n",
      "train loss: 0.021634455770254135\n",
      "train loss: 0.021762888878583908\n",
      "train loss: 0.021547792479395866\n",
      "train loss: 0.0210579801350832\n",
      "train loss: 0.022417908534407616\n",
      "train loss: 0.020711923018097878\n",
      "train loss: 0.02015683799982071\n",
      "train loss: 0.019118482246994972\n",
      "train loss: 0.021608997136354446\n",
      "train loss: 0.01969742774963379\n",
      "train loss: 0.022010741755366325\n",
      "train loss: 0.01955375261604786\n",
      "train loss: 0.018389707431197166\n",
      "train loss: 0.0199002493172884\n",
      "train loss: 0.01782281883060932\n",
      "train loss: 0.019291434437036514\n",
      "train loss: 0.01740103028714657\n",
      "train loss: 0.018267327919602394\n",
      "train loss: 0.018625933676958084\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train loss: 0.01836041733622551\n",
      "train loss: 0.019460175186395645\n",
      "train loss: 0.017178716138005257\n",
      "train loss: 0.01701083406805992\n",
      "train loss: 0.019392183050513268\n",
      "train loss: 0.018410388380289078\n",
      "train loss: 0.017745142802596092\n",
      "train loss: 0.020516321063041687\n",
      "Test loss: 0.016212552785873413\n",
      "Test loss: 0.01715005189180374\n",
      "Test loss: 0.01937827095389366\n",
      "Test loss: 0.01803678087890148\n",
      "Test loss: 0.018411407247185707\n",
      "Test loss: 0.01776549406349659\n",
      "Test loss: 0.01832295022904873\n",
      "Test loss: 0.016899924725294113\n",
      "Test loss: 0.017786718904972076\n",
      "Test loss: 0.017723431810736656\n",
      "Test loss: 0.016886789351701736\n",
      "Test loss: 0.018436189740896225\n",
      "Test loss: 0.020811501890420914\n",
      "Test loss: 0.018572667613625526\n",
      "Test loss: 0.019172506406903267\n",
      "Test loss: 0.01778246834874153\n",
      "Test loss: 0.017399229109287262\n",
      "Test loss: 0.017849158495664597\n",
      "Test loss: 0.017441216856241226\n",
      "Test loss: 0.01732512004673481\n",
      "Test loss: 0.01702185906469822\n",
      "Test loss: 0.019757699221372604\n",
      "Test loss: 0.018219780176877975\n",
      "Test loss: 0.02223796769976616\n",
      "Test loss: 0.01751185767352581\n",
      "Test loss: 0.018189797177910805\n",
      "Test loss: 0.02063167467713356\n",
      "Test loss: 0.020056014880537987\n",
      "Test loss: 0.01733243837952614\n",
      "Test loss: 0.018769167363643646\n",
      "Test loss: 0.017660070210695267\n",
      "Test: 16, loss: 0.018282346427440643\n",
      "Train epoch: 17\n",
      "train loss: 0.018197424709796906\n",
      "train loss: 0.01858760416507721\n",
      "train loss: 0.01727723330259323\n",
      "train loss: 0.019123025238513947\n",
      "train loss: 0.018049512058496475\n",
      "train loss: 0.01749076321721077\n",
      "train loss: 0.017180385068058968\n",
      "train loss: 0.015842529013752937\n",
      "train loss: 0.01769861951470375\n",
      "train loss: 0.018183061853051186\n",
      "train loss: 0.018147243186831474\n",
      "train loss: 0.017546113580465317\n",
      "train loss: 0.016062486916780472\n",
      "train loss: 0.016472160816192627\n",
      "train loss: 0.015546762384474277\n",
      "train loss: 0.016106395050883293\n",
      "train loss: 0.016971245408058167\n",
      "train loss: 0.01672743633389473\n",
      "train loss: 0.016730140894651413\n",
      "train loss: 0.016224246472120285\n",
      "train loss: 0.0175478532910347\n",
      "train loss: 0.016680624336004257\n",
      "train loss: 0.01646009460091591\n",
      "train loss: 0.01831856369972229\n",
      "train loss: 0.01584499329328537\n",
      "train loss: 0.015353783965110779\n",
      "train loss: 0.017495132982730865\n",
      "train loss: 0.019374774768948555\n",
      "train loss: 0.021699562668800354\n",
      "train loss: 0.018783586099743843\n",
      "train loss: 0.01689125970005989\n",
      "Test loss: 0.01581771858036518\n",
      "Test loss: 0.01782018318772316\n",
      "Test loss: 0.016498206183314323\n",
      "Test loss: 0.01663147658109665\n",
      "Test loss: 0.018819598481059074\n",
      "Test loss: 0.01641078107059002\n",
      "Test loss: 0.016184402629733086\n",
      "Test loss: 0.01748095266520977\n",
      "Test loss: 0.01615202985703945\n",
      "Test loss: 0.01640181802213192\n",
      "Test loss: 0.015962112694978714\n",
      "Test loss: 0.016380053013563156\n",
      "Test loss: 0.020083175972104073\n",
      "Test loss: 0.016584614291787148\n",
      "Test loss: 0.017201939597725868\n",
      "Test loss: 0.017750130966305733\n",
      "Test loss: 0.017009416595101357\n",
      "Test loss: 0.017974087968468666\n",
      "Test loss: 0.016960635781288147\n",
      "Test loss: 0.018708351999521255\n",
      "Test loss: 0.017137544229626656\n",
      "Test loss: 0.01763666793704033\n",
      "Test loss: 0.017773065716028214\n",
      "Test loss: 0.016150884330272675\n",
      "Test loss: 0.017017802223563194\n",
      "Test loss: 0.018024709075689316\n",
      "Test loss: 0.017927933484315872\n",
      "Test loss: 0.01746874488890171\n",
      "Test loss: 0.01835075207054615\n",
      "Test loss: 0.01571427844464779\n",
      "Test loss: 0.018018437549471855\n",
      "Test: 17, loss: 0.017227502539753914\n",
      "Train epoch: 18\n",
      "train loss: 0.018317634239792824\n",
      "train loss: 0.015891099348664284\n",
      "train loss: 0.015656081959605217\n",
      "train loss: 0.017503654584288597\n",
      "train loss: 0.016129566356539726\n",
      "train loss: 0.01685313694179058\n",
      "train loss: 0.016360970214009285\n",
      "train loss: 0.01580522023141384\n",
      "train loss: 0.014758051373064518\n",
      "train loss: 0.015841731801629066\n",
      "train loss: 0.0160366240888834\n",
      "train loss: 0.014632875099778175\n",
      "train loss: 0.014596957713365555\n",
      "train loss: 0.014372393488883972\n",
      "train loss: 0.015208646655082703\n",
      "train loss: 0.01550223771482706\n",
      "train loss: 0.0180913507938385\n",
      "train loss: 0.014450302347540855\n",
      "train loss: 0.014507799409329891\n",
      "train loss: 0.013151239603757858\n",
      "train loss: 0.014322379603981972\n",
      "train loss: 0.01472581084817648\n",
      "train loss: 0.018258653581142426\n",
      "train loss: 0.016404638066887856\n",
      "train loss: 0.014287763275206089\n",
      "train loss: 0.014074592851102352\n",
      "train loss: 0.013850183226168156\n",
      "train loss: 0.013697154819965363\n",
      "train loss: 0.016281405463814735\n",
      "train loss: 0.015209944918751717\n",
      "train loss: 0.014939594082534313\n",
      "Test loss: 0.014568758197128773\n",
      "Test loss: 0.014627127908170223\n",
      "Test loss: 0.01595977135002613\n",
      "Test loss: 0.01378670521080494\n",
      "Test loss: 0.017452921718358994\n",
      "Test loss: 0.013796471059322357\n",
      "Test loss: 0.013598746620118618\n",
      "Test loss: 0.015587600879371166\n",
      "Test loss: 0.016719410195946693\n",
      "Test loss: 0.015682386234402657\n",
      "Test loss: 0.013405797071754932\n",
      "Test loss: 0.014385693706572056\n",
      "Test loss: 0.013402502983808517\n",
      "Test loss: 0.01537051610648632\n",
      "Test loss: 0.014029542915523052\n",
      "Test loss: 0.015138768590986729\n",
      "Test loss: 0.01462123729288578\n",
      "Test loss: 0.01520475372672081\n",
      "Test loss: 0.015437185764312744\n",
      "Test loss: 0.01474366057664156\n",
      "Test loss: 0.015035202726721764\n",
      "Test loss: 0.016074223443865776\n",
      "Test loss: 0.01364793162792921\n",
      "Test loss: 0.014973095618188381\n",
      "Test loss: 0.01596340909600258\n",
      "Test loss: 0.01487772911787033\n",
      "Test loss: 0.015236914157867432\n",
      "Test loss: 0.014014944434165955\n",
      "Test loss: 0.014583184383809566\n",
      "Test loss: 0.01414433866739273\n",
      "Test loss: 0.01361064612865448\n",
      "Test: 18, loss: 0.014828423969447613\n",
      "Train epoch: 19\n",
      "train loss: 0.013251010328531265\n",
      "train loss: 0.016529059037566185\n",
      "train loss: 0.014656725339591503\n",
      "train loss: 0.016611000522971153\n",
      "train loss: 0.015160607174038887\n",
      "train loss: 0.01604100689291954\n",
      "train loss: 0.01293194480240345\n",
      "train loss: 0.014190508052706718\n",
      "train loss: 0.013940580189228058\n",
      "train loss: 0.014476470649242401\n",
      "train loss: 0.014395928010344505\n",
      "train loss: 0.014181683771312237\n",
      "train loss: 0.013893920928239822\n",
      "train loss: 0.012808965519070625\n",
      "train loss: 0.013250756077468395\n",
      "train loss: 0.014128925278782845\n",
      "train loss: 0.01455029845237732\n",
      "train loss: 0.01221605483442545\n",
      "train loss: 0.016909074038267136\n",
      "train loss: 0.016093958169221878\n",
      "train loss: 0.014770951122045517\n",
      "train loss: 0.015094039961695671\n",
      "train loss: 0.013537410646677017\n",
      "train loss: 0.016270224004983902\n",
      "train loss: 0.01551027875393629\n",
      "train loss: 0.014161686412990093\n",
      "train loss: 0.01341167837381363\n",
      "train loss: 0.014603002928197384\n",
      "train loss: 0.011894273571670055\n",
      "train loss: 0.01261694636195898\n",
      "train loss: 0.01498181838542223\n",
      "Test loss: 0.014157658442854881\n",
      "Test loss: 0.01394854485988617\n",
      "Test loss: 0.01283141877502203\n",
      "Test loss: 0.015262600965797901\n",
      "Test loss: 0.013012435287237167\n",
      "Test loss: 0.011791631579399109\n",
      "Test loss: 0.013284871354699135\n",
      "Test loss: 0.01552013773471117\n",
      "Test loss: 0.014231346547603607\n",
      "Test loss: 0.013333797454833984\n",
      "Test loss: 0.014002631418406963\n",
      "Test loss: 0.01525205373764038\n",
      "Test loss: 0.01334866788238287\n",
      "Test loss: 0.012392519041895866\n",
      "Test loss: 0.013075863011181355\n",
      "Test loss: 0.012474874965846539\n",
      "Test loss: 0.012134868651628494\n",
      "Test loss: 0.014572208747267723\n",
      "Test loss: 0.01302429847419262\n",
      "Test loss: 0.014227315783500671\n",
      "Test loss: 0.014015902765095234\n",
      "Test loss: 0.016172848641872406\n",
      "Test loss: 0.014300251379609108\n",
      "Test loss: 0.01419804710894823\n",
      "Test loss: 0.012651613913476467\n",
      "Test loss: 0.01628764159977436\n",
      "Test loss: 0.016531240195035934\n",
      "Test loss: 0.015036368742585182\n",
      "Test loss: 0.016097569838166237\n",
      "Test loss: 0.013625032268464565\n",
      "Test loss: 0.013347772881388664\n",
      "Test: 19, loss: 0.014004646800458431\n",
      "Train epoch: 20\n",
      "train loss: 0.012832621112465858\n",
      "train loss: 0.013289201073348522\n",
      "train loss: 0.014078643172979355\n",
      "train loss: 0.014442685060203075\n",
      "train loss: 0.01388630922883749\n",
      "train loss: 0.014219395816326141\n",
      "train loss: 0.013526380993425846\n",
      "train loss: 0.011512304656207561\n",
      "train loss: 0.014213603921234608\n",
      "train loss: 0.012422327883541584\n",
      "train loss: 0.014297300949692726\n",
      "train loss: 0.016600966453552246\n",
      "train loss: 0.014967157505452633\n",
      "train loss: 0.014291446655988693\n",
      "train loss: 0.012352232821285725\n",
      "train loss: 0.013223851099610329\n",
      "train loss: 0.014975924976170063\n",
      "train loss: 0.01280164159834385\n",
      "train loss: 0.014734345488250256\n",
      "train loss: 0.01431943941861391\n",
      "train loss: 0.011759470216929913\n",
      "train loss: 0.011899522505700588\n",
      "train loss: 0.011914439499378204\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train loss: 0.012472527101635933\n",
      "train loss: 0.012278574518859386\n",
      "train loss: 0.014875789172947407\n",
      "train loss: 0.012866473756730556\n",
      "train loss: 0.013363783247768879\n",
      "train loss: 0.013821128755807877\n",
      "train loss: 0.012032074853777885\n",
      "train loss: 0.013034120202064514\n",
      "Test loss: 0.013200458139181137\n",
      "Test loss: 0.013368080370128155\n",
      "Test loss: 0.011945446953177452\n",
      "Test loss: 0.013080067001283169\n",
      "Test loss: 0.013043616898357868\n",
      "Test loss: 0.01132650300860405\n",
      "Test loss: 0.012998594902455807\n",
      "Test loss: 0.014807503670454025\n",
      "Test loss: 0.012670144438743591\n",
      "Test loss: 0.012152676470577717\n",
      "Test loss: 0.012654436752200127\n",
      "Test loss: 0.013465158641338348\n",
      "Test loss: 0.012439928017556667\n",
      "Test loss: 0.01322078425437212\n",
      "Test loss: 0.013470149599015713\n",
      "Test loss: 0.01367259118705988\n",
      "Test loss: 0.012792648747563362\n",
      "Test loss: 0.011523663066327572\n",
      "Test loss: 0.012284825555980206\n",
      "Test loss: 0.014803054742515087\n",
      "Test loss: 0.014346516691148281\n",
      "Test loss: 0.013551857322454453\n",
      "Test loss: 0.013839599676430225\n",
      "Test loss: 0.012738077901303768\n",
      "Test loss: 0.01182970404624939\n",
      "Test loss: 0.011240487918257713\n",
      "Test loss: 0.014233258552849293\n",
      "Test loss: 0.01336144469678402\n",
      "Test loss: 0.012245003134012222\n",
      "Test loss: 0.013318425975739956\n",
      "Test loss: 0.013535449281334877\n",
      "Test: 20, loss: 0.013005166314542294\n",
      "Train epoch: 21\n",
      "train loss: 0.013011662289500237\n",
      "train loss: 0.011898582801222801\n",
      "train loss: 0.012148367241024971\n",
      "train loss: 0.015853436663746834\n",
      "train loss: 0.01362872775644064\n",
      "train loss: 0.014234348200261593\n",
      "train loss: 0.013418779708445072\n",
      "train loss: 0.010928126983344555\n",
      "train loss: 0.012225992977619171\n",
      "train loss: 0.012934521771967411\n",
      "train loss: 0.012752818875014782\n",
      "train loss: 0.0126426350325346\n",
      "train loss: 0.011148162186145782\n",
      "train loss: 0.016207413747906685\n",
      "train loss: 0.012228693813085556\n",
      "train loss: 0.012744765728712082\n",
      "train loss: 0.013298634439706802\n",
      "train loss: 0.014165304601192474\n",
      "train loss: 0.012236355803906918\n",
      "train loss: 0.010639816522598267\n",
      "train loss: 0.011776234023272991\n",
      "train loss: 0.012741843238472939\n",
      "train loss: 0.014010392129421234\n",
      "train loss: 0.012308130040764809\n",
      "train loss: 0.011365099810063839\n",
      "train loss: 0.013782831840217113\n",
      "train loss: 0.013008524663746357\n",
      "train loss: 0.011869143694639206\n",
      "train loss: 0.014362817630171776\n",
      "train loss: 0.013768426142632961\n",
      "train loss: 0.012360408902168274\n",
      "Test loss: 0.010799613781273365\n",
      "Test loss: 0.010918855667114258\n",
      "Test loss: 0.011280248872935772\n",
      "Test loss: 0.012003165669739246\n",
      "Test loss: 0.011440600268542767\n",
      "Test loss: 0.013713995926082134\n",
      "Test loss: 0.01159239374101162\n",
      "Test loss: 0.013578550890088081\n",
      "Test loss: 0.011013816110789776\n",
      "Test loss: 0.011905612424015999\n",
      "Test loss: 0.012161227874457836\n",
      "Test loss: 0.010693793185055256\n",
      "Test loss: 0.010814339853823185\n",
      "Test loss: 0.012245312333106995\n",
      "Test loss: 0.012753705494105816\n",
      "Test loss: 0.0123960692435503\n",
      "Test loss: 0.01373511552810669\n",
      "Test loss: 0.01186454575508833\n",
      "Test loss: 0.011075317859649658\n",
      "Test loss: 0.012939104810357094\n",
      "Test loss: 0.012369382195174694\n",
      "Test loss: 0.013325797393918037\n",
      "Test loss: 0.010067327879369259\n",
      "Test loss: 0.013059727847576141\n",
      "Test loss: 0.011750095523893833\n",
      "Test loss: 0.011990108527243137\n",
      "Test loss: 0.011311592534184456\n",
      "Test loss: 0.011360066011548042\n",
      "Test loss: 0.012668072246015072\n",
      "Test loss: 0.012878833338618279\n",
      "Test loss: 0.012229131534695625\n",
      "Test: 21, loss: 0.011997920460999012\n",
      "Train epoch: 22\n",
      "train loss: 0.010493525303900242\n",
      "train loss: 0.015130740590393543\n",
      "train loss: 0.012447790242731571\n",
      "train loss: 0.010809279978275299\n",
      "train loss: 0.01190122589468956\n",
      "train loss: 0.011864852160215378\n",
      "train loss: 0.012343465350568295\n",
      "train loss: 0.01188259944319725\n",
      "train loss: 0.010906952433288097\n",
      "train loss: 0.012780382297933102\n",
      "train loss: 0.011174023151397705\n",
      "train loss: 0.011693298816680908\n",
      "train loss: 0.010428204201161861\n",
      "train loss: 0.01167227141559124\n",
      "train loss: 0.01185599621385336\n",
      "train loss: 0.011613771319389343\n",
      "train loss: 0.011167596094310284\n",
      "train loss: 0.011448835954070091\n",
      "train loss: 0.012062892317771912\n",
      "train loss: 0.010904138907790184\n",
      "train loss: 0.011568902060389519\n",
      "train loss: 0.012213140726089478\n",
      "train loss: 0.010256697423756123\n",
      "train loss: 0.013400574214756489\n",
      "train loss: 0.01117610838264227\n",
      "train loss: 0.01155609730631113\n",
      "train loss: 0.011216866783797741\n",
      "train loss: 0.013120677322149277\n",
      "train loss: 0.012733181938529015\n",
      "train loss: 0.012945602647960186\n",
      "train loss: 0.013000226579606533\n",
      "Test loss: 0.012424944899976254\n",
      "Test loss: 0.011118009686470032\n",
      "Test loss: 0.011197807267308235\n",
      "Test loss: 0.012191293761134148\n",
      "Test loss: 0.010945453308522701\n",
      "Test loss: 0.012071527540683746\n",
      "Test loss: 0.009905808605253696\n",
      "Test loss: 0.010850382037460804\n",
      "Test loss: 0.013331176713109016\n",
      "Test loss: 0.012137578800320625\n",
      "Test loss: 0.011308346875011921\n",
      "Test loss: 0.011509261094033718\n",
      "Test loss: 0.011855734512209892\n",
      "Test loss: 0.011572647839784622\n",
      "Test loss: 0.01235965359956026\n",
      "Test loss: 0.013483282178640366\n",
      "Test loss: 0.01207357831299305\n",
      "Test loss: 0.010055526159703732\n",
      "Test loss: 0.011101527139544487\n",
      "Test loss: 0.011583272367715836\n",
      "Test loss: 0.012774346396327019\n",
      "Test loss: 0.013703281059861183\n",
      "Test loss: 0.010528769344091415\n",
      "Test loss: 0.013408043421804905\n",
      "Test loss: 0.013769740238785744\n",
      "Test loss: 0.01252532284706831\n",
      "Test loss: 0.011987303383648396\n",
      "Test loss: 0.011209385469555855\n",
      "Test loss: 0.01242752280086279\n",
      "Test loss: 0.011077011935412884\n",
      "Test loss: 0.011432717554271221\n",
      "Test: 22, loss: 0.011868394911289215\n",
      "Train epoch: 23\n",
      "train loss: 0.01085496786981821\n",
      "train loss: 0.012370513752102852\n",
      "train loss: 0.010764758102595806\n",
      "train loss: 0.010744350962340832\n",
      "train loss: 0.012310351245105267\n",
      "train loss: 0.01134363655000925\n",
      "train loss: 0.013354149647057056\n",
      "train loss: 0.011617635376751423\n",
      "train loss: 0.013139461167156696\n",
      "train loss: 0.011428773403167725\n",
      "train loss: 0.011961770243942738\n",
      "train loss: 0.010984768159687519\n",
      "train loss: 0.011622246354818344\n",
      "train loss: 0.01212814450263977\n",
      "train loss: 0.0110047347843647\n",
      "train loss: 0.011112659238278866\n",
      "train loss: 0.010658097453415394\n",
      "train loss: 0.0101496372371912\n",
      "train loss: 0.01106024906039238\n",
      "train loss: 0.01120794378221035\n",
      "train loss: 0.010347054339945316\n",
      "train loss: 0.010016054846346378\n",
      "train loss: 0.008821442723274231\n",
      "train loss: 0.010684525594115257\n",
      "train loss: 0.012237101793289185\n",
      "train loss: 0.011326300911605358\n",
      "train loss: 0.012084643356502056\n",
      "train loss: 0.013230733573436737\n",
      "train loss: 0.009996827691793442\n",
      "train loss: 0.010624765418469906\n",
      "train loss: 0.010632081888616085\n",
      "Test loss: 0.014802394434809685\n",
      "Test loss: 0.01228858157992363\n",
      "Test loss: 0.011871985159814358\n",
      "Test loss: 0.009975984692573547\n",
      "Test loss: 0.01248992420732975\n",
      "Test loss: 0.012332498095929623\n",
      "Test loss: 0.013188892044126987\n",
      "Test loss: 0.012302156537771225\n",
      "Test loss: 0.01153700239956379\n",
      "Test loss: 0.01271950826048851\n",
      "Test loss: 0.012723499909043312\n",
      "Test loss: 0.011517819948494434\n",
      "Test loss: 0.0113137261942029\n",
      "Test loss: 0.010820476338267326\n",
      "Test loss: 0.013114171102643013\n",
      "Test loss: 0.0104744927957654\n",
      "Test loss: 0.009971939958631992\n",
      "Test loss: 0.01005321554839611\n",
      "Test loss: 0.012110794894397259\n",
      "Test loss: 0.011521994136273861\n",
      "Test loss: 0.010683435015380383\n",
      "Test loss: 0.013467908836901188\n",
      "Test loss: 0.013354822993278503\n",
      "Test loss: 0.01136353611946106\n",
      "Test loss: 0.010688417591154575\n",
      "Test loss: 0.010383008979260921\n",
      "Test loss: 0.010897641070187092\n",
      "Test loss: 0.01160083245486021\n",
      "Test loss: 0.010029595345258713\n",
      "Test loss: 0.010925414972007275\n",
      "Test loss: 0.01263190433382988\n",
      "Test: 23, loss: 0.011714759282767773\n",
      "Train epoch: 24\n",
      "train loss: 0.010891745798289776\n",
      "train loss: 0.009823399595916271\n",
      "train loss: 0.011162502691149712\n",
      "train loss: 0.010057765990495682\n",
      "train loss: 0.010808557271957397\n",
      "train loss: 0.008676496334373951\n",
      "train loss: 0.009852471761405468\n",
      "train loss: 0.010816968977451324\n",
      "train loss: 0.010211728513240814\n",
      "train loss: 0.010794498026371002\n",
      "train loss: 0.0122717609629035\n",
      "train loss: 0.01287920493632555\n",
      "train loss: 0.011382940225303173\n",
      "train loss: 0.011890305206179619\n",
      "train loss: 0.010783696547150612\n",
      "train loss: 0.01039910688996315\n",
      "train loss: 0.011441734619438648\n",
      "train loss: 0.010749958455562592\n",
      "train loss: 0.010011831298470497\n",
      "train loss: 0.009813247248530388\n",
      "train loss: 0.011203039437532425\n",
      "train loss: 0.010733239352703094\n",
      "train loss: 0.01154366321861744\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train loss: 0.010346935130655766\n",
      "train loss: 0.009996967390179634\n",
      "train loss: 0.00902729481458664\n",
      "train loss: 0.009561886079609394\n",
      "train loss: 0.010495562106370926\n",
      "train loss: 0.008794263005256653\n",
      "train loss: 0.010190097615122795\n",
      "train loss: 0.008209067396819592\n",
      "Test loss: 0.009585131891071796\n",
      "Test loss: 0.008767666295170784\n",
      "Test loss: 0.010665244422852993\n",
      "Test loss: 0.009193169884383678\n",
      "Test loss: 0.009647413156926632\n",
      "Test loss: 0.011853312142193317\n",
      "Test loss: 0.01116502471268177\n",
      "Test loss: 0.009239798411726952\n",
      "Test loss: 0.008941303007304668\n",
      "Test loss: 0.009535548277199268\n",
      "Test loss: 0.00928928330540657\n",
      "Test loss: 0.00953392032533884\n",
      "Test loss: 0.007942400872707367\n",
      "Test loss: 0.011753573082387447\n",
      "Test loss: 0.009171480312943459\n",
      "Test loss: 0.009412121959030628\n",
      "Test loss: 0.009768130257725716\n",
      "Test loss: 0.009207552298903465\n",
      "Test loss: 0.00976715236902237\n",
      "Test loss: 0.007748878560960293\n",
      "Test loss: 0.009133342653512955\n",
      "Test loss: 0.008376790210604668\n",
      "Test loss: 0.008126960135996342\n",
      "Test loss: 0.008886787109076977\n",
      "Test loss: 0.00918943528085947\n",
      "Test loss: 0.009710337035357952\n",
      "Test loss: 0.009985638782382011\n",
      "Test loss: 0.009741703048348427\n",
      "Test loss: 0.008447653613984585\n",
      "Test loss: 0.009247258305549622\n",
      "Test loss: 0.008481173776090145\n",
      "Test: 24, loss: 0.009403715841472149\n",
      "Train epoch: 25\n",
      "train loss: 0.01004798337817192\n",
      "train loss: 0.00843938160687685\n",
      "train loss: 0.010044943541288376\n",
      "train loss: 0.009953045286238194\n",
      "train loss: 0.01072884164750576\n",
      "train loss: 0.009721034206449986\n",
      "train loss: 0.00933170411735773\n",
      "train loss: 0.010200797580182552\n",
      "train loss: 0.009182831272482872\n",
      "train loss: 0.010000107809901237\n",
      "train loss: 0.010323064401745796\n",
      "train loss: 0.00953887403011322\n",
      "train loss: 0.009170244447886944\n",
      "train loss: 0.009509504772722721\n",
      "train loss: 0.010456553660333157\n",
      "train loss: 0.010873201303184032\n",
      "train loss: 0.010992156341671944\n",
      "train loss: 0.01220752950757742\n",
      "train loss: 0.010545351542532444\n",
      "train loss: 0.008696324191987514\n",
      "train loss: 0.00944749265909195\n",
      "train loss: 0.009924368932843208\n",
      "train loss: 0.008285575546324253\n",
      "train loss: 0.010409058071672916\n",
      "train loss: 0.008617574349045753\n",
      "train loss: 0.009936358779668808\n",
      "train loss: 0.010483886115252972\n",
      "train loss: 0.008051123470067978\n",
      "train loss: 0.011161657981574535\n",
      "train loss: 0.012411752715706825\n",
      "train loss: 0.01061207801103592\n",
      "Test loss: 0.010771326720714569\n",
      "Test loss: 0.010006311349570751\n",
      "Test loss: 0.011190921999514103\n",
      "Test loss: 0.010465280152857304\n",
      "Test loss: 0.010207991115748882\n",
      "Test loss: 0.012223524041473866\n",
      "Test loss: 0.01130890566855669\n",
      "Test loss: 0.011645779013633728\n",
      "Test loss: 0.012522689066827297\n",
      "Test loss: 0.010529306717216969\n",
      "Test loss: 0.012302529998123646\n",
      "Test loss: 0.009990900754928589\n",
      "Test loss: 0.00993164163082838\n",
      "Test loss: 0.009435774758458138\n",
      "Test loss: 0.012231030501425266\n",
      "Test loss: 0.0114692822098732\n",
      "Test loss: 0.01361176185309887\n",
      "Test loss: 0.009947399608790874\n",
      "Test loss: 0.009967371821403503\n",
      "Test loss: 0.010232606902718544\n",
      "Test loss: 0.011662638746201992\n",
      "Test loss: 0.010888614691793919\n",
      "Test loss: 0.00981044303625822\n",
      "Test loss: 0.010997005738317966\n",
      "Test loss: 0.009510490112006664\n",
      "Test loss: 0.012494831345975399\n",
      "Test loss: 0.010977305471897125\n",
      "Test loss: 0.010703503154218197\n",
      "Test loss: 0.01119109895080328\n",
      "Test loss: 0.009162507951259613\n",
      "Test loss: 0.011581816710531712\n",
      "Test: 25, loss: 0.010934599675238132\n",
      "Train epoch: 26\n",
      "train loss: 0.010836574248969555\n",
      "train loss: 0.010174443013966084\n",
      "train loss: 0.009162584319710732\n",
      "train loss: 0.010460998862981796\n",
      "train loss: 0.010055267252027988\n",
      "train loss: 0.009518900886178017\n",
      "train loss: 0.01162755023688078\n",
      "train loss: 0.009969028644263744\n",
      "train loss: 0.010413970798254013\n",
      "train loss: 0.009232532232999802\n",
      "train loss: 0.009704381227493286\n",
      "train loss: 0.008921277709305286\n",
      "train loss: 0.010729724541306496\n",
      "train loss: 0.012783348560333252\n",
      "train loss: 0.00923994556069374\n",
      "train loss: 0.010120732709765434\n",
      "train loss: 0.009800160303711891\n",
      "train loss: 0.008792857639491558\n",
      "train loss: 0.009589259512722492\n",
      "train loss: 0.0090372608974576\n",
      "train loss: 0.008931811898946762\n",
      "train loss: 0.01044966746121645\n",
      "train loss: 0.009597436524927616\n",
      "train loss: 0.009096335619688034\n",
      "train loss: 0.008946207351982594\n",
      "train loss: 0.009988327510654926\n",
      "train loss: 0.010282794013619423\n",
      "train loss: 0.008971807546913624\n",
      "train loss: 0.009935040026903152\n",
      "train loss: 0.011971550062298775\n",
      "train loss: 0.01001749001443386\n",
      "Test loss: 0.009718016721308231\n",
      "Test loss: 0.008937387727200985\n",
      "Test loss: 0.008286755532026291\n",
      "Test loss: 0.008881126530468464\n",
      "Test loss: 0.009988830424845219\n",
      "Test loss: 0.011132237501442432\n",
      "Test loss: 0.009449253790080547\n",
      "Test loss: 0.011311870999634266\n",
      "Test loss: 0.010206801816821098\n",
      "Test loss: 0.008616754785180092\n",
      "Test loss: 0.009109419770538807\n",
      "Test loss: 0.010661915875971317\n",
      "Test loss: 0.008569886907935143\n",
      "Test loss: 0.011167849414050579\n",
      "Test loss: 0.008698510006070137\n",
      "Test loss: 0.008499610237777233\n",
      "Test loss: 0.008485694415867329\n",
      "Test loss: 0.01043121051043272\n",
      "Test loss: 0.011097517795860767\n",
      "Test loss: 0.008870773017406464\n",
      "Test loss: 0.008640103042125702\n",
      "Test loss: 0.00998908281326294\n",
      "Test loss: 0.010293656960129738\n",
      "Test loss: 0.008950076065957546\n",
      "Test loss: 0.009905345737934113\n",
      "Test loss: 0.010132908821105957\n",
      "Test loss: 0.009071498177945614\n",
      "Test loss: 0.008066755719482899\n",
      "Test loss: 0.00974359456449747\n",
      "Test loss: 0.011991282925009727\n",
      "Test loss: 0.008467302657663822\n",
      "Test: 26, loss: 0.00959267932921648\n",
      "Train epoch: 27\n",
      "train loss: 0.010471079498529434\n",
      "train loss: 0.009579809382557869\n",
      "train loss: 0.008254577405750751\n",
      "train loss: 0.009816151112318039\n",
      "train loss: 0.008887313306331635\n",
      "train loss: 0.011505363509058952\n",
      "train loss: 0.00930433813482523\n",
      "train loss: 0.0099094919860363\n",
      "train loss: 0.011167919263243675\n",
      "train loss: 0.010917900130152702\n",
      "train loss: 0.012030280195176601\n",
      "train loss: 0.011514134705066681\n",
      "train loss: 0.011141126044094563\n",
      "train loss: 0.009587740525603294\n",
      "train loss: 0.011285712942481041\n",
      "train loss: 0.009571661241352558\n",
      "train loss: 0.011169741861522198\n",
      "train loss: 0.011617526412010193\n",
      "train loss: 0.008134068921208382\n",
      "train loss: 0.011557317338883877\n",
      "train loss: 0.012961619533598423\n",
      "train loss: 0.010188584215939045\n",
      "train loss: 0.011431754566729069\n",
      "train loss: 0.012194802984595299\n",
      "train loss: 0.008285308256745338\n",
      "train loss: 0.01085396483540535\n",
      "train loss: 0.011739619076251984\n",
      "train loss: 0.009618059732019901\n",
      "train loss: 0.010792331770062447\n",
      "train loss: 0.010931950062513351\n",
      "train loss: 0.009827935136854649\n",
      "Test loss: 0.009109743870794773\n",
      "Test loss: 0.01076540257781744\n",
      "Test loss: 0.009684856981039047\n",
      "Test loss: 0.010464681312441826\n",
      "Test loss: 0.008939414285123348\n",
      "Test loss: 0.01050350908190012\n",
      "Test loss: 0.007881692610681057\n",
      "Test loss: 0.010445237159729004\n",
      "Test loss: 0.01031793188303709\n",
      "Test loss: 0.009115714579820633\n",
      "Test loss: 0.009126291610300541\n",
      "Test loss: 0.007535402663052082\n",
      "Test loss: 0.008345364592969418\n",
      "Test loss: 0.009349655359983444\n",
      "Test loss: 0.008942567743360996\n",
      "Test loss: 0.008721022866666317\n",
      "Test loss: 0.009158773347735405\n",
      "Test loss: 0.009439988993108273\n",
      "Test loss: 0.009264105930924416\n",
      "Test loss: 0.009321043267846107\n",
      "Test loss: 0.00906469114124775\n",
      "Test loss: 0.00809219665825367\n",
      "Test loss: 0.008751168847084045\n",
      "Test loss: 0.008869008161127567\n",
      "Test loss: 0.009227273985743523\n",
      "Test loss: 0.008733135648071766\n",
      "Test loss: 0.00851869024336338\n",
      "Test loss: 0.008264184929430485\n",
      "Test loss: 0.008188746869564056\n",
      "Test loss: 0.009924661368131638\n",
      "Test loss: 0.009037565439939499\n",
      "Test: 27, loss: 0.009132378734648228\n",
      "Train epoch: 28\n",
      "train loss: 0.008323133923113346\n",
      "train loss: 0.010543459095060825\n",
      "train loss: 0.008025757037103176\n",
      "train loss: 0.009873563423752785\n",
      "train loss: 0.007819104008376598\n",
      "train loss: 0.008803552016615868\n",
      "train loss: 0.010482381097972393\n",
      "train loss: 0.011193149723112583\n",
      "train loss: 0.008153406903147697\n",
      "train loss: 0.010636823251843452\n",
      "train loss: 0.010218861512839794\n",
      "train loss: 0.009622307494282722\n",
      "train loss: 0.008646778762340546\n",
      "train loss: 0.009928501211106777\n",
      "train loss: 0.009844421409070492\n",
      "train loss: 0.009341709315776825\n",
      "train loss: 0.008523602969944477\n",
      "train loss: 0.009694995358586311\n",
      "train loss: 0.008933311328291893\n",
      "train loss: 0.007015962619334459\n",
      "train loss: 0.011897976510226727\n",
      "train loss: 0.008634470403194427\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train loss: 0.008667794987559319\n",
      "train loss: 0.009404690936207771\n",
      "train loss: 0.010046910494565964\n",
      "train loss: 0.00893428921699524\n",
      "train loss: 0.009834024123847485\n",
      "train loss: 0.009371097199618816\n",
      "train loss: 0.010035584680736065\n",
      "train loss: 0.009026003070175648\n",
      "train loss: 0.010063382796943188\n",
      "Test loss: 0.008354201912879944\n",
      "Test loss: 0.009605521336197853\n",
      "Test loss: 0.0077550034038722515\n",
      "Test loss: 0.008317037485539913\n",
      "Test loss: 0.00926800724118948\n",
      "Test loss: 0.007519715931266546\n",
      "Test loss: 0.008322987705469131\n",
      "Test loss: 0.008308672346174717\n",
      "Test loss: 0.009057261981070042\n",
      "Test loss: 0.008806880563497543\n",
      "Test loss: 0.009280741214752197\n",
      "Test loss: 0.009631284512579441\n",
      "Test loss: 0.008985280059278011\n",
      "Test loss: 0.010090569965541363\n",
      "Test loss: 0.00948392041027546\n",
      "Test loss: 0.009921375662088394\n",
      "Test loss: 0.008516760542988777\n",
      "Test loss: 0.009334431029856205\n",
      "Test loss: 0.008711101487278938\n",
      "Test loss: 0.009859694167971611\n",
      "Test loss: 0.010221052914857864\n",
      "Test loss: 0.00898077990859747\n",
      "Test loss: 0.00980306789278984\n",
      "Test loss: 0.008234049193561077\n",
      "Test loss: 0.008446984924376011\n",
      "Test loss: 0.009370674379169941\n",
      "Test loss: 0.011087806895375252\n",
      "Test loss: 0.007697144523262978\n",
      "Test loss: 0.01013039518147707\n",
      "Test loss: 0.010013451799750328\n",
      "Test loss: 0.00811165664345026\n",
      "Test: 28, loss: 0.009071855805814266\n",
      "Train epoch: 29\n",
      "train loss: 0.007537494879215956\n",
      "train loss: 0.010571311227977276\n",
      "train loss: 0.009614995680749416\n",
      "train loss: 0.012864204123616219\n",
      "train loss: 0.009330306202173233\n",
      "train loss: 0.010348005220293999\n",
      "train loss: 0.008722441270947456\n",
      "train loss: 0.010458732023835182\n",
      "train loss: 0.008400349877774715\n",
      "train loss: 0.008868498727679253\n",
      "train loss: 0.010906954295933247\n",
      "train loss: 0.008686386980116367\n",
      "train loss: 0.009077029302716255\n",
      "train loss: 0.011093610897660255\n",
      "train loss: 0.009718595072627068\n",
      "train loss: 0.009108482860028744\n",
      "train loss: 0.009698094800114632\n",
      "train loss: 0.009256892837584019\n",
      "train loss: 0.0072668022476136684\n",
      "train loss: 0.009016880765557289\n",
      "train loss: 0.007715241052210331\n",
      "train loss: 0.009634947404265404\n",
      "train loss: 0.009704713709652424\n",
      "train loss: 0.007737179286777973\n",
      "train loss: 0.008959246799349785\n",
      "train loss: 0.008397656492888927\n",
      "train loss: 0.006836255080997944\n",
      "train loss: 0.007552121300250292\n",
      "train loss: 0.0089246341958642\n",
      "train loss: 0.009885871782898903\n",
      "train loss: 0.00859302282333374\n",
      "Test loss: 0.008048738352954388\n",
      "Test loss: 0.0102009242400527\n",
      "Test loss: 0.007900411263108253\n",
      "Test loss: 0.008665845729410648\n",
      "Test loss: 0.008933698758482933\n",
      "Test loss: 0.008728157728910446\n",
      "Test loss: 0.008817607536911964\n",
      "Test loss: 0.008384039625525475\n",
      "Test loss: 0.008355094119906425\n",
      "Test loss: 0.009653368033468723\n",
      "Test loss: 0.008770683780312538\n",
      "Test loss: 0.011963942088186741\n",
      "Test loss: 0.009093688800930977\n",
      "Test loss: 0.008420847356319427\n",
      "Test loss: 0.011232264339923859\n",
      "Test loss: 0.00836622342467308\n",
      "Test loss: 0.010173184797167778\n",
      "Test loss: 0.007716943509876728\n",
      "Test loss: 0.008470817469060421\n",
      "Test loss: 0.008174288086593151\n",
      "Test loss: 0.00737162958830595\n",
      "Test loss: 0.009453834034502506\n",
      "Test loss: 0.009611033834517002\n",
      "Test loss: 0.008680877275764942\n",
      "Test loss: 0.01231388095766306\n",
      "Test loss: 0.00785974133759737\n",
      "Test loss: 0.008682958781719208\n",
      "Test loss: 0.009020352736115456\n",
      "Test loss: 0.00857573188841343\n",
      "Test loss: 0.007644251920282841\n",
      "Test loss: 0.007868180051445961\n",
      "Test: 29, loss: 0.00894042756408453\n",
      "Train epoch: 30\n",
      "train loss: 0.007827742956578732\n",
      "train loss: 0.008749912492930889\n",
      "train loss: 0.007997410371899605\n",
      "train loss: 0.008540470153093338\n",
      "train loss: 0.008934293873608112\n",
      "train loss: 0.00821874849498272\n",
      "train loss: 0.009233691729605198\n",
      "train loss: 0.008534341119229794\n",
      "train loss: 0.009280042722821236\n",
      "train loss: 0.007973014377057552\n",
      "train loss: 0.008866043761372566\n",
      "train loss: 0.010394510813057423\n",
      "train loss: 0.008078241720795631\n",
      "train loss: 0.008449234999716282\n",
      "train loss: 0.009440755471587181\n",
      "train loss: 0.008291374891996384\n",
      "train loss: 0.007641650270670652\n",
      "train loss: 0.008814502507448196\n",
      "train loss: 0.007461804896593094\n",
      "train loss: 0.008218113332986832\n",
      "train loss: 0.009699364192783833\n",
      "train loss: 0.008806672878563404\n",
      "train loss: 0.009175995364785194\n",
      "train loss: 0.009537859819829464\n",
      "train loss: 0.008312386460602283\n",
      "train loss: 0.00913906004279852\n",
      "train loss: 0.008892691694200039\n",
      "train loss: 0.01132810115814209\n",
      "train loss: 0.01074041798710823\n",
      "train loss: 0.009593292139470577\n",
      "train loss: 0.007519076112657785\n",
      "Test loss: 0.008664618246257305\n",
      "Test loss: 0.009038638323545456\n",
      "Test loss: 0.008501136675477028\n",
      "Test loss: 0.009366418235003948\n",
      "Test loss: 0.009307902306318283\n",
      "Test loss: 0.00785007979720831\n",
      "Test loss: 0.008373579941689968\n",
      "Test loss: 0.007877876982092857\n",
      "Test loss: 0.009393497370183468\n",
      "Test loss: 0.009465890005230904\n",
      "Test loss: 0.0087913628667593\n",
      "Test loss: 0.010545247234404087\n",
      "Test loss: 0.009736207313835621\n",
      "Test loss: 0.007678988389670849\n",
      "Test loss: 0.008387037552893162\n",
      "Test loss: 0.0075118993408977985\n",
      "Test loss: 0.010742289945483208\n",
      "Test loss: 0.010422969236969948\n",
      "Test loss: 0.009480093605816364\n",
      "Test loss: 0.010462022386491299\n",
      "Test loss: 0.00869091134518385\n",
      "Test loss: 0.009381385520100594\n",
      "Test loss: 0.008645628578960896\n",
      "Test loss: 0.009992510080337524\n",
      "Test loss: 0.00841500237584114\n",
      "Test loss: 0.008517919108271599\n",
      "Test loss: 0.007836725562810898\n",
      "Test loss: 0.008251825347542763\n",
      "Test loss: 0.007759630214422941\n",
      "Test loss: 0.01034223660826683\n",
      "Test loss: 0.010225147940218449\n",
      "Test: 30, loss: 0.009021183475852013\n",
      "Train epoch: 31\n",
      "train loss: 0.009976179338991642\n",
      "train loss: 0.008029299788177013\n",
      "train loss: 0.0072317966260015965\n",
      "train loss: 0.011141846887767315\n",
      "train loss: 0.007633241359144449\n",
      "train loss: 0.0077550578862428665\n",
      "train loss: 0.009054227732121944\n",
      "train loss: 0.009251631796360016\n",
      "train loss: 0.008002013899385929\n",
      "train loss: 0.009149131365120411\n",
      "train loss: 0.008282151073217392\n",
      "train loss: 0.008337459526956081\n",
      "train loss: 0.008284077048301697\n",
      "train loss: 0.009057512506842613\n",
      "train loss: 0.008854648098349571\n",
      "train loss: 0.007554493844509125\n",
      "train loss: 0.010042304173111916\n",
      "train loss: 0.009524190798401833\n",
      "train loss: 0.009199599735438824\n",
      "train loss: 0.009310634806752205\n",
      "train loss: 0.007988033816218376\n",
      "train loss: 0.007674806285649538\n",
      "train loss: 0.006499669048935175\n",
      "train loss: 0.007765937130898237\n",
      "train loss: 0.009363194927573204\n",
      "train loss: 0.008043074980378151\n",
      "train loss: 0.007569002453237772\n",
      "train loss: 0.008858960121870041\n",
      "train loss: 0.009085636585950851\n",
      "train loss: 0.008320052176713943\n",
      "train loss: 0.008847820572555065\n",
      "Test loss: 0.008053760044276714\n",
      "Test loss: 0.008626946248114109\n",
      "Test loss: 0.009731882251799107\n",
      "Test loss: 0.00786013063043356\n",
      "Test loss: 0.007644623517990112\n",
      "Test loss: 0.00792716909199953\n",
      "Test loss: 0.007079382427036762\n",
      "Test loss: 0.009069017134606838\n",
      "Test loss: 0.008101757615804672\n",
      "Test loss: 0.008146679028868675\n",
      "Test loss: 0.009291511960327625\n",
      "Test loss: 0.008685617707669735\n",
      "Test loss: 0.008866725489497185\n",
      "Test loss: 0.008741875179111958\n",
      "Test loss: 0.006410688161849976\n",
      "Test loss: 0.007491429336369038\n",
      "Test loss: 0.0068152183666825294\n",
      "Test loss: 0.008335075341165066\n",
      "Test loss: 0.006960439961403608\n",
      "Test loss: 0.007978200912475586\n",
      "Test loss: 0.006751839071512222\n",
      "Test loss: 0.00866645760834217\n",
      "Test loss: 0.009434309788048267\n",
      "Test loss: 0.006926873233169317\n",
      "Test loss: 0.009386539459228516\n",
      "Test loss: 0.009126486256718636\n",
      "Test loss: 0.007414200343191624\n",
      "Test loss: 0.007264879997819662\n",
      "Test loss: 0.008133606053888798\n",
      "Test loss: 0.00836312398314476\n",
      "Test loss: 0.007797889411449432\n",
      "Test: 31, loss: 0.008099495433270931\n",
      "Train epoch: 32\n",
      "train loss: 0.007583434693515301\n",
      "train loss: 0.008714905939996243\n",
      "train loss: 0.008142588660120964\n",
      "train loss: 0.008034120313823223\n",
      "train loss: 0.00822522398084402\n",
      "train loss: 0.007851067930459976\n",
      "train loss: 0.007533029653131962\n",
      "train loss: 0.008526468649506569\n",
      "train loss: 0.00994788110256195\n",
      "train loss: 0.010107004083693027\n",
      "train loss: 0.007851683534681797\n",
      "train loss: 0.008397945202887058\n",
      "train loss: 0.008953527547419071\n",
      "train loss: 0.0072359973564744\n",
      "train loss: 0.007166239432990551\n",
      "train loss: 0.008575908839702606\n",
      "train loss: 0.008409745059907436\n",
      "train loss: 0.007319629658013582\n",
      "train loss: 0.008877524174749851\n",
      "train loss: 0.007989185862243176\n",
      "train loss: 0.007176483981311321\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train loss: 0.007455328945070505\n",
      "train loss: 0.007875240407884121\n",
      "train loss: 0.008387877605855465\n",
      "train loss: 0.009030158631503582\n",
      "train loss: 0.009734234772622585\n",
      "train loss: 0.00817213486880064\n",
      "train loss: 0.010795120149850845\n",
      "train loss: 0.008258404210209846\n",
      "train loss: 0.008081409148871899\n",
      "train loss: 0.009231808595359325\n",
      "Test loss: 0.009106673300266266\n",
      "Test loss: 0.00841597095131874\n",
      "Test loss: 0.00740896025672555\n",
      "Test loss: 0.009482652880251408\n",
      "Test loss: 0.008055020123720169\n",
      "Test loss: 0.01071850024163723\n",
      "Test loss: 0.0077223810367286205\n",
      "Test loss: 0.009503494016826153\n",
      "Test loss: 0.007972936145961285\n",
      "Test loss: 0.009000333026051521\n",
      "Test loss: 0.00891917571425438\n",
      "Test loss: 0.008627322502434254\n",
      "Test loss: 0.008462033234536648\n",
      "Test loss: 0.009841570630669594\n",
      "Test loss: 0.00779289286583662\n",
      "Test loss: 0.00911570806056261\n",
      "Test loss: 0.008726746775209904\n",
      "Test loss: 0.007701336871832609\n",
      "Test loss: 0.007639732677489519\n",
      "Test loss: 0.009233196265995502\n",
      "Test loss: 0.008377824909985065\n",
      "Test loss: 0.00885887909680605\n",
      "Test loss: 0.010874802246689796\n",
      "Test loss: 0.008396926335990429\n",
      "Test loss: 0.008824857883155346\n",
      "Test loss: 0.008878583088517189\n",
      "Test loss: 0.0074872784316539764\n",
      "Test loss: 0.008108125999569893\n",
      "Test loss: 0.008092107251286507\n",
      "Test loss: 0.008084644563496113\n",
      "Test loss: 0.007368816528469324\n",
      "Test: 32, loss: 0.008606434799730778\n",
      "Train epoch: 33\n",
      "train loss: 0.007524291984736919\n",
      "train loss: 0.00800471194088459\n",
      "train loss: 0.00986418779939413\n",
      "train loss: 0.010129724629223347\n",
      "train loss: 0.00749946990981698\n",
      "train loss: 0.009175507351756096\n",
      "train loss: 0.007385498844087124\n",
      "train loss: 0.00715100159868598\n",
      "train loss: 0.008310220204293728\n",
      "train loss: 0.009790600277483463\n",
      "train loss: 0.00907002855092287\n",
      "train loss: 0.00856767501682043\n",
      "train loss: 0.0090342927724123\n",
      "train loss: 0.007221544161438942\n",
      "train loss: 0.010204549878835678\n",
      "train loss: 0.00930599495768547\n",
      "train loss: 0.00940187182277441\n",
      "train loss: 0.008475450798869133\n",
      "train loss: 0.007536694873124361\n",
      "train loss: 0.00820754561573267\n",
      "train loss: 0.008318483829498291\n",
      "train loss: 0.0073446547612547874\n",
      "train loss: 0.007740596774965525\n",
      "train loss: 0.008208144456148148\n",
      "train loss: 0.008922506123781204\n",
      "train loss: 0.008104871958494186\n",
      "train loss: 0.008451028726994991\n",
      "train loss: 0.008031070232391357\n",
      "train loss: 0.006366050336509943\n",
      "train loss: 0.0073415013030171394\n",
      "train loss: 0.007060970179736614\n",
      "Test loss: 0.007512358948588371\n",
      "Test loss: 0.007392521947622299\n",
      "Test loss: 0.006813271902501583\n",
      "Test loss: 0.007522535510361195\n",
      "Test loss: 0.00720618013292551\n",
      "Test loss: 0.007080604322254658\n",
      "Test loss: 0.006937026511877775\n",
      "Test loss: 0.007341550197452307\n",
      "Test loss: 0.0070436978712677956\n",
      "Test loss: 0.007833298295736313\n",
      "Test loss: 0.00658315047621727\n",
      "Test loss: 0.007307445164769888\n",
      "Test loss: 0.008342701010406017\n",
      "Test loss: 0.006749620195478201\n",
      "Test loss: 0.008923961780965328\n",
      "Test loss: 0.010063599795103073\n",
      "Test loss: 0.006079972255975008\n",
      "Test loss: 0.007505816873162985\n",
      "Test loss: 0.007022456731647253\n",
      "Test loss: 0.00635044788941741\n",
      "Test loss: 0.006215349771082401\n",
      "Test loss: 0.009261597879230976\n",
      "Test loss: 0.006554254796355963\n",
      "Test loss: 0.0067327083088457584\n",
      "Test loss: 0.007158930879086256\n",
      "Test loss: 0.006866689771413803\n",
      "Test loss: 0.006743723060935736\n",
      "Test loss: 0.007350353989750147\n",
      "Test loss: 0.007162088993936777\n",
      "Test loss: 0.007666019722819328\n",
      "Test loss: 0.00709120836108923\n",
      "Test: 33, loss: 0.007303715217858553\n",
      "Train epoch: 34\n",
      "train loss: 0.0058634658344089985\n",
      "train loss: 0.007046365179121494\n",
      "train loss: 0.009020066820085049\n",
      "train loss: 0.009202221408486366\n",
      "train loss: 0.010004578158259392\n",
      "train loss: 0.00851904321461916\n",
      "train loss: 0.008719340898096561\n",
      "train loss: 0.006423397455364466\n",
      "train loss: 0.00806095264852047\n",
      "train loss: 0.008563880808651447\n",
      "train loss: 0.008546375669538975\n",
      "train loss: 0.00902380421757698\n",
      "train loss: 0.008336668834090233\n",
      "train loss: 0.00916930753737688\n",
      "train loss: 0.0076456135138869286\n",
      "train loss: 0.008085374720394611\n",
      "train loss: 0.00891162920743227\n",
      "train loss: 0.009080592542886734\n",
      "train loss: 0.007872349582612514\n",
      "train loss: 0.00705953361466527\n",
      "train loss: 0.007343295495957136\n",
      "train loss: 0.007020686753094196\n",
      "train loss: 0.0064038969576358795\n",
      "train loss: 0.009130289778113365\n",
      "train loss: 0.008511967957019806\n",
      "train loss: 0.007795057725161314\n",
      "train loss: 0.008821642026305199\n",
      "train loss: 0.007061600219458342\n",
      "train loss: 0.008735772222280502\n",
      "train loss: 0.007968415506184101\n",
      "train loss: 0.008449451997876167\n",
      "Test loss: 0.008802173659205437\n",
      "Test loss: 0.007375126238912344\n",
      "Test loss: 0.0073712412267923355\n",
      "Test loss: 0.011435831896960735\n",
      "Test loss: 0.008406927809119225\n",
      "Test loss: 0.008073021657764912\n",
      "Test loss: 0.00861081387847662\n",
      "Test loss: 0.010227466933429241\n",
      "Test loss: 0.010476858355104923\n",
      "Test loss: 0.007160591892898083\n",
      "Test loss: 0.00801500491797924\n",
      "Test loss: 0.009105843491852283\n",
      "Test loss: 0.007160626817494631\n",
      "Test loss: 0.01083671860396862\n",
      "Test loss: 0.008782576769590378\n",
      "Test loss: 0.008644361048936844\n",
      "Test loss: 0.010232302360236645\n",
      "Test loss: 0.00847144890576601\n",
      "Test loss: 0.007756038568913937\n",
      "Test loss: 0.007673593703657389\n",
      "Test loss: 0.008098535239696503\n",
      "Test loss: 0.00742583442479372\n",
      "Test loss: 0.007416297681629658\n",
      "Test loss: 0.008062239736318588\n",
      "Test loss: 0.00982605293393135\n",
      "Test loss: 0.010008443146944046\n",
      "Test loss: 0.006551394239068031\n",
      "Test loss: 0.007421765010803938\n",
      "Test loss: 0.008121768943965435\n",
      "Test loss: 0.009223253466188908\n",
      "Test loss: 0.007419581059366465\n",
      "Test: 34, loss: 0.008522379212081432\n",
      "Train epoch: 35\n",
      "train loss: 0.00897329393774271\n",
      "train loss: 0.008679043501615524\n",
      "train loss: 0.007204075809568167\n",
      "train loss: 0.007672335021197796\n",
      "train loss: 0.00954701192677021\n",
      "train loss: 0.00708372425287962\n",
      "train loss: 0.007529900409281254\n",
      "train loss: 0.008271817117929459\n",
      "train loss: 0.007971180602908134\n",
      "train loss: 0.007441340014338493\n",
      "train loss: 0.006745369639247656\n",
      "train loss: 0.008553525432944298\n",
      "train loss: 0.007542027626186609\n",
      "train loss: 0.007663718890398741\n",
      "train loss: 0.00805142056196928\n",
      "train loss: 0.007046374026685953\n",
      "train loss: 0.006321964785456657\n",
      "train loss: 0.006706529296934605\n",
      "train loss: 0.008276465348899364\n",
      "train loss: 0.007220753468573093\n",
      "train loss: 0.008906867355108261\n",
      "train loss: 0.00929083488881588\n",
      "train loss: 0.00860677007585764\n",
      "train loss: 0.008186079561710358\n",
      "train loss: 0.007895035669207573\n",
      "train loss: 0.007885630242526531\n",
      "train loss: 0.007386626675724983\n",
      "train loss: 0.007988540455698967\n",
      "train loss: 0.008373371325433254\n",
      "train loss: 0.007207744289189577\n",
      "train loss: 0.007527836132794619\n",
      "Test loss: 0.007812485098838806\n",
      "Test loss: 0.0071768853813409805\n",
      "Test loss: 0.0069337510503828526\n",
      "Test loss: 0.006345257628709078\n",
      "Test loss: 0.0071236612275242805\n",
      "Test loss: 0.00584702892228961\n",
      "Test loss: 0.008470316417515278\n",
      "Test loss: 0.006614415906369686\n",
      "Test loss: 0.006779635325074196\n",
      "Test loss: 0.006081847473978996\n",
      "Test loss: 0.00821280013769865\n",
      "Test loss: 0.009200971573591232\n",
      "Test loss: 0.008701317943632603\n",
      "Test loss: 0.007157562300562859\n",
      "Test loss: 0.0070996214635670185\n",
      "Test loss: 0.006651479750871658\n",
      "Test loss: 0.0066405655816197395\n",
      "Test loss: 0.006765030790120363\n",
      "Test loss: 0.0067663658410310745\n",
      "Test loss: 0.00788030494004488\n",
      "Test loss: 0.0070112599059939384\n",
      "Test loss: 0.006919213104993105\n",
      "Test loss: 0.007422957569360733\n",
      "Test loss: 0.0075411079451441765\n",
      "Test loss: 0.006289019715040922\n",
      "Test loss: 0.008324289694428444\n",
      "Test loss: 0.006202284246683121\n",
      "Test loss: 0.0065370239317417145\n",
      "Test loss: 0.007185598369687796\n",
      "Test loss: 0.009456421248614788\n",
      "Test loss: 0.005826643202453852\n",
      "Test: 35, loss: 0.007192810066044331\n",
      "Train epoch: 36\n",
      "train loss: 0.007102394476532936\n",
      "train loss: 0.005647717975080013\n",
      "train loss: 0.007508391048759222\n",
      "train loss: 0.00700409198179841\n",
      "train loss: 0.007827313616871834\n",
      "train loss: 0.007353674620389938\n",
      "train loss: 0.007665196433663368\n",
      "train loss: 0.008716222830116749\n",
      "train loss: 0.006956229917705059\n",
      "train loss: 0.006612907163798809\n",
      "train loss: 0.007677007466554642\n",
      "train loss: 0.008569651283323765\n",
      "train loss: 0.008573934435844421\n",
      "train loss: 0.006910734809935093\n",
      "train loss: 0.009178816340863705\n",
      "train loss: 0.009376989677548409\n",
      "train loss: 0.007812973111867905\n",
      "train loss: 0.007321466691792011\n",
      "train loss: 0.00774232717230916\n",
      "train loss: 0.007546670734882355\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train loss: 0.00839553214609623\n",
      "train loss: 0.007753540761768818\n",
      "train loss: 0.006853464990854263\n",
      "train loss: 0.006461611483246088\n",
      "train loss: 0.006926833651959896\n",
      "train loss: 0.008309048600494862\n",
      "train loss: 0.006913456134498119\n",
      "train loss: 0.007608543150126934\n",
      "train loss: 0.006974884774535894\n",
      "train loss: 0.007003235165029764\n",
      "train loss: 0.007498663384467363\n",
      "Test loss: 0.006772313732653856\n",
      "Test loss: 0.006576873362064362\n",
      "Test loss: 0.007355535868555307\n",
      "Test loss: 0.006476379930973053\n",
      "Test loss: 0.007545615080744028\n",
      "Test loss: 0.007450878154486418\n",
      "Test loss: 0.006609152536839247\n",
      "Test loss: 0.006000003311783075\n",
      "Test loss: 0.007995889522135258\n",
      "Test loss: 0.006868417840451002\n",
      "Test loss: 0.006801814306527376\n",
      "Test loss: 0.005444263573735952\n",
      "Test loss: 0.007043400313705206\n",
      "Test loss: 0.005521868355572224\n",
      "Test loss: 0.006925547495484352\n",
      "Test loss: 0.007090270519256592\n",
      "Test loss: 0.006623877678066492\n",
      "Test loss: 0.008565333671867847\n",
      "Test loss: 0.005458960775285959\n",
      "Test loss: 0.006237165071070194\n",
      "Test loss: 0.007058506365865469\n",
      "Test loss: 0.007003026083111763\n",
      "Test loss: 0.00769015634432435\n",
      "Test loss: 0.008094335906207561\n",
      "Test loss: 0.007637285627424717\n",
      "Test loss: 0.007563804741948843\n",
      "Test loss: 0.005701704882085323\n",
      "Test loss: 0.007664268836379051\n",
      "Test loss: 0.00610531959682703\n",
      "Test loss: 0.00519314780831337\n",
      "Test loss: 0.0071058813482522964\n",
      "Test: 36, loss: 0.006844548508524895\n",
      "Train epoch: 37\n",
      "train loss: 0.007226640358567238\n",
      "train loss: 0.0081123486161232\n",
      "train loss: 0.00837777927517891\n",
      "train loss: 0.007865275256335735\n",
      "train loss: 0.007724168710410595\n",
      "train loss: 0.007096122484654188\n",
      "train loss: 0.006920349318534136\n",
      "train loss: 0.007733534090220928\n",
      "train loss: 0.0060177939012646675\n",
      "train loss: 0.006971031427383423\n",
      "train loss: 0.006933823227882385\n",
      "train loss: 0.005085983779281378\n",
      "train loss: 0.008187400177121162\n",
      "train loss: 0.006164337042719126\n",
      "train loss: 0.007143962197005749\n",
      "train loss: 0.008271542377769947\n",
      "train loss: 0.00647520087659359\n",
      "train loss: 0.006883928086608648\n",
      "train loss: 0.0078309690579772\n",
      "train loss: 0.00854787603020668\n",
      "train loss: 0.00601964583620429\n",
      "train loss: 0.0065087564289569855\n",
      "train loss: 0.005528683308511972\n",
      "train loss: 0.005722362548112869\n",
      "train loss: 0.0056845746003091335\n",
      "train loss: 0.006924700457602739\n",
      "train loss: 0.0062906090170145035\n",
      "train loss: 0.006801184266805649\n",
      "train loss: 0.006328729912638664\n",
      "train loss: 0.007408132776618004\n",
      "train loss: 0.007385586854070425\n",
      "Test loss: 0.0077102091163396835\n",
      "Test loss: 0.006527594290673733\n",
      "Test loss: 0.007366358768194914\n",
      "Test loss: 0.007901608943939209\n",
      "Test loss: 0.007608304265886545\n",
      "Test loss: 0.0073300693184137344\n",
      "Test loss: 0.005579792894423008\n",
      "Test loss: 0.007828276604413986\n",
      "Test loss: 0.007861599326133728\n",
      "Test loss: 0.007207863964140415\n",
      "Test loss: 0.0059718722477555275\n",
      "Test loss: 0.0071309576742351055\n",
      "Test loss: 0.006027780473232269\n",
      "Test loss: 0.0061148302629590034\n",
      "Test loss: 0.00725338002666831\n",
      "Test loss: 0.007614153437316418\n",
      "Test loss: 0.006682186853140593\n",
      "Test loss: 0.007157875690609217\n",
      "Test loss: 0.007003339473158121\n",
      "Test loss: 0.005969857331365347\n",
      "Test loss: 0.005790778435766697\n",
      "Test loss: 0.005769216455519199\n",
      "Test loss: 0.006439656484872103\n",
      "Test loss: 0.007183884736150503\n",
      "Test loss: 0.006556480657309294\n",
      "Test loss: 0.00783639308065176\n",
      "Test loss: 0.00653443206101656\n",
      "Test loss: 0.006684461142867804\n",
      "Test loss: 0.005844975356012583\n",
      "Test loss: 0.007499932311475277\n",
      "Test loss: 0.00618103938177228\n",
      "Test: 37, loss: 0.006844166666269302\n",
      "Train epoch: 38\n",
      "train loss: 0.008205964230000973\n",
      "train loss: 0.007342240307480097\n",
      "train loss: 0.006145459599792957\n",
      "train loss: 0.007789240218698978\n",
      "train loss: 0.007483143359422684\n",
      "train loss: 0.006595162209123373\n",
      "train loss: 0.007558273617178202\n",
      "train loss: 0.007181419525295496\n",
      "train loss: 0.006558330729603767\n",
      "train loss: 0.00719981687143445\n",
      "train loss: 0.006113714072853327\n",
      "train loss: 0.007961206138134003\n",
      "train loss: 0.00659189885482192\n",
      "train loss: 0.009189598262310028\n",
      "train loss: 0.007045637350529432\n",
      "train loss: 0.006720238830894232\n",
      "train loss: 0.008266140706837177\n",
      "train loss: 0.006425037980079651\n",
      "train loss: 0.0061036269180476665\n",
      "train loss: 0.007963052950799465\n",
      "train loss: 0.006519624963402748\n",
      "train loss: 0.006839472334831953\n",
      "train loss: 0.0065745091997087\n",
      "train loss: 0.006515807006508112\n",
      "train loss: 0.0063051506876945496\n",
      "train loss: 0.006724948529154062\n",
      "train loss: 0.0076667251996695995\n",
      "train loss: 0.0074708485044538975\n",
      "train loss: 0.0062582059763371944\n",
      "train loss: 0.006578068248927593\n",
      "train loss: 0.005620173644274473\n",
      "Test loss: 0.0070501454174518585\n",
      "Test loss: 0.007033165078610182\n",
      "Test loss: 0.005778217222541571\n",
      "Test loss: 0.007168768905103207\n",
      "Test loss: 0.0062933024019002914\n",
      "Test loss: 0.006289508193731308\n",
      "Test loss: 0.006737942341715097\n",
      "Test loss: 0.006457154639065266\n",
      "Test loss: 0.006365066859871149\n",
      "Test loss: 0.0070105320774018764\n",
      "Test loss: 0.008066302165389061\n",
      "Test loss: 0.006492311134934425\n",
      "Test loss: 0.0071379016153514385\n",
      "Test loss: 0.008259125985205173\n",
      "Test loss: 0.006212878972291946\n",
      "Test loss: 0.006800851318985224\n",
      "Test loss: 0.00744461826980114\n",
      "Test loss: 0.0072679282166063786\n",
      "Test loss: 0.0066794962622225285\n",
      "Test loss: 0.006688270252197981\n",
      "Test loss: 0.006419951096177101\n",
      "Test loss: 0.0072351195849478245\n",
      "Test loss: 0.006202021148055792\n",
      "Test loss: 0.006355935242027044\n",
      "Test loss: 0.007081599906086922\n",
      "Test loss: 0.006448725238442421\n",
      "Test loss: 0.006724951323121786\n",
      "Test loss: 0.006498973350971937\n",
      "Test loss: 0.005981705151498318\n",
      "Test loss: 0.0058748372830450535\n",
      "Test loss: 0.005387534853070974\n",
      "Test: 38, loss: 0.006691769231110811\n",
      "Train epoch: 39\n",
      "train loss: 0.007650767918676138\n",
      "train loss: 0.007799845654517412\n",
      "train loss: 0.008788153529167175\n",
      "train loss: 0.00743234483525157\n",
      "train loss: 0.0071973553858697414\n",
      "train loss: 0.005387323442846537\n",
      "train loss: 0.006630041636526585\n",
      "train loss: 0.006988076493144035\n",
      "train loss: 0.0088953897356987\n",
      "train loss: 0.008354687131941319\n",
      "train loss: 0.006520840339362621\n",
      "train loss: 0.006709894631057978\n",
      "train loss: 0.008735034614801407\n",
      "train loss: 0.006497297435998917\n",
      "train loss: 0.006898555438965559\n",
      "train loss: 0.0069541954435408115\n",
      "train loss: 0.006701333913952112\n",
      "train loss: 0.006669677793979645\n",
      "train loss: 0.007036267779767513\n",
      "train loss: 0.0063518439419567585\n",
      "train loss: 0.0063814278692007065\n",
      "train loss: 0.006172873545438051\n",
      "train loss: 0.006699686869978905\n",
      "train loss: 0.007362211123108864\n",
      "train loss: 0.006672936957329512\n",
      "train loss: 0.006738577038049698\n",
      "train loss: 0.004902636632323265\n",
      "train loss: 0.0055975052528083324\n",
      "train loss: 0.0064858258701860905\n",
      "train loss: 0.006735547445714474\n",
      "train loss: 0.006770178209990263\n",
      "Test loss: 0.005327147897332907\n",
      "Test loss: 0.0052751279436051846\n",
      "Test loss: 0.00626591918990016\n",
      "Test loss: 0.005479309242218733\n",
      "Test loss: 0.007437727879732847\n",
      "Test loss: 0.005953413899987936\n",
      "Test loss: 0.005519454833120108\n",
      "Test loss: 0.006705897394567728\n",
      "Test loss: 0.006525480188429356\n",
      "Test loss: 0.006398782134056091\n",
      "Test loss: 0.00743985315784812\n",
      "Test loss: 0.0062882183119654655\n",
      "Test loss: 0.006166028790175915\n",
      "Test loss: 0.006742068100720644\n",
      "Test loss: 0.007430411875247955\n",
      "Test loss: 0.006727584637701511\n",
      "Test loss: 0.00838268082588911\n",
      "Test loss: 0.007307862862944603\n",
      "Test loss: 0.005944548174738884\n",
      "Test loss: 0.007052064873278141\n",
      "Test loss: 0.0073987399227917194\n",
      "Test loss: 0.005562236066907644\n",
      "Test loss: 0.006918147671967745\n",
      "Test loss: 0.00576040567830205\n",
      "Test loss: 0.006702249404042959\n",
      "Test loss: 0.006336257793009281\n",
      "Test loss: 0.005859547294676304\n",
      "Test loss: 0.007081456016749144\n",
      "Test loss: 0.006873223464936018\n",
      "Test loss: 0.006099761463701725\n",
      "Test loss: 0.007595849689096212\n",
      "Test: 39, loss: 0.006534111220389605\n",
      "Train epoch: 40\n",
      "train loss: 0.006044003646820784\n",
      "train loss: 0.007061105687171221\n",
      "train loss: 0.0055412049405276775\n",
      "train loss: 0.006528025958687067\n",
      "train loss: 0.007119317073374987\n",
      "train loss: 0.0069786012172698975\n",
      "train loss: 0.006379099562764168\n",
      "train loss: 0.0060662939213216305\n",
      "train loss: 0.005759033374488354\n",
      "train loss: 0.006597805302590132\n",
      "train loss: 0.005238420329988003\n",
      "train loss: 0.007168399635702372\n",
      "train loss: 0.007350760977715254\n",
      "train loss: 0.007802238222211599\n",
      "train loss: 0.00548814982175827\n",
      "train loss: 0.006968002766370773\n",
      "train loss: 0.007973394356667995\n",
      "train loss: 0.008638055063784122\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train loss: 0.006562059745192528\n",
      "train loss: 0.006548344157636166\n",
      "train loss: 0.006361778359860182\n",
      "train loss: 0.0077500250190496445\n",
      "train loss: 0.007157389540225267\n",
      "train loss: 0.007722760550677776\n",
      "train loss: 0.0066139274276793\n",
      "train loss: 0.007372363470494747\n",
      "train loss: 0.00888800248503685\n",
      "train loss: 0.005422062240540981\n",
      "train loss: 0.006085806060582399\n",
      "train loss: 0.008131775073707104\n",
      "train loss: 0.007586036808788776\n",
      "Test loss: 0.0071795228868722916\n",
      "Test loss: 0.0070890565402805805\n",
      "Test loss: 0.007877538911998272\n",
      "Test loss: 0.00913314800709486\n",
      "Test loss: 0.006646875292062759\n",
      "Test loss: 0.007266296539455652\n",
      "Test loss: 0.007666338700801134\n",
      "Test loss: 0.007037234026938677\n",
      "Test loss: 0.007353870198130608\n",
      "Test loss: 0.008185585960745811\n",
      "Test loss: 0.008799298666417599\n",
      "Test loss: 0.00836874172091484\n",
      "Test loss: 0.008101342245936394\n",
      "Test loss: 0.007803122978657484\n",
      "Test loss: 0.007161456160247326\n",
      "Test loss: 0.008730252273380756\n",
      "Test loss: 0.0069244359619915485\n",
      "Test loss: 0.007915613241493702\n",
      "Test loss: 0.006641832180321217\n",
      "Test loss: 0.008827803656458855\n",
      "Test loss: 0.0073189204558730125\n",
      "Test loss: 0.007323469966650009\n",
      "Test loss: 0.006659097503870726\n",
      "Test loss: 0.007546532433480024\n",
      "Test loss: 0.0077709597535431385\n",
      "Test loss: 0.008156701922416687\n",
      "Test loss: 0.00701806228607893\n",
      "Test loss: 0.006597056984901428\n",
      "Test loss: 0.00746147008612752\n",
      "Test loss: 0.006565703544765711\n",
      "Test loss: 0.008689225651323795\n",
      "Test: 40, loss: 0.007606985978782177\n",
      "Train epoch: 41\n",
      "train loss: 0.0071119521744549274\n",
      "train loss: 0.008025696501135826\n",
      "train loss: 0.006981278769671917\n",
      "train loss: 0.006874792743474245\n",
      "train loss: 0.00757583836093545\n",
      "train loss: 0.005421306937932968\n",
      "train loss: 0.006858113221824169\n",
      "train loss: 0.006548125762492418\n",
      "train loss: 0.006854651495814323\n",
      "train loss: 0.0057573928497731686\n",
      "train loss: 0.007995261810719967\n",
      "train loss: 0.006621574051678181\n",
      "train loss: 0.007556543219834566\n",
      "train loss: 0.005601332522928715\n",
      "train loss: 0.006996816024184227\n",
      "train loss: 0.005929912906140089\n",
      "train loss: 0.0056947143748402596\n",
      "train loss: 0.00628762599080801\n",
      "train loss: 0.0063025145791471004\n",
      "train loss: 0.006762003526091576\n",
      "train loss: 0.00875686015933752\n",
      "train loss: 0.006216756068170071\n",
      "train loss: 0.007261297665536404\n",
      "train loss: 0.005810920614749193\n",
      "train loss: 0.007951030507683754\n",
      "train loss: 0.006847269833087921\n",
      "train loss: 0.006236795336008072\n",
      "train loss: 0.006654479540884495\n",
      "train loss: 0.006741678807884455\n",
      "train loss: 0.0059113092720508575\n",
      "train loss: 0.007396677043288946\n",
      "Test loss: 0.007122264243662357\n",
      "Test loss: 0.006010989658534527\n",
      "Test loss: 0.007182067725807428\n",
      "Test loss: 0.006302720867097378\n",
      "Test loss: 0.006135208066552877\n",
      "Test loss: 0.004795325919985771\n",
      "Test loss: 0.006168142892420292\n",
      "Test loss: 0.0066095502115786076\n",
      "Test loss: 0.00797792337834835\n",
      "Test loss: 0.008024267852306366\n",
      "Test loss: 0.005568350665271282\n",
      "Test loss: 0.005767466966062784\n",
      "Test loss: 0.005710706114768982\n",
      "Test loss: 0.0063169063068926334\n",
      "Test loss: 0.006746018771082163\n",
      "Test loss: 0.0058220080099999905\n",
      "Test loss: 0.007239791098982096\n",
      "Test loss: 0.006728039588779211\n",
      "Test loss: 0.006615083664655685\n",
      "Test loss: 0.006508509162813425\n",
      "Test loss: 0.006728350184857845\n",
      "Test loss: 0.005530518013983965\n",
      "Test loss: 0.0076285009272396564\n",
      "Test loss: 0.005638344679027796\n",
      "Test loss: 0.00544380210340023\n",
      "Test loss: 0.0049021583981812\n",
      "Test loss: 0.006883725989609957\n",
      "Test loss: 0.006387566681951284\n",
      "Test loss: 0.0070183067582547665\n",
      "Test loss: 0.005759550724178553\n",
      "Test loss: 0.006076464429497719\n",
      "Test: 41, loss: 0.006366084795445204\n",
      "Train epoch: 42\n",
      "train loss: 0.0057112062349915504\n",
      "train loss: 0.007383265532553196\n",
      "train loss: 0.006671905051916838\n",
      "train loss: 0.00732222618535161\n",
      "train loss: 0.006456136237829924\n",
      "train loss: 0.006135998759418726\n",
      "train loss: 0.005689252633601427\n",
      "train loss: 0.006933802738785744\n",
      "train loss: 0.006166345905512571\n",
      "train loss: 0.008553811348974705\n",
      "train loss: 0.006632758304476738\n",
      "train loss: 0.006660250015556812\n",
      "train loss: 0.008600451983511448\n",
      "train loss: 0.0069707902148365974\n",
      "train loss: 0.008811918087303638\n",
      "train loss: 0.008053739555180073\n",
      "train loss: 0.006141677498817444\n",
      "train loss: 0.006541483569890261\n",
      "train loss: 0.0065773059614002705\n",
      "train loss: 0.006849905010312796\n",
      "train loss: 0.007035982329398394\n",
      "train loss: 0.008639417588710785\n",
      "train loss: 0.006887455470860004\n",
      "train loss: 0.006353900767862797\n",
      "train loss: 0.006760239135473967\n",
      "train loss: 0.007281776517629623\n",
      "train loss: 0.007027745246887207\n",
      "train loss: 0.006921237800270319\n",
      "train loss: 0.006182506214827299\n",
      "train loss: 0.006351187359541655\n",
      "train loss: 0.0067190146073699\n",
      "Test loss: 0.00679368618875742\n",
      "Test loss: 0.0067430948838591576\n",
      "Test loss: 0.006568320095539093\n",
      "Test loss: 0.006055905483663082\n",
      "Test loss: 0.006814639549702406\n",
      "Test loss: 0.007622655015438795\n",
      "Test loss: 0.006886359304189682\n",
      "Test loss: 0.005641319323331118\n",
      "Test loss: 0.00609178002923727\n",
      "Test loss: 0.006337381899356842\n",
      "Test loss: 0.005702517461031675\n",
      "Test loss: 0.007260122336447239\n",
      "Test loss: 0.007821409963071346\n",
      "Test loss: 0.006744231563061476\n",
      "Test loss: 0.005941924639046192\n",
      "Test loss: 0.007465241011232138\n",
      "Test loss: 0.006105809472501278\n",
      "Test loss: 0.007098710164427757\n",
      "Test loss: 0.006149629596620798\n",
      "Test loss: 0.0066006965935230255\n",
      "Test loss: 0.006506196688860655\n",
      "Test loss: 0.005949223414063454\n",
      "Test loss: 0.0075488039292395115\n",
      "Test loss: 0.007979607209563255\n",
      "Test loss: 0.006834373809397221\n",
      "Test loss: 0.006166117265820503\n",
      "Test loss: 0.006593547761440277\n",
      "Test loss: 0.007232871372252703\n",
      "Test loss: 0.005910916719585657\n",
      "Test loss: 0.0062864175997674465\n",
      "Test loss: 0.006608886644244194\n",
      "Test: 42, loss: 0.006647174246609211\n",
      "Train epoch: 43\n",
      "train loss: 0.006220144219696522\n",
      "train loss: 0.007963583804666996\n",
      "train loss: 0.0058341724798083305\n",
      "train loss: 0.006556153763085604\n",
      "train loss: 0.005822749342769384\n",
      "train loss: 0.006073760334402323\n",
      "train loss: 0.006525309290736914\n",
      "train loss: 0.006669270806014538\n",
      "train loss: 0.005787377245724201\n",
      "train loss: 0.005158640909940004\n",
      "train loss: 0.0071223038248717785\n",
      "train loss: 0.0067167324014008045\n",
      "train loss: 0.007354474160820246\n",
      "train loss: 0.005419227294623852\n",
      "train loss: 0.006359693594276905\n",
      "train loss: 0.00543440505862236\n",
      "train loss: 0.005641131196171045\n",
      "train loss: 0.0056084091775119305\n",
      "train loss: 0.006129208952188492\n",
      "train loss: 0.006354017648845911\n",
      "train loss: 0.007055368274450302\n",
      "train loss: 0.00492179486900568\n",
      "train loss: 0.006592884659767151\n",
      "train loss: 0.007539569865912199\n",
      "train loss: 0.005934159271419048\n",
      "train loss: 0.005496793892234564\n",
      "train loss: 0.007162139285355806\n",
      "train loss: 0.006599061656743288\n",
      "train loss: 0.006713260896503925\n",
      "train loss: 0.006865366827696562\n",
      "train loss: 0.006577800959348679\n",
      "Test loss: 0.0069374567829072475\n",
      "Test loss: 0.007545062806457281\n",
      "Test loss: 0.008478701114654541\n",
      "Test loss: 0.006065319757908583\n",
      "Test loss: 0.0072313169948756695\n",
      "Test loss: 0.005265613552182913\n",
      "Test loss: 0.006019566673785448\n",
      "Test loss: 0.006564062088727951\n",
      "Test loss: 0.005156639497727156\n",
      "Test loss: 0.006097439210861921\n",
      "Test loss: 0.006635013967752457\n",
      "Test loss: 0.006559640169143677\n",
      "Test loss: 0.007740219589322805\n",
      "Test loss: 0.005664834287017584\n",
      "Test loss: 0.006682748906314373\n",
      "Test loss: 0.005616848822683096\n",
      "Test loss: 0.006403208710253239\n",
      "Test loss: 0.006376154255121946\n",
      "Test loss: 0.005505908280611038\n",
      "Test loss: 0.005928375758230686\n",
      "Test loss: 0.006154067814350128\n",
      "Test loss: 0.006482528522610664\n",
      "Test loss: 0.006367502734065056\n",
      "Test loss: 0.00665283203125\n",
      "Test loss: 0.006344346795231104\n",
      "Test loss: 0.00681188004091382\n",
      "Test loss: 0.006446441635489464\n",
      "Test loss: 0.005443225149065256\n",
      "Test loss: 0.006341276224702597\n",
      "Test loss: 0.0055656577460467815\n",
      "Test loss: 0.006085433065891266\n",
      "Test: 43, loss: 0.006360301282256842\n",
      "Train epoch: 44\n",
      "train loss: 0.00689665786921978\n",
      "train loss: 0.007967132143676281\n",
      "train loss: 0.006979570258408785\n",
      "train loss: 0.005452918354421854\n",
      "train loss: 0.006064156070351601\n",
      "train loss: 0.007545218337327242\n",
      "train loss: 0.006967117078602314\n",
      "train loss: 0.006125897169113159\n",
      "train loss: 0.006992583628743887\n",
      "train loss: 0.006333288736641407\n",
      "train loss: 0.006872188299894333\n",
      "train loss: 0.006746523082256317\n",
      "train loss: 0.0059178113006055355\n",
      "train loss: 0.007806751877069473\n",
      "train loss: 0.006235538981854916\n",
      "train loss: 0.006696485914289951\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train loss: 0.006604725029319525\n",
      "train loss: 0.00611573550850153\n",
      "train loss: 0.00789414532482624\n",
      "train loss: 0.006087972782552242\n",
      "train loss: 0.006441647652536631\n",
      "train loss: 0.006359410472214222\n",
      "train loss: 0.007080773822963238\n",
      "train loss: 0.006479667965322733\n",
      "train loss: 0.006645166780799627\n",
      "train loss: 0.00654037157073617\n",
      "train loss: 0.006264044903218746\n",
      "train loss: 0.006000131834298372\n",
      "train loss: 0.006215366534888744\n",
      "train loss: 0.005803689826279879\n",
      "train loss: 0.006683063693344593\n",
      "Test loss: 0.006299334578216076\n",
      "Test loss: 0.007513787131756544\n",
      "Test loss: 0.006161304656416178\n",
      "Test loss: 0.006692445371299982\n",
      "Test loss: 0.006154058966785669\n",
      "Test loss: 0.005999528802931309\n",
      "Test loss: 0.006724587641656399\n",
      "Test loss: 0.005781811662018299\n",
      "Test loss: 0.006356827914714813\n",
      "Test loss: 0.007524408865720034\n",
      "Test loss: 0.007039742078632116\n",
      "Test loss: 0.005711077246814966\n",
      "Test loss: 0.006535041611641645\n",
      "Test loss: 0.006370037328451872\n",
      "Test loss: 0.006994927302002907\n",
      "Test loss: 0.007544671185314655\n",
      "Test loss: 0.007407509256154299\n",
      "Test loss: 0.007528586778789759\n",
      "Test loss: 0.005936477333307266\n",
      "Test loss: 0.005402218550443649\n",
      "Test loss: 0.0068451641127467155\n",
      "Test loss: 0.005732922349125147\n",
      "Test loss: 0.006384213455021381\n",
      "Test loss: 0.006539364345371723\n",
      "Test loss: 0.0063008517026901245\n",
      "Test loss: 0.0068224407732486725\n",
      "Test loss: 0.0060914624482393265\n",
      "Test loss: 0.007031232584267855\n",
      "Test loss: 0.006150542758405209\n",
      "Test loss: 0.006395143922418356\n",
      "Test loss: 0.005667392164468765\n",
      "Test: 44, loss: 0.006504487246274948\n",
      "Train epoch: 45\n",
      "train loss: 0.006152622401714325\n",
      "train loss: 0.006610678043216467\n",
      "train loss: 0.0053145610727369785\n",
      "train loss: 0.005603819619864225\n",
      "train loss: 0.0068621705286204815\n",
      "train loss: 0.00564350001513958\n",
      "train loss: 0.005144717637449503\n",
      "train loss: 0.005206912290304899\n",
      "train loss: 0.005994134582579136\n",
      "train loss: 0.0053170532919466496\n",
      "train loss: 0.00717134727165103\n",
      "train loss: 0.007366112899035215\n",
      "train loss: 0.005537657532840967\n",
      "train loss: 0.006158686708658934\n",
      "train loss: 0.007389256730675697\n",
      "train loss: 0.005105737131088972\n",
      "train loss: 0.005673771724104881\n",
      "train loss: 0.0056309388019144535\n",
      "train loss: 0.005832958966493607\n",
      "train loss: 0.00644201785326004\n",
      "train loss: 0.006975442171096802\n",
      "train loss: 0.006482094991952181\n",
      "train loss: 0.005606049206107855\n",
      "train loss: 0.006566398777067661\n",
      "train loss: 0.008026108145713806\n",
      "train loss: 0.004790383856743574\n",
      "train loss: 0.006784315221011639\n",
      "train loss: 0.0058930981904268265\n",
      "train loss: 0.005522333551198244\n",
      "train loss: 0.008099963888525963\n",
      "train loss: 0.006437019445002079\n",
      "Test loss: 0.006760102231055498\n",
      "Test loss: 0.005751380696892738\n",
      "Test loss: 0.006228737998753786\n",
      "Test loss: 0.005124161019921303\n",
      "Test loss: 0.00624379888176918\n",
      "Test loss: 0.0060181934386491776\n",
      "Test loss: 0.005930350627750158\n",
      "Test loss: 0.0059070708230137825\n",
      "Test loss: 0.0057470528408885\n",
      "Test loss: 0.0051062447018921375\n",
      "Test loss: 0.005688820965588093\n",
      "Test loss: 0.005026586819440126\n",
      "Test loss: 0.004867443814873695\n",
      "Test loss: 0.00750141404569149\n",
      "Test loss: 0.0052070715464651585\n",
      "Test loss: 0.006012630183249712\n",
      "Test loss: 0.004998680669814348\n",
      "Test loss: 0.005402883980423212\n",
      "Test loss: 0.005005845334380865\n",
      "Test loss: 0.005778750404715538\n",
      "Test loss: 0.0049091302789747715\n",
      "Test loss: 0.0051064882427453995\n",
      "Test loss: 0.00519127631559968\n",
      "Test loss: 0.0077791884541511536\n",
      "Test loss: 0.006751146633177996\n",
      "Test loss: 0.005805444438010454\n",
      "Test loss: 0.004876244347542524\n",
      "Test loss: 0.0057228077203035355\n",
      "Test loss: 0.0058625019155442715\n",
      "Test loss: 0.0051107220351696014\n",
      "Test loss: 0.006743773818016052\n",
      "Test: 45, loss: 0.005747288931161165\n",
      "Train epoch: 46\n",
      "train loss: 0.006899605505168438\n",
      "train loss: 0.005059492774307728\n",
      "train loss: 0.005458660423755646\n",
      "train loss: 0.00603632815182209\n",
      "train loss: 0.007125480100512505\n",
      "train loss: 0.006680149585008621\n",
      "train loss: 0.00557786924764514\n",
      "train loss: 0.007157130166888237\n",
      "train loss: 0.006576870568096638\n",
      "train loss: 0.005760862957686186\n",
      "train loss: 0.005536411423236132\n",
      "train loss: 0.006698806304484606\n",
      "train loss: 0.005396576132625341\n",
      "train loss: 0.007397042121738195\n",
      "train loss: 0.00519530987367034\n",
      "train loss: 0.005754361394792795\n",
      "train loss: 0.004971692804247141\n",
      "train loss: 0.0059168157167732716\n",
      "train loss: 0.006042294204235077\n",
      "train loss: 0.005355836823582649\n",
      "train loss: 0.004974011331796646\n",
      "train loss: 0.006971655413508415\n",
      "train loss: 0.007259910926222801\n",
      "train loss: 0.007330466993153095\n",
      "train loss: 0.006746705621480942\n",
      "train loss: 0.006455339957028627\n",
      "train loss: 0.00790312234312296\n",
      "train loss: 0.0056952922604978085\n",
      "train loss: 0.007188998628407717\n",
      "train loss: 0.006273151375353336\n",
      "train loss: 0.005722693633288145\n",
      "Test loss: 0.006221057381480932\n",
      "Test loss: 0.006201681215316057\n",
      "Test loss: 0.007066345773637295\n",
      "Test loss: 0.005918737035244703\n",
      "Test loss: 0.006127335596829653\n",
      "Test loss: 0.0068514044396579266\n",
      "Test loss: 0.006285545416176319\n",
      "Test loss: 0.006139592733234167\n",
      "Test loss: 0.007340022828429937\n",
      "Test loss: 0.005101141054183245\n",
      "Test loss: 0.005973921623080969\n",
      "Test loss: 0.006430733948945999\n",
      "Test loss: 0.007096060086041689\n",
      "Test loss: 0.0075539047829806805\n",
      "Test loss: 0.009180868975818157\n",
      "Test loss: 0.007437778636813164\n",
      "Test loss: 0.00640112766996026\n",
      "Test loss: 0.00698543107137084\n",
      "Test loss: 0.005858490243554115\n",
      "Test loss: 0.0059967441484332085\n",
      "Test loss: 0.008561564609408379\n",
      "Test loss: 0.006145489402115345\n",
      "Test loss: 0.006830297876149416\n",
      "Test loss: 0.005901984870433807\n",
      "Test loss: 0.008188717998564243\n",
      "Test loss: 0.00600534537807107\n",
      "Test loss: 0.007015299052000046\n",
      "Test loss: 0.0068948338739573956\n",
      "Test loss: 0.006023497320711613\n",
      "Test loss: 0.005832788068801165\n",
      "Test loss: 0.0067002917639911175\n",
      "Test: 46, loss: 0.00665380759164691\n",
      "Train epoch: 47\n",
      "train loss: 0.005359139293432236\n",
      "train loss: 0.006439663469791412\n",
      "train loss: 0.005072254687547684\n",
      "train loss: 0.007136684842407703\n",
      "train loss: 0.007209131494164467\n",
      "train loss: 0.00780188012868166\n",
      "train loss: 0.005922467913478613\n",
      "train loss: 0.006377714220434427\n",
      "train loss: 0.007114427629858255\n",
      "train loss: 0.006711476016789675\n",
      "train loss: 0.007723924703896046\n",
      "train loss: 0.005572187714278698\n",
      "train loss: 0.00541923101991415\n",
      "train loss: 0.006626775022596121\n",
      "train loss: 0.0060298326425254345\n",
      "train loss: 0.006946248468011618\n",
      "train loss: 0.006821862887591124\n",
      "train loss: 0.005308427847921848\n",
      "train loss: 0.005594856105744839\n",
      "train loss: 0.006412234157323837\n",
      "train loss: 0.006036091595888138\n",
      "train loss: 0.00552993081510067\n",
      "train loss: 0.0052221473306417465\n",
      "train loss: 0.006563045550137758\n",
      "train loss: 0.005840201862156391\n",
      "train loss: 0.00610697316005826\n",
      "train loss: 0.005055555142462254\n",
      "train loss: 0.006191983353346586\n",
      "train loss: 0.0044948263093829155\n",
      "train loss: 0.0053309123031795025\n",
      "train loss: 0.005758750252425671\n",
      "Test loss: 0.005775203462690115\n",
      "Test loss: 0.006166783161461353\n",
      "Test loss: 0.006665111053735018\n",
      "Test loss: 0.00659437058493495\n",
      "Test loss: 0.006295269820839167\n",
      "Test loss: 0.005561401601880789\n",
      "Test loss: 0.006868792697787285\n",
      "Test loss: 0.005493154749274254\n",
      "Test loss: 0.005129951983690262\n",
      "Test loss: 0.006143637467175722\n",
      "Test loss: 0.005694219376891851\n",
      "Test loss: 0.006817422341555357\n",
      "Test loss: 0.006407538428902626\n",
      "Test loss: 0.00647002225741744\n",
      "Test loss: 0.005593885667622089\n",
      "Test loss: 0.006196573842316866\n",
      "Test loss: 0.008013193495571613\n",
      "Test loss: 0.0052433074451982975\n",
      "Test loss: 0.006521421950310469\n",
      "Test loss: 0.00591939315199852\n",
      "Test loss: 0.007692249957472086\n",
      "Test loss: 0.005642511881887913\n",
      "Test loss: 0.0067462008446455\n",
      "Test loss: 0.006059238221496344\n",
      "Test loss: 0.007966753095388412\n",
      "Test loss: 0.006187646649777889\n",
      "Test loss: 0.007253312040120363\n",
      "Test loss: 0.006699522491544485\n",
      "Test loss: 0.005445876158773899\n",
      "Test loss: 0.006114618852734566\n",
      "Test loss: 0.005608356557786465\n",
      "Test: 47, loss: 0.006289901677519083\n",
      "Train epoch: 48\n",
      "train loss: 0.005298226606100798\n",
      "train loss: 0.006061146967113018\n",
      "train loss: 0.00497310608625412\n",
      "train loss: 0.0075954413041472435\n",
      "train loss: 0.006419315002858639\n",
      "train loss: 0.004934949334710836\n",
      "train loss: 0.005614614114165306\n",
      "train loss: 0.007374993525445461\n",
      "train loss: 0.0064054690301418304\n",
      "train loss: 0.005244274158030748\n",
      "train loss: 0.005880645010620356\n",
      "train loss: 0.0055539486929774284\n",
      "train loss: 0.006597642786800861\n",
      "train loss: 0.0054800845682621\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train loss: 0.004960918799042702\n",
      "train loss: 0.006023683585226536\n",
      "train loss: 0.005235166288912296\n",
      "train loss: 0.005970697849988937\n",
      "train loss: 0.004505226854234934\n",
      "train loss: 0.00603155791759491\n",
      "train loss: 0.006038228515535593\n",
      "train loss: 0.006145105231553316\n",
      "train loss: 0.006520228926092386\n",
      "train loss: 0.0049859448336064816\n",
      "train loss: 0.005176258739084005\n",
      "train loss: 0.006125892046838999\n",
      "train loss: 0.005856992211192846\n",
      "train loss: 0.005834012757986784\n",
      "train loss: 0.005929422564804554\n",
      "train loss: 0.005634748842567205\n",
      "train loss: 0.004346650093793869\n",
      "Test loss: 0.004821218550205231\n",
      "Test loss: 0.006283246446400881\n",
      "Test loss: 0.006210365332663059\n",
      "Test loss: 0.005900388117879629\n",
      "Test loss: 0.0060539525002241135\n",
      "Test loss: 0.0066900174133479595\n",
      "Test loss: 0.005890707951039076\n",
      "Test loss: 0.005456237122416496\n",
      "Test loss: 0.00490589952096343\n",
      "Test loss: 0.005561914294958115\n",
      "Test loss: 0.004988784436136484\n",
      "Test loss: 0.004215448163449764\n",
      "Test loss: 0.0048723286017775536\n",
      "Test loss: 0.004578131716698408\n",
      "Test loss: 0.004346128087490797\n",
      "Test loss: 0.005191259551793337\n",
      "Test loss: 0.004501756280660629\n",
      "Test loss: 0.005891922861337662\n",
      "Test loss: 0.0054431795142591\n",
      "Test loss: 0.004597758874297142\n",
      "Test loss: 0.005095772445201874\n",
      "Test loss: 0.007206991780549288\n",
      "Test loss: 0.00512893358245492\n",
      "Test loss: 0.005463763605803251\n",
      "Test loss: 0.005060774274170399\n",
      "Test loss: 0.005088779143989086\n",
      "Test loss: 0.0061731417663395405\n",
      "Test loss: 0.005535649135708809\n",
      "Test loss: 0.005196712911128998\n",
      "Test loss: 0.005171018652617931\n",
      "Test loss: 0.005350807216018438\n",
      "Test: 48, loss: 0.005382999312132597\n",
      "Train epoch: 49\n",
      "train loss: 0.006236660294234753\n",
      "train loss: 0.006035293452441692\n",
      "train loss: 0.0054338239133358\n",
      "train loss: 0.005817188881337643\n",
      "train loss: 0.0059042442589998245\n",
      "train loss: 0.004873174242675304\n",
      "train loss: 0.005641550291329622\n",
      "train loss: 0.007321215700358152\n",
      "train loss: 0.005055767018347979\n",
      "train loss: 0.00556557672098279\n",
      "train loss: 0.00690876692533493\n",
      "train loss: 0.00618203729391098\n",
      "train loss: 0.00417323550209403\n",
      "train loss: 0.006667133420705795\n",
      "train loss: 0.005686769727617502\n",
      "train loss: 0.0056023988872766495\n",
      "train loss: 0.0058952257968485355\n",
      "train loss: 0.0070578791201114655\n",
      "train loss: 0.006236301269382238\n",
      "train loss: 0.005012973211705685\n",
      "train loss: 0.006587653420865536\n",
      "train loss: 0.005840921308845282\n",
      "train loss: 0.005986406002193689\n",
      "train loss: 0.0061302464455366135\n",
      "train loss: 0.006173756439238787\n",
      "train loss: 0.005475564859807491\n",
      "train loss: 0.006703225430101156\n",
      "train loss: 0.0057062129490077496\n",
      "train loss: 0.005504074972122908\n",
      "train loss: 0.005000862292945385\n",
      "train loss: 0.005006789695471525\n",
      "Test loss: 0.005291348788887262\n",
      "Test loss: 0.005076928995549679\n",
      "Test loss: 0.005338813178241253\n",
      "Test loss: 0.005288147833198309\n",
      "Test loss: 0.0057677896693348885\n",
      "Test loss: 0.005042423959821463\n",
      "Test loss: 0.005084089003503323\n",
      "Test loss: 0.0071830302476882935\n",
      "Test loss: 0.006061533000320196\n",
      "Test loss: 0.00728157302364707\n",
      "Test loss: 0.005944631062448025\n",
      "Test loss: 0.006500903517007828\n",
      "Test loss: 0.0053785028867423534\n",
      "Test loss: 0.00697516743093729\n",
      "Test loss: 0.005500185769051313\n",
      "Test loss: 0.006701492238789797\n",
      "Test loss: 0.0055543603375554085\n",
      "Test loss: 0.006866275332868099\n",
      "Test loss: 0.005261174403131008\n",
      "Test loss: 0.00554236164316535\n",
      "Test loss: 0.0058988225646317005\n",
      "Test loss: 0.005047924350947142\n",
      "Test loss: 0.0052592698484659195\n",
      "Test loss: 0.005045117344707251\n",
      "Test loss: 0.006164678372442722\n",
      "Test loss: 0.005356321111321449\n",
      "Test loss: 0.0062286569736897945\n",
      "Test loss: 0.005227174609899521\n",
      "Test loss: 0.005388293415307999\n",
      "Test loss: 0.00552928913384676\n",
      "Test loss: 0.006219759117811918\n",
      "Test: 49, loss: 0.005774389021098614\n",
      "Train epoch: 50\n",
      "train loss: 0.006004260387271643\n",
      "train loss: 0.008333857171237469\n",
      "train loss: 0.00647731963545084\n",
      "train loss: 0.00589168444275856\n",
      "train loss: 0.005027136765420437\n",
      "train loss: 0.0060561653226614\n",
      "train loss: 0.004860866814851761\n",
      "train loss: 0.004824697971343994\n",
      "train loss: 0.004068414214998484\n",
      "train loss: 0.007530029863119125\n",
      "train loss: 0.004107033368200064\n",
      "train loss: 0.005106505937874317\n",
      "train loss: 0.00396912032738328\n",
      "train loss: 0.006088262423872948\n",
      "train loss: 0.004442770034074783\n",
      "train loss: 0.004321679938584566\n",
      "train loss: 0.004197434987872839\n",
      "train loss: 0.004841062240302563\n",
      "train loss: 0.004779122769832611\n",
      "train loss: 0.0041299303993582726\n",
      "train loss: 0.004953434225171804\n",
      "train loss: 0.004039091989398003\n",
      "train loss: 0.007074498105794191\n",
      "train loss: 0.005398430861532688\n",
      "train loss: 0.004527823068201542\n",
      "train loss: 0.00406509218737483\n",
      "train loss: 0.004498730413615704\n",
      "train loss: 0.00412529893219471\n",
      "train loss: 0.0036608416121453047\n",
      "train loss: 0.00480997608974576\n",
      "train loss: 0.004559945315122604\n",
      "Test loss: 0.0053218649700284\n",
      "Test loss: 0.004950638860464096\n",
      "Test loss: 0.004020105581730604\n",
      "Test loss: 0.0044036926701664925\n",
      "Test loss: 0.0037452203687280416\n",
      "Test loss: 0.005663693882524967\n",
      "Test loss: 0.004752591252326965\n",
      "Test loss: 0.004973266739398241\n",
      "Test loss: 0.004338699858635664\n",
      "Test loss: 0.004186070524156094\n",
      "Test loss: 0.0037531396374106407\n",
      "Test loss: 0.00452745845541358\n",
      "Test loss: 0.004697899799793959\n",
      "Test loss: 0.004001436289399862\n",
      "Test loss: 0.005956853739917278\n",
      "Test loss: 0.004356569144874811\n",
      "Test loss: 0.004226109478622675\n",
      "Test loss: 0.004088126588612795\n",
      "Test loss: 0.004797655623406172\n",
      "Test loss: 0.003943442367017269\n",
      "Test loss: 0.004229061305522919\n",
      "Test loss: 0.004844311624765396\n",
      "Test loss: 0.003830614499747753\n",
      "Test loss: 0.0044918665662407875\n",
      "Test loss: 0.004925855901092291\n",
      "Test loss: 0.004651943687349558\n",
      "Test loss: 0.00407533161342144\n",
      "Test loss: 0.003957160282880068\n",
      "Test loss: 0.004293973557651043\n",
      "Test loss: 0.004361162893474102\n",
      "Test loss: 0.00622521573677659\n",
      "Test: 50, loss: 0.004535194952040911\n",
      "Train epoch: 51\n",
      "train loss: 0.0038996171206235886\n",
      "train loss: 0.0042174807749688625\n",
      "train loss: 0.005215718876570463\n",
      "train loss: 0.004049669951200485\n",
      "train loss: 0.00445218663662672\n",
      "train loss: 0.0042730169370770454\n",
      "train loss: 0.0038304091431200504\n",
      "train loss: 0.004429813474416733\n",
      "train loss: 0.004038793500512838\n",
      "train loss: 0.00454800296574831\n",
      "train loss: 0.003966443706303835\n",
      "train loss: 0.004657757002860308\n",
      "train loss: 0.00580322602763772\n",
      "train loss: 0.00349142262712121\n",
      "train loss: 0.004665895830839872\n",
      "train loss: 0.005189993418753147\n",
      "train loss: 0.0037903490010648966\n",
      "train loss: 0.004344575572758913\n",
      "train loss: 0.004817203618586063\n",
      "train loss: 0.006132792681455612\n",
      "train loss: 0.004513743333518505\n",
      "train loss: 0.004343522246927023\n",
      "train loss: 0.00395193649455905\n",
      "train loss: 0.004636032972484827\n",
      "train loss: 0.0034848544746637344\n",
      "train loss: 0.0045144869945943356\n",
      "train loss: 0.004427120555192232\n",
      "train loss: 0.00511369202286005\n",
      "train loss: 0.0038543823175132275\n",
      "train loss: 0.004314261022955179\n",
      "train loss: 0.004140721168369055\n",
      "Test loss: 0.0035909023135900497\n",
      "Test loss: 0.004296735394746065\n",
      "Test loss: 0.005106877535581589\n",
      "Test loss: 0.0036988735664635897\n",
      "Test loss: 0.0035571439657360315\n",
      "Test loss: 0.004143617581576109\n",
      "Test loss: 0.004691907204687595\n",
      "Test loss: 0.004841420333832502\n",
      "Test loss: 0.004970056004822254\n",
      "Test loss: 0.003614494577050209\n",
      "Test loss: 0.005863913334906101\n",
      "Test loss: 0.0035830321721732616\n",
      "Test loss: 0.003831462236121297\n",
      "Test loss: 0.003988566342741251\n",
      "Test loss: 0.0040058265440166\n",
      "Test loss: 0.004207257647067308\n",
      "Test loss: 0.004275158513337374\n",
      "Test loss: 0.0036901296116411686\n",
      "Test loss: 0.006294570863246918\n",
      "Test loss: 0.003781481646001339\n",
      "Test loss: 0.004600328393280506\n",
      "Test loss: 0.005092427134513855\n",
      "Test loss: 0.003925154451280832\n",
      "Test loss: 0.00459900451824069\n",
      "Test loss: 0.004191301297396421\n",
      "Test loss: 0.003573554800823331\n",
      "Test loss: 0.004981812089681625\n",
      "Test loss: 0.0041936179623007774\n",
      "Test loss: 0.004408947192132473\n",
      "Test loss: 0.003518040059134364\n",
      "Test loss: 0.004689013119786978\n",
      "Test: 51, loss: 0.004316342528909445\n",
      "Train epoch: 52\n",
      "train loss: 0.0037108955439180136\n",
      "train loss: 0.004731722176074982\n",
      "train loss: 0.00439877063035965\n",
      "train loss: 0.0043291919864714146\n",
      "train loss: 0.005720552522689104\n",
      "train loss: 0.0041233934462070465\n",
      "train loss: 0.005844266153872013\n",
      "train loss: 0.004449967760592699\n",
      "train loss: 0.004821346141397953\n",
      "train loss: 0.004323436878621578\n",
      "train loss: 0.004255418200045824\n",
      "train loss: 0.0047314041294157505\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train loss: 0.0047834813594818115\n",
      "train loss: 0.005143065936863422\n",
      "train loss: 0.0039316629990935326\n",
      "train loss: 0.003921871539205313\n",
      "train loss: 0.005540701560676098\n",
      "train loss: 0.0041940463706851006\n",
      "train loss: 0.003593452973291278\n",
      "train loss: 0.0063250442035496235\n",
      "train loss: 0.004807496443390846\n",
      "train loss: 0.00448449794203043\n",
      "train loss: 0.0037640470545738935\n",
      "train loss: 0.005013321526348591\n",
      "train loss: 0.00502997450530529\n",
      "train loss: 0.00442543625831604\n",
      "train loss: 0.003826381405815482\n",
      "train loss: 0.004046151414513588\n",
      "train loss: 0.0035782528575509787\n",
      "train loss: 0.004258526489138603\n",
      "train loss: 0.006255675107240677\n",
      "Test loss: 0.004539660178124905\n",
      "Test loss: 0.003535729367285967\n",
      "Test loss: 0.003851494751870632\n",
      "Test loss: 0.004286901094019413\n",
      "Test loss: 0.0050828102976083755\n",
      "Test loss: 0.0051887910813093185\n",
      "Test loss: 0.004707669839262962\n",
      "Test loss: 0.003896412206813693\n",
      "Test loss: 0.0035834324080497026\n",
      "Test loss: 0.003964180126786232\n",
      "Test loss: 0.004523875191807747\n",
      "Test loss: 0.004129980690777302\n",
      "Test loss: 0.004174939822405577\n",
      "Test loss: 0.004942602477967739\n",
      "Test loss: 0.005361279938369989\n",
      "Test loss: 0.0033411188051104546\n",
      "Test loss: 0.0051813265308737755\n",
      "Test loss: 0.0044647869653999805\n",
      "Test loss: 0.004218738991767168\n",
      "Test loss: 0.00580877810716629\n",
      "Test loss: 0.0039442721754312515\n",
      "Test loss: 0.005778786726295948\n",
      "Test loss: 0.004910354968160391\n",
      "Test loss: 0.005080284550786018\n",
      "Test loss: 0.003929737955331802\n",
      "Test loss: 0.003824970219284296\n",
      "Test loss: 0.0048692235723137856\n",
      "Test loss: 0.003940983675420284\n",
      "Test loss: 0.004736666567623615\n",
      "Test loss: 0.005245990585535765\n",
      "Test loss: 0.003512627212330699\n",
      "Test: 52, loss: 0.004469625186175108\n",
      "Train epoch: 53\n",
      "train loss: 0.004151689354330301\n",
      "train loss: 0.004641230683773756\n",
      "train loss: 0.004910200834274292\n",
      "train loss: 0.004585386719554663\n",
      "train loss: 0.00395930465310812\n",
      "train loss: 0.004196654539555311\n",
      "train loss: 0.005172278266400099\n",
      "train loss: 0.003946208860725164\n",
      "train loss: 0.003914488013833761\n",
      "train loss: 0.004660200327634811\n",
      "train loss: 0.003529246663674712\n",
      "train loss: 0.0046128095127642155\n",
      "train loss: 0.0044585177674889565\n",
      "train loss: 0.003114752471446991\n",
      "train loss: 0.003804717445746064\n",
      "train loss: 0.004730928689241409\n",
      "train loss: 0.0042879097163677216\n",
      "train loss: 0.004641979932785034\n",
      "train loss: 0.004806408658623695\n",
      "train loss: 0.0038551772013306618\n",
      "train loss: 0.004122291225939989\n",
      "train loss: 0.003994827624410391\n",
      "train loss: 0.004328414797782898\n",
      "train loss: 0.005231544841080904\n",
      "train loss: 0.0034495957661420107\n",
      "train loss: 0.004916247446089983\n",
      "train loss: 0.004213128238916397\n",
      "train loss: 0.004497619345784187\n",
      "train loss: 0.00406057620421052\n",
      "train loss: 0.004065347835421562\n",
      "train loss: 0.005382230039685965\n",
      "Test loss: 0.004678196739405394\n",
      "Test loss: 0.0047234902158379555\n",
      "Test loss: 0.004093070514500141\n",
      "Test loss: 0.005270913243293762\n",
      "Test loss: 0.004947125446051359\n",
      "Test loss: 0.003772485302761197\n",
      "Test loss: 0.0037892048712819815\n",
      "Test loss: 0.00427202507853508\n",
      "Test loss: 0.0036033475771546364\n",
      "Test loss: 0.0044602579437196255\n",
      "Test loss: 0.003962268121540546\n",
      "Test loss: 0.004102371633052826\n",
      "Test loss: 0.004443848971277475\n",
      "Test loss: 0.004020423628389835\n",
      "Test loss: 0.005114932078868151\n",
      "Test loss: 0.004698870237916708\n",
      "Test loss: 0.004068715963512659\n",
      "Test loss: 0.004600443411618471\n",
      "Test loss: 0.004394343588501215\n",
      "Test loss: 0.004530561156570911\n",
      "Test loss: 0.0035126160364598036\n",
      "Test loss: 0.004975566174834967\n",
      "Test loss: 0.004136727657169104\n",
      "Test loss: 0.0048109558410942554\n",
      "Test loss: 0.00383369205519557\n",
      "Test loss: 0.005182438530027866\n",
      "Test loss: 0.005349914077669382\n",
      "Test loss: 0.004737310577183962\n",
      "Test loss: 0.004697434138506651\n",
      "Test loss: 0.004217699635773897\n",
      "Test loss: 0.003465320449322462\n",
      "Test: 53, loss: 0.004402147140353918\n",
      "Train epoch: 54\n",
      "train loss: 0.0049176644533872604\n",
      "train loss: 0.004580403678119183\n",
      "train loss: 0.004031300079077482\n",
      "train loss: 0.003939762245863676\n",
      "train loss: 0.0041115861386060715\n",
      "train loss: 0.0036784112453460693\n",
      "train loss: 0.0041046226397156715\n",
      "train loss: 0.004406830761581659\n",
      "train loss: 0.004261434078216553\n",
      "train loss: 0.00341134169138968\n",
      "train loss: 0.005368643440306187\n",
      "train loss: 0.003754714271053672\n",
      "train loss: 0.004190473817288876\n",
      "train loss: 0.004542587790638208\n",
      "train loss: 0.0041253287345170975\n",
      "train loss: 0.0038453307934105396\n",
      "train loss: 0.005251001566648483\n",
      "train loss: 0.00422144727781415\n",
      "train loss: 0.004714801907539368\n",
      "train loss: 0.003982191439718008\n",
      "train loss: 0.004758836701512337\n",
      "train loss: 0.0036612977273762226\n",
      "train loss: 0.004030684940516949\n",
      "train loss: 0.004075268749147654\n",
      "train loss: 0.0035382069181650877\n",
      "train loss: 0.005592669826000929\n",
      "train loss: 0.003760281018912792\n",
      "train loss: 0.004108271095901728\n",
      "train loss: 0.003708286676555872\n",
      "train loss: 0.00566828902810812\n",
      "train loss: 0.005019834265112877\n",
      "Test loss: 0.004300619941204786\n",
      "Test loss: 0.004580559208989143\n",
      "Test loss: 0.004170347936451435\n",
      "Test loss: 0.004310145974159241\n",
      "Test loss: 0.004575221799314022\n",
      "Test loss: 0.004391181748360395\n",
      "Test loss: 0.004795788321644068\n",
      "Test loss: 0.005710026249289513\n",
      "Test loss: 0.005262455902993679\n",
      "Test loss: 0.003993616439402103\n",
      "Test loss: 0.004043577704578638\n",
      "Test loss: 0.005202076397836208\n",
      "Test loss: 0.0038361861370503902\n",
      "Test loss: 0.005307708866894245\n",
      "Test loss: 0.0052062394097447395\n",
      "Test loss: 0.004386796150356531\n",
      "Test loss: 0.004379102494567633\n",
      "Test loss: 0.004176081623882055\n",
      "Test loss: 0.0036772999446839094\n",
      "Test loss: 0.003847541054710746\n",
      "Test loss: 0.0051729194819927216\n",
      "Test loss: 0.004506769590079784\n",
      "Test loss: 0.0044899494387209415\n",
      "Test loss: 0.0038212721701711416\n",
      "Test loss: 0.005046232603490353\n",
      "Test loss: 0.004543102812021971\n",
      "Test loss: 0.004519715439528227\n",
      "Test loss: 0.0038076555356383324\n",
      "Test loss: 0.00474132364615798\n",
      "Test loss: 0.005220124032348394\n",
      "Test loss: 0.004543847870081663\n",
      "Test: 54, loss: 0.004534370265901089\n",
      "Train epoch: 55\n",
      "train loss: 0.003211052855476737\n",
      "train loss: 0.005023120436817408\n",
      "train loss: 0.003732918994501233\n",
      "train loss: 0.004512947518378496\n",
      "train loss: 0.004228590987622738\n",
      "train loss: 0.005142704583704472\n",
      "train loss: 0.00401780940592289\n",
      "train loss: 0.003926584962755442\n",
      "train loss: 0.003705807263031602\n",
      "train loss: 0.00359958503395319\n",
      "train loss: 0.004360886290669441\n",
      "train loss: 0.003929048776626587\n",
      "train loss: 0.004047898575663567\n",
      "train loss: 0.004694330971688032\n",
      "train loss: 0.0040759434923529625\n",
      "train loss: 0.004153333138674498\n",
      "train loss: 0.0032895670738071203\n",
      "train loss: 0.005190642084926367\n",
      "train loss: 0.0035175124648958445\n",
      "train loss: 0.004354126285761595\n",
      "train loss: 0.0035330287646502256\n",
      "train loss: 0.004202589392662048\n",
      "train loss: 0.0040232837200164795\n",
      "train loss: 0.0045195333659648895\n",
      "train loss: 0.003604558063670993\n",
      "train loss: 0.005159775726497173\n",
      "train loss: 0.004919032100588083\n",
      "train loss: 0.004584119189530611\n",
      "train loss: 0.004019534680992365\n",
      "train loss: 0.00420819828286767\n",
      "train loss: 0.0038992715999484062\n",
      "Test loss: 0.005335139576345682\n",
      "Test loss: 0.004665545653551817\n",
      "Test loss: 0.004349607974290848\n",
      "Test loss: 0.00418505072593689\n",
      "Test loss: 0.004963075742125511\n",
      "Test loss: 0.004424826242029667\n",
      "Test loss: 0.0032223439775407314\n",
      "Test loss: 0.004180500283837318\n",
      "Test loss: 0.003796784207224846\n",
      "Test loss: 0.004676565062254667\n",
      "Test loss: 0.004149667453020811\n",
      "Test loss: 0.004545942880213261\n",
      "Test loss: 0.004389772657305002\n",
      "Test loss: 0.0055202883668243885\n",
      "Test loss: 0.004673399496823549\n",
      "Test loss: 0.004655999597162008\n",
      "Test loss: 0.0046286508440971375\n",
      "Test loss: 0.00376236904412508\n",
      "Test loss: 0.003996057901531458\n",
      "Test loss: 0.005063510499894619\n",
      "Test loss: 0.0037052910774946213\n",
      "Test loss: 0.004840195644646883\n",
      "Test loss: 0.0033063001465052366\n",
      "Test loss: 0.003732517594471574\n",
      "Test loss: 0.00421565305441618\n",
      "Test loss: 0.004608303774148226\n",
      "Test loss: 0.004307842347770929\n",
      "Test loss: 0.0040684835985302925\n",
      "Test loss: 0.0041307141073048115\n",
      "Test loss: 0.0031705601140856743\n",
      "Test loss: 0.004458066541701555\n",
      "Test: 55, loss: 0.004313840065151453\n",
      "Train epoch: 56\n",
      "train loss: 0.0040261405520141125\n",
      "train loss: 0.003547164611518383\n",
      "train loss: 0.005349928047508001\n",
      "train loss: 0.004299416672438383\n",
      "train loss: 0.0034044142812490463\n",
      "train loss: 0.005507138557732105\n",
      "train loss: 0.003527246881276369\n",
      "train loss: 0.004275265149772167\n",
      "train loss: 0.004916215781122446\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train loss: 0.003980686888098717\n",
      "train loss: 0.0052066645585000515\n",
      "train loss: 0.006163208279758692\n",
      "train loss: 0.00438696751371026\n",
      "train loss: 0.004843531176447868\n",
      "train loss: 0.0038890272844582796\n",
      "train loss: 0.0049445126205682755\n",
      "train loss: 0.005250873044133186\n",
      "train loss: 0.003724797163158655\n",
      "train loss: 0.005006090737879276\n",
      "train loss: 0.003854127135127783\n",
      "train loss: 0.004685407504439354\n",
      "train loss: 0.0035612969659268856\n",
      "train loss: 0.004064892418682575\n",
      "train loss: 0.00379953277297318\n",
      "train loss: 0.0036181786563247442\n",
      "train loss: 0.004587670788168907\n",
      "train loss: 0.004356806166470051\n",
      "train loss: 0.004125804174691439\n",
      "train loss: 0.0038755813147872686\n",
      "train loss: 0.003436486469581723\n",
      "train loss: 0.004410731606185436\n",
      "Test loss: 0.005467098206281662\n",
      "Test loss: 0.0037888865917921066\n",
      "Test loss: 0.0048662046901881695\n",
      "Test loss: 0.004049465525895357\n",
      "Test loss: 0.005566614679992199\n",
      "Test loss: 0.004433800000697374\n",
      "Test loss: 0.004078216385096312\n",
      "Test loss: 0.0031075843144208193\n",
      "Test loss: 0.005140695255249739\n",
      "Test loss: 0.005620288196951151\n",
      "Test loss: 0.0037356694228947163\n",
      "Test loss: 0.0042838421650230885\n",
      "Test loss: 0.003668135032057762\n",
      "Test loss: 0.005069132894277573\n",
      "Test loss: 0.00421113008633256\n",
      "Test loss: 0.0032840745989233255\n",
      "Test loss: 0.004754409193992615\n",
      "Test loss: 0.004430496599525213\n",
      "Test loss: 0.005119235720485449\n",
      "Test loss: 0.004829770419746637\n",
      "Test loss: 0.005293270107358694\n",
      "Test loss: 0.003418685868382454\n",
      "Test loss: 0.004358816426247358\n",
      "Test loss: 0.004696746356785297\n",
      "Test loss: 0.0043484000489115715\n",
      "Test loss: 0.00451277382671833\n",
      "Test loss: 0.0038230856880545616\n",
      "Test loss: 0.00548880361020565\n",
      "Test loss: 0.0044574858620762825\n",
      "Test loss: 0.003576823975890875\n",
      "Test loss: 0.003924461547285318\n",
      "Test: 56, loss: 0.004432390443980694\n",
      "Train epoch: 57\n",
      "train loss: 0.0041099293157458305\n",
      "train loss: 0.0038494595792144537\n",
      "train loss: 0.004227971658110619\n",
      "train loss: 0.0038443414960056543\n",
      "train loss: 0.004019762855023146\n",
      "train loss: 0.003899191739037633\n",
      "train loss: 0.005050053820014\n",
      "train loss: 0.004578220192342997\n",
      "train loss: 0.0034316077362746\n",
      "train loss: 0.004373535048216581\n",
      "train loss: 0.004004864487797022\n",
      "train loss: 0.005450367461889982\n",
      "train loss: 0.004979334771633148\n",
      "train loss: 0.0042801061645150185\n",
      "train loss: 0.0041876379400491714\n",
      "train loss: 0.004716394003480673\n",
      "train loss: 0.004459785297513008\n",
      "train loss: 0.004502834286540747\n",
      "train loss: 0.004228347446769476\n",
      "train loss: 0.004811449442058802\n",
      "train loss: 0.0037064040079712868\n",
      "train loss: 0.004055012483149767\n",
      "train loss: 0.004716955125331879\n",
      "train loss: 0.003953096456825733\n",
      "train loss: 0.0034643348772078753\n",
      "train loss: 0.00416890624910593\n",
      "train loss: 0.004335846286267042\n",
      "train loss: 0.004491679836064577\n",
      "train loss: 0.002948060166090727\n",
      "train loss: 0.0035622413270175457\n",
      "train loss: 0.004651371389627457\n",
      "Test loss: 0.003762295935302973\n",
      "Test loss: 0.00369841861538589\n",
      "Test loss: 0.003642949741333723\n",
      "Test loss: 0.003549038665369153\n",
      "Test loss: 0.0033369252923876047\n",
      "Test loss: 0.0041941446252167225\n",
      "Test loss: 0.004583978094160557\n",
      "Test loss: 0.004916975740343332\n",
      "Test loss: 0.003455108031630516\n",
      "Test loss: 0.003921924624592066\n",
      "Test loss: 0.00423744972795248\n",
      "Test loss: 0.0035609700717031956\n",
      "Test loss: 0.004958040546625853\n",
      "Test loss: 0.0033450510818511248\n",
      "Test loss: 0.003352551721036434\n",
      "Test loss: 0.00437169848009944\n",
      "Test loss: 0.0038243825547397137\n",
      "Test loss: 0.0049710045568645\n",
      "Test loss: 0.004408441949635744\n",
      "Test loss: 0.0038562859408557415\n",
      "Test loss: 0.0037506893277168274\n",
      "Test loss: 0.005073122680187225\n",
      "Test loss: 0.0030021434649825096\n",
      "Test loss: 0.003517066827043891\n",
      "Test loss: 0.0036426635924726725\n",
      "Test loss: 0.005431083962321281\n",
      "Test loss: 0.0038606601301580667\n",
      "Test loss: 0.00563297513872385\n",
      "Test loss: 0.004082574509084225\n",
      "Test loss: 0.004980362951755524\n",
      "Test loss: 0.004033820237964392\n",
      "Test: 57, loss: 0.004095316398888826\n",
      "Train epoch: 58\n",
      "train loss: 0.0038492826279252768\n",
      "train loss: 0.0038764220662415028\n",
      "train loss: 0.0036490997299551964\n",
      "train loss: 0.0037176618352532387\n",
      "train loss: 0.004286294803023338\n",
      "train loss: 0.00396464578807354\n",
      "train loss: 0.004979366902261972\n",
      "train loss: 0.004209091421216726\n",
      "train loss: 0.0035698357969522476\n",
      "train loss: 0.004155028145760298\n",
      "train loss: 0.005050207022577524\n",
      "train loss: 0.004436860326677561\n",
      "train loss: 0.004630285780876875\n",
      "train loss: 0.0037681504618376493\n",
      "train loss: 0.0036157905124127865\n",
      "train loss: 0.004087931476533413\n",
      "train loss: 0.004934356547892094\n",
      "train loss: 0.004576731473207474\n",
      "train loss: 0.00381108233705163\n",
      "train loss: 0.005039709620177746\n",
      "train loss: 0.004145725630223751\n",
      "train loss: 0.0041865152306854725\n",
      "train loss: 0.0032559747342020273\n",
      "train loss: 0.004147709347307682\n",
      "train loss: 0.004424866288900375\n",
      "train loss: 0.003680667607113719\n",
      "train loss: 0.0050811185501515865\n",
      "train loss: 0.004268616437911987\n",
      "train loss: 0.004328429698944092\n",
      "train loss: 0.004386167507618666\n",
      "train loss: 0.00394826103001833\n",
      "Test loss: 0.004694758448749781\n",
      "Test loss: 0.004874933045357466\n",
      "Test loss: 0.004430653993040323\n",
      "Test loss: 0.005007805768400431\n",
      "Test loss: 0.004471783991903067\n",
      "Test loss: 0.0034733335487544537\n",
      "Test loss: 0.0037859687581658363\n",
      "Test loss: 0.004645275417715311\n",
      "Test loss: 0.006489253137260675\n",
      "Test loss: 0.005676628556102514\n",
      "Test loss: 0.003859586315229535\n",
      "Test loss: 0.0031881676986813545\n",
      "Test loss: 0.004804216790944338\n",
      "Test loss: 0.005141472909599543\n",
      "Test loss: 0.004757882561534643\n",
      "Test loss: 0.0043329522013664246\n",
      "Test loss: 0.003950710874050856\n",
      "Test loss: 0.00418056221678853\n",
      "Test loss: 0.003745576599612832\n",
      "Test loss: 0.003474125172942877\n",
      "Test loss: 0.0033665618393570185\n",
      "Test loss: 0.003766514826565981\n",
      "Test loss: 0.004073264542967081\n",
      "Test loss: 0.004735256545245647\n",
      "Test loss: 0.005297285504639149\n",
      "Test loss: 0.0032967787701636553\n",
      "Test loss: 0.004607176873832941\n",
      "Test loss: 0.004706228151917458\n",
      "Test loss: 0.004104343242943287\n",
      "Test loss: 0.0037240255624055862\n",
      "Test loss: 0.0041381292976439\n",
      "Test: 58, loss: 0.004348426125943661\n",
      "Train epoch: 59\n",
      "train loss: 0.004649613983929157\n",
      "train loss: 0.004918881691992283\n",
      "train loss: 0.003980642184615135\n",
      "train loss: 0.0046413373202085495\n",
      "train loss: 0.003460669657215476\n",
      "train loss: 0.005136413499712944\n",
      "train loss: 0.0037802141159772873\n",
      "train loss: 0.00423743762075901\n",
      "train loss: 0.003779592690989375\n",
      "train loss: 0.004021095111966133\n",
      "train loss: 0.004795474000275135\n",
      "train loss: 0.004007827490568161\n",
      "train loss: 0.004140407312661409\n",
      "train loss: 0.003169842530041933\n",
      "train loss: 0.003989533521234989\n",
      "train loss: 0.004105197731405497\n",
      "train loss: 0.004473336506634951\n",
      "train loss: 0.0030130622908473015\n",
      "train loss: 0.004134703427553177\n",
      "train loss: 0.003343792399391532\n",
      "train loss: 0.00406649149954319\n",
      "train loss: 0.004108051303774118\n",
      "train loss: 0.004825616255402565\n",
      "train loss: 0.00400445144623518\n",
      "train loss: 0.003288011997938156\n",
      "train loss: 0.00380796380341053\n",
      "train loss: 0.00436479551717639\n",
      "train loss: 0.005284104961901903\n",
      "train loss: 0.004845012910664082\n",
      "train loss: 0.003837516298517585\n",
      "train loss: 0.003906607627868652\n",
      "Test loss: 0.003997067920863628\n",
      "Test loss: 0.004495862405747175\n",
      "Test loss: 0.004261632449924946\n",
      "Test loss: 0.004237339831888676\n",
      "Test loss: 0.0029090819880366325\n",
      "Test loss: 0.003898054827004671\n",
      "Test loss: 0.0043223220854997635\n",
      "Test loss: 0.004325388465076685\n",
      "Test loss: 0.004028148017823696\n",
      "Test loss: 0.004593170713633299\n",
      "Test loss: 0.004820395261049271\n",
      "Test loss: 0.003618288319557905\n",
      "Test loss: 0.00401077838614583\n",
      "Test loss: 0.004160996060818434\n",
      "Test loss: 0.004136547911912203\n",
      "Test loss: 0.003932856488972902\n",
      "Test loss: 0.003323412500321865\n",
      "Test loss: 0.003583731362596154\n",
      "Test loss: 0.003611758816987276\n",
      "Test loss: 0.004334515891969204\n",
      "Test loss: 0.0042342147789895535\n",
      "Test loss: 0.0031480835750699043\n",
      "Test loss: 0.004829862620681524\n",
      "Test loss: 0.005208415910601616\n",
      "Test loss: 0.004614314995706081\n",
      "Test loss: 0.003989502787590027\n",
      "Test loss: 0.004312966018915176\n",
      "Test loss: 0.005198548547923565\n",
      "Test loss: 0.0037788141053169966\n",
      "Test loss: 0.00478340731933713\n",
      "Test loss: 0.004560652654618025\n",
      "Test: 59, loss: 0.004169682040810585\n",
      "Train epoch: 60\n",
      "train loss: 0.0045988415367901325\n",
      "train loss: 0.004996486473828554\n",
      "train loss: 0.005028394516557455\n",
      "train loss: 0.0038216933608055115\n",
      "train loss: 0.003091338789090514\n",
      "train loss: 0.003814105177298188\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train loss: 0.004081399645656347\n",
      "train loss: 0.0036570122465491295\n",
      "train loss: 0.003923614043742418\n",
      "train loss: 0.003782568732276559\n",
      "train loss: 0.004319809377193451\n",
      "train loss: 0.0038816439919173717\n",
      "train loss: 0.004529211670160294\n",
      "train loss: 0.004370000679045916\n",
      "train loss: 0.004679519217461348\n",
      "train loss: 0.004667991306632757\n",
      "train loss: 0.0038925427943468094\n",
      "train loss: 0.004976814612746239\n",
      "train loss: 0.004630440846085548\n",
      "train loss: 0.004786061588674784\n",
      "train loss: 0.005992555059492588\n",
      "train loss: 0.0036278788466006517\n",
      "train loss: 0.003369356272742152\n",
      "train loss: 0.003958068322390318\n",
      "train loss: 0.004368561785668135\n",
      "train loss: 0.0036324334796518087\n",
      "train loss: 0.0039707934483885765\n",
      "train loss: 0.004766962956637144\n",
      "train loss: 0.004466949962079525\n",
      "train loss: 0.00401523569598794\n",
      "train loss: 0.004477162845432758\n",
      "Test loss: 0.0034545513335615396\n",
      "Test loss: 0.0036001536063849926\n",
      "Test loss: 0.0036845128051936626\n",
      "Test loss: 0.005000634118914604\n",
      "Test loss: 0.0043141902424395084\n",
      "Test loss: 0.003491361625492573\n",
      "Test loss: 0.004693243652582169\n",
      "Test loss: 0.0036171970423310995\n",
      "Test loss: 0.0038245103787630796\n",
      "Test loss: 0.0033550297375768423\n",
      "Test loss: 0.004517114721238613\n",
      "Test loss: 0.003918062895536423\n",
      "Test loss: 0.004647755064070225\n",
      "Test loss: 0.0035880357027053833\n",
      "Test loss: 0.004407616797834635\n",
      "Test loss: 0.005439295899122953\n",
      "Test loss: 0.003382720984518528\n",
      "Test loss: 0.0038932550232857466\n",
      "Test loss: 0.004229888319969177\n",
      "Test loss: 0.0036458333488553762\n",
      "Test loss: 0.00543266860768199\n",
      "Test loss: 0.004019512794911861\n",
      "Test loss: 0.00382737023755908\n",
      "Test loss: 0.0060247755609452724\n",
      "Test loss: 0.004474744200706482\n",
      "Test loss: 0.0037117968313395977\n",
      "Test loss: 0.004694096278399229\n",
      "Test loss: 0.003878172254189849\n",
      "Test loss: 0.004647029563784599\n",
      "Test loss: 0.004080703482031822\n",
      "Test loss: 0.004903174005448818\n",
      "Test: 60, loss: 0.0042064194567501545\n",
      "Train epoch: 61\n",
      "train loss: 0.00367911858484149\n",
      "train loss: 0.004099564161151648\n",
      "train loss: 0.004579069558531046\n",
      "train loss: 0.0038630645722150803\n",
      "train loss: 0.003849050495773554\n",
      "train loss: 0.004416794516146183\n",
      "train loss: 0.00478241965174675\n",
      "train loss: 0.004809129051864147\n",
      "train loss: 0.003203693311661482\n",
      "train loss: 0.004880593623965979\n",
      "train loss: 0.0034023032058030367\n",
      "train loss: 0.0036236592568457127\n",
      "train loss: 0.004713703878223896\n",
      "train loss: 0.0045608170330524445\n",
      "train loss: 0.004352073650807142\n",
      "train loss: 0.004347245208919048\n",
      "train loss: 0.005063876509666443\n",
      "train loss: 0.003734673373401165\n",
      "train loss: 0.004423138219863176\n",
      "train loss: 0.004108458757400513\n",
      "train loss: 0.004396476782858372\n",
      "train loss: 0.004595351871103048\n",
      "train loss: 0.0036792049650102854\n",
      "train loss: 0.005017057526856661\n",
      "train loss: 0.0054384563118219376\n",
      "train loss: 0.004896066151559353\n",
      "train loss: 0.004307348746806383\n",
      "train loss: 0.00388628919608891\n",
      "train loss: 0.004082537721842527\n",
      "train loss: 0.004191564861685038\n",
      "train loss: 0.003537381300702691\n",
      "Test loss: 0.00385418813675642\n",
      "Test loss: 0.0035491930320858955\n",
      "Test loss: 0.003238019533455372\n",
      "Test loss: 0.004217036999762058\n",
      "Test loss: 0.004086893051862717\n",
      "Test loss: 0.004359351471066475\n",
      "Test loss: 0.005806871224194765\n",
      "Test loss: 0.005059492774307728\n",
      "Test loss: 0.004803050309419632\n",
      "Test loss: 0.005010389722883701\n",
      "Test loss: 0.004779641050845385\n",
      "Test loss: 0.006704458501189947\n",
      "Test loss: 0.0033266025129705667\n",
      "Test loss: 0.004648166708648205\n",
      "Test loss: 0.003582137171179056\n",
      "Test loss: 0.004420338664203882\n",
      "Test loss: 0.003866171929985285\n",
      "Test loss: 0.004578010644763708\n",
      "Test loss: 0.0034882330801337957\n",
      "Test loss: 0.004401518031954765\n",
      "Test loss: 0.003594977781176567\n",
      "Test loss: 0.004105555824935436\n",
      "Test loss: 0.005558876786381006\n",
      "Test loss: 0.0051234811544418335\n",
      "Test loss: 0.004082501400262117\n",
      "Test loss: 0.004526791628450155\n",
      "Test loss: 0.0049993982538580894\n",
      "Test loss: 0.005072289612144232\n",
      "Test loss: 0.004508800804615021\n",
      "Test loss: 0.005383491516113281\n",
      "Test loss: 0.004483576398342848\n",
      "Test: 61, loss: 0.004490952473133802\n",
      "Train epoch: 62\n",
      "train loss: 0.0049808332696557045\n",
      "train loss: 0.003555776784196496\n",
      "train loss: 0.004484192468225956\n",
      "train loss: 0.004021584987640381\n",
      "train loss: 0.004088126122951508\n",
      "train loss: 0.003676119726151228\n",
      "train loss: 0.0033862185664474964\n",
      "train loss: 0.003912773448973894\n",
      "train loss: 0.004474546294659376\n",
      "train loss: 0.004074666649103165\n",
      "train loss: 0.0038555378559976816\n",
      "train loss: 0.0043107871897518635\n",
      "train loss: 0.003792820032685995\n",
      "train loss: 0.004686260130256414\n",
      "train loss: 0.005583952181041241\n",
      "train loss: 0.003971074242144823\n",
      "train loss: 0.0045921276323497295\n",
      "train loss: 0.0045899562537670135\n",
      "train loss: 0.004164574202150106\n",
      "train loss: 0.004711522255092859\n",
      "train loss: 0.004850936122238636\n",
      "train loss: 0.004286362323909998\n",
      "train loss: 0.005227926652878523\n",
      "train loss: 0.0037500537000596523\n",
      "train loss: 0.003194973338395357\n",
      "train loss: 0.004611511714756489\n",
      "train loss: 0.0036780775990337133\n",
      "train loss: 0.00508688110858202\n",
      "train loss: 0.004020183812826872\n",
      "train loss: 0.004860850516706705\n",
      "train loss: 0.0034596247132867575\n",
      "Test loss: 0.00380452792160213\n",
      "Test loss: 0.005714431405067444\n",
      "Test loss: 0.004343697801232338\n",
      "Test loss: 0.0035059810616075993\n",
      "Test loss: 0.0038742907345294952\n",
      "Test loss: 0.004759490955621004\n",
      "Test loss: 0.0055396221578121185\n",
      "Test loss: 0.004184151999652386\n",
      "Test loss: 0.004351056646555662\n",
      "Test loss: 0.00411090487614274\n",
      "Test loss: 0.004716638475656509\n",
      "Test loss: 0.0030645765364170074\n",
      "Test loss: 0.005610536318272352\n",
      "Test loss: 0.0038104418199509382\n",
      "Test loss: 0.004472263157367706\n",
      "Test loss: 0.0037147407419979572\n",
      "Test loss: 0.004029758274555206\n",
      "Test loss: 0.004107033833861351\n",
      "Test loss: 0.004001701716333628\n",
      "Test loss: 0.004312137607485056\n",
      "Test loss: 0.003930444363504648\n",
      "Test loss: 0.006084445398300886\n",
      "Test loss: 0.0048775807954370975\n",
      "Test loss: 0.003800023114308715\n",
      "Test loss: 0.004470248240977526\n",
      "Test loss: 0.004254080820828676\n",
      "Test loss: 0.004393543116748333\n",
      "Test loss: 0.005328793544322252\n",
      "Test loss: 0.003566233441233635\n",
      "Test loss: 0.006975560449063778\n",
      "Test loss: 0.004702719859778881\n",
      "Test: 62, loss: 0.004464892204850912\n",
      "Train epoch: 63\n",
      "train loss: 0.0037970636039972305\n",
      "train loss: 0.0037697895895689726\n",
      "train loss: 0.0036325317341834307\n",
      "train loss: 0.005187161732465029\n",
      "train loss: 0.004330375697463751\n",
      "train loss: 0.004435549955815077\n",
      "train loss: 0.0038599036633968353\n",
      "train loss: 0.0063850185833871365\n",
      "train loss: 0.004650482442229986\n",
      "train loss: 0.003939259331673384\n",
      "train loss: 0.004785223864018917\n",
      "train loss: 0.0038569101598113775\n",
      "train loss: 0.003410154487937689\n",
      "train loss: 0.004575046245008707\n",
      "train loss: 0.003441751468926668\n",
      "train loss: 0.003962614107877016\n",
      "train loss: 0.0037725677248090506\n",
      "train loss: 0.0043564909137785435\n",
      "train loss: 0.006830472964793444\n",
      "train loss: 0.0038190684281289577\n",
      "train loss: 0.0045701004564762115\n",
      "train loss: 0.004916278645396233\n",
      "train loss: 0.0048785340040922165\n",
      "train loss: 0.0049707465805113316\n",
      "train loss: 0.0038159105461090803\n",
      "train loss: 0.004569048061966896\n",
      "train loss: 0.003983344882726669\n",
      "train loss: 0.004284626804292202\n",
      "train loss: 0.004503773525357246\n",
      "train loss: 0.0037826653569936752\n",
      "train loss: 0.003918414935469627\n",
      "Test loss: 0.003599468618631363\n",
      "Test loss: 0.0044000712223351\n",
      "Test loss: 0.004272690042853355\n",
      "Test loss: 0.003444670932367444\n",
      "Test loss: 0.0033040614798665047\n",
      "Test loss: 0.004645931534469128\n",
      "Test loss: 0.003405269468203187\n",
      "Test loss: 0.0037196276243776083\n",
      "Test loss: 0.004270130768418312\n",
      "Test loss: 0.0030657993629574776\n",
      "Test loss: 0.004170367494225502\n",
      "Test loss: 0.0038301318418234587\n",
      "Test loss: 0.0034072704147547483\n",
      "Test loss: 0.005770223680883646\n",
      "Test loss: 0.00485657574608922\n",
      "Test loss: 0.005472688004374504\n",
      "Test loss: 0.004065121058374643\n",
      "Test loss: 0.0034827664494514465\n",
      "Test loss: 0.00376832764595747\n",
      "Test loss: 0.00434418860822916\n",
      "Test loss: 0.005096936598420143\n",
      "Test loss: 0.004768364131450653\n",
      "Test loss: 0.004138415679335594\n",
      "Test loss: 0.004640372935682535\n",
      "Test loss: 0.004671740345656872\n",
      "Test loss: 0.004604296293109655\n",
      "Test loss: 0.005470513366162777\n",
      "Test loss: 0.004287147428840399\n",
      "Test loss: 0.004365930333733559\n",
      "Test loss: 0.004906836897134781\n",
      "Test loss: 0.004077700432389975\n",
      "Test: 63, loss: 0.004268504213541746\n",
      "Train epoch: 64\n",
      "train loss: 0.003924598451703787\n",
      "train loss: 0.003946583718061447\n",
      "train loss: 0.003477733815088868\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train loss: 0.004116883035749197\n",
      "train loss: 0.004204737022519112\n",
      "train loss: 0.005840721540153027\n",
      "train loss: 0.003948376048356295\n",
      "train loss: 0.0037744625005871058\n",
      "train loss: 0.004396623931825161\n",
      "train loss: 0.004704742692410946\n",
      "train loss: 0.003335455432534218\n",
      "train loss: 0.00459043774753809\n",
      "train loss: 0.003929093014448881\n",
      "train loss: 0.0047302087768912315\n",
      "train loss: 0.0038885651156306267\n",
      "train loss: 0.0043587577529251575\n",
      "train loss: 0.004298130515962839\n",
      "train loss: 0.004718064796179533\n",
      "train loss: 0.0034221287351101637\n",
      "train loss: 0.003633096581324935\n",
      "train loss: 0.0045877620577812195\n",
      "train loss: 0.004373742267489433\n",
      "train loss: 0.0035545348655432463\n",
      "train loss: 0.0034142995718866587\n",
      "train loss: 0.0035664313472807407\n",
      "train loss: 0.0035919584333896637\n",
      "train loss: 0.0039173089899122715\n",
      "train loss: 0.002986162668094039\n",
      "train loss: 0.003980713430792093\n",
      "train loss: 0.004688888322561979\n",
      "train loss: 0.004500693175941706\n",
      "Test loss: 0.004262696485966444\n",
      "Test loss: 0.004178672097623348\n",
      "Test loss: 0.0038997819647192955\n",
      "Test loss: 0.005684862844645977\n",
      "Test loss: 0.0047687930054962635\n",
      "Test loss: 0.00316607509739697\n",
      "Test loss: 0.0043680607341229916\n",
      "Test loss: 0.004539360757917166\n",
      "Test loss: 0.004418392665684223\n",
      "Test loss: 0.0038858160842210054\n",
      "Test loss: 0.004307330586016178\n",
      "Test loss: 0.0036683587823063135\n",
      "Test loss: 0.004354375880211592\n",
      "Test loss: 0.0038990648463368416\n",
      "Test loss: 0.004816360305994749\n",
      "Test loss: 0.004723388701677322\n",
      "Test loss: 0.0037340763956308365\n",
      "Test loss: 0.0037256707437336445\n",
      "Test loss: 0.003903160337358713\n",
      "Test loss: 0.0038609385956078768\n",
      "Test loss: 0.004282551817595959\n",
      "Test loss: 0.003679431974887848\n",
      "Test loss: 0.0044013578444719315\n",
      "Test loss: 0.003594524459913373\n",
      "Test loss: 0.005241684149950743\n",
      "Test loss: 0.0034298489335924387\n",
      "Test loss: 0.005080253351479769\n",
      "Test loss: 0.004680402111262083\n",
      "Test loss: 0.004357481375336647\n",
      "Test loss: 0.003244732040911913\n",
      "Test loss: 0.003798100398853421\n",
      "Test: 64, loss: 0.00419211620464921\n",
      "Train epoch: 65\n",
      "train loss: 0.0032981846015900373\n",
      "train loss: 0.0036075240932404995\n",
      "train loss: 0.003770981216803193\n",
      "train loss: 0.004331777337938547\n",
      "train loss: 0.00381009210832417\n",
      "train loss: 0.0039047030732035637\n",
      "train loss: 0.003462507389485836\n",
      "train loss: 0.004847344476729631\n",
      "train loss: 0.003805028274655342\n",
      "train loss: 0.002956496784463525\n",
      "train loss: 0.003943267278373241\n",
      "train loss: 0.00533063942566514\n",
      "train loss: 0.0046312022022902966\n",
      "train loss: 0.00343164405785501\n",
      "train loss: 0.0038450038991868496\n",
      "train loss: 0.003990452736616135\n",
      "train loss: 0.004596902523189783\n",
      "train loss: 0.004613797180354595\n",
      "train loss: 0.003955491352826357\n",
      "train loss: 0.003184061497449875\n",
      "train loss: 0.004588267765939236\n",
      "train loss: 0.00412073964253068\n",
      "train loss: 0.0034728948958218098\n",
      "train loss: 0.0031968739349395037\n",
      "train loss: 0.004215252585709095\n",
      "train loss: 0.004295751452445984\n",
      "train loss: 0.004000688903033733\n",
      "train loss: 0.0041586593724787235\n",
      "train loss: 0.0043188356794416904\n",
      "train loss: 0.004852028097957373\n",
      "train loss: 0.004222292918711901\n",
      "Test loss: 0.0035004685632884502\n",
      "Test loss: 0.0036590497475117445\n",
      "Test loss: 0.004474388435482979\n",
      "Test loss: 0.004471613094210625\n",
      "Test loss: 0.005007525905966759\n",
      "Test loss: 0.003976735286414623\n",
      "Test loss: 0.005514864344149828\n",
      "Test loss: 0.0054223984479904175\n",
      "Test loss: 0.00426179263740778\n",
      "Test loss: 0.005415270570665598\n",
      "Test loss: 0.004414211958646774\n",
      "Test loss: 0.00501379556953907\n",
      "Test loss: 0.0043299938552081585\n",
      "Test loss: 0.0035355116706341505\n",
      "Test loss: 0.004115361720323563\n",
      "Test loss: 0.005538744851946831\n",
      "Test loss: 0.004388585686683655\n",
      "Test loss: 0.004344122484326363\n",
      "Test loss: 0.0033242092467844486\n",
      "Test loss: 0.0038856822066009045\n",
      "Test loss: 0.003759288229048252\n",
      "Test loss: 0.004077104385942221\n",
      "Test loss: 0.004304461646825075\n",
      "Test loss: 0.0036259975749999285\n",
      "Test loss: 0.004035799764096737\n",
      "Test loss: 0.0040911599062383175\n",
      "Test loss: 0.0035973740741610527\n",
      "Test loss: 0.003674386302009225\n",
      "Test loss: 0.004076799843460321\n",
      "Test loss: 0.0038302121683955193\n",
      "Test loss: 0.003923431504517794\n",
      "Test: 65, loss: 0.004244849551469088\n",
      "Train epoch: 66\n",
      "train loss: 0.004532085731625557\n",
      "train loss: 0.0036422517150640488\n",
      "train loss: 0.004824814386665821\n",
      "train loss: 0.00572586664929986\n",
      "train loss: 0.004873392637819052\n",
      "train loss: 0.0034454348497092724\n",
      "train loss: 0.003905175719410181\n",
      "train loss: 0.004068508744239807\n",
      "train loss: 0.00432478915899992\n",
      "train loss: 0.004599643871188164\n",
      "train loss: 0.0033249983098357916\n",
      "train loss: 0.005473630968481302\n",
      "train loss: 0.0039373221807181835\n",
      "train loss: 0.0033459740225225687\n",
      "train loss: 0.004581134300678968\n",
      "train loss: 0.005495583638548851\n",
      "train loss: 0.0036539495922625065\n",
      "train loss: 0.004196661524474621\n",
      "train loss: 0.004117798060178757\n",
      "train loss: 0.0035275383852422237\n",
      "train loss: 0.0038810893893241882\n",
      "train loss: 0.0038529683370143175\n",
      "train loss: 0.004740000702440739\n",
      "train loss: 0.003831565147265792\n",
      "train loss: 0.004525901284068823\n",
      "train loss: 0.0032537842635065317\n",
      "train loss: 0.0037484276108443737\n",
      "train loss: 0.004057556390762329\n",
      "train loss: 0.004126043524593115\n",
      "train loss: 0.004753347486257553\n",
      "train loss: 0.004229962360113859\n",
      "Test loss: 0.0046778349205851555\n",
      "Test loss: 0.0044560846872627735\n",
      "Test loss: 0.004140456207096577\n",
      "Test loss: 0.003994410391896963\n",
      "Test loss: 0.003769242437556386\n",
      "Test loss: 0.004412992857396603\n",
      "Test loss: 0.004179747309535742\n",
      "Test loss: 0.0037039497401565313\n",
      "Test loss: 0.004296117462217808\n",
      "Test loss: 0.003590801963582635\n",
      "Test loss: 0.003730335272848606\n",
      "Test loss: 0.003563039470463991\n",
      "Test loss: 0.004140987992286682\n",
      "Test loss: 0.004872268531471491\n",
      "Test loss: 0.005136358086019754\n",
      "Test loss: 0.004482760559767485\n",
      "Test loss: 0.0034172837622463703\n",
      "Test loss: 0.005083146039396524\n",
      "Test loss: 0.00495620584115386\n",
      "Test loss: 0.0037307008169591427\n",
      "Test loss: 0.004484978038817644\n",
      "Test loss: 0.004506922326982021\n",
      "Test loss: 0.003909435123205185\n",
      "Test loss: 0.0037242805119603872\n",
      "Test loss: 0.0034585304092615843\n",
      "Test loss: 0.003931733779609203\n",
      "Test loss: 0.004900457337498665\n",
      "Test loss: 0.003899443196132779\n",
      "Test loss: 0.004447773564606905\n",
      "Test loss: 0.003995642997324467\n",
      "Test loss: 0.00440059881657362\n",
      "Test: 66, loss: 0.004193371161818504\n",
      "Train epoch: 67\n",
      "train loss: 0.004233589395880699\n",
      "train loss: 0.004562871064990759\n",
      "train loss: 0.00432436540722847\n",
      "train loss: 0.004302594345062971\n",
      "train loss: 0.003790380200371146\n",
      "train loss: 0.005670544691383839\n",
      "train loss: 0.0037454983685165644\n",
      "train loss: 0.004511099308729172\n",
      "train loss: 0.00391815509647131\n",
      "train loss: 0.003918591421097517\n",
      "train loss: 0.004287597723305225\n",
      "train loss: 0.004048704635351896\n",
      "train loss: 0.003865589387714863\n",
      "train loss: 0.004693055525422096\n",
      "train loss: 0.004321219865232706\n",
      "train loss: 0.004172979854047298\n",
      "train loss: 0.003830119501799345\n",
      "train loss: 0.005073159467428923\n",
      "train loss: 0.0036977471318095922\n",
      "train loss: 0.0038878063205629587\n",
      "train loss: 0.004046294372528791\n",
      "train loss: 0.004422912839800119\n",
      "train loss: 0.004303837660700083\n",
      "train loss: 0.0040685860440135\n",
      "train loss: 0.004252708051353693\n",
      "train loss: 0.004808264318853617\n",
      "train loss: 0.0038792332634329796\n",
      "train loss: 0.004140931647270918\n",
      "train loss: 0.004663793370127678\n",
      "train loss: 0.003718514693900943\n",
      "train loss: 0.0048788562417030334\n",
      "Test loss: 0.003960540983825922\n",
      "Test loss: 0.004226986318826675\n",
      "Test loss: 0.004146924242377281\n",
      "Test loss: 0.003922692500054836\n",
      "Test loss: 0.004838800989091396\n",
      "Test loss: 0.003360094502568245\n",
      "Test loss: 0.004723190795630217\n",
      "Test loss: 0.004448344465345144\n",
      "Test loss: 0.004456663504242897\n",
      "Test loss: 0.003991167061030865\n",
      "Test loss: 0.004261319525539875\n",
      "Test loss: 0.003563453210517764\n",
      "Test loss: 0.003916949033737183\n",
      "Test loss: 0.004741983953863382\n",
      "Test loss: 0.0037782443687319756\n",
      "Test loss: 0.004430933855473995\n",
      "Test loss: 0.004543088376522064\n",
      "Test loss: 0.003907186910510063\n",
      "Test loss: 0.0034507066011428833\n",
      "Test loss: 0.004151232074946165\n",
      "Test loss: 0.004256993066519499\n",
      "Test loss: 0.004863484762609005\n",
      "Test loss: 0.0044382717460393906\n",
      "Test loss: 0.0041221752762794495\n",
      "Test loss: 0.0037761502899229527\n",
      "Test loss: 0.004522574134171009\n",
      "Test loss: 0.004998215939849615\n",
      "Test loss: 0.0035385338123887777\n",
      "Test loss: 0.0033940186258405447\n",
      "Test loss: 0.0033548083156347275\n",
      "Test loss: 0.004435880575329065\n",
      "Test: 67, loss: 0.004145858343690634\n",
      "Train epoch: 68\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train loss: 0.004199552815407515\n",
      "train loss: 0.004649385344237089\n",
      "train loss: 0.003933952189981937\n",
      "train loss: 0.003635488683357835\n",
      "train loss: 0.004248722922056913\n",
      "train loss: 0.00398310087621212\n",
      "train loss: 0.004378700628876686\n",
      "train loss: 0.004195345100015402\n",
      "train loss: 0.005595651920884848\n",
      "train loss: 0.0041581206023693085\n",
      "train loss: 0.0037348130717873573\n",
      "train loss: 0.0037836579140275717\n",
      "train loss: 0.00384368933737278\n",
      "train loss: 0.0037814127281308174\n",
      "train loss: 0.005185388494282961\n",
      "train loss: 0.0036959548015147448\n",
      "train loss: 0.005219802726060152\n",
      "train loss: 0.004609353374689817\n",
      "train loss: 0.004257616586983204\n",
      "train loss: 0.0045088669285178185\n",
      "train loss: 0.004012661520391703\n",
      "train loss: 0.003833820577710867\n",
      "train loss: 0.004563852213323116\n",
      "train loss: 0.004322886001318693\n",
      "train loss: 0.00409337505698204\n",
      "train loss: 0.003842147532850504\n",
      "train loss: 0.003997923340648413\n",
      "train loss: 0.0040962048806250095\n",
      "train loss: 0.0042459601536393166\n",
      "train loss: 0.004570245277136564\n",
      "train loss: 0.003646120661869645\n",
      "Test loss: 0.0044454592280089855\n",
      "Test loss: 0.003714790800586343\n",
      "Test loss: 0.004211544059216976\n",
      "Test loss: 0.003812409471720457\n",
      "Test loss: 0.004205374978482723\n",
      "Test loss: 0.0039453390054404736\n",
      "Test loss: 0.00445663882419467\n",
      "Test loss: 0.0034280321560800076\n",
      "Test loss: 0.004198080860078335\n",
      "Test loss: 0.00436556339263916\n",
      "Test loss: 0.0037648349534720182\n",
      "Test loss: 0.00427603954449296\n",
      "Test loss: 0.004857312887907028\n",
      "Test loss: 0.00405123783275485\n",
      "Test loss: 0.004275113809853792\n",
      "Test loss: 0.0038393069989979267\n",
      "Test loss: 0.004231688566505909\n",
      "Test loss: 0.004083509556949139\n",
      "Test loss: 0.004233719781041145\n",
      "Test loss: 0.005158537067472935\n",
      "Test loss: 0.003967199940234423\n",
      "Test loss: 0.005090357735753059\n",
      "Test loss: 0.00472051277756691\n",
      "Test loss: 0.0035091445315629244\n",
      "Test loss: 0.003633426036685705\n",
      "Test loss: 0.004907379858195782\n",
      "Test loss: 0.003958078566938639\n",
      "Test loss: 0.0043420917354524136\n",
      "Test loss: 0.003237359691411257\n",
      "Test loss: 0.004233682062476873\n",
      "Test loss: 0.003930911421775818\n",
      "Test: 68, loss: 0.0041640219278633595\n",
      "Train epoch: 69\n",
      "train loss: 0.005444711539894342\n",
      "train loss: 0.005644469056278467\n",
      "train loss: 0.0037841147277504206\n",
      "train loss: 0.003980251029133797\n",
      "train loss: 0.004174035973846912\n",
      "train loss: 0.003650994971394539\n",
      "train loss: 0.00444511603564024\n",
      "train loss: 0.005441176239401102\n",
      "train loss: 0.0039001095574349165\n",
      "train loss: 0.0035866580437868834\n",
      "train loss: 0.004644979722797871\n",
      "train loss: 0.003783079795539379\n",
      "train loss: 0.004348509944975376\n",
      "train loss: 0.00393278943374753\n",
      "train loss: 0.0036061767023056746\n",
      "train loss: 0.004215853754431009\n",
      "train loss: 0.0032550417818129063\n",
      "train loss: 0.00460739154368639\n",
      "train loss: 0.0035499080549925566\n",
      "train loss: 0.0038667244371026754\n",
      "train loss: 0.003609319683164358\n",
      "train loss: 0.00356067786924541\n",
      "train loss: 0.0034999102354049683\n",
      "train loss: 0.003973687998950481\n",
      "train loss: 0.004117557313293219\n",
      "train loss: 0.0035677829291671515\n",
      "train loss: 0.005001350771635771\n",
      "train loss: 0.00370758562348783\n",
      "train loss: 0.0038987097796052694\n",
      "train loss: 0.0039701261557638645\n",
      "train loss: 0.004106175620108843\n",
      "Test loss: 0.0041350144892930984\n",
      "Test loss: 0.004175552166998386\n",
      "Test loss: 0.003740064799785614\n",
      "Test loss: 0.004481825046241283\n",
      "Test loss: 0.005333805922418833\n",
      "Test loss: 0.00424882210791111\n",
      "Test loss: 0.005388339515775442\n",
      "Test loss: 0.0033628526143729687\n",
      "Test loss: 0.00415352638810873\n",
      "Test loss: 0.003691545221954584\n",
      "Test loss: 0.005754263140261173\n",
      "Test loss: 0.0035947184078395367\n",
      "Test loss: 0.003840713994577527\n",
      "Test loss: 0.003727539209648967\n",
      "Test loss: 0.004395980853587389\n",
      "Test loss: 0.003960581962019205\n",
      "Test loss: 0.0046666450798511505\n",
      "Test loss: 0.0037108969409018755\n",
      "Test loss: 0.004637301899492741\n",
      "Test loss: 0.0055237505584955215\n",
      "Test loss: 0.0050813909620046616\n",
      "Test loss: 0.0036434249486774206\n",
      "Test loss: 0.003574671922251582\n",
      "Test loss: 0.005056807305663824\n",
      "Test loss: 0.003812868380919099\n",
      "Test loss: 0.004029167350381613\n",
      "Test loss: 0.003857496427372098\n",
      "Test loss: 0.004211935214698315\n",
      "Test loss: 0.0036255689337849617\n",
      "Test loss: 0.0038663295563310385\n",
      "Test loss: 0.003207739209756255\n",
      "Test: 69, loss: 0.0042093913070857525\n",
      "Train epoch: 70\n",
      "train loss: 0.004527658689767122\n",
      "train loss: 0.004596257116645575\n",
      "train loss: 0.004403109662234783\n",
      "train loss: 0.004884202033281326\n",
      "train loss: 0.005179402884095907\n",
      "train loss: 0.003433900885283947\n",
      "train loss: 0.0035377549938857555\n",
      "train loss: 0.004765872377902269\n",
      "train loss: 0.0036851565819233656\n",
      "train loss: 0.003931362181901932\n",
      "train loss: 0.003669202094897628\n",
      "train loss: 0.003789110342040658\n",
      "train loss: 0.0046221413649618626\n",
      "train loss: 0.004240336362272501\n",
      "train loss: 0.0047841682098805904\n",
      "train loss: 0.00467017712071538\n",
      "train loss: 0.0051760138012468815\n",
      "train loss: 0.003816111944615841\n",
      "train loss: 0.004106708336621523\n",
      "train loss: 0.0036675625015050173\n",
      "train loss: 0.00379524496383965\n",
      "train loss: 0.003735403763130307\n",
      "train loss: 0.004393288400024176\n",
      "train loss: 0.0035918420180678368\n",
      "train loss: 0.003506697714328766\n",
      "train loss: 0.003415460931137204\n",
      "train loss: 0.003946551121771336\n",
      "train loss: 0.004222077317535877\n",
      "train loss: 0.0031336599495261908\n",
      "train loss: 0.004281692206859589\n",
      "train loss: 0.005223199725151062\n",
      "Test loss: 0.00406804820522666\n",
      "Test loss: 0.004818129353225231\n",
      "Test loss: 0.004125708248466253\n",
      "Test loss: 0.0034133908338844776\n",
      "Test loss: 0.0033509887289255857\n",
      "Test loss: 0.003319262992590666\n",
      "Test loss: 0.003651739563792944\n",
      "Test loss: 0.004160915967077017\n",
      "Test loss: 0.004450251813977957\n",
      "Test loss: 0.004541607107967138\n",
      "Test loss: 0.004309894982725382\n",
      "Test loss: 0.003870690707117319\n",
      "Test loss: 0.00391803402453661\n",
      "Test loss: 0.003519027028232813\n",
      "Test loss: 0.004229079931974411\n",
      "Test loss: 0.004610843025147915\n",
      "Test loss: 0.00348017574287951\n",
      "Test loss: 0.003987313713878393\n",
      "Test loss: 0.0035874885506927967\n",
      "Test loss: 0.004998300690203905\n",
      "Test loss: 0.004068531561642885\n",
      "Test loss: 0.003479955717921257\n",
      "Test loss: 0.004313793499022722\n",
      "Test loss: 0.005528336390852928\n",
      "Test loss: 0.0042881956323981285\n",
      "Test loss: 0.003618909977376461\n",
      "Test loss: 0.0035467420239001513\n",
      "Test loss: 0.005156889092177153\n",
      "Test loss: 0.003585416590794921\n",
      "Test loss: 0.005020991899073124\n",
      "Test loss: 0.004567279014736414\n",
      "Test: 70, loss: 0.004115675110369921\n",
      "Train epoch: 71\n",
      "train loss: 0.0029480422381311655\n",
      "train loss: 0.004244448617100716\n",
      "train loss: 0.0054457043297588825\n",
      "train loss: 0.005312033928930759\n",
      "train loss: 0.003872840665280819\n",
      "train loss: 0.00503976596519351\n",
      "train loss: 0.00481875566765666\n",
      "train loss: 0.004864680580794811\n",
      "train loss: 0.0046591125428676605\n",
      "train loss: 0.003327116370201111\n",
      "train loss: 0.004409825429320335\n",
      "train loss: 0.00594458170235157\n",
      "train loss: 0.004141201265156269\n",
      "train loss: 0.003899205243214965\n",
      "train loss: 0.00505498331040144\n",
      "train loss: 0.004057880491018295\n",
      "train loss: 0.0035389431286603212\n",
      "train loss: 0.00351278786547482\n",
      "train loss: 0.0041537219658494\n",
      "train loss: 0.004192589782178402\n",
      "train loss: 0.0035334588028490543\n",
      "train loss: 0.004837518557906151\n",
      "train loss: 0.004522113129496574\n",
      "train loss: 0.005130815785378218\n",
      "train loss: 0.004012025892734528\n",
      "train loss: 0.00408064853399992\n",
      "train loss: 0.003811815520748496\n",
      "train loss: 0.0032352355774492025\n",
      "train loss: 0.003434143029153347\n",
      "train loss: 0.004146644379943609\n",
      "train loss: 0.0038723598700016737\n",
      "Test loss: 0.003671396290883422\n",
      "Test loss: 0.003928523510694504\n",
      "Test loss: 0.003869196167215705\n",
      "Test loss: 0.0047251335345208645\n",
      "Test loss: 0.0037521582562476397\n",
      "Test loss: 0.003625286975875497\n",
      "Test loss: 0.004697369411587715\n",
      "Test loss: 0.0039797211065888405\n",
      "Test loss: 0.0036816983483731747\n",
      "Test loss: 0.00426820944994688\n",
      "Test loss: 0.0035512566100806\n",
      "Test loss: 0.0038351016119122505\n",
      "Test loss: 0.0029559859540313482\n",
      "Test loss: 0.0035573982167989016\n",
      "Test loss: 0.004809733014553785\n",
      "Test loss: 0.004939931444823742\n",
      "Test loss: 0.0041640750132501125\n",
      "Test loss: 0.00407901918515563\n",
      "Test loss: 0.004210575949400663\n",
      "Test loss: 0.003766832873225212\n",
      "Test loss: 0.0037746645975857973\n",
      "Test loss: 0.003903026459738612\n",
      "Test loss: 0.003970321733504534\n",
      "Test loss: 0.005788414739072323\n",
      "Test loss: 0.004689742345362902\n",
      "Test loss: 0.003898628056049347\n",
      "Test loss: 0.005730326287448406\n",
      "Test loss: 0.0041524372063577175\n",
      "Test loss: 0.0031362511217594147\n",
      "Test loss: 0.0038367784582078457\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss: 0.004232177976518869\n",
      "Test: 71, loss: 0.004102625418454409\n",
      "Train epoch: 72\n",
      "train loss: 0.005358594935387373\n",
      "train loss: 0.0038978084921836853\n",
      "train loss: 0.004506617784500122\n",
      "train loss: 0.004392911214381456\n",
      "train loss: 0.004303630907088518\n",
      "train loss: 0.004721225239336491\n",
      "train loss: 0.0037977504543960094\n",
      "train loss: 0.00334246177226305\n",
      "train loss: 0.004641063045710325\n",
      "train loss: 0.005522729363292456\n",
      "train loss: 0.003634347813203931\n",
      "train loss: 0.0039247204549610615\n",
      "train loss: 0.0032910131849348545\n",
      "train loss: 0.00406388845294714\n",
      "train loss: 0.0036239521577954292\n",
      "train loss: 0.004705248866230249\n",
      "train loss: 0.00452058482915163\n",
      "train loss: 0.0037873019464313984\n",
      "train loss: 0.004273045342415571\n",
      "train loss: 0.004399548284709454\n",
      "train loss: 0.0037990231066942215\n",
      "train loss: 0.004819598980247974\n",
      "train loss: 0.003892233595252037\n",
      "train loss: 0.0036751769948750734\n",
      "train loss: 0.0031751112546771765\n",
      "train loss: 0.004312638193368912\n",
      "train loss: 0.004223241936415434\n",
      "train loss: 0.004017012193799019\n",
      "train loss: 0.0035758016165345907\n",
      "train loss: 0.003565504215657711\n",
      "train loss: 0.005258698482066393\n",
      "Test loss: 0.003973114769905806\n",
      "Test loss: 0.0046789939515292645\n",
      "Test loss: 0.004667243920266628\n",
      "Test loss: 0.003983853384852409\n",
      "Test loss: 0.004239290952682495\n",
      "Test loss: 0.004729053005576134\n",
      "Test loss: 0.003985564690083265\n",
      "Test loss: 0.0042924522422254086\n",
      "Test loss: 0.003153880126774311\n",
      "Test loss: 0.004464802797883749\n",
      "Test loss: 0.005068925209343433\n",
      "Test loss: 0.003560576355084777\n",
      "Test loss: 0.0034177256748080254\n",
      "Test loss: 0.004750497173517942\n",
      "Test loss: 0.0034269122406840324\n",
      "Test loss: 0.004169780295342207\n",
      "Test loss: 0.005041306372731924\n",
      "Test loss: 0.003927017096430063\n",
      "Test loss: 0.004389078356325626\n",
      "Test loss: 0.004195632413029671\n",
      "Test loss: 0.003417794592678547\n",
      "Test loss: 0.003265696344897151\n",
      "Test loss: 0.0033317124471068382\n",
      "Test loss: 0.003786760149523616\n",
      "Test loss: 0.003228902816772461\n",
      "Test loss: 0.0049247220158576965\n",
      "Test loss: 0.004550539888441563\n",
      "Test loss: 0.005399462301284075\n",
      "Test loss: 0.00489902775734663\n",
      "Test loss: 0.004475320689380169\n",
      "Test loss: 0.003935020416975021\n",
      "Test: 72, loss: 0.004171956796199083\n",
      "Train epoch: 73\n",
      "train loss: 0.003590908134356141\n",
      "train loss: 0.003943979740142822\n",
      "train loss: 0.004037700593471527\n",
      "train loss: 0.004785963334143162\n",
      "train loss: 0.00429457938298583\n",
      "train loss: 0.004667334258556366\n",
      "train loss: 0.0033514308743178844\n",
      "train loss: 0.0038512919563800097\n",
      "train loss: 0.00432195421308279\n",
      "train loss: 0.003922660835087299\n",
      "train loss: 0.004432084038853645\n",
      "train loss: 0.003773259464651346\n",
      "train loss: 0.004245941527187824\n",
      "train loss: 0.004064798820763826\n",
      "train loss: 0.004374867770820856\n",
      "train loss: 0.0034914903808385134\n",
      "train loss: 0.005088153760880232\n",
      "train loss: 0.003866189392283559\n",
      "train loss: 0.0035526177380234003\n",
      "train loss: 0.003848865395411849\n",
      "train loss: 0.004139717668294907\n",
      "train loss: 0.004698839504271746\n",
      "train loss: 0.004095087293535471\n",
      "train loss: 0.00416755024343729\n",
      "train loss: 0.004979627672582865\n",
      "train loss: 0.003987457137554884\n",
      "train loss: 0.005299302749335766\n",
      "train loss: 0.003502156585454941\n",
      "train loss: 0.005116267129778862\n",
      "train loss: 0.004638051614165306\n",
      "train loss: 0.003357467008754611\n",
      "Test loss: 0.003211650997400284\n",
      "Test loss: 0.004681561142206192\n",
      "Test loss: 0.003631643019616604\n",
      "Test loss: 0.004342956934124231\n",
      "Test loss: 0.004946874920278788\n",
      "Test loss: 0.00411247881129384\n",
      "Test loss: 0.004542757291346788\n",
      "Test loss: 0.004760670010000467\n",
      "Test loss: 0.0037164653185755014\n",
      "Test loss: 0.004925300367176533\n",
      "Test loss: 0.004649339709430933\n",
      "Test loss: 0.004282668698579073\n",
      "Test loss: 0.0044140261597931385\n",
      "Test loss: 0.003207471454516053\n",
      "Test loss: 0.004047758877277374\n",
      "Test loss: 0.003656335873529315\n",
      "Test loss: 0.0044615790247917175\n",
      "Test loss: 0.00544129591435194\n",
      "Test loss: 0.0031577509362250566\n",
      "Test loss: 0.003878503106534481\n",
      "Test loss: 0.004775963723659515\n",
      "Test loss: 0.00431338045746088\n",
      "Test loss: 0.0035296983551234007\n",
      "Test loss: 0.004962833598256111\n",
      "Test loss: 0.005543299950659275\n",
      "Test loss: 0.003625320503488183\n",
      "Test loss: 0.0035174100194126368\n",
      "Test loss: 0.0033132389653474092\n",
      "Test loss: 0.005082833115011454\n",
      "Test loss: 0.004273107275366783\n",
      "Test loss: 0.004247117787599564\n",
      "Test: 73, loss: 0.004233976360410452\n",
      "Train epoch: 74\n",
      "train loss: 0.004331767093390226\n",
      "train loss: 0.003828862914815545\n",
      "train loss: 0.0036024099681526423\n",
      "train loss: 0.004029861185699701\n",
      "train loss: 0.0039636557921767235\n",
      "train loss: 0.0046999636106193066\n",
      "train loss: 0.00517713138833642\n",
      "train loss: 0.0037205927073955536\n",
      "train loss: 0.003368514124304056\n",
      "train loss: 0.003588357474654913\n",
      "train loss: 0.004351911600679159\n",
      "train loss: 0.003912519663572311\n",
      "train loss: 0.004823349881917238\n",
      "train loss: 0.004516461864113808\n",
      "train loss: 0.0035836738534271717\n",
      "train loss: 0.00488449027761817\n",
      "train loss: 0.004556415602564812\n",
      "train loss: 0.00343611859716475\n",
      "train loss: 0.003979804925620556\n",
      "train loss: 0.00399034982547164\n",
      "train loss: 0.0036010497715324163\n",
      "train loss: 0.005056213587522507\n",
      "train loss: 0.004963987041264772\n",
      "train loss: 0.004290459677577019\n",
      "train loss: 0.003994372207671404\n",
      "train loss: 0.0038067963905632496\n",
      "train loss: 0.003536628559231758\n",
      "train loss: 0.003299941774457693\n",
      "train loss: 0.003190097399055958\n",
      "train loss: 0.003372753271833062\n",
      "train loss: 0.00416615279391408\n",
      "Test loss: 0.003648526733741164\n",
      "Test loss: 0.003801654325798154\n",
      "Test loss: 0.004794532433152199\n",
      "Test loss: 0.003962623421102762\n",
      "Test loss: 0.005249815061688423\n",
      "Test loss: 0.0049787783063948154\n",
      "Test loss: 0.003787251189351082\n",
      "Test loss: 0.004393272567540407\n",
      "Test loss: 0.0036816687788814306\n",
      "Test loss: 0.003624281147494912\n",
      "Test loss: 0.003455444937571883\n",
      "Test loss: 0.0040020933374762535\n",
      "Test loss: 0.004802452400326729\n",
      "Test loss: 0.004161190707236528\n",
      "Test loss: 0.003888343460857868\n",
      "Test loss: 0.004788365680724382\n",
      "Test loss: 0.003887586761265993\n",
      "Test loss: 0.003504956839606166\n",
      "Test loss: 0.004065234214067459\n",
      "Test loss: 0.004142689984291792\n",
      "Test loss: 0.0033521775621920824\n",
      "Test loss: 0.004634743090718985\n",
      "Test loss: 0.004721005447208881\n",
      "Test loss: 0.00450147595256567\n",
      "Test loss: 0.0038060671649873257\n",
      "Test loss: 0.004557410255074501\n",
      "Test loss: 0.0038135736249387264\n",
      "Test loss: 0.0037079891189932823\n",
      "Test loss: 0.005724548362195492\n",
      "Test loss: 0.00457983510568738\n",
      "Test loss: 0.003966205753386021\n",
      "Test: 74, loss: 0.004193090833723545\n",
      "Train epoch: 75\n",
      "train loss: 0.00449763610959053\n",
      "train loss: 0.004125469829887152\n",
      "train loss: 0.004594943951815367\n",
      "train loss: 0.0045614768750965595\n",
      "train loss: 0.0037451661191880703\n",
      "train loss: 0.0034725344739854336\n",
      "train loss: 0.003379901172593236\n",
      "train loss: 0.0031130663119256496\n",
      "train loss: 0.004518041852861643\n",
      "train loss: 0.004059419967234135\n",
      "train loss: 0.004683472216129303\n",
      "train loss: 0.003966520074754953\n",
      "train loss: 0.004889287054538727\n",
      "train loss: 0.003242253325879574\n",
      "train loss: 0.004231266677379608\n",
      "train loss: 0.00499167712405324\n",
      "train loss: 0.0032294271513819695\n",
      "train loss: 0.0032997450325638056\n",
      "train loss: 0.004199782852083445\n",
      "train loss: 0.004759500734508038\n",
      "train loss: 0.005389863159507513\n",
      "train loss: 0.004921772517263889\n",
      "train loss: 0.004465935751795769\n",
      "train loss: 0.005910197738558054\n",
      "train loss: 0.004666217602789402\n",
      "train loss: 0.0037209508009254932\n",
      "train loss: 0.0041887410916388035\n",
      "train loss: 0.004502404946833849\n",
      "train loss: 0.004921081941574812\n",
      "train loss: 0.0045400201342999935\n",
      "train loss: 0.0035467231646180153\n",
      "Test loss: 0.0042948112823069096\n",
      "Test loss: 0.005144985858350992\n",
      "Test loss: 0.0038247511256486177\n",
      "Test loss: 0.00472366763278842\n",
      "Test loss: 0.0039006273727864027\n",
      "Test loss: 0.004864778835326433\n",
      "Test loss: 0.0039847055450081825\n",
      "Test loss: 0.0048452201299369335\n",
      "Test loss: 0.003941067028790712\n",
      "Test loss: 0.0029452568851411343\n",
      "Test loss: 0.005307001993060112\n",
      "Test loss: 0.004046827554702759\n",
      "Test loss: 0.0036104153841733932\n",
      "Test loss: 0.0041706436313688755\n",
      "Test loss: 0.004543516784906387\n",
      "Test loss: 0.004484398756176233\n",
      "Test loss: 0.0047063217498362064\n",
      "Test loss: 0.003761625848710537\n",
      "Test loss: 0.004265938885509968\n",
      "Test loss: 0.004919623490422964\n",
      "Test loss: 0.003458117600530386\n",
      "Test loss: 0.0044305454939603806\n",
      "Test loss: 0.003394674276933074\n",
      "Test loss: 0.003348118392750621\n",
      "Test loss: 0.004166791681200266\n",
      "Test loss: 0.0034728380851447582\n",
      "Test loss: 0.00394040159881115\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss: 0.003418851178139448\n",
      "Test loss: 0.00472797267138958\n",
      "Test loss: 0.004771594423800707\n",
      "Test loss: 0.004237431567162275\n",
      "Test: 75, loss: 0.004182371310889721\n",
      "Train epoch: 76\n",
      "train loss: 0.003908843267709017\n",
      "train loss: 0.004009320866316557\n",
      "train loss: 0.004204240161925554\n",
      "train loss: 0.0034570787101984024\n",
      "train loss: 0.0052268290892243385\n",
      "train loss: 0.00407826853916049\n",
      "train loss: 0.0036355124320834875\n",
      "train loss: 0.004059843719005585\n",
      "train loss: 0.0047282869927585125\n",
      "train loss: 0.004404840990900993\n",
      "train loss: 0.0042887525632977486\n",
      "train loss: 0.003656599670648575\n",
      "train loss: 0.0036512441001832485\n",
      "train loss: 0.0034970147535204887\n",
      "train loss: 0.004647538997232914\n",
      "train loss: 0.004671336617320776\n",
      "train loss: 0.004661311861127615\n",
      "train loss: 0.004027392249554396\n",
      "train loss: 0.0043887849897146225\n",
      "train loss: 0.003702458692714572\n",
      "train loss: 0.003981335554271936\n",
      "train loss: 0.0049377065151929855\n",
      "train loss: 0.004344513639807701\n",
      "train loss: 0.005213846452534199\n",
      "train loss: 0.004330924712121487\n",
      "train loss: 0.004228414036333561\n",
      "train loss: 0.005038632545620203\n",
      "train loss: 0.004425382707268\n",
      "train loss: 0.003579701529815793\n",
      "train loss: 0.003546410473063588\n",
      "train loss: 0.0037387132178992033\n",
      "Test loss: 0.0030364845879375935\n",
      "Test loss: 0.0036460880655795336\n",
      "Test loss: 0.004321271553635597\n",
      "Test loss: 0.0031977826729416847\n",
      "Test loss: 0.0032156335655599833\n",
      "Test loss: 0.005235034041106701\n",
      "Test loss: 0.003915041219443083\n",
      "Test loss: 0.004523745272308588\n",
      "Test loss: 0.003683486022055149\n",
      "Test loss: 0.004461329430341721\n",
      "Test loss: 0.004339220002293587\n",
      "Test loss: 0.004581849090754986\n",
      "Test loss: 0.004086577333509922\n",
      "Test loss: 0.0035274773836135864\n",
      "Test loss: 0.004053822718560696\n",
      "Test loss: 0.003836308140307665\n",
      "Test loss: 0.004216419067233801\n",
      "Test loss: 0.0042573013342916965\n",
      "Test loss: 0.004082818049937487\n",
      "Test loss: 0.005328294355422258\n",
      "Test loss: 0.0036621135659515858\n",
      "Test loss: 0.00418290076777339\n",
      "Test loss: 0.0037822253070771694\n",
      "Test loss: 0.004449871834367514\n",
      "Test loss: 0.003134729340672493\n",
      "Test loss: 0.0035401848144829273\n",
      "Test loss: 0.0032271326053887606\n",
      "Test loss: 0.004646929446607828\n",
      "Test loss: 0.0035605784505605698\n",
      "Test loss: 0.003543501254171133\n",
      "Test loss: 0.004246986471116543\n",
      "Test: 76, loss: 0.003984617535024881\n",
      "Train epoch: 77\n",
      "train loss: 0.0035995489452034235\n",
      "train loss: 0.003991020377725363\n",
      "train loss: 0.004295677877962589\n",
      "train loss: 0.004171646665781736\n",
      "train loss: 0.003244050545617938\n",
      "train loss: 0.0035743541084229946\n",
      "train loss: 0.00392677029594779\n",
      "train loss: 0.003966593649238348\n",
      "train loss: 0.003876732662320137\n",
      "train loss: 0.004935919772833586\n",
      "train loss: 0.004039113409817219\n",
      "train loss: 0.004334111697971821\n",
      "train loss: 0.003822825150564313\n",
      "train loss: 0.004549775738269091\n",
      "train loss: 0.004111738409847021\n",
      "train loss: 0.0029753493145108223\n",
      "train loss: 0.00340650649741292\n",
      "train loss: 0.003643969539552927\n",
      "train loss: 0.004015440121293068\n",
      "train loss: 0.004109908826649189\n",
      "train loss: 0.004991238005459309\n",
      "train loss: 0.0034350152127444744\n",
      "train loss: 0.0037357276305556297\n",
      "train loss: 0.0034318752586841583\n",
      "train loss: 0.0053457519970834255\n",
      "train loss: 0.0036018609534949064\n",
      "train loss: 0.003418887034058571\n",
      "train loss: 0.0037637173663824797\n",
      "train loss: 0.0038052103482186794\n",
      "train loss: 0.004076276905834675\n",
      "train loss: 0.003817833261564374\n",
      "Test loss: 0.004124654922634363\n",
      "Test loss: 0.00408413028344512\n",
      "Test loss: 0.003099271096289158\n",
      "Test loss: 0.004654202144593\n",
      "Test loss: 0.00457235099747777\n",
      "Test loss: 0.003964878153055906\n",
      "Test loss: 0.003787154797464609\n",
      "Test loss: 0.003899502335116267\n",
      "Test loss: 0.004460202530026436\n",
      "Test loss: 0.005889387335628271\n",
      "Test loss: 0.00457238033413887\n",
      "Test loss: 0.0036964972969144583\n",
      "Test loss: 0.0031287071760743856\n",
      "Test loss: 0.004376009106636047\n",
      "Test loss: 0.003434124868363142\n",
      "Test loss: 0.003965669311583042\n",
      "Test loss: 0.00462318817153573\n",
      "Test loss: 0.003361868904903531\n",
      "Test loss: 0.003112128470093012\n",
      "Test loss: 0.0037240255624055862\n",
      "Test loss: 0.0034343143925070763\n",
      "Test loss: 0.0036760363727808\n",
      "Test loss: 0.004201662726700306\n",
      "Test loss: 0.003559290897101164\n",
      "Test loss: 0.003161801490932703\n",
      "Test loss: 0.003606679616495967\n",
      "Test loss: 0.005162839312106371\n",
      "Test loss: 0.004253616090863943\n",
      "Test loss: 0.0033504038583487272\n",
      "Test loss: 0.004534231498837471\n",
      "Test loss: 0.004070770926773548\n",
      "Test: 77, loss: 0.003985225223004818\n",
      "Train epoch: 78\n",
      "train loss: 0.004890544340014458\n",
      "train loss: 0.004836954176425934\n",
      "train loss: 0.003495423123240471\n",
      "train loss: 0.003948911093175411\n",
      "train loss: 0.0040095895528793335\n",
      "train loss: 0.004345825873315334\n",
      "train loss: 0.0035967447329312563\n",
      "train loss: 0.004474344663321972\n",
      "train loss: 0.003389575984328985\n",
      "train loss: 0.003652967046946287\n",
      "train loss: 0.004892793018370867\n",
      "train loss: 0.004049805458635092\n",
      "train loss: 0.003693867940455675\n",
      "train loss: 0.004890178795903921\n",
      "train loss: 0.005302238278090954\n",
      "train loss: 0.0031696290243417025\n",
      "train loss: 0.0035358453169465065\n",
      "train loss: 0.004090372938662767\n",
      "train loss: 0.0030732648447155952\n",
      "train loss: 0.00581813370808959\n",
      "train loss: 0.004134354181587696\n",
      "train loss: 0.003857643576338887\n",
      "train loss: 0.004022569861263037\n",
      "train loss: 0.003872493514791131\n",
      "train loss: 0.003579678013920784\n",
      "train loss: 0.003432887839153409\n",
      "train loss: 0.004302669316530228\n",
      "train loss: 0.0056403097696602345\n",
      "train loss: 0.004597899969667196\n",
      "train loss: 0.003761646570637822\n",
      "train loss: 0.0035933905746787786\n",
      "Test loss: 0.004985386971384287\n",
      "Test loss: 0.003866212908178568\n",
      "Test loss: 0.004169192165136337\n",
      "Test loss: 0.00500644464045763\n",
      "Test loss: 0.0032820163760334253\n",
      "Test loss: 0.005291595123708248\n",
      "Test loss: 0.005043108016252518\n",
      "Test loss: 0.0033982719760388136\n",
      "Test loss: 0.004294395912438631\n",
      "Test loss: 0.00374826742336154\n",
      "Test loss: 0.0037844995968043804\n",
      "Test loss: 0.005407515447586775\n",
      "Test loss: 0.003884753445163369\n",
      "Test loss: 0.0032421923242509365\n",
      "Test loss: 0.003892968874424696\n",
      "Test loss: 0.0048603154718875885\n",
      "Test loss: 0.0038155517540872097\n",
      "Test loss: 0.003731045639142394\n",
      "Test loss: 0.004414449445903301\n",
      "Test loss: 0.006715422961860895\n",
      "Test loss: 0.004236969631165266\n",
      "Test loss: 0.0042078192345798016\n",
      "Test loss: 0.003276243107393384\n",
      "Test loss: 0.004459042567759752\n",
      "Test loss: 0.0035180901177227497\n",
      "Test loss: 0.004123328719288111\n",
      "Test loss: 0.0035164279397577047\n",
      "Test loss: 0.005183183588087559\n",
      "Test loss: 0.0036009165924042463\n",
      "Test loss: 0.0036657354794442654\n",
      "Test loss: 0.003771302057430148\n",
      "Test: 78, loss: 0.004206215031445026\n",
      "Train epoch: 79\n",
      "train loss: 0.004300245083868504\n",
      "train loss: 0.004168630577623844\n",
      "train loss: 0.0036015042569488287\n",
      "train loss: 0.0041708676144480705\n",
      "train loss: 0.0041131386533379555\n",
      "train loss: 0.003950329963117838\n",
      "train loss: 0.0036623936612159014\n",
      "train loss: 0.00353485275991261\n",
      "train loss: 0.003930361475795507\n",
      "train loss: 0.0044807144440710545\n",
      "train loss: 0.003832089714705944\n",
      "train loss: 0.004773767665028572\n",
      "train loss: 0.004549646750092506\n",
      "train loss: 0.00357077713124454\n",
      "train loss: 0.004165712743997574\n",
      "train loss: 0.0035456407349556684\n",
      "train loss: 0.003685982199385762\n",
      "train loss: 0.004708855412900448\n",
      "train loss: 0.0041593145579099655\n",
      "train loss: 0.00463572284206748\n",
      "train loss: 0.003600634867325425\n",
      "train loss: 0.0034106371458619833\n",
      "train loss: 0.003927447833120823\n",
      "train loss: 0.0040857926942408085\n",
      "train loss: 0.003390839323401451\n",
      "train loss: 0.0033542364835739136\n",
      "train loss: 0.003941213246434927\n",
      "train loss: 0.004143820609897375\n",
      "train loss: 0.0033355895429849625\n",
      "train loss: 0.004356075078248978\n",
      "train loss: 0.005766890477389097\n",
      "Test loss: 0.0044623869471251965\n",
      "Test loss: 0.004203448072075844\n",
      "Test loss: 0.003631636267527938\n",
      "Test loss: 0.003773854346945882\n",
      "Test loss: 0.004549393430352211\n",
      "Test loss: 0.0034664056729525328\n",
      "Test loss: 0.0032838755287230015\n",
      "Test loss: 0.003541446989402175\n",
      "Test loss: 0.0041762059554457664\n",
      "Test loss: 0.003640231676399708\n",
      "Test loss: 0.0036824713461101055\n",
      "Test loss: 0.0056974622420966625\n",
      "Test loss: 0.004141500685364008\n",
      "Test loss: 0.004110496491193771\n",
      "Test loss: 0.0044680191203951836\n",
      "Test loss: 0.0033179032616317272\n",
      "Test loss: 0.004095927812159061\n",
      "Test loss: 0.003502410836517811\n",
      "Test loss: 0.003784094238653779\n",
      "Test loss: 0.003104829927906394\n",
      "Test loss: 0.003495449898764491\n",
      "Test loss: 0.0035375540610402822\n",
      "Test loss: 0.004246184136718512\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss: 0.0033014100044965744\n",
      "Test loss: 0.0033447151072323322\n",
      "Test loss: 0.0032442593947052956\n",
      "Test loss: 0.004438549280166626\n",
      "Test loss: 0.005127688404172659\n",
      "Test loss: 0.0037483011838048697\n",
      "Test loss: 0.0035433045122772455\n",
      "Test loss: 0.0041477978229522705\n",
      "Test: 79, loss: 0.003897071350365877\n",
      "Train epoch: 80\n",
      "train loss: 0.0034614333417266607\n",
      "train loss: 0.00430708983913064\n",
      "train loss: 0.004128694999963045\n",
      "train loss: 0.00344707234762609\n",
      "train loss: 0.0031221245881170034\n",
      "train loss: 0.004150166176259518\n",
      "train loss: 0.004574325401335955\n",
      "train loss: 0.0038187026511877775\n",
      "train loss: 0.003909693565219641\n",
      "train loss: 0.0033084466122090816\n",
      "train loss: 0.005591466557234526\n",
      "train loss: 0.0036530490033328533\n",
      "train loss: 0.003853384405374527\n",
      "train loss: 0.004693109076470137\n",
      "train loss: 0.004420123063027859\n",
      "train loss: 0.0042117503471672535\n",
      "train loss: 0.004736198112368584\n",
      "train loss: 0.0038716355338692665\n",
      "train loss: 0.003102498361840844\n",
      "train loss: 0.0042992644011974335\n",
      "train loss: 0.004865521099418402\n",
      "train loss: 0.004887615330517292\n",
      "train loss: 0.0039321305230259895\n",
      "train loss: 0.0035644646268337965\n",
      "train loss: 0.004002560395747423\n",
      "train loss: 0.0040910677053034306\n",
      "train loss: 0.004746477119624615\n",
      "train loss: 0.0038750881794840097\n",
      "train loss: 0.004065230954438448\n",
      "train loss: 0.004565809387713671\n",
      "train loss: 0.004386685788631439\n",
      "Test loss: 0.0055709341540932655\n",
      "Test loss: 0.0032665750477463007\n",
      "Test loss: 0.0038615763187408447\n",
      "Test loss: 0.004035318270325661\n",
      "Test loss: 0.0030312109738588333\n",
      "Test loss: 0.003752815769985318\n",
      "Test loss: 0.0038487380370497704\n",
      "Test loss: 0.0034502139315009117\n",
      "Test loss: 0.004821659065783024\n",
      "Test loss: 0.0029510329477488995\n",
      "Test loss: 0.0033486029133200645\n",
      "Test loss: 0.003828698769211769\n",
      "Test loss: 0.003801860846579075\n",
      "Test loss: 0.0034043302293866873\n",
      "Test loss: 0.004427896346896887\n",
      "Test loss: 0.0036538979038596153\n",
      "Test loss: 0.0032932953909039497\n",
      "Test loss: 0.0034702213015407324\n",
      "Test loss: 0.00443975068628788\n",
      "Test loss: 0.00291250622831285\n",
      "Test loss: 0.0038775058928877115\n",
      "Test loss: 0.004698525182902813\n",
      "Test loss: 0.004244839772582054\n",
      "Test loss: 0.003988547716289759\n",
      "Test loss: 0.0038477813359349966\n",
      "Test loss: 0.00298455567099154\n",
      "Test loss: 0.002980517689138651\n",
      "Test loss: 0.004076147451996803\n",
      "Test loss: 0.004487635102123022\n",
      "Test loss: 0.003944713622331619\n",
      "Test loss: 0.004239072557538748\n",
      "Test: 80, loss: 0.003823901992291212\n",
      "Train epoch: 81\n",
      "train loss: 0.0034894654527306557\n",
      "train loss: 0.003720651613548398\n",
      "train loss: 0.0033954556565731764\n",
      "train loss: 0.006580858025699854\n",
      "train loss: 0.003870287211611867\n",
      "train loss: 0.003457201411947608\n",
      "train loss: 0.0036298339255154133\n",
      "train loss: 0.003907403908669949\n",
      "train loss: 0.003983700647950172\n",
      "train loss: 0.004063504282385111\n",
      "train loss: 0.003988135606050491\n",
      "train loss: 0.004460630472749472\n",
      "train loss: 0.004687250126153231\n",
      "train loss: 0.003449567360803485\n",
      "train loss: 0.004733574111014605\n",
      "train loss: 0.003067541169002652\n",
      "train loss: 0.0037043497432023287\n",
      "train loss: 0.003812558250501752\n",
      "train loss: 0.004808810073882341\n",
      "train loss: 0.004023282788693905\n",
      "train loss: 0.0034935218282043934\n",
      "train loss: 0.004277972970157862\n",
      "train loss: 0.0045195710845291615\n",
      "train loss: 0.004576258827000856\n",
      "train loss: 0.005597337149083614\n",
      "train loss: 0.005546306259930134\n",
      "train loss: 0.0046049910597503185\n",
      "train loss: 0.0036289694253355265\n",
      "train loss: 0.0043558343313634396\n",
      "train loss: 0.00376917258836329\n",
      "train loss: 0.004738930147141218\n",
      "Test loss: 0.004107112064957619\n",
      "Test loss: 0.0038994213100522757\n",
      "Test loss: 0.003923892043530941\n",
      "Test loss: 0.004016747232526541\n",
      "Test loss: 0.003882467746734619\n",
      "Test loss: 0.003242985112592578\n",
      "Test loss: 0.004458871204406023\n",
      "Test loss: 0.0043050083331763744\n",
      "Test loss: 0.0037248979788273573\n",
      "Test loss: 0.004641905426979065\n",
      "Test loss: 0.0036409511230885983\n",
      "Test loss: 0.004975855350494385\n",
      "Test loss: 0.0039127301424741745\n",
      "Test loss: 0.005037278402596712\n",
      "Test loss: 0.0040328591130673885\n",
      "Test loss: 0.003672183258458972\n",
      "Test loss: 0.003421102650463581\n",
      "Test loss: 0.005400065798312426\n",
      "Test loss: 0.0047621941193938255\n",
      "Test loss: 0.0041371784172952175\n",
      "Test loss: 0.003997222986072302\n",
      "Test loss: 0.0036137858405709267\n",
      "Test loss: 0.0033429113682359457\n",
      "Test loss: 0.0038118381053209305\n",
      "Test loss: 0.0032806003000587225\n",
      "Test loss: 0.0037157966289669275\n",
      "Test loss: 0.003635505447164178\n",
      "Test loss: 0.003999012056738138\n",
      "Test loss: 0.0039711082354187965\n",
      "Test loss: 0.004124883562326431\n",
      "Test loss: 0.004083265084773302\n",
      "Test: 81, loss: 0.004024891648441553\n",
      "Train epoch: 82\n",
      "train loss: 0.003911927808076143\n",
      "train loss: 0.0045652855187654495\n",
      "train loss: 0.0031097489409148693\n",
      "train loss: 0.0036769681610167027\n",
      "train loss: 0.005496248137205839\n",
      "train loss: 0.00406620092689991\n",
      "train loss: 0.0040108319371938705\n",
      "train loss: 0.003302567172795534\n",
      "train loss: 0.004245308693498373\n",
      "train loss: 0.004337762016803026\n",
      "train loss: 0.003466489026322961\n",
      "train loss: 0.003548910142853856\n",
      "train loss: 0.0040576099418103695\n",
      "train loss: 0.003171412507072091\n",
      "train loss: 0.004833970218896866\n",
      "train loss: 0.003306012600660324\n",
      "train loss: 0.0038183594588190317\n",
      "train loss: 0.004190810024738312\n",
      "train loss: 0.00389291113242507\n",
      "train loss: 0.004352814983576536\n",
      "train loss: 0.0037788336630910635\n",
      "train loss: 0.005894297268241644\n",
      "train loss: 0.004985359031707048\n",
      "train loss: 0.003483173903077841\n",
      "train loss: 0.00477970065549016\n",
      "train loss: 0.0047821952030062675\n",
      "train loss: 0.0035531201865524054\n",
      "train loss: 0.004878307227045298\n",
      "train loss: 0.0039674509316682816\n",
      "train loss: 0.005392939783632755\n",
      "train loss: 0.0037068496458232403\n",
      "Test loss: 0.0036394172348082066\n",
      "Test loss: 0.004099396523088217\n",
      "Test loss: 0.0032367503736168146\n",
      "Test loss: 0.005033859983086586\n",
      "Test loss: 0.004537722561508417\n",
      "Test loss: 0.004129126667976379\n",
      "Test loss: 0.0033548916690051556\n",
      "Test loss: 0.003674545558169484\n",
      "Test loss: 0.0035064364783465862\n",
      "Test loss: 0.003990907222032547\n",
      "Test loss: 0.004262273665517569\n",
      "Test loss: 0.002989798318594694\n",
      "Test loss: 0.00443900004029274\n",
      "Test loss: 0.0039294688031077385\n",
      "Test loss: 0.004017401486635208\n",
      "Test loss: 0.0038824137300252914\n",
      "Test loss: 0.004070769529789686\n",
      "Test loss: 0.003607373684644699\n",
      "Test loss: 0.004792975727468729\n",
      "Test loss: 0.004285302013158798\n",
      "Test loss: 0.0037905850913375616\n",
      "Test loss: 0.0036905910819768906\n",
      "Test loss: 0.004693168215453625\n",
      "Test loss: 0.004233529791235924\n",
      "Test loss: 0.0035606571473181248\n",
      "Test loss: 0.0035567698068916798\n",
      "Test loss: 0.0042449408210814\n",
      "Test loss: 0.004565689247101545\n",
      "Test loss: 0.0034551385324448347\n",
      "Test loss: 0.0048259347677230835\n",
      "Test loss: 0.004271798301488161\n",
      "Test: 82, loss: 0.004011891316622496\n",
      "Train epoch: 83\n",
      "train loss: 0.00448672054335475\n",
      "train loss: 0.004132410511374474\n",
      "train loss: 0.00455284072086215\n",
      "train loss: 0.005329018458724022\n",
      "train loss: 0.004307341296225786\n",
      "train loss: 0.0037296435330063105\n",
      "train loss: 0.004461517557501793\n",
      "train loss: 0.004713909234851599\n",
      "train loss: 0.0040771919302642345\n",
      "train loss: 0.003976501524448395\n",
      "train loss: 0.00394901679828763\n",
      "train loss: 0.004295860417187214\n",
      "train loss: 0.004952050279825926\n",
      "train loss: 0.0063505773432552814\n",
      "train loss: 0.003568092128261924\n",
      "train loss: 0.004609564784914255\n",
      "train loss: 0.0037805039901286364\n",
      "train loss: 0.004604725632816553\n",
      "train loss: 0.00421645725145936\n",
      "train loss: 0.004803809802979231\n",
      "train loss: 0.0038670937065035105\n",
      "train loss: 0.004008305259048939\n",
      "train loss: 0.0041150618344545364\n",
      "train loss: 0.003465482732281089\n",
      "train loss: 0.0034528260584920645\n",
      "train loss: 0.004227465484291315\n",
      "train loss: 0.003989878576248884\n",
      "train loss: 0.0037669066805392504\n",
      "train loss: 0.00339277065359056\n",
      "train loss: 0.00352791347540915\n",
      "train loss: 0.004367014393210411\n",
      "Test loss: 0.004870452452450991\n",
      "Test loss: 0.004025549162179232\n",
      "Test loss: 0.002987490501254797\n",
      "Test loss: 0.0033525191247463226\n",
      "Test loss: 0.003361728275194764\n",
      "Test loss: 0.0032994216307997704\n",
      "Test loss: 0.0041999369859695435\n",
      "Test loss: 0.0046707382425665855\n",
      "Test loss: 0.004619916435331106\n",
      "Test loss: 0.0053816866129636765\n",
      "Test loss: 0.004394959192723036\n",
      "Test loss: 0.004215956199914217\n",
      "Test loss: 0.0037569417618215084\n",
      "Test loss: 0.003975352738052607\n",
      "Test loss: 0.004233060870319605\n",
      "Test loss: 0.0036999552976340055\n",
      "Test loss: 0.004616418853402138\n",
      "Test loss: 0.004783837124705315\n",
      "Test loss: 0.00410253182053566\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss: 0.0048082321882247925\n",
      "Test loss: 0.0037246188148856163\n",
      "Test loss: 0.004499560222029686\n",
      "Test loss: 0.00373618071898818\n",
      "Test loss: 0.0042308056727051735\n",
      "Test loss: 0.0038392599672079086\n",
      "Test loss: 0.00412960397079587\n",
      "Test loss: 0.005238379817456007\n",
      "Test loss: 0.003141292603686452\n",
      "Test loss: 0.004691274370998144\n",
      "Test loss: 0.003119422821328044\n",
      "Test loss: 0.003727274015545845\n",
      "Test: 83, loss: 0.0041107856668531895\n",
      "Train epoch: 84\n",
      "train loss: 0.0034761035349220037\n",
      "train loss: 0.003675841959193349\n",
      "train loss: 0.0038103603292256594\n",
      "train loss: 0.004521042108535767\n",
      "train loss: 0.003707696683704853\n",
      "train loss: 0.004121172707527876\n",
      "train loss: 0.0051488145254552364\n",
      "train loss: 0.00464163301512599\n",
      "train loss: 0.005273598246276379\n",
      "train loss: 0.0039428723976016045\n",
      "train loss: 0.0035482551902532578\n",
      "train loss: 0.004026375710964203\n",
      "train loss: 0.004130552057176828\n",
      "train loss: 0.0037690105382353067\n",
      "train loss: 0.004459441639482975\n",
      "train loss: 0.0035813504364341497\n",
      "train loss: 0.0038212090730667114\n",
      "train loss: 0.00418687192723155\n",
      "train loss: 0.004394026473164558\n",
      "train loss: 0.003329564817249775\n",
      "train loss: 0.0049529606476426125\n",
      "train loss: 0.005091142375022173\n",
      "train loss: 0.0031739736441522837\n",
      "train loss: 0.003487429115921259\n",
      "train loss: 0.003947784658521414\n",
      "train loss: 0.004317580256611109\n",
      "train loss: 0.0038697575218975544\n",
      "train loss: 0.005454779602587223\n",
      "train loss: 0.0034665127750486135\n",
      "train loss: 0.005238138604909182\n",
      "train loss: 0.004363836720585823\n",
      "Test loss: 0.004548501688987017\n",
      "Test loss: 0.0040731970220804214\n",
      "Test loss: 0.003592538880184293\n",
      "Test loss: 0.0034692715853452682\n",
      "Test loss: 0.003877786220982671\n",
      "Test loss: 0.004442843142896891\n",
      "Test loss: 0.0034602743107825518\n",
      "Test loss: 0.003251677844673395\n",
      "Test loss: 0.004641397390514612\n",
      "Test loss: 0.0038761983159929514\n",
      "Test loss: 0.0044097257778048515\n",
      "Test loss: 0.0044243899174034595\n",
      "Test loss: 0.0038342152256518602\n",
      "Test loss: 0.0043222312815487385\n",
      "Test loss: 0.003961586393415928\n",
      "Test loss: 0.0036400523968040943\n",
      "Test loss: 0.0034791412763297558\n",
      "Test loss: 0.004453206434845924\n",
      "Test loss: 0.003532694885507226\n",
      "Test loss: 0.004205492325127125\n",
      "Test loss: 0.003453665180131793\n",
      "Test loss: 0.003268103813752532\n",
      "Test loss: 0.0038101663812994957\n",
      "Test loss: 0.0035028993152081966\n",
      "Test loss: 0.0056627183221280575\n",
      "Test loss: 0.004887002054601908\n",
      "Test loss: 0.00423330208286643\n",
      "Test loss: 0.0035089165903627872\n",
      "Test loss: 0.003568632761016488\n",
      "Test loss: 0.0054474626667797565\n",
      "Test loss: 0.004209431819617748\n",
      "Test: 84, loss: 0.004033830016851425\n",
      "Train epoch: 85\n",
      "train loss: 0.004131916910409927\n",
      "train loss: 0.0042685410007834435\n",
      "train loss: 0.0041690184734761715\n",
      "train loss: 0.0036699986085295677\n",
      "train loss: 0.00424949498847127\n",
      "train loss: 0.004018633626401424\n",
      "train loss: 0.004198606591671705\n",
      "train loss: 0.0037081819027662277\n",
      "train loss: 0.004684621002525091\n",
      "train loss: 0.003708885982632637\n",
      "train loss: 0.003695129416882992\n",
      "train loss: 0.0033632663544267416\n",
      "train loss: 0.0038834521546959877\n",
      "train loss: 0.004947781562805176\n",
      "train loss: 0.004114622715860605\n",
      "train loss: 0.003935534041374922\n",
      "train loss: 0.005803985055536032\n",
      "train loss: 0.003950331825762987\n",
      "train loss: 0.003757631406188011\n",
      "train loss: 0.0037417493294924498\n",
      "train loss: 0.0037520925980061293\n",
      "train loss: 0.004579010885208845\n",
      "train loss: 0.005070890299975872\n",
      "train loss: 0.004760399926453829\n",
      "train loss: 0.0032222196459770203\n",
      "train loss: 0.0035510084126144648\n",
      "train loss: 0.003793564857915044\n",
      "train loss: 0.00400000112131238\n",
      "train loss: 0.0037106985691934824\n",
      "train loss: 0.0032333231065422297\n",
      "train loss: 0.004717485047876835\n",
      "Test loss: 0.005998615175485611\n",
      "Test loss: 0.0035709948278963566\n",
      "Test loss: 0.004022388719022274\n",
      "Test loss: 0.0038195529486984015\n",
      "Test loss: 0.0038714546244591475\n",
      "Test loss: 0.004021187778562307\n",
      "Test loss: 0.004940773360431194\n",
      "Test loss: 0.005294010974466801\n",
      "Test loss: 0.0036387061700224876\n",
      "Test loss: 0.0037537170574069023\n",
      "Test loss: 0.003629181534051895\n",
      "Test loss: 0.004751984030008316\n",
      "Test loss: 0.0038635788951069117\n",
      "Test loss: 0.004066301044076681\n",
      "Test loss: 0.0031759703997522593\n",
      "Test loss: 0.0052437433041632175\n",
      "Test loss: 0.003415341954678297\n",
      "Test loss: 0.0041330247186124325\n",
      "Test loss: 0.00403304910287261\n",
      "Test loss: 0.0033813740592449903\n",
      "Test loss: 0.0038036485202610493\n",
      "Test loss: 0.0047699767164886\n",
      "Test loss: 0.003886088728904724\n",
      "Test loss: 0.0035330448299646378\n",
      "Test loss: 0.004029178060591221\n",
      "Test loss: 0.004343956243246794\n",
      "Test loss: 0.00382155436091125\n",
      "Test loss: 0.003541202750056982\n",
      "Test loss: 0.004591324366629124\n",
      "Test loss: 0.004512856714427471\n",
      "Test loss: 0.0045225354842841625\n",
      "Test: 85, loss: 0.004128396976739168\n",
      "Train epoch: 86\n",
      "train loss: 0.004203948192298412\n",
      "train loss: 0.005439083091914654\n",
      "train loss: 0.002816237276419997\n",
      "train loss: 0.0042460327968001366\n",
      "train loss: 0.006175803020596504\n",
      "train loss: 0.003534220391884446\n",
      "train loss: 0.0033271387219429016\n",
      "train loss: 0.004749505780637264\n",
      "train loss: 0.00412707170471549\n",
      "train loss: 0.004549155943095684\n",
      "train loss: 0.0037758992984890938\n",
      "train loss: 0.0033675741869956255\n",
      "train loss: 0.003855557646602392\n",
      "train loss: 0.0036809381563216448\n",
      "train loss: 0.0037565932143479586\n",
      "train loss: 0.0034542768262326717\n",
      "train loss: 0.00449735950678587\n",
      "train loss: 0.003339493880048394\n",
      "train loss: 0.003854232607409358\n",
      "train loss: 0.004554834682494402\n",
      "train loss: 0.0038293085526674986\n",
      "train loss: 0.0034820917062461376\n",
      "train loss: 0.005938399117439985\n",
      "train loss: 0.0039001978002488613\n",
      "train loss: 0.003188840812072158\n",
      "train loss: 0.00330151105299592\n",
      "train loss: 0.003934885840862989\n",
      "train loss: 0.003678201697766781\n",
      "train loss: 0.005606410559266806\n",
      "train loss: 0.0036843817215412855\n",
      "train loss: 0.003933715634047985\n",
      "Test loss: 0.0034674995113164186\n",
      "Test loss: 0.004272747319191694\n",
      "Test loss: 0.0038159722462296486\n",
      "Test loss: 0.004748134408146143\n",
      "Test loss: 0.0032606255263090134\n",
      "Test loss: 0.004323146305978298\n",
      "Test loss: 0.003309491788968444\n",
      "Test loss: 0.0029856211040169\n",
      "Test loss: 0.00439768610522151\n",
      "Test loss: 0.0036103306338191032\n",
      "Test loss: 0.003217295976355672\n",
      "Test loss: 0.0036210808902978897\n",
      "Test loss: 0.004905015230178833\n",
      "Test loss: 0.004644316155463457\n",
      "Test loss: 0.005849730223417282\n",
      "Test loss: 0.004496066831052303\n",
      "Test loss: 0.003430867800489068\n",
      "Test loss: 0.0031945162918418646\n",
      "Test loss: 0.004019563086330891\n",
      "Test loss: 0.0033885501325130463\n",
      "Test loss: 0.004165968392044306\n",
      "Test loss: 0.004337426740676165\n",
      "Test loss: 0.0035089647863060236\n",
      "Test loss: 0.004626317415386438\n",
      "Test loss: 0.003527318360283971\n",
      "Test loss: 0.00447138212621212\n",
      "Test loss: 0.0041656214743852615\n",
      "Test loss: 0.004201342351734638\n",
      "Test loss: 0.0039016336668282747\n",
      "Test loss: 0.0033672479912638664\n",
      "Test loss: 0.004479190334677696\n",
      "Test: 86, loss: 0.003990666940808296\n",
      "Train epoch: 87\n",
      "train loss: 0.005048722494393587\n",
      "train loss: 0.003792754840105772\n",
      "train loss: 0.004891266580671072\n",
      "train loss: 0.003893498331308365\n",
      "train loss: 0.0033018412068486214\n",
      "train loss: 0.0036011089105159044\n",
      "train loss: 0.003702818416059017\n",
      "train loss: 0.0054168542847037315\n",
      "train loss: 0.003067211015149951\n",
      "train loss: 0.003859942080453038\n",
      "train loss: 0.0037183379754424095\n",
      "train loss: 0.0034111186396330595\n",
      "train loss: 0.004593230783939362\n",
      "train loss: 0.0036110286600887775\n",
      "train loss: 0.004240902606397867\n",
      "train loss: 0.004128919914364815\n",
      "train loss: 0.003684146562591195\n",
      "train loss: 0.004180191084742546\n",
      "train loss: 0.00438232347369194\n",
      "train loss: 0.0031843308825045824\n",
      "train loss: 0.0034119628835469484\n",
      "train loss: 0.0042242249473929405\n",
      "train loss: 0.004685180261731148\n",
      "train loss: 0.0038291069213300943\n",
      "train loss: 0.0038425414822995663\n",
      "train loss: 0.004965611733496189\n",
      "train loss: 0.004521328490227461\n",
      "train loss: 0.0036822573747485876\n",
      "train loss: 0.00387963536195457\n",
      "train loss: 0.003925349097698927\n",
      "train loss: 0.0034503096248954535\n",
      "Test loss: 0.004096736200153828\n",
      "Test loss: 0.004093969240784645\n",
      "Test loss: 0.004463201854377985\n",
      "Test loss: 0.0037545976229012012\n",
      "Test loss: 0.004587744362652302\n",
      "Test loss: 0.005169251002371311\n",
      "Test loss: 0.004197976551949978\n",
      "Test loss: 0.003715592436492443\n",
      "Test loss: 0.003541774582117796\n",
      "Test loss: 0.005497801583260298\n",
      "Test loss: 0.004360928200185299\n",
      "Test loss: 0.004234603606164455\n",
      "Test loss: 0.004784247372299433\n",
      "Test loss: 0.003794668009504676\n",
      "Test loss: 0.00471818633377552\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss: 0.003023384604603052\n",
      "Test loss: 0.003913113381713629\n",
      "Test loss: 0.004570224322378635\n",
      "Test loss: 0.00372413732111454\n",
      "Test loss: 0.0031828999053686857\n",
      "Test loss: 0.0037519754841923714\n",
      "Test loss: 0.003845843253657222\n",
      "Test loss: 0.0034862179309129715\n",
      "Test loss: 0.004063216503709555\n",
      "Test loss: 0.004216432571411133\n",
      "Test loss: 0.003840900491923094\n",
      "Test loss: 0.00397276459261775\n",
      "Test loss: 0.0049002827145159245\n",
      "Test loss: 0.0034713500645011663\n",
      "Test loss: 0.004109727218747139\n",
      "Test loss: 0.003600779687985778\n",
      "Test: 87, loss: 0.004086597822606564\n",
      "Train epoch: 88\n",
      "train loss: 0.0033766392152756453\n",
      "train loss: 0.0037174972239881754\n",
      "train loss: 0.0038637237157672644\n",
      "train loss: 0.004116219934076071\n",
      "train loss: 0.003356280969455838\n",
      "train loss: 0.004440633114427328\n",
      "train loss: 0.004803314805030823\n",
      "train loss: 0.004812735598534346\n",
      "train loss: 0.006163276731967926\n",
      "train loss: 0.00422314740717411\n",
      "train loss: 0.0035594310611486435\n",
      "train loss: 0.004302531015127897\n",
      "train loss: 0.0035436470061540604\n",
      "train loss: 0.0052540176548063755\n",
      "train loss: 0.003274122718721628\n",
      "train loss: 0.004618474747985601\n",
      "train loss: 0.003855771152302623\n",
      "train loss: 0.004548140801489353\n",
      "train loss: 0.00318130268715322\n",
      "train loss: 0.004886202979832888\n",
      "train loss: 0.0032033706083893776\n",
      "train loss: 0.004111478570848703\n",
      "train loss: 0.003594359615817666\n",
      "train loss: 0.0032360758632421494\n",
      "train loss: 0.003843530546873808\n",
      "train loss: 0.0038153708446770906\n",
      "train loss: 0.0038892123848199844\n",
      "train loss: 0.0030953085515648127\n",
      "train loss: 0.0038071812596172094\n",
      "train loss: 0.004197473172098398\n",
      "train loss: 0.0037210131995379925\n",
      "Test loss: 0.003652070416137576\n",
      "Test loss: 0.004482611548155546\n",
      "Test loss: 0.004340250510722399\n",
      "Test loss: 0.0036858650855720043\n",
      "Test loss: 0.004105603322386742\n",
      "Test loss: 0.0042737508192658424\n",
      "Test loss: 0.003368583507835865\n",
      "Test loss: 0.0041198148392140865\n",
      "Test loss: 0.0033647932577878237\n",
      "Test loss: 0.004446146544069052\n",
      "Test loss: 0.004085029009729624\n",
      "Test loss: 0.003350375685840845\n",
      "Test loss: 0.0034534756559878588\n",
      "Test loss: 0.0036541768349707127\n",
      "Test loss: 0.005014019552618265\n",
      "Test loss: 0.003472489770501852\n",
      "Test loss: 0.003932362422347069\n",
      "Test loss: 0.003554216818884015\n",
      "Test loss: 0.004206087905913591\n",
      "Test loss: 0.0040411027148365974\n",
      "Test loss: 0.003671785583719611\n",
      "Test loss: 0.003737226128578186\n",
      "Test loss: 0.00481628580018878\n",
      "Test loss: 0.0056226979941129684\n",
      "Test loss: 0.003438555635511875\n",
      "Test loss: 0.004010138567537069\n",
      "Test loss: 0.004806397017091513\n",
      "Test loss: 0.004070529714226723\n",
      "Test loss: 0.005044386722147465\n",
      "Test loss: 0.003447729628533125\n",
      "Test loss: 0.0032769227400422096\n",
      "Test: 88, loss: 0.004017596133053303\n",
      "Train epoch: 89\n",
      "train loss: 0.0036518010310828686\n",
      "train loss: 0.004416323266923428\n",
      "train loss: 0.005277436226606369\n",
      "train loss: 0.003525790059939027\n",
      "train loss: 0.006106254644691944\n",
      "train loss: 0.003916879650205374\n",
      "train loss: 0.004223449621349573\n",
      "train loss: 0.005564333871006966\n",
      "train loss: 0.0038371901027858257\n",
      "train loss: 0.003782839747145772\n",
      "train loss: 0.003445153124630451\n",
      "train loss: 0.003554176539182663\n",
      "train loss: 0.003927296027541161\n",
      "train loss: 0.00344933127053082\n",
      "train loss: 0.00410939147695899\n",
      "train loss: 0.003673378610983491\n",
      "train loss: 0.0040581487119197845\n",
      "train loss: 0.003858645912259817\n",
      "train loss: 0.0037012803368270397\n",
      "train loss: 0.0038098564837127924\n",
      "train loss: 0.004909921437501907\n",
      "train loss: 0.004253576509654522\n",
      "train loss: 0.003464584704488516\n",
      "train loss: 0.0029395348392426968\n",
      "train loss: 0.0034488034434616566\n",
      "train loss: 0.004530527163296938\n",
      "train loss: 0.004159841220825911\n",
      "train loss: 0.004356802441179752\n",
      "train loss: 0.004534254781901836\n",
      "train loss: 0.003993418533354998\n",
      "train loss: 0.0037693872582167387\n",
      "Test loss: 0.003630568739026785\n",
      "Test loss: 0.004725094884634018\n",
      "Test loss: 0.0033916409593075514\n",
      "Test loss: 0.0037010617088526487\n",
      "Test loss: 0.0033644335344433784\n",
      "Test loss: 0.0031871795654296875\n",
      "Test loss: 0.004285277798771858\n",
      "Test loss: 0.003846528474241495\n",
      "Test loss: 0.003713025012984872\n",
      "Test loss: 0.004059406463056803\n",
      "Test loss: 0.004196637775748968\n",
      "Test loss: 0.00343987625092268\n",
      "Test loss: 0.004011745098978281\n",
      "Test loss: 0.003990659955888987\n",
      "Test loss: 0.003437690669670701\n",
      "Test loss: 0.004878371488302946\n",
      "Test loss: 0.00417535612359643\n",
      "Test loss: 0.004176916088908911\n",
      "Test loss: 0.003930815029889345\n",
      "Test loss: 0.0032382274512201548\n",
      "Test loss: 0.0037772529758512974\n",
      "Test loss: 0.00332399713806808\n",
      "Test loss: 0.003989954013377428\n",
      "Test loss: 0.004018889274448156\n",
      "Test loss: 0.004040000960230827\n",
      "Test loss: 0.0030459912959486246\n",
      "Test loss: 0.0035745701752603054\n",
      "Test loss: 0.003933062311261892\n",
      "Test loss: 0.003658437402918935\n",
      "Test loss: 0.003119755769148469\n",
      "Test loss: 0.0029419446364045143\n",
      "Test: 89, loss: 0.0037678831722587347\n",
      "Train epoch: 90\n",
      "train loss: 0.004542505368590355\n",
      "train loss: 0.0031169711146503687\n",
      "train loss: 0.003412021556869149\n",
      "train loss: 0.0038611781783401966\n",
      "train loss: 0.004068498499691486\n",
      "train loss: 0.003317034337669611\n",
      "train loss: 0.0046665421687066555\n",
      "train loss: 0.003050709841772914\n",
      "train loss: 0.0032879256177693605\n",
      "train loss: 0.0046867383643984795\n",
      "train loss: 0.0034032142721116543\n",
      "train loss: 0.003544293576851487\n",
      "train loss: 0.0035579081159085035\n",
      "train loss: 0.003767828457057476\n",
      "train loss: 0.003260525641962886\n",
      "train loss: 0.0036677021998912096\n",
      "train loss: 0.005998668260872364\n",
      "train loss: 0.0031264580320566893\n",
      "train loss: 0.0036241787020117044\n",
      "train loss: 0.005026719998568296\n",
      "train loss: 0.003923754207789898\n",
      "train loss: 0.0037202041130512953\n",
      "train loss: 0.003656691173091531\n",
      "train loss: 0.004722428973764181\n",
      "train loss: 0.004307104740291834\n",
      "train loss: 0.004501246381551027\n",
      "train loss: 0.0036461083218455315\n",
      "train loss: 0.00509118614718318\n",
      "train loss: 0.005226216744631529\n",
      "train loss: 0.00354302441701293\n",
      "train loss: 0.004148046486079693\n",
      "Test loss: 0.004022439010441303\n",
      "Test loss: 0.0029661611188203096\n",
      "Test loss: 0.0035506533458828926\n",
      "Test loss: 0.003447870025411248\n",
      "Test loss: 0.002940298756584525\n",
      "Test loss: 0.0034635362681001425\n",
      "Test loss: 0.0045723700895905495\n",
      "Test loss: 0.0039927763864398\n",
      "Test loss: 0.0042992765083909035\n",
      "Test loss: 0.003731327597051859\n",
      "Test loss: 0.006295884493738413\n",
      "Test loss: 0.0036905761808156967\n",
      "Test loss: 0.0032254818361252546\n",
      "Test loss: 0.004400651436299086\n",
      "Test loss: 0.0033311699517071247\n",
      "Test loss: 0.004458082374185324\n",
      "Test loss: 0.0036212485283613205\n",
      "Test loss: 0.004716997034847736\n",
      "Test loss: 0.002778760390356183\n",
      "Test loss: 0.003776877885684371\n",
      "Test loss: 0.004130238201469183\n",
      "Test loss: 0.003627790603786707\n",
      "Test loss: 0.00424711499363184\n",
      "Test loss: 0.0035920811351388693\n",
      "Test loss: 0.0032389871776103973\n",
      "Test loss: 0.0035898468922823668\n",
      "Test loss: 0.0036243346985429525\n",
      "Test loss: 0.003605079837143421\n",
      "Test loss: 0.003734088269993663\n",
      "Test loss: 0.0038330520037561655\n",
      "Test loss: 0.003217329503968358\n",
      "Test: 90, loss: 0.0037974966689944267\n",
      "Train epoch: 91\n",
      "train loss: 0.0038500186055898666\n",
      "train loss: 0.0035780256148427725\n",
      "train loss: 0.0030174851417541504\n",
      "train loss: 0.005166664719581604\n",
      "train loss: 0.0038759494200348854\n",
      "train loss: 0.003301020711660385\n",
      "train loss: 0.003939843270927668\n",
      "train loss: 0.004902176093310118\n",
      "train loss: 0.0035455054603517056\n",
      "train loss: 0.0034690278116613626\n",
      "train loss: 0.003629859769716859\n",
      "train loss: 0.004052337724715471\n",
      "train loss: 0.0031085095833986998\n",
      "train loss: 0.003197331679984927\n",
      "train loss: 0.0036502578295767307\n",
      "train loss: 0.004584404174238443\n",
      "train loss: 0.003344545606523752\n",
      "train loss: 0.004660675767809153\n",
      "train loss: 0.004138332325965166\n",
      "train loss: 0.0031352115329355\n",
      "train loss: 0.0032335538417100906\n",
      "train loss: 0.003136603394523263\n",
      "train loss: 0.004217908252030611\n",
      "train loss: 0.0038987770676612854\n",
      "train loss: 0.003610201645642519\n",
      "train loss: 0.00349780754186213\n",
      "train loss: 0.0034738897811621428\n",
      "train loss: 0.004963939543813467\n",
      "train loss: 0.0036927838809788227\n",
      "train loss: 0.0041840109042823315\n",
      "train loss: 0.0036634046118706465\n",
      "Test loss: 0.004133645445108414\n",
      "Test loss: 0.0035939589142799377\n",
      "Test loss: 0.0040231384336948395\n",
      "Test loss: 0.0038537802174687386\n",
      "Test loss: 0.003607220249250531\n",
      "Test loss: 0.004539430141448975\n",
      "Test loss: 0.004053497686982155\n",
      "Test loss: 0.0034985861275345087\n",
      "Test loss: 0.004083242733031511\n",
      "Test loss: 0.0034406017512083054\n",
      "Test loss: 0.004142485558986664\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss: 0.00445768516510725\n",
      "Test loss: 0.004348495043814182\n",
      "Test loss: 0.003724735928699374\n",
      "Test loss: 0.004122462123632431\n",
      "Test loss: 0.002950109541416168\n",
      "Test loss: 0.004403457976877689\n",
      "Test loss: 0.0039642308838665485\n",
      "Test loss: 0.004490491468459368\n",
      "Test loss: 0.003879455616697669\n",
      "Test loss: 0.0036914239171892405\n",
      "Test loss: 0.003777232486754656\n",
      "Test loss: 0.00489745894446969\n",
      "Test loss: 0.0036253167781978846\n",
      "Test loss: 0.0034563466906547546\n",
      "Test loss: 0.0038786581717431545\n",
      "Test loss: 0.0036232853308320045\n",
      "Test loss: 0.00339215574786067\n",
      "Test loss: 0.004841381683945656\n",
      "Test loss: 0.0039061608258634806\n",
      "Test loss: 0.004153602756559849\n",
      "Test: 91, loss: 0.00395334605127573\n",
      "Train epoch: 92\n",
      "train loss: 0.0036389685701578856\n",
      "train loss: 0.0041333516128361225\n",
      "train loss: 0.003540374105796218\n",
      "train loss: 0.004067845176905394\n",
      "train loss: 0.00352105381898582\n",
      "train loss: 0.0043918355368077755\n",
      "train loss: 0.0035025898832827806\n",
      "train loss: 0.003727705217897892\n",
      "train loss: 0.003412202699109912\n",
      "train loss: 0.0035744078923016787\n",
      "train loss: 0.004631577525287867\n",
      "train loss: 0.0053768400102853775\n",
      "train loss: 0.003154459409415722\n",
      "train loss: 0.0036355273332446814\n",
      "train loss: 0.0038791950792074203\n",
      "train loss: 0.004378955345600843\n",
      "train loss: 0.003535094205290079\n",
      "train loss: 0.004346086177974939\n",
      "train loss: 0.003002146491780877\n",
      "train loss: 0.00393593218177557\n",
      "train loss: 0.003794439136981964\n",
      "train loss: 0.005060963798314333\n",
      "train loss: 0.0038416660390794277\n",
      "train loss: 0.004083987791091204\n",
      "train loss: 0.003406979376450181\n",
      "train loss: 0.005942704156041145\n",
      "train loss: 0.003451169468462467\n",
      "train loss: 0.003901933552697301\n",
      "train loss: 0.003765976754948497\n",
      "train loss: 0.004729213658720255\n",
      "train loss: 0.0040377420373260975\n",
      "Test loss: 0.0036743367090821266\n",
      "Test loss: 0.003073198487982154\n",
      "Test loss: 0.0033444745931774378\n",
      "Test loss: 0.004486371297389269\n",
      "Test loss: 0.00356704811565578\n",
      "Test loss: 0.003915443085134029\n",
      "Test loss: 0.0036119362339377403\n",
      "Test loss: 0.00468253530561924\n",
      "Test loss: 0.004106719978153706\n",
      "Test loss: 0.0033564914483577013\n",
      "Test loss: 0.0037420783191919327\n",
      "Test loss: 0.0035242459271103144\n",
      "Test loss: 0.004495423287153244\n",
      "Test loss: 0.0028236843645572662\n",
      "Test loss: 0.0036292443983256817\n",
      "Test loss: 0.003520460333675146\n",
      "Test loss: 0.0035711033269762993\n",
      "Test loss: 0.0037967402022331953\n",
      "Test loss: 0.003245825879275799\n",
      "Test loss: 0.0037986580282449722\n",
      "Test loss: 0.0035084073897451162\n",
      "Test loss: 0.004844482988119125\n",
      "Test loss: 0.003650240832939744\n",
      "Test loss: 0.003855686401948333\n",
      "Test loss: 0.0035009924322366714\n",
      "Test loss: 0.003238049102947116\n",
      "Test loss: 0.003396612359210849\n",
      "Test loss: 0.003240233287215233\n",
      "Test loss: 0.003644669195637107\n",
      "Test loss: 0.0037815789692103863\n",
      "Test loss: 0.0037089884281158447\n",
      "Test: 92, loss: 0.0036882569547742605\n",
      "Train epoch: 93\n",
      "train loss: 0.0035191092174500227\n",
      "train loss: 0.004081335384398699\n",
      "train loss: 0.0038979037199169397\n",
      "train loss: 0.003927990794181824\n",
      "train loss: 0.0036894234362989664\n",
      "train loss: 0.005397154483944178\n",
      "train loss: 0.0033712186850607395\n",
      "train loss: 0.0033916921820491552\n",
      "train loss: 0.004078385885804892\n",
      "train loss: 0.004770029801875353\n",
      "train loss: 0.003539563622325659\n",
      "train loss: 0.003760459367185831\n",
      "train loss: 0.004152185749262571\n",
      "train loss: 0.00491543160751462\n",
      "train loss: 0.0049699340015649796\n",
      "train loss: 0.0037557620089501143\n",
      "train loss: 0.004000471439212561\n",
      "train loss: 0.005988134071230888\n",
      "train loss: 0.005018547642976046\n",
      "train loss: 0.003779471619054675\n",
      "train loss: 0.004201395902782679\n",
      "train loss: 0.0036687981337308884\n",
      "train loss: 0.003459976287558675\n",
      "train loss: 0.0038992706686258316\n",
      "train loss: 0.004472596105188131\n",
      "train loss: 0.004042454529553652\n",
      "train loss: 0.003343552118167281\n",
      "train loss: 0.0041019064374268055\n",
      "train loss: 0.003638216760009527\n",
      "train loss: 0.0038986322470009327\n",
      "train loss: 0.003109419485554099\n",
      "Test loss: 0.0031936841551214457\n",
      "Test loss: 0.003857171628624201\n",
      "Test loss: 0.004538675304502249\n",
      "Test loss: 0.004991469439119101\n",
      "Test loss: 0.0049661085940897465\n",
      "Test loss: 0.003446483053267002\n",
      "Test loss: 0.003601481905207038\n",
      "Test loss: 0.00356830982491374\n",
      "Test loss: 0.003988480661064386\n",
      "Test loss: 0.00361045403406024\n",
      "Test loss: 0.00456606550142169\n",
      "Test loss: 0.003626363817602396\n",
      "Test loss: 0.0035026089753955603\n",
      "Test loss: 0.0037475936114788055\n",
      "Test loss: 0.004493548069149256\n",
      "Test loss: 0.0031661498360335827\n",
      "Test loss: 0.0044423965737223625\n",
      "Test loss: 0.004000155255198479\n",
      "Test loss: 0.004007833078503609\n",
      "Test loss: 0.004325220361351967\n",
      "Test loss: 0.0038390392437577248\n",
      "Test loss: 0.004740238655358553\n",
      "Test loss: 0.003781923558562994\n",
      "Test loss: 0.003038075752556324\n",
      "Test loss: 0.004115663468837738\n",
      "Test loss: 0.004383582156151533\n",
      "Test loss: 0.003511439310386777\n",
      "Test loss: 0.0038920396473258734\n",
      "Test loss: 0.0038458502385765314\n",
      "Test loss: 0.0037103600334376097\n",
      "Test loss: 0.004197161179035902\n",
      "Test: 93, loss: 0.003957923036068678\n",
      "Train epoch: 94\n",
      "train loss: 0.0037280593533068895\n",
      "train loss: 0.004173451103270054\n",
      "train loss: 0.004249900579452515\n",
      "train loss: 0.003954451065510511\n",
      "train loss: 0.004661334212869406\n",
      "train loss: 0.003804311854764819\n",
      "train loss: 0.0036324721295386553\n",
      "train loss: 0.0033841377589851618\n",
      "train loss: 0.0047336420975625515\n",
      "train loss: 0.004565808922052383\n",
      "train loss: 0.003306978615000844\n",
      "train loss: 0.005560833960771561\n",
      "train loss: 0.0037320726551115513\n",
      "train loss: 0.003425422590225935\n",
      "train loss: 0.005089355632662773\n",
      "train loss: 0.0037235659547150135\n",
      "train loss: 0.004131863359361887\n",
      "train loss: 0.003931220155209303\n",
      "train loss: 0.004524982534348965\n",
      "train loss: 0.00483199767768383\n",
      "train loss: 0.004097548313438892\n",
      "train loss: 0.0045998976565897465\n",
      "train loss: 0.0033603983465582132\n",
      "train loss: 0.0037181328516453505\n",
      "train loss: 0.004319414496421814\n",
      "train loss: 0.004265023861080408\n",
      "train loss: 0.0050115445628762245\n",
      "train loss: 0.00459959777072072\n",
      "train loss: 0.004505934193730354\n",
      "train loss: 0.004486085847020149\n",
      "train loss: 0.004119128920137882\n",
      "Test loss: 0.0035122931003570557\n",
      "Test loss: 0.0036149881780147552\n",
      "Test loss: 0.00459012808278203\n",
      "Test loss: 0.0039648995734751225\n",
      "Test loss: 0.005149481818079948\n",
      "Test loss: 0.004272273276001215\n",
      "Test loss: 0.003810793161392212\n",
      "Test loss: 0.003720667213201523\n",
      "Test loss: 0.004121204372495413\n",
      "Test loss: 0.004699021577835083\n",
      "Test loss: 0.005137765314429998\n",
      "Test loss: 0.0040819053538143635\n",
      "Test loss: 0.004867750220000744\n",
      "Test loss: 0.0037584968376904726\n",
      "Test loss: 0.0041521950624883175\n",
      "Test loss: 0.0033489814959466457\n",
      "Test loss: 0.0032781169284135103\n",
      "Test loss: 0.004623422399163246\n",
      "Test loss: 0.0037090936675667763\n",
      "Test loss: 0.0032322171609848738\n",
      "Test loss: 0.00341813825070858\n",
      "Test loss: 0.0036440182011574507\n",
      "Test loss: 0.003620671574026346\n",
      "Test loss: 0.005047148559242487\n",
      "Test loss: 0.003606298938393593\n",
      "Test loss: 0.003964236006140709\n",
      "Test loss: 0.004165287595242262\n",
      "Test loss: 0.004078720696270466\n",
      "Test loss: 0.003251273650676012\n",
      "Test loss: 0.005728472489863634\n",
      "Test loss: 0.0035301174502819777\n",
      "Test: 94, loss: 0.004054841585457325\n",
      "Train epoch: 95\n",
      "train loss: 0.0036038316320627928\n",
      "train loss: 0.003419952467083931\n",
      "train loss: 0.004438130650669336\n",
      "train loss: 0.003955492749810219\n",
      "train loss: 0.005267517641186714\n",
      "train loss: 0.0034193319734185934\n",
      "train loss: 0.0038554498460143805\n",
      "train loss: 0.0033895622473210096\n",
      "train loss: 0.004259556066244841\n",
      "train loss: 0.0047657364048063755\n",
      "train loss: 0.004524235147982836\n",
      "train loss: 0.004948075860738754\n",
      "train loss: 0.003613097593188286\n",
      "train loss: 0.004267677199095488\n",
      "train loss: 0.0051571945659816265\n",
      "train loss: 0.003963975235819817\n",
      "train loss: 0.004125313367694616\n",
      "train loss: 0.003457835875451565\n",
      "train loss: 0.0031769340857863426\n",
      "train loss: 0.0050632706843316555\n",
      "train loss: 0.003726270981132984\n",
      "train loss: 0.005038280505686998\n",
      "train loss: 0.004387010354548693\n",
      "train loss: 0.002726761857047677\n",
      "train loss: 0.003223155392333865\n",
      "train loss: 0.0032986451406031847\n",
      "train loss: 0.004500268492847681\n",
      "train loss: 0.004330669529736042\n",
      "train loss: 0.0041680969297885895\n",
      "train loss: 0.003832272719591856\n",
      "train loss: 0.003940668888390064\n",
      "Test loss: 0.004230253864079714\n",
      "Test loss: 0.0034804223105311394\n",
      "Test loss: 0.004516508895903826\n",
      "Test loss: 0.005876755341887474\n",
      "Test loss: 0.003669171826913953\n",
      "Test loss: 0.004620199091732502\n",
      "Test loss: 0.005068160593509674\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss: 0.003913696855306625\n",
      "Test loss: 0.0034862966276705265\n",
      "Test loss: 0.003252248279750347\n",
      "Test loss: 0.0036302232183516026\n",
      "Test loss: 0.0038508388679474592\n",
      "Test loss: 0.003708477132022381\n",
      "Test loss: 0.003724301466718316\n",
      "Test loss: 0.005033375229686499\n",
      "Test loss: 0.0033515947870910168\n",
      "Test loss: 0.002865471178665757\n",
      "Test loss: 0.004646637476980686\n",
      "Test loss: 0.005041217897087336\n",
      "Test loss: 0.0042295316234230995\n",
      "Test loss: 0.003288418985903263\n",
      "Test loss: 0.0038218770641833544\n",
      "Test loss: 0.0034363409504294395\n",
      "Test loss: 0.003657624591141939\n",
      "Test loss: 0.0034963495563715696\n",
      "Test loss: 0.00428585521876812\n",
      "Test loss: 0.004711122252047062\n",
      "Test loss: 0.0032112731132656336\n",
      "Test loss: 0.0035471527371555567\n",
      "Test loss: 0.004558107350021601\n",
      "Test loss: 0.003555537201464176\n",
      "Test: 95, loss: 0.003992420621216297\n",
      "Train epoch: 96\n",
      "train loss: 0.004457457456737757\n",
      "train loss: 0.0045006051659584045\n",
      "train loss: 0.00368689582683146\n",
      "train loss: 0.004250470548868179\n",
      "train loss: 0.004797560628503561\n",
      "train loss: 0.004828717093914747\n",
      "train loss: 0.003983736503869295\n",
      "train loss: 0.0038291565142571926\n",
      "train loss: 0.00393178965896368\n",
      "train loss: 0.004082987550646067\n",
      "train loss: 0.0038211012724786997\n",
      "train loss: 0.004096855875104666\n",
      "train loss: 0.003332300577312708\n",
      "train loss: 0.005197320133447647\n",
      "train loss: 0.0028910019900649786\n",
      "train loss: 0.003754987381398678\n",
      "train loss: 0.00468077789992094\n",
      "train loss: 0.0034091072157025337\n",
      "train loss: 0.004129419103264809\n",
      "train loss: 0.004953640978783369\n",
      "train loss: 0.0031336364336311817\n",
      "train loss: 0.0037173181772232056\n",
      "train loss: 0.004401805344969034\n",
      "train loss: 0.004452754743397236\n",
      "train loss: 0.0034467007499188185\n",
      "train loss: 0.0032924336846917868\n",
      "train loss: 0.004906601272523403\n",
      "train loss: 0.003670166013762355\n",
      "train loss: 0.003211948787793517\n",
      "train loss: 0.0033937825355678797\n",
      "train loss: 0.00497702369466424\n",
      "Test loss: 0.004343518987298012\n",
      "Test loss: 0.0031619772780686617\n",
      "Test loss: 0.004109144676476717\n",
      "Test loss: 0.00323250493966043\n",
      "Test loss: 0.0037893904373049736\n",
      "Test loss: 0.0033527296036481857\n",
      "Test loss: 0.00419249152764678\n",
      "Test loss: 0.0033992649987339973\n",
      "Test loss: 0.003906394354999065\n",
      "Test loss: 0.003679323708638549\n",
      "Test loss: 0.0035017537884414196\n",
      "Test loss: 0.0037221089005470276\n",
      "Test loss: 0.0034553096629679203\n",
      "Test loss: 0.003671160899102688\n",
      "Test loss: 0.00393644068390131\n",
      "Test loss: 0.0031714627984911203\n",
      "Test loss: 0.003058968111872673\n",
      "Test loss: 0.0036495456006377935\n",
      "Test loss: 0.0039545465260744095\n",
      "Test loss: 0.0033855473157018423\n",
      "Test loss: 0.004935045726597309\n",
      "Test loss: 0.0033682577777653933\n",
      "Test loss: 0.003709933953359723\n",
      "Test loss: 0.003999623004347086\n",
      "Test loss: 0.004236151464283466\n",
      "Test loss: 0.00451282225549221\n",
      "Test loss: 0.004048073198646307\n",
      "Test loss: 0.003103467868641019\n",
      "Test loss: 0.002976442454382777\n",
      "Test loss: 0.003767599817365408\n",
      "Test loss: 0.0031552619766443968\n",
      "Test: 96, loss: 0.003693105187267065\n",
      "Train epoch: 97\n",
      "train loss: 0.0035768093075603247\n",
      "train loss: 0.003303683828562498\n",
      "train loss: 0.004966937471181154\n",
      "train loss: 0.004419711884111166\n",
      "train loss: 0.003782263956964016\n",
      "train loss: 0.0033627396915107965\n",
      "train loss: 0.00350011233240366\n",
      "train loss: 0.0040647778660058975\n",
      "train loss: 0.005103773903101683\n",
      "train loss: 0.0035221914295107126\n",
      "train loss: 0.0043072630651295185\n",
      "train loss: 0.00365457427687943\n",
      "train loss: 0.0042165108025074005\n",
      "train loss: 0.003250472480431199\n",
      "train loss: 0.0044717430137097836\n",
      "train loss: 0.004115567542612553\n",
      "train loss: 0.004204983357340097\n",
      "train loss: 0.0033953210804611444\n",
      "train loss: 0.0035789096727967262\n",
      "train loss: 0.0050915260799229145\n",
      "train loss: 0.0036201593466103077\n",
      "train loss: 0.0036738321650773287\n",
      "train loss: 0.004511399194598198\n",
      "train loss: 0.0031799785792827606\n",
      "train loss: 0.004331605974584818\n",
      "train loss: 0.004212831612676382\n",
      "train loss: 0.0054954164661467075\n",
      "train loss: 0.0032711171079427004\n",
      "train loss: 0.004715494345873594\n",
      "train loss: 0.003910069353878498\n",
      "train loss: 0.004431464709341526\n",
      "Test loss: 0.004405816551297903\n",
      "Test loss: 0.0034018508158624172\n",
      "Test loss: 0.003296412294730544\n",
      "Test loss: 0.0038639830891042948\n",
      "Test loss: 0.00388878071680665\n",
      "Test loss: 0.004409404005855322\n",
      "Test loss: 0.004381487611681223\n",
      "Test loss: 0.003652232000604272\n",
      "Test loss: 0.004279153421521187\n",
      "Test loss: 0.0033205372747033834\n",
      "Test loss: 0.005549433641135693\n",
      "Test loss: 0.003135927952826023\n",
      "Test loss: 0.003938740584999323\n",
      "Test loss: 0.003823470324277878\n",
      "Test loss: 0.0035628038458526134\n",
      "Test loss: 0.00370554206892848\n",
      "Test loss: 0.0033607392106205225\n",
      "Test loss: 0.002918069716542959\n",
      "Test loss: 0.00367877003736794\n",
      "Test loss: 0.0036602576728910208\n",
      "Test loss: 0.0031658331863582134\n",
      "Test loss: 0.004286852199584246\n",
      "Test loss: 0.003865538164973259\n",
      "Test loss: 0.00442434661090374\n",
      "Test loss: 0.003781119827181101\n",
      "Test loss: 0.003225067164748907\n",
      "Test loss: 0.005345392040908337\n",
      "Test loss: 0.004291218239814043\n",
      "Test loss: 0.0043761273846030235\n",
      "Test loss: 0.003968506120145321\n",
      "Test loss: 0.00483043072745204\n",
      "Test: 97, loss: 0.003928833641111851\n",
      "Train epoch: 98\n",
      "train loss: 0.004177596885710955\n",
      "train loss: 0.004122238606214523\n",
      "train loss: 0.003772319993004203\n",
      "train loss: 0.0044014607556164265\n",
      "train loss: 0.004416303709149361\n",
      "train loss: 0.003511162009090185\n",
      "train loss: 0.004319236613810062\n",
      "train loss: 0.003748679533600807\n",
      "train loss: 0.005245641339570284\n",
      "train loss: 0.004446167498826981\n",
      "train loss: 0.004011929035186768\n",
      "train loss: 0.003943498246371746\n",
      "train loss: 0.0037079821340739727\n",
      "train loss: 0.003707305993884802\n",
      "train loss: 0.005043309647589922\n",
      "train loss: 0.003679536981508136\n",
      "train loss: 0.0036741008516401052\n",
      "train loss: 0.002906060079112649\n",
      "train loss: 0.003279808908700943\n",
      "train loss: 0.003471702104434371\n",
      "train loss: 0.004083970561623573\n",
      "train loss: 0.004374406766146421\n",
      "train loss: 0.003347899066284299\n",
      "train loss: 0.003666358068585396\n",
      "train loss: 0.00354758370667696\n",
      "train loss: 0.00349382939748466\n",
      "train loss: 0.0032359508331865072\n",
      "train loss: 0.003325849771499634\n",
      "train loss: 0.0035421731881797314\n",
      "train loss: 0.002948635257780552\n",
      "train loss: 0.0035503751132637262\n",
      "Test loss: 0.004188994411379099\n",
      "Test loss: 0.0051375096663832664\n",
      "Test loss: 0.0030084375757724047\n",
      "Test loss: 0.003842143574729562\n",
      "Test loss: 0.004321899265050888\n",
      "Test loss: 0.00371755613014102\n",
      "Test loss: 0.005153949372470379\n",
      "Test loss: 0.005471993237733841\n",
      "Test loss: 0.005074929445981979\n",
      "Test loss: 0.0038675544783473015\n",
      "Test loss: 0.004963824991136789\n",
      "Test loss: 0.003498838748782873\n",
      "Test loss: 0.004791520535945892\n",
      "Test loss: 0.004050838761031628\n",
      "Test loss: 0.004068595357239246\n",
      "Test loss: 0.0038723209872841835\n",
      "Test loss: 0.00342434155754745\n",
      "Test loss: 0.004161073826253414\n",
      "Test loss: 0.0038218912668526173\n",
      "Test loss: 0.003482968546450138\n",
      "Test loss: 0.0034702441189438105\n",
      "Test loss: 0.004817906301468611\n",
      "Test loss: 0.0037449707742780447\n",
      "Test loss: 0.005558152683079243\n",
      "Test loss: 0.004534522071480751\n",
      "Test loss: 0.0037942782510071993\n",
      "Test loss: 0.00404268316924572\n",
      "Test loss: 0.0035850603599101305\n",
      "Test loss: 0.004672531969845295\n",
      "Test loss: 0.003050012281164527\n",
      "Test loss: 0.004553600214421749\n",
      "Test: 98, loss: 0.0041853273287415504\n",
      "Train epoch: 99\n",
      "train loss: 0.003814955009147525\n",
      "train loss: 0.0031466481741517782\n",
      "train loss: 0.0044649941846728325\n",
      "train loss: 0.003444004338234663\n",
      "train loss: 0.004181126598268747\n",
      "train loss: 0.003819255856797099\n",
      "train loss: 0.0034573967568576336\n",
      "train loss: 0.004450937267392874\n",
      "train loss: 0.00413527199998498\n",
      "train loss: 0.003798678983002901\n",
      "train loss: 0.004111698362976313\n",
      "train loss: 0.004154609516263008\n",
      "train loss: 0.003385761985555291\n",
      "train loss: 0.003729607444256544\n",
      "train loss: 0.0039191716350615025\n",
      "train loss: 0.003556782379746437\n",
      "train loss: 0.0033782748505473137\n",
      "train loss: 0.002881578868255019\n",
      "train loss: 0.003972774837166071\n",
      "train loss: 0.003614050569012761\n",
      "train loss: 0.0028901775367558002\n",
      "train loss: 0.0031706502195447683\n",
      "train loss: 0.004436795599758625\n",
      "train loss: 0.003592260181903839\n",
      "train loss: 0.003523858031257987\n",
      "train loss: 0.003995999228209257\n",
      "train loss: 0.003890097374096513\n",
      "train loss: 0.003781792474910617\n",
      "train loss: 0.0031056979205459356\n",
      "train loss: 0.004618421662598848\n",
      "train loss: 0.004898538812994957\n",
      "Test loss: 0.005049499217420816\n",
      "Test loss: 0.0037075169384479523\n",
      "Test loss: 0.0039508952759206295\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss: 0.0042497748509049416\n",
      "Test loss: 0.003280807286500931\n",
      "Test loss: 0.003594351466745138\n",
      "Test loss: 0.0052370429039001465\n",
      "Test loss: 0.003201701445505023\n",
      "Test loss: 0.004206730984151363\n",
      "Test loss: 0.0035991936456412077\n",
      "Test loss: 0.004516194574534893\n",
      "Test loss: 0.003939182497560978\n",
      "Test loss: 0.004385407082736492\n",
      "Test loss: 0.0030964768957346678\n",
      "Test loss: 0.00410750787705183\n",
      "Test loss: 0.0032824822701513767\n",
      "Test loss: 0.0034029053058475256\n",
      "Test loss: 0.0034738241229206324\n",
      "Test loss: 0.004014310892671347\n",
      "Test loss: 0.004178137984126806\n",
      "Test loss: 0.0035167778842151165\n",
      "Test loss: 0.003522291546687484\n",
      "Test loss: 0.0040511772967875\n",
      "Test loss: 0.0044252341613173485\n",
      "Test loss: 0.004581873770803213\n",
      "Test loss: 0.003921181429177523\n",
      "Test loss: 0.004694170318543911\n",
      "Test loss: 0.004534372128546238\n",
      "Test loss: 0.003887925297021866\n",
      "Test loss: 0.0035385247319936752\n",
      "Test loss: 0.004971656482666731\n",
      "Test: 99, loss: 0.004003842826932669\n",
      "Best epoch: 92\n"
     ]
    }
   ],
   "source": [
    "best_epoch = 0\n",
    "lowest_loss = 100\n",
    "train_loss_list = []\n",
    "dev_loss_list = []\n",
    "\n",
    "for j in range(epoch):\n",
    "    print('Train epoch: {}'.format(j))\n",
    "\n",
    "    snn.train()\n",
    "\n",
    "    train_loss = train(snn, optimizer, scheduler, train_dataloader)\n",
    "    train_loss_list.append(train_loss)\n",
    "\n",
    "    save_checkpoint = True\n",
    "    if save_checkpoint:\n",
    "        checkpoint_path = os.path.join(checkpoint_base_path, checkpoint_base_name + str(j))\n",
    "        torch.save({\n",
    "            'epoch': j,\n",
    "            'snn_state_dict': snn.state_dict(),\n",
    "            'optimizer_state_dict': optimizer.state_dict(),\n",
    "            'loss': train_loss,\n",
    "        }, checkpoint_path)\n",
    "    \n",
    "    snn.eval()\n",
    "    average_loss = test(snn, dev_dataloader)\n",
    "    dev_loss_list.append(average_loss)\n",
    "    if lowest_loss > average_loss:\n",
    "        lowest_loss = average_loss\n",
    "        best_epoch = j\n",
    "    \n",
    "    print('Test: {}, loss: {}'.format(j, average_loss))\n",
    "\n",
    "print('Best epoch: {}'.format(best_epoch))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "92 0.003688257\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(train_loss_list, label='train')\n",
    "plt.plot(dev_loss_list, label='dev')\n",
    "dev_index = np.argmin(dev_loss_list)\n",
    "print(dev_index, dev_loss_list[dev_index])\n",
    "plt.plot(dev_index, dev_loss_list[dev_index], 'ro')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(93, tensor(0.0031, device='cuda:2', grad_fn=<MseLossBackward>))"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_index = np.argmin(train_loss_list)\n",
    "train_index, train_loss_list[train_index]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Run the best epoch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Checkpoint: ./associative_memory_checkpoint/associative_memory_checkpoint_92 loaded\n",
      "0 tensor(0.0045, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "1 tensor(0.0041, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "2 tensor(0.0044, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "3 tensor(0.0048, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "4 tensor(0.0041, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "5 tensor(0.0035, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "6 tensor(0.0049, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "7 tensor(0.0037, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "8 tensor(0.0039, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "9 tensor(0.0040, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "10 tensor(0.0030, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "11 tensor(0.0043, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "12 tensor(0.0049, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "13 tensor(0.0044, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "14 tensor(0.0042, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "15 tensor(0.0044, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "16 tensor(0.0040, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "17 tensor(0.0039, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "18 tensor(0.0044, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "19 tensor(0.0046, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "20 tensor(0.0050, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "21 tensor(0.0034, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "22 tensor(0.0043, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "23 tensor(0.0042, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "24 tensor(0.0045, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "25 tensor(0.0041, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "26 tensor(0.0046, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "27 tensor(0.0036, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "28 tensor(0.0038, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "29 tensor(0.0041, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "30 tensor(0.0049, device='cuda:2', grad_fn=<MseLossBackward>)\n",
      "test loss tensor(0.0042, device='cuda:2', grad_fn=<DivBackward0>)\n"
     ]
    }
   ],
   "source": [
    "input_list = []\n",
    "output_list = []\n",
    "\n",
    "checkpoint_path = os.path.join(checkpoint_base_path, checkpoint_base_name + str(best_epoch))\n",
    "# checkpoint_path = './associative_memory_checkpoint/pretrained_associative_memory'\n",
    "checkpoint = torch.load(checkpoint_path)\n",
    "\n",
    "snn.load_state_dict(checkpoint[\"snn_state_dict\"])\n",
    "\n",
    "print('Checkpoint: {} loaded'.format(checkpoint_path))\n",
    "\n",
    "snn.eval()\n",
    "\n",
    "test_loss_list = []\n",
    "\n",
    "for i_batch, sample_batched in enumerate(test_dataloader):\n",
    "\n",
    "    criterion = torch.nn.MSELoss()\n",
    "    x_test = sample_batched[0].to(device)\n",
    "    label = sample_batched[1].to(device)\n",
    "    target_pattern = sample_batched[2].to(device)\n",
    "    filtered_target_pattern = sample_batched[3].to(device)\n",
    "\n",
    "    out_spike, filtered_out_spike = snn(x_test)\n",
    "\n",
    "    loss = criterion(filtered_out_spike, filtered_target_pattern)\n",
    "    test_loss_list.append(loss)\n",
    "    print(i_batch, loss)\n",
    "\n",
    "    input_list.append(x_test.cpu().detach().numpy())\n",
    "    output_list.append(out_spike.cpu().detach().numpy())\n",
    "print('test loss', sum(test_loss_list) / len(test_loss_list))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Plot some samples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 8 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 8 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axs = plt.subplots(2, 4)\n",
    "axs[0, 0].imshow(input_list[0][0])\n",
    "axs[0, 1].imshow(output_list[0][0])\n",
    "\n",
    "axs[1, 0].imshow(input_list[0][1])\n",
    "axs[1, 1].imshow(output_list[0][1])\n",
    "\n",
    "axs[0, 2].imshow(input_list[0][2])\n",
    "axs[0, 3].imshow(output_list[0][2])\n",
    "\n",
    "axs[1, 2].imshow(input_list[0][3])\n",
    "axs[1, 3].imshow(output_list[0][3])\n",
    "\n",
    "fig, axs = plt.subplots(2, 4)\n",
    "axs[0, 0].imshow(input_list[0][25])\n",
    "axs[0, 1].imshow(output_list[0][25])\n",
    "\n",
    "axs[1, 0].imshow(input_list[0][26])\n",
    "axs[1, 1].imshow(output_list[0][26])\n",
    "\n",
    "axs[0, 2].imshow(input_list[0][23])\n",
    "axs[0, 3].imshow(output_list[0][23])\n",
    "\n",
    "axs[1, 2].imshow(input_list[0][27])\n",
    "axs[1, 3].imshow(output_list[0][27])\n",
    "\n",
    "plt.figure()\n",
    "utilities.plot_raster_dot(output_list[0][0])\n",
    "plt.figure()\n",
    "utilities.plot_raster_dot(input_list[0][0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(31, 32, 31, 32)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(input_list), len(input_list[0]), len(output_list), len(output_list[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "utilities.plot_raster_dot(input_list[0][0])\n",
    "utilities.plot_raster_dot(output_list[0][0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAD8CAYAAAB+UHOxAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAAoiklEQVR4nO2db6xfxXnnv0+JSxEmKmxvsBccBbNeUtx4neTKGylLVlXacoOqNUGKZF5k/QLhfeFIySqVFrfSKn2ByFZK+qqJZAtUa5U1shQi0DbxlkWpQqQ0rumC4eKyGJMNDja+Da3AFaKBPPvinp/53eOZufOcmTl/vx/p6v7u+Z157sw5c+aZeZ5nniOqCkIIIdPjV7quACGEkG6gAiCEkIlCBUAIIROFCoAQQiYKFQAhhEwUKgBCCJko6yoAEfk1ETkuIs+IyLKI/HF1/DoReVxEXqx+XztX5oCInBaRF0Tk9pINIIQQ0gxZbx+AiAiAq1X1oohsAPBDAF8EcBeA11X1qyJyH4BrVfW/iMitAI4A2AXgXwL43wD+taq+W7IhhBBCbKy7AtBVLlZ/bqh+FMBuAIer44cB3Fl93g3gYVV9W1VfBnAaq8qAEEJIj3hfzEkicgWApwD8KwB/pqo/FpHrVfUcAKjqORH5QHX6DQD+eq742epYXeY+APsA4Oqrr/74hz/84eatIISQCfLUU0/9vaouNC0fpQAq881OEfl1AN8Rkd8KnC4uEQ6ZBwEcBIDFxUU9ceJETFUIIYRUiMj/SylvigJS1X8E8FcAlgC8JiKbq0psBnChOu0sgC1zxW4E8GpKJQkhhOQnJgpooZr5Q0SuAvA7AP4OwGMA9lan7QXwaPX5MQB7RORKEbkJwDYAxzPXmxBCSCIxJqDNAA5XfoBfAXBUVf+niPwIwFERuQfATwF8DgBUdVlEjgJ4HsA7APYzAogQQvrHumGgbUAfACGE2BGRp1R1sWl57gQmhJCJQgVACCEThQqAEEImChUAIYRMFCoAQgiZKFQAhBAyUagAJsKZlYs48MhJnFm56PzbWj6XDKuc+XPbbkPo3FzXgpA2oQKYCIeePIMjx1/BoSfPOP+e4RuYXOe7joUGNt//rB+PleGT55MR2wafjNC5ua6FCyoLUoqoZHBk+Nx729Y1v5e2b8Lxl1/H0vZNa86bDUwA8MBdO7zlfTJ85X0yzqxcxBtv/QK//5HNl45bZLz51jt4461f4MzKRWxd2BisR2wbfDJc5X1taHItZrIOPXkG99629VJ7QnIISYErgJEzmz0Cq4PHbFA5tnweL638E44tn19z/tL2Tbh54eo1A6JrUPLJcJUPyTj05Bn8xbPncc1V77t0PFbG1oWNuOaq9+Evnj1/2UzaJ6OO7zq4ZGxd2Ih7b9uKQ0+euWxlUW9Dk2sxk1VfDbmUy+y7mJUBVxDEBxXAyPGZGnyDk2tAtMjwDailZKS2497btuL3P7L50iqiiQxLHULXwrcacikXlxyL+Y4QgApg9FgHJ9f5Fhm+c30DbaqM1HZYVhG+2bilDqHjltWQ6zursieECmDkWAcn1/muY9bB0DfQWlYAdRkh80hsOyzXwjcbtyg3az1CJqr6d1Zl74LmomlBBTByrIPCvbdtxd27tlzmKK3LCA2G9fKATWHEygiZR2LbYbkWvvI+5eZrh6Uelutpuac0FxGACmD0WGe9sTKsZoX6YD0bgJa2b3IOcDEyQvb72HZY2mJxGJesB2A3F7nKW8xFXBmMEyqAkZPDQVnC2Ts779jy+TXRSRYZIft9CWdtDodxSE6bDvjZqmBp+6Y1A7v1vpJhw30AI2f+gf73t3zg0nFfHLorLt4lw1LeJcNX3iIjRzssMmYK58jxV/D+qzY02icRkmORkXotti5sxAN37cCBR05G7XUI7Zcgw4UrgJFjNdXEmnt8MfGx5o3ZAATgMtNCrIxQPH0Os1WqDF87Qs7rWBmp18JqggvJoFlouFABjJw+x+WnygiZJfq8R8HnvG7zWvhMcJZrQbPQ8KECGDnWWHTLrNcyQJaI7bfa5Pt+LVJkdHEtfP+TK4PhQAUwckrGolsGFsv+gFgZ1lQQfb8W9eOWfQ5dXAvf/7QmxiPdQSfwyEl11lplWB2t68WzryejVDssMmL2ODRxOs9MRXfv2uLc59D1tbDUgwnt+glXACPH52y1zvRiHb5WR2tqPLvVzFJChjUZXOz1bHufg1VGjiSBpFuoACZCfVleylkbWz4UhWKRYUkFUSoxnjUVROz1LLnPoS9JAkm30AQ0cmaztHlnIWBb0ltMHBbzhjVXfpsmEpcM33WwxPWXup6WduS4FlYZvn7lC10l7cAVwMjxhftZluRWE4cLl3nDtRPVKqNNE0ko91DX1zNH+VIyQu+CYBhpt1ABjJxSph6fDEsahwfu2oFjy+ejX9PYponEJSNXKoi2zSx93a9hVeAkP+sqABHZIiLfF5FTIrIsIl+sjn9FRH4mIk9XP3fMlTkgIqdF5AURub1kA0gY60wvNZ7dOrCkyvA5J0vk7beGWvbhWviOt9kvQs5w1/VkyGh7xKwA3gHwZVX9TQCfALBfRG6tvvtTVd1Z/XwXAKrv9gDYDmAJwDdE5IoCdScRWAeh1Hh2n1yLozQkI8ZhbE1VbbkWFhklr4VlQE69p6kyQg5glwyahtpjXSewqp4DcK76/KaInAJwQ6DIbgAPq+rbAF4WkdMAdgH4UYb6EiMWpx5QzrlocZT66jYf0ho611I+dL7ruEVGqB5dJ4OzOL5dQQQWGb42WNtB8mPyAYjIhwB8FMCPq0NfEJGTIvKQiFxbHbsBwCtzxc4irDBIQazx16Wciz45swF1fqbuOuYjtXzo/JL1cK0kXOeXctZaHN++tN2xMqx9grRHtAIQkY0Avg3gS6r6BoBvArgZwE6srhC+NjvVUVwd8vaJyAkRObGysmKtN4nE4tQDyjkXQ3KmSKxysZicStzTULRW285zkp8oBSAiG7A6+H9LVR8BAFV9TVXfVdVfAjiEVTMPsDrj3zJX/EYAr9ZlqupBVV1U1cWFhYWUNhAPPmeodQNVDsegJQUyeQ+fonApBouy8A3slmittp3nJD8xUUAC4EEAp1T163PHN8+d9lkAz1WfHwOwR0SuFJGbAGwDcDxflUksobTDlnj2HI7B0P8kdiwmp5DprT6w98Hp7DrGyKAyxOwE/iSAzwN4VkSero79IYC7RWQnVs07PwHwnwBAVZdF5CiA57EaQbRfVd/NW20SQ8i5+eZb71xakvtszm0kgyPl8d0TIN5h22a/cJVnMrkyrLsCUNUfqqqo6o75kE9V/byqfqQ6/h+qaKFZmftV9WZVvUVVv1e2CcOgTzMY65LcR6rDmLSDzzTkomunc47yJB7uBG6JLhxbvv9ZMoma1elMyhNyOKckCWwzkIABBGVgMriW6MIE4lvO9ynPPOmW+v3KsefD0i9c5kjuDWgPrgBGjHUnqlVOrv0BpDvqq4Mc99QiI/aNcL53KJA0qABaogsTiDUHi6+epcwCpH/k2PNh6RfrvRFuvTrMZPTFvzY0qABawuKIy4V1B2aOJGrWHDOkX/j6qcUub+kXljfChWTQv9QMKoCWsKYnyIF1V641T31sDPd6dSH9wddPS8T2+2RY9heE6kHWhwpgxIRm3b6dpJZXG9ZlWKOLyHCwDNSp/cLVN9m3ykAFMGKsMyOfb8CSOCzHG7NI/4hdAbje/pWjX7BvlYEKYMSEZkYlYriZCG68xPYLV7/KkQyOfasM3AcwYkKx0yVi+y0pB8iwiO0Xrrh+3/sPfDIsqSR8MkgcXAFMlBKx/VyKj5fYKLZUMyLg9wvw/cH5oQIYMaHwuBKx/U1MTmQYuKKD2kwFYdm7wn0B8VABjBir7RVIi+0PzfK4OhgflkmEJbbfujqtH+dkIx4qgBHTJOtnSgx3aAVAR934KBXbb91MWD9Oc1E8VAAjx/owuWy9sTJCdmKuAMaHpW/l2GEcKyM08SFroQIYOSUTubkePF/CLq4AxkepvmWdLOSQMVWoAEaOdQXAZHAkllL9wpoMziKDrIUKYOSkLqetMpgMbjqU6heWvmmVQdZCBTBiQvlTfKQup5mwazq03S/Yt/JDBTBiQvlT2ozhDskgw6UvZkSLDLIWKoAREwqH60sMNxkupfqFdRJhkUHWQgUwYkLhcH2J4SbDpVS/sE4iLDLIWpgMbuRYX9LuSrplkeEqH5JBhkupfmHpm1YZZC1cAUyUvsRwk/HRh0ACrgDioAIYOSUdajliuMlw6UsggfX9AeQ9qABGTkmHWqrDmAwbyytESwYScCNYc6gARk5Jh1qqw5gMG0vu/5KBBDRFNocKYORYH6a2d3GS4eLbaJjaL0ruXidrWVcBiMgWEfm+iJwSkWUR+WJ1/DoReVxEXqx+XztX5oCInBaRF0Tk9pINIGH6sgOTD+T48G00bDuQgCuA5sSsAN4B8GVV/U0AnwCwX0RuBXAfgCdUdRuAJ6q/UX23B8B2AEsAviEiV5SoPFmfkg61HLs4yXDpSyABVwDNWVcBqOo5Vf3b6vObAE4BuAHAbgCHq9MOA7iz+rwbwMOq+raqvgzgNIBdmetNIim5nGYyuGnTFzOiRQZZi8kHICIfAvBRAD8GcL2qngNWlQSA2c6MGwC8MlfsbHWsLmufiJwQkRMrKysNqk5imD0IS9s3Rb0ntQ8x3GQY9MWMyL7VnGgFICIbAXwbwJdU9Y3QqY5jetkB1YOquqiqiwsLC7HVIEZmL/M+tnx+zfK5LzHcZLj0xYxokUHWEqUARGQDVgf/b6nqI9Xh10Rkc/X9ZgAXquNnAWyZK34jgFfzVJc0pf7wlIrtD5kF+J7WcVEqtr/k3hWylpgoIAHwIIBTqvr1ua8eA7C3+rwXwKNzx/eIyJUichOAbQCO56syaUL94SkV2+978Pie1vFRKrbfdSz0bgvL3hWylphkcJ8E8HkAz4rI09WxPwTwVQBHReQeAD8F8DkAUNVlETkK4HmsRhDtV9V3c1ec2KgnzFra3m4yuJAcMkzaTAY3Czm9e9eWy95twWRwzYmJAvqhqoqq7lDVndXPd1X156r6aVXdVv1+fa7M/ap6s6reoqrfK9sE0oRSTjlf+TMrF3HoyTO497atlz3AZFyUCCSwmhC5AoiDO4EnQt0plsOh5noorTHcZLi0GUgQMiEyGVxz+D6AkTObec8ehNkyOXVJDrz3UB45/gref9UGPHDXDvP7B8hwSTUjxqSSiHmnhNUUSd6DK4CRM5sdHVs+jwfu2nHJ/JIj/try0nkuycdHqhkxRyoJXz24AoiDCmDklEwF4XqAmQpiOqSaEXP0TUtCOnI5VAAjxzpLS43t50aw6ZAal5+jb+ZaRUwVKoCR40sFUSq2nxvBpkNqbL/lhTK5ZJC1UAGMHF8qiFy51etyuBFsnJxZuXhZLqmQso9JBufrE5a+aZVB1kIFMHJmD+7S9k1RD0RqbH8OhzHpH6mhlkwG10+oAEaOLwqo7WRwPlstGQZtJglkMrj24D6AkWONyy+VCoJpIIZNm/3C0jd9Mtjf4uAKYKKUWpL7nM5bFzbi3tu24tCTZ+gEHiClUoTElrfKmPm+uNoMQwUwckoup31OOZfTOSSH9AeXsxeI7xfWuPy6DGvWT189SBxUACPHGgYaG8Ptk+FzOtMJPAwsyj42Lj907+syfL4iiwwSDxXAyPHNyFNjuH0yQk5nOoH7T2q/sGzWcsnwxfVbZJB4qAAmQv3BTI3h9smwvtCbdEvd5JPaLyx9wiWjSVw/+1ZzqAAmQv3BLOWUm604AEQNLKRb6iYfy47t2J25oQCA2FWEVQaJgwpgIsSuAHK96D32/QOkW2Jn4Kl5+0vtOwG46zcFKoCJELsCyJG0y/UdZ2nt4IviiY3u8TlbU/uFZRJh6ZsAQz5ToAKYCLEDsuWBDA3qsc490hzXoO6bJcfOwH3O1tR+YVEWlr5J0uBO4IlQ30VZendvXY7r7WEkjdmgDuDS9bS8eQtor1+U2nlO0uAKYCLE2uBTd3H65HAfQF5c19OVnA/wx+aXShIYWz6XDNIcKoCJUH94SjrlXHK4D6A5PlNPytvY+pIk0LLznM7e/NAENBHqy+rSS/JY08KY8M3AU+XEmnpK3tNUGTnMRfMhxiQPXAFMlFJLcqtpwYU1kiVGhqVsUxmuvDZN2jEvx2I6K2lmSZVBU08/oQKYCLFx+ZYleQ7TQkokS2ggrZ8bMlf55MTK8Ck9a0SOS47PdNbnnPulzEUkP1QAEyE2DLTtGG7LAODyK/gG0vqsORSGGjuQrdeGutJrei3m5ZTar2EN7W2zX3AF0B5UABMhdiNY2zHcroE5VkZoIK3PmkPvJI69FpY2tH0tLO0Iye1Dv6Cztz3WVQAi8pCIXBCR5+aOfUVEfiYiT1c/d8x9d0BETovICyJye6mKExuxK4C2k8G5BuZYGZY2NMkzn/rC+zavhfV6WFcAbfYL7uxtj5gVwJ8DWHIc/1NV3Vn9fBcARORWAHsAbK/KfENErshVWdKc2NlzTNkZOWSkODl9M29X4rBQGGpsO0KmiVxOzpS69OWe0uE7HNZVAKr6AwCvR8rbDeBhVX1bVV8GcBrAroT6kUzEzp7bdi6mxLOHzDqWLJex18Iqw3otXHL6EJefQwZj+/tJyj6AL4jIfwRwAsCXVfUfANwA4K/nzjlbHbsMEdkHYB8AfPCDH0yoBomhHofdhy3/OWTEpj4IpaKIvRZWGdZr4ZKTei18Miz3I4cMn1zG9ndLUyfwNwHcDGAngHMAvlYdF8e56hKgqgdVdVFVFxcWFhpWg8Timj278qsPybRgTX1gTUUxpGtRsh05ZNCu308aKQBVfU1V31XVXwI4hPfMPGcBbJk79UYAr6ZVkeQgdlk+JNNCyKxQN9WEfAB9MbO45JQyv1nakUsG6R+NFICIbJ7787MAZhFCjwHYIyJXishNALYBOJ5WRZID1+zZNSO2zPQsg0IOGfWBPeQEdkXT+Oz3sYOs9cXk1mvhkm+RYVEWlvvhk+Gz39OuPxxiwkCPAPgRgFtE5KyI3APgT0TkWRE5CeC3AfxnAFDVZQBHATwP4BiA/ar6brHak2hcs2fXjDg1hts667XIqA/sPidwyAfgchjHDrKhVUSOa+GSb5FhURaW++GT4TPr0NwzHNZ1Aqvq3Y7DDwbOvx/A/SmVIvkZg3OxLsc3I58NpHfv2rJmECqVK7+La9GHZHBk+HAn8ESxzvRKOIytMlxZMl0z8pC5p06O/QHWdpSSUdLpTLPOOKECmAgpzkVXeZ+MNp2LFh+Arw6W2P6Qc7PraxFqi8Vh7BvoadYZJ3wfwESoL+0tpoUYh3EXeeYtcfm+Nlhk3HvbVrz51juXBljfO3PXa4dPTur19LWFOfeJD64AJkJ9FpjDTNIH80Rs+RymHktCudDLYXxyUq9FSMnVTXi++9HkvQlkuFABTIT64JLDTJIai55DRqq5yCLD0o713j8QG4JruRaWvQ6+djCGf1pQAUyE2NmzzwkcO2O1+hZSZPgGUmsYaIl9DiGFY1lR5YjtT91fMJPBlcH4oAKYCLGDWWosuXVgSZHhG0itqSBK7HPI8f4B6/Ustb9gJqO+MqBSGD50Ak+EWAdjqrPWGkeeIsPiZPXtDfDVwVePUnH5OWSUaMfMlzG/spkxUwoA6DgeKFwBTIRYB2Pq/oBQHHms0zk1Ft3yYhVrPUrG1MdeT2tiu1Tnuet1l7N2uO4pVwbDgQpgIsQ6GFNNA7PBbWn7pjWDgNXpHBOLHnJY1mWE4tgtMfE5HMaWulicvaWc5677OWtH7D0l/YQKYCK4whRjI1Fc5X0yZoPbseXzawYB3/9zDbKxm45Cs3rLxiXL5idLAjTftQxRv84uGb5BOYfD2BUt5rqfgP+eWlZJpFuoACaCK0zREtvvW0HExtb7zk3ZYZprd6pFjiUBWg6zk8/05hqUcziMXZFWBx45iaXtmy5ri++e+mTQJNQ/6ASeCCnJ4FzlQzJcckKJ1MZKk1219evkux+uc0vsMA45enPIIN3CFcBEsYZrxsoIzRjJ+tRXEpZVRIkdxk3MOXUZlgAA0i5UABMhdmdsqpPTFzVCx2AzrM7rWHNPrMM45MeIlRHaE0G6hSagiVBfllsSh7nKA3mSwZHmuK517v0aIdNdjj0KoZxJpDxcAUwEl2OuRCz5vN173vHXJCKGhIl1PFv2a+TY5+CKOPMN8lwZdgsVwESwpFFIjSV3yaEduB18JqOm74OwvgMhlwzSDlQAEyHWMZcjltz1He3A3ZGSDC60ESx2l3Kon4QCCThRKA8VwESIHZBzxJK7vrOanEg+UpLBhTaCxe5SDvWTUCABJwrloRN4IuR2DvrK++SEkrGRssQ6i33nAmlBAL79CRYZpAxcAUyEWMdcDuegSw5tvd3h8guk3k+LjCapsUk7UAFMhFjHnDUVhC9HfH0jGKOA+oVlvweQtr/A+i4GmoDagyagiZCSNsBVHnCbBXzb/rms7xeWdykAaeYe67sYuGekPbgCmCiWtAEWc5FvBeGKRSf9IzWdhKt8DnMTKQMVwESoL6tLvZi8SeoA0j6+e1GiX1j7BP0C7UEFMBFiN4K5Hr4cewMYBtovfEnmSvQLy34RV3lSjnUVgIg8JCIXROS5uWPXicjjIvJi9fvaue8OiMhpEXlBRG4vVXFiI/ahdD18OfYGhN4dQNrHt2O4RL+w7BcJ1YHkJ2YF8OcAlmrH7gPwhKpuA/BE9TdE5FYAewBsr8p8Q0SuyFZb0pjYh9I1M7RkDvXNLPlQDwNLv4hVFr4+YX0rHcnPugpAVX8A4PXa4d0ADlefDwO4c+74w6r6tqq+DOA0gF15qkpSSInLL5FnnvSTEsngfFjMTUwPUYamPoDrVfUcAFS/Z/GCNwB4Ze68s9WxyxCRfSJyQkROrKysNKwGiSV2BWCJ4051GJP+YekXqcngLCtLBhCUIbcTWBzH1HWiqh5U1UVVXVxYWMhcDVIndgUQm9/Fd67Fhkz6h6VfxK4AQqHBsStLTiDK0FQBvCYimwGg+n2hOn4WwJa5824E8Grz6pFcxK4ALE65VIcx6R8l+oU1GZylX5E0miqAxwDsrT7vBfDo3PE9InKliNwEYBuA42lVJDmIdcxZZnopjkHST0r0i5Dyjw0k4ASiDDFhoEcA/AjALSJyVkTuAfBVAL8rIi8C+N3qb6jqMoCjAJ4HcAzAflV9t1TlSTyxD5DvgYx9mQtXAMPGFx6aEq7p6xMu35I1SSFJIyYK6G5V3ayqG1T1RlV9UFV/rqqfVtVt1e/X586/X1VvVtVbVPV7ZatPYklJBuez1aY4BsmwsO75iHn7mMu3xCCCdmEyuIlQT7rlS7iVmgzOUp4Mh9hkcMDl/cLSJ6xJCkkaTAUxUXKYariLczqk9AtLgsAce05IPFQAEyH2Je2pceCW8mQ4dJ0MjibEMlABTIT6Q2WdacU+lJbyZDiUSAZn2WDI/lMGKoCJEPtQpcaBW8qT4VAiGZxlgyH7TxnoBJ4IsQ5bl2MuZqZmdQySYWF5c1e9X5R6+xhJhyuAiVCfQVny81tmarH/nwyf1EACOny7hwpgItQfKt+g7nLK5UjaxYd62JRIBueTwWRw7UEFMBFSUkH4ZmpMBTEdmApinFABTITUbfttlSf9pEQqiFwySHOoACZCSioIV3mfDK4AxkmJVBA+GXxRfHtQAUyE+gNkGah9DmOrWSA2oRzpHyl7PqwrAIaBtgcVwESoP0AWU43PYWwZFHx+BDIMUvZ8hJR/7P4CrgDKwH0AEyElGZwltj+UtIux3MMlpV/MlP+R46/g/VdtwAN37bh0bmqSQZIGVwATxbICcCXt8skIbft35Xknw6DNJIFcAbQHFcBEYDI4EsOZlYs48MjJKFONJZCgVL8iadAENBFil+Wp5h7rtn/SL2YDMoBGphqgjBmR5sMycAUwEbpeltMJPAxypPeoywiZ/2ju6RYqgImQuuU+x5Z9Ptj9J9VU45LBVBD9hQpgIsTOtCz7AyzlfTJIv/DdP5fPKHbPiDUM1NqvSHOoACZC7EzLEh1kKe+TQfqF7/65THixe0ZC5r8c/Yo0h07giRDrWEt17PnK+2SQfuFztrpm9rF9IpR6PEaGJXU5scEVwEQokczNtT+AK4Bh4Av39FGf2ed4n0SsDF95kg4VwEQolcwtNfEX6QafUzU2CMDyPglfn4iVwTxS5aACmAiuh8iy4cZ17npmgTpcAfSHWc79pe2b1qwEYicAvkE51lkcklGvA0OIy0EFMBFcD5ElBtu3gqjP4HwDi08G6YatCxvxwF07cGz5/JrZduxg7RuUXcd9A71PBvcGtAedwBMiJiGcz1Eb6/CdDSwHHjl52Y5S7ubsH7FJAl07x5kMbvgkrQBE5Cci8qyIPC0iJ6pj14nI4yLyYvX72jxVJSm4dmPGzrR8Ozlp7hkOdafv7G9gVUGH+sTs/Finb4ld575XmJI0cpiAfltVd6rqYvX3fQCeUNVtAJ6o/iYdk5J0q8mL3rmbs1/Ur731nrrMfW0mg5utLBkFlJcSPoDdAA5Xnw8DuLPA/yBGYh/KXC9654u9+0Xsi9pDPoCUfuGbwTPzZ7ekKgAF8Jci8pSI7KuOXa+q5wCg+k2DXQ+oP1SpS3rAby5ymRZCckg+XPH9seY/wBZyaekXFhmcKLRHqgL4pKp+DMBnAOwXkU/FFhSRfSJyQkROrKysJFaD+JgNCEvbN62ZgeWI42YyuG5xDfau6+865ovW8kXmlOoXXAF0S5ICUNVXq98XAHwHwC4Ar4nIZgCofl/wlD2oqouquriwsJBSDRJg9uAdWz6/ZkYeG4MN2GZ63AjWHN/uXN/x+qDqW9W57rUvDNQnI7VfWF5AxIlCezRWACJytYhcM/sM4PcAPAfgMQB7q9P2Ang0tZKkOb6HKUcMNpPBNSd29h46Hrs713WvrSvD1H7h629cAXRLyj6A6wF8R0Rmcv6Hqh4Tkb8BcFRE7gHwUwCfS68maUoofjo1BpvJ4NbHF0LrevNWaiI1SyI3y5u/gPR+4WtHbEI5UobGKwBVPaOq/6b62a6q91fHf66qn1bVbdXv1/NVl1jx2Xotsf05HIO+AcCSlMwXy96kbOx3qTJcs/c2E6n5ZFhXY6mmGlcdXG0IvT2M5IepIEaOz9ZrccrlcAxanM6xg2nI4VyXETrX912qDN8gXSqRWqm4/BwyXOUtbSBlYCqIiRC75T/V3GMp75MRa54IvWi+LiNkVvBdi1QZswHu7l1b1tStlJnFck8tdcghI7Y8zT/twhXAROlLDHeKDMubplzvLgjVIYeMkAPeJadECoW+yPCtLOvXgTt+24UKYCLE5lhPXdb3wbTgk2FNf1Aynr3E9RzitaDJp1uoACaCazYbG5bnO26ZsVoGltQ6+M611CGHDN+5MRExU7kWTPLWLVQAE6H+UFoGId9xyyDk+v+uOHSrjJgQyfXkWgYyi4zQrNflBE65niXbUfJa0OTTLXQCT4S6Y66UgzK2vM/Ra5XhaoNPRinHd+q1yCWjVDtKXgvSLVwBTIT6DMwS2+8qD6Q5BkNvDouVkcPhnNqOkjJCMfF9dRhb9ieQ7qECmAixPoBUx15sed/+BIuMHE7WkJyuHa3WxHql2mGRYdnjQLqHJqCJELssT13Wt2lasMblW+qQQ0bqtfCV98ko0Y7ZKmRWtmkaB5qA+glXACPH52y1pIIo9e6ANmWEUlHEmlmsMlLbEVrhpF7P2PK+bLLWOtDZ20+oAEaO7wG2mCdypC4oYZ6wmEiamCZSZaRei5CfJtXkZKmDz1djqQPpJ1QAIycUf52an90ysJSQYRkgLe8/yCXD145YOZadztaQ2Nj7EfLVuGL4Gdc/LKgARk4o/jo1P7tlYCkhw5oKIvb9Bz6zmUVGqB0l3sXgW534ZMTej9B+DZdZh6aeYUEn8Mix5IgH2o1FzyGjhNPaukfBl5TO1w5LXWKvRSgxXso9DV0LMny4Apgolt2oPlJj0XPIKOFo9dm9fc5ea1oNi9M4th2W1ZDletCkM26oAEZO35N+pciw2r1jHc7WdyhY02r0OTFefcCnSWfcUAGMHKtN3nV+KYdxqgyL3ds6SLtk+OplXU3FXouQgkt1wPtm9hzwpwUVwMixmkhc55dyGKfKsETwNDF51WXkuJYWOSEFl+qA50BPADqBR4/FSQq0lwwuhwyLk9V6HXLIWG9X7MzHMPMF1OWEds+6vrPWgxCuAEaONRbdIidkPwewxoFqqUfsG7MsbUh1FlvbEGqHz8dgsb+7vqNZh1jhCmDk+GZ/M7POkeOv4P1Xbbg0aPvC/iwzTpecHPWoy/CVtfz/km0IXU9fe1JCLVPLkwmiqp3/fPzjH1fSPi9deFPv+/Yz+tKFN4PHLOVzyYkt7zsvtXyONlhlEGIFwAlNGHtlVUa3LC4u6okTJ7quBiGEDAoReUpVF5uWpw+AEEImChUAIYRMFCoAQgiZKFQAhBAyUYopABFZEpEXROS0iNxX6v8QQghpRhEFICJXAPgzAJ8BcCuAu0Xk1hL/ixBCSDNKrQB2ATitqmdU9Z8BPAxgd6H/RQghpAGldgLfAOCVub/PAvi38yeIyD4A+6o/3xaR5wrVpQ/8BoC/77oSBWH7hs2Y2zfmtgHALSmFSykAcRxbs+NMVQ8COAgAInIiZTND32H7hg3bN1zG3DZgtX0p5UuZgM4C2DL3940AXi30vwghhDSglAL4GwDbROQmEflVAHsAPFbofxFCCGlAEROQqr4jIl8A8L8AXAHgIVVdDhQ5WKIePYLtGzZs33AZc9uAxPb1IhkcIYSQ9uFOYEIImShUAIQQMlE6VwBjTBkhIj8RkWdF5OlZmJaIXCcij4vIi9Xva7uuZwwi8pCIXJjfpxFqi4gcqO7lCyJyeze1jsfTvq+IyM+q+/e0iNwx993Q2rdFRL4vIqdEZFlEvlgdH8U9DLRv8PdQRH5NRI6LyDNV2/64Op7v3qW8TSb1B6sO4pcAbAXwqwCeAXBrl3XK1K6fAPiN2rE/AXBf9fk+AP+t63pGtuVTAD4G4Ln12oLVtB/PALgSwE3Vvb2i6zY0aN9XAPyB49whtm8zgI9Vn68B8H+rdoziHgbaN/h7iNX9VBurzxsA/BjAJ3Leu65XAFNKGbEbwOHq82EAd3ZXlXhU9QcAXq8d9rVlN4CHVfVtVX0ZwGms3uPe4mmfjyG275yq/m31+U0Ap7C6U38U9zDQPh+DaZ+ucrH6c0P1o8h477pWAK6UEaGbNxQUwF+KyFNVygsAuF5VzwGrnRbAB7yl+4+vLWO6n18QkZOViWi2xB50+0TkQwA+itWZ5OjuYa19wAjuoYhcISJPA7gA4HFVzXrvulYA66aMGCifVNWPYTUb6n4R+VTXFWqJsdzPbwK4GcBOAOcAfK06Ptj2ichGAN8G8CVVfSN0quNY79voaN8o7qGqvquqO7GaTWGXiPxW4HRz27pWAKNMGaGqr1a/LwD4DlaXYa+JyGYAqH5f6K6GyfjaMor7qaqvVQ/eLwEcwnvL6EG2T0Q2YHVw/JaqPlIdHs09dLVvbPdQVf8RwF8BWELGe9e1AhhdyggRuVpErpl9BvB7AJ7Darv2VqftBfBoNzXMgq8tjwHYIyJXishNALYBON5B/ZKYPVwVn8Xq/QMG2D4REQAPAjilql+f+2oU99DXvjHcQxFZEJFfrz5fBeB3APwdct67Hni678Cq5/4lAH/UdX0ytGcrVj3xzwBYnrUJwL8A8ASAF6vf13Vd18j2HMHqEvoXWJ1h3BNqC4A/qu7lCwA+03X9G7bvvwN4FsDJ6qHaPOD2/TusmgFOAni6+rljLPcw0L7B30MAOwD8n6oNzwH4r9XxbPeOqSAIIWSidG0CIoQQ0hFUAIQQMlGoAAghZKJQARBCyEShAiCEkIlCBUAIIROFCoAQQibK/wd/hfqDSBeVbwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "utilities.plot_raster_dot(input_list[0][2])\n",
    "utilities.plot_raster_dot(output_list[0][2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAD8CAYAAAB+UHOxAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAAuZElEQVR4nO2df6weV3nnv0+dNPHaQUmWS+wmpsRpCI2BNXCVRWLDZrdVMBRtAAmUqIL8EdmVNmhBykrYVKIUKYJdLe4/BSRboIZdmihVgohMN00akZJKFPdmN4njhCSOw2ITJ76sQbHB9eL02T/ujDM+9zxnzjMz733fmfl+JOtez51z3nPeOXOec55fR1QVhBBCxsdvTLsBhBBCpgMFACGEjBQKAEIIGSkUAIQQMlIoAAghZKRQABBCyEipFQAicr6I7BWRx0Vkv4j8aXH9YhF5UESeK35eVCmzQ0QOiMgzIvK+SXaAEEJIM6QuDkBEBMAaVT0hIucC+HsAnwLwEQDHVPVLIrIdwEWq+hkRuRrAnQCuAfBbAP4WwJtV9dVJdoQQQoiP2h2ALnGi+O+5xT8FcAOAO4rrdwD4UPH7DQDuUtVTqvoCgANYEgaEEEJmiHNybhKRVQAeBfA7AL6iqj8UkUtU9QgAqOoREXlDcfulAP6hUvxwcS2scxuAbQCwZs2ad73lLW9p3gtCCBkhjz766M9Uda5p+SwBUKhvNovIhQC+LSJvTdwusSoide4CsAsA5ufndWFhIacphBBCCkTk/7Qp7/ICUtVfAHgYwBYAL4vI+qIR6wEcLW47DGBDpdhlAF5s00hCCCHdk+MFNFes/CEiqwH8PoAfAbgPwM3FbTcD+E7x+30AbhSR80TkcgBXAtjbcbsJIYS0JEcFtB7AHYUd4DcA3K2qe0TkBwDuFpFbAPwEwEcBQFX3i8jdAJ4CcBrArfQAIoSQ2aPWDXQloA2AEEL8iMijqjrftDwjgQkhZKRQABBCyEihACCEkJFCAUAIGR0HF09gx71P4ODiifqbBwwFwAQJBxkHHSGzwe5HDuLOvYew+5GD027KVKEAmCDhIOOgI2Q22LJpHa6YW4Mtm9ZNuylThQJggoSDjIOOkJUjteO+f/9LeH7xl7h//0u19w4ZCoAJEg6y8P8hYx2EhEyC1I47XIyNdXeelQyONGPrtRvP+rll0zrsfeGYuQMoByEAfPEjb1+ZRhIyUFLvW3Ux9m+vesOyd3UscAfQAbkr99QO4ODiCbxy8tf44NvWj24QEjIJUu9bdQdwcPEEdj9yEFuv3YiNc2un0NLpQQHQAdb2MbyesgHsfuQgvrvvJVyw+pzRDUJCJkHqfasKB+v9HYNKlgKgA6yBFl7PXZGUjGEAEtIFsXcl9b5tvXYjbrpmA7ZeuxFbNq3DhotW48gv/ums8mOwC1AANCAcbNZAC6/nrkhKxjAACemC2LuSet82zq3FFz/ydmycW4v797+EQz8/iYefXcwuPxQoABoQDrbqaqKKZwcQq2MMA5CQLth67UZ88G3r8crJX9cuzGJlr3vzHDZctPqsdy23fJ+hAGhA7sQcDqBykt+yad2yCOGYEcoagFQNkbFijf2Nc2txwepz8N19LyVtbrHyG+fWYv2F5+PQz0+e9a5Z7+uQoABoQDgxW6qacACW287797+UFSFsCRqqhshYSRlsQy86j1o19q5Z7+uQYBxAA3L9+0Nf40mVJ2QsWO9K6UV30zUbzuyiY++J912z6qnSZzdS7gCcHFw8gS8/8AyOnzx95pqlqonpJWO0LU/IUMl1uGiqlq0rnzO593lHTgHgpFxp7Nl3pNa/P6aXLOsIjcixiT63fAltA2Ro5MbS5Kp7rHfNEgw5k3ufnTUoAJzEPAZSxtpYdG/MNhCb6HPLl/R5JUKGRxcLkhxPOs97EnvXrPK50fl99haiAHAS8xhITcix6N7YgInV4Smfagch06CLBUlOLE3b98Qqnxud3+f3jgKgAbn+/bnxAVYdnvKpdhDShLYreGv8ej4/x7On7Xtilc91A+3ze0cB0ABPhG9OeW8dnh0A7QKkKW1X8NVo2xzCsRpbgcd0+Bvn1mLrtRux+5GDWXr9sA6rfK4bKHcAIyNcMXgNSLk7AG98ANNJkC5pu4L3Uo7VnQ88ix33PoEtm9Yt+/xcx4iU/j5Whye+IKTPOwDGATSgXBmUWH7CHp/jLnyWPXUQUkc4zidJdaJVaPJcjJxxHosLSNVhvcN19aTK9gHuADqg+qJM02fZUwdVQ2QlqRtvVXXPbddfZerec9OmpGJoYnVY6qqcd82r6polKAA6pGufZa+x11NHE9UQhQZpSmq8hWqWlO7dk3Ylpiqqa0tIn9U7OdQKABHZICLfE5GnRWS/iHyquP55EfmpiDxW/PtApcwOETkgIs+IyPsm2YFZoJwYQ52lZwWfaxew7vXUkaPXjE32sReHQoHk0OQwpLBMatzmOlZ4T95baTvISpOzAzgN4DZV/V0A7wZwq4hcXfztz1R1c/HvrwGg+NuNADYB2ALgqyKyagJtnxnKifH+/S+d2Qp6B6vHMyj3/AGrjhz/ZiuKMnwZaGQmOdSlQs+JzrXGrfWuWTttqw4ry2hf1Ts51BqBVfUIgCPF78dF5GkAlyaK3ADgLlU9BeAFETkA4BoAP+igvTNJzAiUMh7F7t967UYcP3n6LNc0bzK43DpyjFaxe2JGwT4bwMjKkXJGKNU1d+49hNetPvfMGMtNmmi9a7nvWVlHyvA8VFw2ABF5E4B3APhhcemTIvKEiHxDRC4qrl0K4FCl2GGkBUbvia0SvInccvOZe+uIBbNYfs91fWpzHxkPnuMZY+rT8hqAs8aWN/4lZiy2bAN99uVvQ7YAEJG1AO4B8GlVfQXA1wBcAWAzlnYIXy5vjRTXSH3bRGRBRBYWFxe97Z55mhihciOMU2qXtmcQUKdP2tLEGaGqPp10/IvH3jZ0sgSAiJyLpcn/W6p6LwCo6suq+qqq/jOA3VhS8wBLK/4NleKXAXgxrFNVd6nqvKrOz83NtenDzNJFiHpsF5FareSukjzeQRQKxCI2NjzHM8Z2qFYKBs8OILUD96STGDo5XkAC4OsAnlbVnZXr6yu3fRjAk8Xv9wG4UUTOE5HLAVwJYG93Te4Plnok1zPH2kV4s4/Gyni8g2joJRaxsWGpIq2U5+EO1dq1enYAqR24lSV0jKrMnEjg9wD4OIB9IvJYce2zAG4Skc1YUu/8GMAfAYCq7heRuwE8hSUPoltV9dVum91vYsYsjyHLawyL1dO2DVX6fCISaYfHISFm6E3VE45Rj1NE7mIoPP1rbNTuAFT171VVVPXtVZdPVf24qr6tuP4fCm+hssztqnqFql6lqv9zsl3oH5NKBhdbZXkMalb5IZ+IRCZDuAL3+t8D7ZIuplydx6ruicFI4CkQG4Aeg6/nBDFrcrayKuaWr+sPGQe5SdTqYk+q9XgDKz3vCDBedU8MJoObAh5/+piqJrWdzvWdbut7XdcfMnxSq/pQnZhSI8aERcwnv6t3hLwGdwAzjsdYG1PXNDEY131+Wb7OM4jeQ8PBSg1irerbqGtiXkApVWSTpIkcm0tQAMwIHt/n1BF2Hv/rnDraJJOjbWA4eMYWEE/j0CZ2xevXH2tDXX/GCAXAjGDp0T0GY4//dW5COY/BuUoTox+ZPuHK2NLHA77dZa6wqH5+9Z3oIvalCu1WS1AAzAiefOTWgPekk8hNKGeVT/lZA/kHapPZIlwZxyJ1Szy7y9RqvFpP9fOr70Ru+SqpMjQEL0Ej8IzjTebWxnfaU96qo9TVluXHvsLqG9YJWaU+3rIv1Y0tKwlbWI8nriBWPmeMk9fgDmDG8W5VY+kkrNOVcst7ktqlVoxk9qmujKuG1zAqF0jvDsNkg7mRuakD2q3y1hjnKr8eCoAZJzaIJ5kMLlbek9RurFkVh0j1+XrsS2HZklxVjSd2BbDHOKmHAqCHNDGIhTsJj97Uk9RurFkVh0i4MrdSjVuqyNyxEdaTWtGn7E407PqhAOghHoNYF7nVPe2wcrPT53p2sZ5PnTowda5EaiFRp45MrehTkzxVPn4oAHpI6iXI3U53kVs9d5VHn+vZJnw+lutnbqoQywU4Vx2Zcj3lJN8t9ALqIanUC6EXj+XFYXlOxMLqrXQQYR3Wi5+TToJMHiuaNvSWsVIxVJ9vOZ5i3l6xrLLVz65LN2J9Puke7gAGRrgCt1ZdlkEttoLPVRdZvv9WOouu1EJUMeVh7cRCz5+6/D4XrD7nzHjKjQ+w/PuBdlk/STu4AxgY4erK8svPTQbnyavuycve5SqPK8Y8LL/46urcc6aExy+/i9gV0j0UAAMjVA+lJsdUcE21vDUh5Kh2rPI5ZXMPmmHATz2p77I6RqznEitvZdu0snOW94R15SwkyGSgCmjgpDwvcox6bRJ+eY6ojGEZGGP9oGEwTW7siCfnfpMUDLG6rNgVPs/Jwx3AwClfph33PpGVYz3cFXhSQcSMebnqBGD5yjB2D9U9zWibiiG2qvemYLDUkVT5TA8KgJGQM+ECy20DOWqiktRWvvpZAKLqiHByj3k70aMojaXqSR2QkvOdxiZp77OwhDef6fSgCmgk5KaUCNU2nlQQqa189V5vbEIVT6TxGL2DrO82Vx3n8cn3qIvaqgPJZKAAGDGWu17oyhlzGY290LkRyk11x94zBoYegBYTcLkpmnPz7lvJ/VLPIndclTCFw/SgABgxuakcLGNx+ELnGoybpqHwnjEwdH/ymIDzxGzk5N3PFShVcsdVCY2+04M2gBETswvkGvZitgHv2QFhrMLxk6eTxkCve+DQjYux7yM3ajvX2Nsk177HYEymC3cAIya28rKSuVnG4pzTx8Ly4eeWq9ETp04nD/HOiQmoMvQdQIzcHUDVwF5VI+XuAJomgqOaZ7agACBn4UnmFtoGPEbBKuUkA8B9iHfK0Dt042IXvvl1hvzUd+gx5lPNM5tQAJCziL28qVVgdRdQXeVVJ+a6lXg5yaw5b5UZtJbSRVvCxcp31CdSAi43G6vHYOvJy+MVNmP0ypp1KADIWXiSwQHLDw2JuX2mylcnp9uuv8rMA9/EcFx3gEgfSAk4S9USCj2PwdazA/A+k6F7ZfURGoHJWXgMi0C+kdAKJotFC7cxbub0p0+kAvhiqZhjAV+eOjx5ebzPpO/PYojU7gBEZIOIfE9EnhaR/SLyqeL6xSLyoIg8V/y8qFJmh4gcEJFnROR9k+wA6ZaYrjZ3B2DVU3egd7li/btnjkZPL0u1IWV4tuqaVXLzHFm++WUdObESsTpiZS1jMZA+4jF1dm8fnsVYyFEBnQZwm6r+LoB3A7hVRK4GsB3AQ6p6JYCHiv+j+NuNADYB2ALgqyKyahKNJytDyoMj19CaY0f4wp6n3Pr8pobnWSS3zakEfzF1T66xOKUq8hich6B6Gwu1KiBVPQLgSPH7cRF5GsClAG4AcF1x2x0AHgbwmeL6Xap6CsALInIAwDUAftB148nKkDqBLHdbnxMjsGXTOvzVwuGoqij3/IKSPuaXyf0uUwn+ctVnbXP2W3V4+0Kmi8sILCJvAvAOAD8EcEkhHEohUY6ASwEcqhQ7XFwL69omIgsisrC4uNig6WQWSPmDV8nxyNlw8b9Irhw9fv05aSX67pESfh9WrESdX/7BxRO49VuPYucDz5pxFh7f/iYxG2Q6ZAsAEVkL4B4An1bVV1K3Rq7psguqu1R1XlXn5+bmcptBZpAc1UVOUrmdDzyb1F+3iVFo0uaVxtum8PvwJGELPba+u+8l7Nl3pDZWI8e3fxa/WxInSwCIyLlYmvy/par3FpdfFpH1xd/XAzhaXD8MYEOl+GUAXuymuWQWyV2ZW5NyWf7EqdPJXD9h+ZTBMyVwvEnlcvHuKsL7vXEL1e89dfRm7DsNk8H9wdvW4bo3z5mf7fHtZ9Rvf8jxAhIAXwfwtKrurPzpPgA3F7/fDOA7les3ish5InI5gCsB7O2uyWTWyJ0crEk5DAQLA8lKwvJ1yeGsCdWbVC4X78o3vL9OaKW8cKw+WcF5YTK4r/zhu7D+wvNN9ZtleI71md4+/SEnDuA9AD4OYJ+IPFZc+yyALwG4W0RuAfATAB8FAFXdLyJ3A3gKSx5Et6rqq103nMwOlsEvdgBIKilc9bCY4ydPY8++I2eVtcqXk1LuISiTMlB667W+i1jMhOe7DHXwVTfOso7Uc4h9n5bhmcbenqOqU//3rne9S8nweP7ocd1+z+P6/NHj0f/H7v/3/+17+tuf2aP/8X8sLLvXKr/9nsf1tz+zR7ff8/iyex/+0cvZ9XTRx9y/paj2J9aPunrD78P7XXi/TzI9ACxoi7mXkcBkYoTuo3Xn+e5+5CCeX/wlXnf+OfjY/IZlroWeIwVTn2X9rYn3SpPPqaO6qo7VYZ3vbEX3etvo/T5Jf6EAICtGnW/+1ms3Yu8Lx0zfck/qgSYpDJpMcqnPSf0tJWyqgrOtmqiuHV3ECJD+wmRwZMXIiRp+y7oLTG+UpieJ5bajyfkBKYNnKkaiajxNeQ/F6rCMxakUDN7c/WE9NOwOEwoAsmLkHvl46Oe/qs0bVJcOIuWR401jHOJx90zFKZTeOXXeQ+HfU779dbEW1Tqsw96ZymE8UACQFaNugi0n5g9vvtSVY8ZzhoHVjtikak30ue6edXEK5YracrG0JumUG6vVb+tM4FhCOYC+/GOBAoCsGHWTSjkxf/uxn5pxBbEJtc0ZBiW5SdRy+pGqM0YpDMJzEKxJOhUwZvU7vF7XB6p8xgGNwGTFSCWVA14zMG6+7ELsqnizlMTODqiWq0uAVpJr5LTqqOtHtV3WOQip/ld/xspb8Q1WP2J9ye0DGTbcAZCZ45HnftbaSOvZAXiSqMXUQpaqyBvZG5J7jkJVZWR5Fg39fGTSDAoAMjOUKg+Fts7977EBeBLKxe61yqdsADn1eHMdNWkHGTdUAZGZIRYAlUrhkDoaMTdXfc4kW3e8YsovP6ayivUFWK6mSZWP1ZGKb0jVQ8YLdwBkZqgaHnPUPSlPllj52DGRKUNtKBwsFYvHOynsK/DaUYsxQ20qO2hoqE0d0WgZfIdwLgJpDgUAmUlyfPu9rp5eVVEoHLyxBTl692qdYR1ef/yc85lTn0/GBwUAmUliq9ZwRZxKQx1T68T87VOG3tD/3ut+WXdgfZmLv/yMpnWk2tDkeybjgTYAMpPE3BRDnbxX5x1LaexJfJZyv/ScmRvWn0rCllsH0CxfD91Bxw0FAOkN4cq+zB4aTo7WRBgzGsfqSJW3DMaeBGpdJLWL1cHJnHihCoj0hlAn703eFjMae+rwpmCwkrB5ktF56yDEAwUA6Q2hvtoTFwDkG2o95VP1WHV5k9F5jdeE5EIBQHpD6M1iTYKerKGxOqzkbN78QtZ1bzrqph5GhNRBAUB6izUJxtwnPYnkrORsTXYAVr792EHtTesgpCk0ApPe4jnpypNIzqojN7q4JJW3P/eg9lg7wjoIaQp3AKS3eE66Sp2WFUbkWnU0MTqnoozDWAMafMlKQwFAek3MQBpbeddF1eYYWlMTcay8peqp3h+msairh5AuoQqI9JpYIrYu1D3AcpVPqnysHZaqx0oql1MPIV3CHQDpHdXVdGxlX5cTB0ire6xUECmVk7XDSNUVawvP4yUrCXcApHeEaRCq6ZB33PsEjp88bRprY+XLsuUOwEoFYZUtsXYj1v1WfU1SOpSkDoUhJIQ7ANI7Ylkzq26bJ06dThpr6/zqU4e0pw5W8exGrJ1BtT/VCTw3bTOzexIPFACkd8TUNdXMmmvOW5X0mgmNueHEnjqk3ePVE/usktRZBrHJPndip8GYeKhVAYnINwB8EMBRVX1rce3zALYCWCxu+6yq/nXxtx0AbgHwKoD/pKp/M4F2kxFTVZHsfOBZ7Nl3BMdPnsaf/+E78cWPvB0HF0/gdavPNSfBmMoox2gclquqWSxVT8rYG7tu1ZWrFqLBmHjI2QH8BYAtket/pqqbi3/l5H81gBsBbCrKfFVEVnXVWEKAs1UkJ06dBoAzP8u/W8baavlclZFVrk1untQhLVZSOOt+QppSKwBU9fsAjmXWdwOAu1T1lKq+AOAAgGtatI+QJGvOW3XWz5KUyiRU11gqo9wYA8CfmC6l088N/OJxjqQtbbyAPikinwCwAOA2Vf05gEsB/EPlnsPFtWWIyDYA2wDgjW98Y4tmkDFz2/VXRdU9Vs59YLmKxVIZWYfF5MYYpIzGdd5E1Z+WZ0+dhxEhdTQ1An8NwBUANgM4AuDLxXWJ3KuxClR1l6rOq+r83Nxcw2aQMZKz8k2tousOW0/VkYoxCNVO3vMDLKxdBA2+pC2NdgCq+nL5u4jsBrCn+O9hABsqt14G4MXGrSMkQnXlCyC6Ck7tAMKjHT11pBLCVdu19dqNSZfRLuqhwZe0pdEOQETWV/77YQBPFr/fB+BGETlPRC4HcCWAve2aSMgSMVdLT+7/aj3VSbXt+QGxv1mr/5T/f6xtsXrqdkC0DZBcctxA7wRwHYDXi8hhAH8C4DoR2Ywl9c6PAfwRAKjqfhG5G8BTAE4DuFVVX51Iy8ngCXXfls47dlB7LCq3JKbHz60jtXIPg8lin58TGVxtW1fRxYTEqBUAqnpT5PLXE/ffDuD2No0iBLBTPuTmy49N6tXy1Xpy68g9MN76/JQ/fyqLaU4brDoIsWAkMJlZQiNnbr78OjVLjJxjGlN5dmKqGs8h71YdllDw1EGIBQUAmVliwU85eftTaRZy6yipCoZUbIHn3F7PofPWhO6JUSDEgtlASa/I0cvXpU3wePeEdVl2hVh5T87/LuqwYhQIseAOgMw0oUdLNePmzgeedeXaL6lLBlelqm4BYObq9/j1Wzn/wzpSKidP5lFCLCgAyEwTqjqqk7VC3aoeIK4ySunNq/XkqIrqPj/WhlgddeksQqHFs4OJFwoAMnPEUjyXk1x1sv7Y/AZzxZta1eceGB+rx5pkY2cIpFbksXrCdqTKx4RWXT8YG0BCKADIzFFd+YbqnKq6p+6QdmtVH07WgK3aCeuxJtlYplBvOopQrZMqb3kYpfrBg2JICI3AZOaoGj5LPfjxk6exZ98RAHk58lO5+8vJunpQu8fYasUXxOqxjMa5cQIeo3OqH6nUGGS8cAdAZo6q+2e5clVoVgK3sI5Y7v6SnGRq4ao651jIsu11B7zXGX1T5esOvg93PdZugqqhcUMBQGaW6mR72/VXLZsMc9QaKT16TNDUCQpPoFWdsKgz+nrsGE3bQdXQuKEAICuKZ8VZp3/PcXvMWfl6VvV1u4Zqvd6EcGH/mqSTts4TTtlDPDsrMixoAyArSipRWagCCfXZsTTOVmK2sq5ygkwdyAIgO4CqLgVzmMq57rOrdYX9S+ntLRuA9zzhOpsGGTYUAGRF8RyGHptsq+UPHftV9qlfOe3oIn1CtV5LWMSM3DGht/uRg6aA85xAVie0Uu0hw4YqILKieA9DLylVG8BrwsHrJhlrB4BOJ7xq/yx1V/Wzt35z4SwdfEzdVI0tSJFS9VhtsVJu0yYwDigAyFSITUh1fv25Cdzq6qqrtw2hDSBVd7nCv2JuzZlVeLV8ypMpVndKn2+1JbzOdBLjgiogsqJ06defOpylLiFcnY2gKTk2gFgbUwfGtDm/IPZZVcJ6Ut8pGR7cAZDWeD17Sr9+z4HmsdVwnYunlTO/2g4rj1BTql5CKXVXtU+p3P+WTj7n/IIcwnq4AxgXFAAki9Qkn6NGCV0fb7v+qqxc/2EdOXl5curLCQJrQt2knyKmw/ecHQD4zyAI1UZMKDcuqAIiWVheNSkf+urqNcf9s6qOia18w3z3Vl79kpQbZZ1nTNekvGtS6ihLdRNT1YTPovqZVj11rqf0Cho2FAAkC2syTR1CUp30U5NxTDhU8/SkDmlP+bDPkj471seUPaQkx5W0+hnVZxF+h6HgjgmH0PW0zp2W9BsKAJKFN/lY+Lccn/aqgTdWb5tD2qeNNWHfufcQ/uBt9tnF1gq8Lkai+jOWEC+c2C2Dc6oO0n9oAyC1NJ1Mq/rwHIPtF/Y8tSwNdJhCIbcdnpw9K0GsP6UdIrSH5LqSxk5Li9kf/mrhcNRlNBZjENaRk1QvbAcTzPUHCgASJSenDVB/alVZR47BtvSJtwKfPEbSSRl525A7YVf7lOpHndF85wPP4s69h3Di1OlldXgndk87GEzWH6gCIlFy/dlThliPT3xpI/jcB68+MykBqPVltz5/pY28OeTo08NdTqofVjqIUnAoFACw5rxVrjrKerZ+cwHPL/7yTHtz66h71mR2oAAgUaoT66FjvzLvaxJ8FKPcIXxhz1P43Aevzi7Xp2RmOd9HyqgeEhMOVSFz2/VX4XWrz13mEVTelxIwYZSyZYuIXZ9F4UviUACMnJSRsZxYf/TS8bNWgmFZK5o2PM6xuvqNJUHb+8KxM2oia1LK3RnMIjkTY9uTu6qCG4h7VQFwueTGPIpKUt5NNBjPPrQBjBxLX1tVRWy7dmPUgFsXTVvVIYdG4PBzN86txe5PzJ/RM8cMid5TsPqI96CX8DuKnSIWqpVycgaFzzT23VtGedoA+kOtABCRb4jIURF5snLtYhF5UESeK35eVPnbDhE5ICLPiMj7JtVw0p66k6JKg+tjh3/RKO1AdSLISTlQd0LXGKJUPakYcqOEQ+N5k6MmY9+95RyQEjBktshRAf0FgD8H8M3Kte0AHlLVL4nI9uL/nxGRqwHcCGATgN8C8Lci8mZVfbXbZpMuSOmbq2oVK+9+XaBVqg5P2dS1oeEJXsuNEk59l6F/vyfeI2WA74tdZuzU7gBU9fsAjgWXbwBwR/H7HQA+VLl+l6qeUtUXABwAcE03TSVdY23rc/Pu1+Xvr+qBc5OOxT6/ZEiqHguP+2purEQsMV7MDdQb72HtJKx6GB8wezS1AVyiqkcAoPhZLhUuBXCoct/h4toyRGSbiCyIyMLi4mLDZpA6Ui+dta3Pzbufm78/NiF4k5bl9GcItBVyllomR13UJN4jNja8bSDTo2sjsESuaexGVd2lqvOqOj83N9dxM0hJXUbMUFcbu2ZN1ildb93EYgkVj12BLCfcQVgH0JdUn21qR5hTvqTpsyUrT1MB8LKIrAeA4ufR4vphABsq910G4MXmzSNtqUvBEG7hY9dS3jeWMbFuYrGESp2hdxYjfGeJcAdR56lVfTapHWGsvKXqafJsh76zm1WaxgHcB+BmAF8qfn6ncv0vRWQnlozAVwLY27aRJI+Y/7XX2FqnrkmdvJWbYTL2uXX1McioGbHna9lm6gzvsfODY04E1rP1ZoQlk6dWAIjInQCuA/B6ETkM4E+wNPHfLSK3APgJgI8CgKruF5G7ATwF4DSAW+kBtHLEXiLrpau+zGFgT/hS59SRSj/sCW5KBY+R9pRj5PjJ07hg9TnLAr6slNGx5+sNWgsXEjnnFZDJUisAVPUm40+/Z9x/O4Db2zSKNCP2Qlqrd2vF1aYOyy0wVj5nxcdJoT2xZwQAr5z8de33X5bd+8Ix7P7E/LLn0XZsWSmpycrBVBA9JOec2DpVi7V6a1OH5f/d1Kef6p52WC6hZSqOMk+QxZZN6/DdJ47g+cVfYvcjB5dN0m3HVioRHVNJrAxMBdFDUimQQ6+cUKVS59nj8dQI6/D4kY/Bp3/ahN5XVUNr7ED60Ah7//6X8Mo/nT4rIVyYzjqMLyjL5Ywta2yO3RV4JaEA6BmpSTbllVPnl1+SGxsQq4P+37NF6DGVeg7l33Y+8OyyMwB2f2LeTM/RRdxIWAddgVcOCoCekQrWAfJ8sK06LMGQG+xD/+/ZIlzlp1xoy78p9KxT2aor/FiMQY4baCqWIDdIMPdwGuKDAqBn1CXayvHBtuqwBEOsTk/A2BiSuPWBlNqt/NvH5jeYWVtjMQax8WJlfo3FEuQuJFLtIM2hEXiGsfzgU4m2LL/v0N0v11hrrfJidaR8+61Tw8hskWuoBdobe8tr4djwOCJUofHYD3cAM0yu7j2njpz87lb5HHVR6uVL2SbIbOFRr1iqmjYJ5VILjpjBuQptA34oAGYYT6Kt6t9zDGptE8GF7ah7+ai37QdWKoncXFKx8ZmqI5zwmySki9VD8qAKaEbITeOQ2gp70jjE6vEEgYXly59hfvkS+vT3k1QqiFw1YEoFGEaep2IDUhN86mwLYsMdwIyQu/pObYU9njmxejyZHcPysfzypP+kTmnznD2QewKZNb5TO4OVWP0PNfaAAmBG8GTM9ORmz83Z30V5gKqeIROLK8iN+2irikx5v9WpRbtgqPYFCoAZoYu86p46cv2vU8bi2GRPF73hEj5bz/jMjQ6u8zrznGPc5ap9qAsbCoAZoYuc+d5dRI66KLUD4GQ/bjzjM3fR4g1STLWjzmjsEQ5DHes0Ak+BtgbfSSRy85YnpO3YitVhje2UkdcTv1Ktj6nGuQOYCh4daQxrxVMXJVxXh6c8IbnJ4OryV1Xr6CJJIdA8ZcrYoACYArmqFo+xF4jrST11eMoTAuQtZjyxK95EcrE25HgFMT3JElQBTYhUZGyuqsWj6inJ3VJ7/fupAiIxYuPDm/e/mqZk9yPLjw+1PqckdtBMXUyAlTJlbGkkuAOYEKmVc+7207sdrrMt1NWROoN3TC8FySc2PnJVibE0JU3UkOEYT9VRGn8BZEc8V8tZ9fVVZUoBMCFSk3zbNAy5Xg9d5P0nxEuoSrTGVWyittw9PQuq3HMxwrak3Dytcn1/ZygAJkRKx5jr++zdAXh0r7E6hurrTFae6liyVuSxidpzJkVJFzE0qZ1uaiHV93eGAmBCpLahub7P3h1AeN370lDdQ7qiOpY8QVzWu2CNee/iKVWPpcpJLaT6/s7QCNwB3rz9MQNrLFmWJ5EbsNwY5jEW5/SJkKZY4y3XYOyND2hSjxUXMGRHCO4AOsDrrhlirZBiu4hc7yKrfF0ddX0ipAnWSjk3GZxXHeqpZ8yppCkAOqCtsdaqw+uXn2sMy5nc+67bJP2gbTI4b7LC2HvmiVMYGhQATmK6wrbGWk8dHtuC9Vnh5B7rU991m6QflGNxy6Z1yw59r17z7JJLct+z1E596LsDCgAnXbhrtkmW5c2KmGPsHfIKh8w2sXMkYteaZAnNfc/qdhHh7qDvvv9VaAR2EjPY5hprY5GPJR6DbxfJ4MI+1d1DyCSpjkErQtiTDA7oJuFh7H0fUiK5VjsAEfmxiOwTkcdEZKG4drGIPCgizxU/L+qmqbNBbMWRG1Ubi3ws6SIZnGVHqFPlUN1Dpk3s5LGcCGGPUwTgTwKX+773lS5UQP9OVTer6nzx/+0AHlLVKwE8VPx/UIQDoG1kr1WHtx4muCJDwGMDsNSXnviAunMDcg5O6iuTsAHcAOCO4vc7AHxoAp8xVWLulqFR1ROcEqujxKOzpOcOGQLW+dKx8W0thizPnlgdKUNyrB7rPeujbaCtAFAAD4jIoyKyrbh2iaoeAYDi5xvM0j2lbgvojexN0cU2lpA+kpMYLrUYyk0qZ+0uUobnWIxBH50p2gqA96jqOwG8H8CtIvLe3IIisk1EFkRkYXFxsWUzVpZYKofYSqW6hY2VqxIbPN6dRB8HICEWOUnlPHmGAN97looPsNrStx14KwGgqi8WP48C+DaAawC8LCLrAaD4edQou0tV51V1fm5urk0zVpzqCjw2eOq2sKFgCOss8e4k+jgACUkRJpXLtQ1Yk3pMYHjfs9TOoG/OFI0FgIisEZELyt8BXA/gSQD3Abi5uO1mAN9p28hZo7oC92TctARDWKdVPnVvtf4+DUBCUoRJ5WLvj2fx5PHqSe20U5HDfaJNHMAlAL4tImU9f6mq94vIPwK4W0RuAfATAB9t38zZIvQbjiVyA9r79nuTwZUwkRsZMuH49/j213n1NEm62Gca7wBU9aCq/qvi3yZVvb24/n9V9fdU9cri57HumjsbNE11myK22kjpN1MrfdoCyJAJx791sIxlqA1X710klOsrTAXRAV6jb5tkcDnQS4iMCY/Bt0kqiCEYey2YCqIDyhXJjnufOCtEvG2IerWsZ7ClzgEgZIjE0q7UedA1URcBr73vQ4A7gA7JXRlM2rd/SCsUQnKIJThsG4szJGOvBQVAh+Rm2fRsQ5vo8+kNRMaORy2bOig+N5isr1AATBCPexl9+wnpDo/LqDeYbEjQBjBBQn1jk3TQQ9I3ErLS5LiM1p3fbbl5DwEKgAkSDr6xHjxNyLQIF1Be3/6UcBgCVAFNkFAXnzJAUW9PyORp4ts/ZDUsdwArCN0zCZk+sZ14agc+ZDUsBcAKQjUPIdMn9h4OeZJPIao67TZgfn5eFxYWpt0MQgjpFSLyaOU0Rje0ARBCyEihACCEkJFCAUAIISOFAoAQQkYKBQAhhIwUCgBCCBkpFACEEDJSKAAIIWSkUAAQQshIoQAghJCRQgFACCEjhQKAEEJGCgUAIYSMFAoAQggZKRQAhBAyUigACCFkpFAAEELISJmYABCRLSLyjIgcEJHtk/ocQgghzZiIABCRVQC+AuD9AK4GcJOIXD2JzyKEENKMSe0ArgFwQFUPqur/A3AXgBsm9FmEEEIacM6E6r0UwKHK/w8D+NfVG0RkG4BtxX9PiciTE2rLLPB6AD+bdiMmCPvXb4bcvyH3DQCualN4UgJAItf0rP+o7gKwCwBEZKHNyfazDvvXb9i//jLkvgFL/WtTflIqoMMANlT+fxmAFyf0WYQQQhowKQHwjwCuFJHLReQ3AdwI4L4JfRYhhJAGTEQFpKqnReSTAP4GwCoA31DV/YkiuybRjhmC/es37F9/GXLfgJb9E1Wtv4sQQsjgYCQwIYSMFAoAQggZKVMXAENMGSEiPxaRfSLyWOmmJSIXi8iDIvJc8fOiabczBxH5hogcrcZppPoiIjuKZ/mMiLxvOq3Ox+jf50Xkp8Xze0xEPlD5W9/6t0FEviciT4vIfhH5VHF9EM8w0b/eP0MROV9E9orI40Xf/rS43t2zU9Wp/cOSgfh5ABsB/CaAxwFcPc02ddSvHwN4fXDtvwLYXvy+HcB/mXY7M/vyXgDvBPBkXV+wlPbjcQDnAbi8eLarpt2HBv37PID/HLm3j/1bD+Cdxe8XAHi26McgnmGif71/hliKp1pb/H4ugB8CeHeXz27aO4AxpYy4AcAdxe93APjQ9JqSj6p+H8Cx4LLVlxsA3KWqp1T1BQAHsPSMZxajfxZ97N8RVf1fxe/HATyNpUj9QTzDRP8setM/XeJE8d9zi3+KDp/dtAVALGVE6uH1BQXwgIg8WqS8AIBLVPUIsDRoAbxhaq1rj9WXIT3PT4rIE4WKqNxi97p/IvImAO/A0kpycM8w6B8wgGcoIqtE5DEARwE8qKqdPrtpC4DalBE95T2q+k4sZUO9VUTeO+0GrRBDeZ5fA3AFgM0AjgD4cnG9t/0TkbUA7gHwaVV9JXVr5NrM9zHSv0E8Q1V9VVU3YymbwjUi8tbE7e6+TVsADDJlhKq+WPw8CuDbWNqGvSwi6wGg+Hl0ei1sjdWXQTxPVX25ePH+GcBuvLaN7mX/RORcLE2O31LVe4vLg3mGsf4N7Rmq6i8APAxgCzp8dtMWAINLGSEia0TkgvJ3ANcDeBJL/bq5uO1mAN+ZTgs7werLfQBuFJHzRORyAFcC2DuF9rWifLkKPoyl5wf0sH8iIgC+DuBpVd1Z+dMgnqHVvyE8QxGZE5ELi99XA/h9AD9Cl89uBizdH8CS5f55AH887fZ00J+NWLLEPw5gf9knAP8SwEMAnit+Xjzttmb2504sbaF/jaUVxi2pvgD44+JZPgPg/dNuf8P+/XcA+wA8UbxU63vcv3+DJTXAEwAeK/59YCjPMNG/3j9DAG8H8L+LPjwJ4HPF9c6eHVNBEELISJm2CogQQsiUoAAghJCRQgFACCEjhQKAEEJGCgUAIYSMFAoAQggZKRQAhBAyUv4/jSXXjMKmiq4AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "utilities.plot_raster_dot(input_list[0][8])\n",
    "utilities.plot_raster_dot(output_list[0][8])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAD8CAYAAAB+UHOxAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAAzLElEQVR4nO2df2xexbnnv09jmgQ72STLC0mJq/LmhvQmpetSK1updUu3t5D2Rktb3SJQVfiDtf8JUnuXlUr2Sqg3EiqtBFcr9Ydkb6sLKwoClapR6GbhokJTqdvUdN0Ehw0kThe7JORlg9c4OGkMz/7hc8zx8fk1Z2bOzJwzHymyc/ye553fzzzPzDxDzAyPx+PxNI/3mU6Ax+PxeMzgFYDH4/E0FK8APB6Pp6F4BeDxeDwNxSsAj8fjaSheAXg8Hk9DyVUARLSKiA4T0R+JaJyI/jF4voGIniGiV4Kf6yPv7CWiE0R0nIhu0pkBj8fj8ZSD8s4BEBEB6GbmWSK6DMBvAHwDwFcAnGPm+4noHgDrmflbRLQdwKMAdgL4AIB/AXAtM7+jMyMej8fjESPXAuAFZoP/Xhb8YwA3A3goeP4QgC8Fv98M4DFmvsjMpwCcwIIy8Hg8Ho9FdBX5EBGtAPACgL8C8ANm/h0RXcXMpwGAmU8T0ZXBx68G8D8jr08Fz+IyhwAMAUB3d/fHP/zhD5fPhWVcnH8Xb8xexBU9K7Gyq9plliLfrTJ9JvPq8TSdF1544Q1mbpV9v5ACCNw3fUS0DsDPiegjGR+nJBEJMocBDANAf38/j46OFkmKE+x98ggePTyJK1vdGLm9H+1WT+XfvXtnL77zlY+W/ozK7/N46s5EZxYjhyYwONCutL8T0f+ReV9oysbM0wCeA7ALwOtEtClIxCYAZ4OPTQHojby2GcBrMol0jcGBNra0unGycx4jhya0ftdEZxZ7nzyCic7s4nfftrMXgwPtzPTlfaYoKmXFiefN46mCMu1u5NAEHj08qbS/p6VDZb/ItQCIqAXgEjNPE9FqAH8D4LsA9gO4A8D9wc9fBK/sB/BTInoQC4vAWwEclk6pQ7RbPRi5vX9xRqCTsOEBwHe+8lG0Wz25M/EinymKSllx4nnzeKqgTLsL+7nK/p6WjuhzWYq4gDYBeChYB3gfgMeZ+QAR/RbA40R0J4BXAXwVAJh5nIgeB3AMwDyAPU3cAaRzYIwi2vBMmaplyMubS3nxuEOZwVxHf09LR/T5/ZLfkbsNtArqtgYQxbZBKvTZ32aRzz6tjPLKzsa8eOzEtn6oCiJ6gZn7y77vt21oRodvUAadPvuypJVRXtnZmBdPOibXdGzrh7ZQaBeQpzw6fIMyVOWaEqGIqZuEjXnxpGNyTce2fmgLXgFoxg9S+aSVkS+7euEHYfvwLqAa4LdLelwgVOgmfPAuuIBM9GOvAGqAycbtivJxJZ0ePbiwXmSiH3sXUA0waVq7slfflXR69OCCO3FwoI235uYxM3cJE53ZSiwlrwAsRWTbmsnG7Ypf18Z01nVroqcc7VYP1qzuwqOHJ7F29WWV9GmvACzFlRmrKeUTDp67dmzEwfEzuYOojTNAV+rYUx1VT1S8ArAUG2esNhEOnodPncPJznkA7g2iTa5jb/0kU/VExS8CW0a4WAnA2I4JFwgX9e7dvd3o4p7M4nKRXTF1WbyO58OFXTlNoBEKwKVO5DtGMcLB8zPbrjSqKHXXl63tQbRPxfPhwq6cJtAIF5BLvtYmuwVcRHd9ReXb5DYR7VPxcqrC1WFTedlKI4LB+YbgqQM2Bb9zoU/ZVF66kA0G54wFINPgbNwB4vGIYpN16EKfsqm8bMWZNQBbfaE249LaRx51yktZTIZScJEy5dW0duaMAvCLRuLIKk3RzqCz8/gJgL3UadBsWjtzxgXkgslpG7ImcHShb3CgneuC07nY7s354lTtn3dpk0UeTWtnzigAjziySjPaGYp0cp2dR/UEwNZFTBXpqnpArtOg2bSJplcAnlSinaFIJ2+3egpZCjZg66xVRbqqHpCbNmiWwdYJhxVrAH+enquF/1AE1/ymRRfUXPGh2rqmpCJdfrE4vX+Z6ndp/cL0OGCFAjh3/i/WDxiqcWWgFMXWgTVOfJBU0RFNd2bPe5S9Z1oXaf3C9DhghQtoQ/f7rR8wVFMnv2kUF9wBSea4iOslzZy3RYan/D3TukjrF6bHASssgKvXrW5cA/ZmujpEzf2kWdeuHRuxpdWNXTs25spIm7UlyRBJg4iMic4sBh8eXSbDNtdH1eQFU7St35lOjxUKQCVNaehNRWRAFhlkD46fwcnOeRwcP5MrI01ZJMkQSYOIjJFDEzjZOY+1q7qWyBAti7r1F9MuFdewwgWkElt3d2ThTfniJNVvmhm9a8dGHD51LnOQ/cy2K4VlJL2fJkMkDSIyBgfai3chPDE6tXgpjmhZuNhfsjDtUnGN2lkAMouQtu0Q8CwnbeacRNqsXlaGDWlot3pw7+7t2NLqxuzF+dz2I5IOl60C0y4V16idApBpALbtEGgyaYNQlW6WNBm63EVlZXSvXLHYfnSWhcuKwZNMrgIgol4i+hURvURE40T0jeD5t4noz0Q0Fvz7YuSdvUR0goiOE9FNOjOgElMDcZrSUtnhbOi8IguUIgOZ6CArKyPtsyLKQqWMW/p7F9uPzrJIqpOmLzq7ThELYB7A3cz81wA+AWAPEW0P/vZPzNwX/PslAAR/uxXADgC7APyQiFZoSHvtiXc4mU5l0s0UpvvBp18uvEApMpDFn4Xf17d5XWkZWelI+6yIstAlQ1ThyMoQWXT2SsE+cheBmfk0gNPB728R0UsArs545WYAjzHzRQCniOgEgJ0AfqsgvVqxbUEsvqAlkz6Ti2Nhuv/2uo2JFpbqhdb4hfGqF2t1LRiXlbFrx0bsffIIBgfa2hadVZSFbf3LI7gGQEQfAvAxAL8LHt1FREeI6CdEtD54djWAychrU8hWGEqRmWWILMxVQdQ1NNGZxczcJey+blOpQbzqxbFoPYSutVv6exM/q3qhNfy+oYG28cXatPcnz72Nw6fOYfLc26VlhDwxOrU429aVDxUybOtfOtBp5eiQXVgBEFEPgJ8B+CYzzwD4EYAtAPqwYCE8EH404fVl904S0RARjRLRaKfTEU13KmVcHWHBPjE6ldjwbWDk0ASeOnoGa1Z3WbXDochBp1D5HBw/U8lCa/h9Y1PTxvf2p+Vj34FjONk5j30HjkmXBYMXLasqFp1Vl0Wd0Olq1SG7kAIgosuwMPg/wsxPAgAzv87M7zDzuwBGsODmARZm/NGp3mYAr8VlMvMwM/czc3+r1ZLJwxLKLOQmdSTbsHWnkK7FWtsWWlXnY2igjbWrujAUq88kGX2b12Htqi70bV6XmLboIrCLZVEndPZTHbKL7AIiAD8G8BIzPxh5vinysS8DeDH4fT+AW4loJRFdA2ArgMPKUpxDGVdHWLB337jN2j3Etu5v1rW4WJeF1rTPjk1NY+bCPMampnNlDB+awMyFeQzHlGxdyqJO6OynOmQXsQA+CeDrAP5dbMvn94joKBEdAfBZAH8PAMw8DuBxAMcAHASwh5nfUZZiDdg6uNqEyL58IHm2IjKwpM12RAYWXTJ05SNNRnjg697d23NluFgWHnPkKgBm/g0zEzN/NLrlk5m/zszXBc//fbBbKHznPmbewszbmPm/682CpwpEDxglYcvioupFZ90yejdcjp3XbEDvhsuV5kOFDBVl4TFH7U4Ce/Tg4uKiDad7bT6lXEVZPDE6tWg5Zi1i+jMCZvAKwLOMpM7oomvBBpeTqOKsW1kweHHQz7IKbIqHVUQZ1UVheQXgWYbITC8NG1wLNricbM6HChl579/S37uoTLIWgZOUjq3BGdPuYnARrwA8y3DBteCKm8Vm11kVZXFw/ExunCIgeSOGrcEZw7sYtrS6nV/Q9gqggYQzq+ePny28syet84oMLCIyRGasooOsrIxdOzaid/1qnJ6+sKTsZPOR9jw8oPjE6JSVZSFbp2nYFpwxnq6R2/u17Bys0vLxCqCBhDOrfQeOSc/0RAYWERkig5DoICsr4+D4GUy+OYfnXu4sKTvZfKQ9n704v+SnbWUhW6dp2Lo9W3e6qrR8anEjmMkbtVy8zSucUfVtXofhYHEuSrTzlgn6pUJG0vtpMkTSoELG4EAbM3OXQKAlMpLkiqQh7Xn3yhVLftpWFkn5EKlTz1KqDNxYCwvA5A4Cm3YvFCUvVo4Li4siaejbvC7XpBbZldNu9eAHX/s4vv+165co/aSZYZobI20WmfT87hu3LZ5Uz/usSD5Ey0K2PEUtgCLUZTdOlCotn1pYACZDHbt8B6nITC8tlK8uGWmzxSQZeWkYDhbtwndE7hUOO2NZZN8XlSGaD9nyBCBdpzL4ENNy1EIBqOhkLn63CEmuKtvdLPH302TkpWHXjo04OH5mmcz4rN6FesxDNB9l3FbR8gzDWsvUqQxRmfE27oJ71nQaa+ECyqOOZqIosnv7dbl6VMhIez8cDD+z7colJrWti4smEHFbhfRuuLx09FGd6Y+3cRfcs6ZvTnNSAYgWkAsNQTd13s8uur3Qk02aglRdp6oHunhaXAg+l5TGKscrJxWAaAG50BBUIRK1U9e+fBUykp5NdGYXL1KJyihTv7pmWXW1NtNupJOpU9UDXTwtLlh6ZSwwlTipAEQLyIWGoAoVM2dd+9nTDlAV3YuedgKzTP3qmmWpkmubIkm7kS6tLxapU9UDXV0meo3dBVR0QaQuC3Y6sGFfflRGuGUwjAUz+eYcJt+cw6ZDE5k7cJLSEP2cbOeQWZDMaqeDA228NTePmblLi4N3mUU+23a3iO4uKlKnqvuxHxfEscoCqIOvPjpzMzGLs2FfflTGcKROQwvghmtbuQNvUhpUzoyislSuKbVbPVizugtPHT2DkUMTpdu0bbNZ0bIXOaMA2GfxAHamSTVWWQBZszLT26WKEp25Acl7pHViw778qIyhgTbGpqYxONDGyKEJTL45h09tvSIx6FdURpXnK0Rn23lpS/q7aD7qOJvNKjfbLB5AT5psG8esUgBZjd7GBpKEis5fhLSGpHJf/r4DxzCy4XK0Wz2lQ0GMTU3nDupJMqocAEWVTV7a4n+3ub1WSVa5lVX4OgdUHZMQ28Yxq1xAWagwiasw6aKmr87FHNGwv0mEg3T01qZQxtpVXTjZOZ/rutB5PqAqXN8kYNrtqIKydaDTbSzbLpLqwjbXnjMKIF4ZZRp6HdYYQlTuy5+9OL/kbwfHz2DmwvyS3TY648zb1ilcI1quMm3cReVhc9tJqgvbJhvOKIA4ZRq6jY2lbKdTuS+/e+WKJeUSfvbe3dsXG6qowim6tx+wr1PYgEi7iLZrmTZepE/F06VSaZSRVVXbKZM2G8ebOM4qgDKFa+NAU3bGpmpf/pZWN27p711SLkX35Yumw7SrxyVE2oUqt2ORPqUz3ILNFnqZtNk43sSxahFYBJFFQttW3qOUXWgquy+/7N5+FcHgqtzZ4zomyqpIn0oLqKcinTa3D1Np0z12WW0BqDIvbZ5ZRDudSF7L7ssvu7dfxWKv6hmRafeDTmydPcbTpetshm2YSpvusctqBaAq8y744kSjAoos9gJ2BINTTVPcD03DNmVskqSxS2X5WO0CUmV22XyoJjTxohEMQ7L2DKe5ZERcNSIy0lw9oulQiYr2kVX+HjPYtldeFWXcOUljV/ywqQxWWwBVmF0mZxsTnVkMPjyKRw9P4uD4mdSogLt2bFyWRpHFXsCOYHCqUdE+ws6UVP4eM7hgsZfBRo+G1RZAFZicbaRFtwwJB7i9Tx4pHCrBlmBw8ee2LsTrWtwzlV9by1kEmy12GWz0aORaAETUS0S/IqKXiGiciL4RPN9ARM8Q0SvBz/WRd/YS0QkiOk5ENylJaQmKzO5NzjbC7x65vT+zs1a9WPvE6NTiCeEy7wPLZ+e2+th1WZmm8mtbObviz6/ifggbF7mLuIDmAdzNzH8N4BMA9hDRdgD3AHiWmbcCeDb4P4K/3QpgB4BdAH5IRCt0JD6PIp3BZKUU/W7ZG7NEZcxenF/yM+39LBlx6mrWpw0cpvJrWznbppDSsP1+CF3kuoCY+TSA08HvbxHRSwCuBnAzgBuCjz0E4DkA3wqeP8bMFwGcIqITAHYC+K3qxOdh875iEXTty0+T0b1yxZKfae9nyYhTV7M+zYVoKr+2lbMrfVBXOm3Pv9AiMBF9CMDHAPwOwFWBcgiVRDjaXA0gukQ9FTyLyxoiolEiGu10OonfJ2uW2WhyqUJnELZb+nsXTwiLyrDV5NeVLttm3LbhSh/Ulc6ick31m8IKgIh6APwMwDeZeSbrownPeNkD5mFm7mfm/larlShI1nyqy72vWfvywxu3wrTo2ttf1OVkq8mrK12uDHBZTHRmseeRF3DXI38o3KZtVfRFsDHtpvpNIQVARJdhYfB/hJmfDB6/TkSbgr9vAnA2eD4FIDp13AzgtTKJk51duebXS2uYWVs4h2NpqfKS9iQZts6IbU2XDYwcWrjv98DR04XbtK2Kvgg2pt1U+8xdAyAiAvBjAC8x84ORP+0HcAeA+4Ofv4g8/ykRPQjgAwC2AjhcJnGy/kzTfj3RLXlp/uSsQ1zRG7fSPpsno8g6QtG1Bdt80CEq0lWHLZZJDA60MTN3CQQq3Fds921nUcUhRVFM9Zsi5wA+CeDrAI4S0Vjw7D9jYeB/nIjuBPAqgK8CADOPE9HjAI5hYQfRHmZ+R3XCi6CrUIvKVXXVYNa+/OiNW2VlyCwY2zrg66CuJ1TbrR784GsfF37H1TJIa/dNpMguoN8g2a8PAJ9Leec+APdJpKsWqJoliczedcnwncb8rNe0BWL6+1Vhuh5twupQEK4jukCoIpDbg0+/jEcPT+LBp18uLUNkwbhJmF7wNe271vn9VS7Mmq5Hm3BSAdi4iq8CkV05aYtGSYe4RGWIKAtPdUTry0Qf0LlQaVq5NRUnFYDqxmKLQlGxtz/pEJeoDNm7A+qOqfYSnbmaGDB1zpxllItIfdRla7gqnFQAqmcitsw+VOzLTzvEJRtOwlsA72FDe6nbtlYZ5SJSH65tDdeNk9FAVe9AsGVRSCTmvmgoCBkZE51ZvDY9hw+uX+0tACxtL6YWRm3ZhWPDwrBI/zW9Ndw2nLQAVGPLopCKC9Z1yBg5NIHnX34Dr7455y0AmHfF2IQN+Rfpv6ZDPtiGkxZAXan6kvaiMrIOCtkwAzSJqzM/VTQ9/65jnQXg6mKKCCIhH9LQGQwuLiM8KPT9r12/bJCvegZoqn2kfa+rMz9VRF1Rde+3dcQ6BWCDSambtDxWfUm7qIwkql6MLNM+VCiNJrRLGXz5uIl1LqAmmJRFr1EMUXFFowoZSVS9GFmmfagI4dCEdilDXvnU1VWoMl8mysg6C6AJJrVoHlXsy3dhb3/STD3+rEz7GBxoY/d1mzAzd8nfLVGSPCsqr3zqaiGozJeJMrJOATQZFfvydcrQTdL3iqQly0+/ZnUXnjp6pnYDUFXItgmVrkKb1glV5svE2Q6vAAyR1IhFfPIicf9VydBNUgcQ6RRZg5TqzmXTIFQFsuWn0oKyyZpIylfZtmHCyvQKwBBJjVhkZ0/aZ3XK0E1SBxDpFFmDlOrOZdMgVAU2ucBsPwXtUtuwbhG4Kcju7Re9pF2FDNupckHaLwqbw5ZT0Gm41Da8BWARZfb2T557u9CdwCrOB8jQNJdJWVwqJ5fSWiVZ1pJtZeYVgCFUXdK+78CxQncCywaD05Ffl/FBxdxKqy3YVmbeBWQIkX35WYHcit4JrCKgnAwumcVF8EHF3EqrLdhWZt4CMITIrpysQG7hncChuakzoJwMNi0iqqCqoGJFzkaYom51WgW2lZlXAIYQ2ZWTtOtBNIyDChm6sC2+j03Ino3weLLwCsAQsidzqwwGpxubDp7ZhuzZiLrigvJ2Aa8ADKHrkvYqgsGp7nymBjQXBlLZsxF1xQXl7QJeARhC1yXtOmWEqO58pgY0P5C6S9hGd+3Y2ChLQPXkyysAQ1R9SbuIjLyBsczMuU4me53y4iphGz04fqZRloDqyZfTCsDljqjrkvYqLnovM3Ouk8lep7y4jgtuPJWozq/TCsDljhgdkENF1rd5XaXB4Po2r8PaVV3o27yudD6eP34Wn3vgOTx//Gzm51SEZLaFpg06NlOFG8+miabq/DqtAGzuiHmNJjogh4ps+NBEpcHghg9NYObCPIYlFOi+A8dwsnMe+w4cy/xcnUIy+7WDZuHyRDOPXAVARD8horNE9GLk2beJ6M9ENBb8+2Lkb3uJ6AQRHSeim3QlHLC7I+Y1mqjyCn8fGmgXnr2LLPamybh393ZsaXXj3t3bS+dTRIbNCruu2DR7dZU6t9siFsA/A9iV8PyfmLkv+PdLACCi7QBuBbAjeOeHRLRCVWJdQqTRhIpsbGq60r39vRsux85rNqB3w+UFc7Wcz2y7Es/efUOhkBE2K+y6UufZa1XUud3mKgBm/jWAcwXl3QzgMWa+yMynAJwAsFMifc5S5oo8nYHcRGS4ip/tLqfOs1fA17ksMmsAdxHRkcBFtD54djWAychnpoJnyyCiISIaJaLRTqeT+AVFK9eFRhBPo82hIFylbgpNBXWevQJm6tyF8aYoZRXAjwBsAdAH4DSAB4LnlPBZThLAzMPM3M/M/a1WK/FLilauikagu1LjabQ5FISr1H2261mOiTqv00SjVDhoZn49/J2IRgAcCP47BaA38tHNAF4rm7iioVNVhFgNKxWAltuG4mkUCQedlDbRUM4iMlzF9puiPOoxUee2hXSWoZQFQESbIv/9MoBwh9B+ALcS0UoiugbAVgCHyyauqPmqwszVPZOIp1F2b7+oBaDyIJjHk4eMRZ32ri2uF9nxxpZ8AMW2gT4K4LcAthHRFBHdCeB7RHSUiI4A+CyAvwcAZh4H8DiAYwAOAtjDzO9oS71CdPpKkypcdm+/6BqAinASHk9RZNwkae/WxfViUz5yXUDMfFvC4x9nfP4+APfJJKpuJLlfZC9pT3P1qJDh0cNEZxYjhyYwONCu7aJsiIybJO3durhebMqH0yeBXUF2B044UO87cCxzF1ERGS5ZADaZyrJMdGYx+PCoNTM/3chY1Gnv1mVHk035sFIB1KnjA8kVLrq3f+2qLpzsnM/cRZQnY0urG32b1y2Wre1rADaZyrKMBGE+trS6rZj5NZG6jSsqsPJSeN07cmwgyQzM2tkzc2F+yeAh4uoJZZzsnF+MN5QlwxZsT58I0bzYMPNrIk0YV0Sx0gIwtZ9bdIagekaRt7Pn3t3bcwePPBlDDu2Vt8lUlqVOeXEVm86JRMcOk5aJlQrAVIhXUZeD6p0OVYSCGJuaXizbOrlYPJ48bFLC0b5nsh9a6QKqgiRzMHpIa++TR3LNdRkXhciunKTPih4EE5FRlCbtavF4VJI0dpiwTKy0AKogyRwUvWZOZkYhu7df9CCYiIyieAvCPfxCqB1Exw6TlkljLYCsI+RVLD7K7u0XtQBEZBSlTou0TcEvhHqiNNYCyKIKjawrGJzINY+yFoBNPlVPMWxaCPWYx1kF4LopK3vRe5qySLvmUUSGp740VWm7Pl7owlkF4Lr/WVcwuKGBNtau6sJQgasibT8I1kT8QKUH18cLXTirAFw3ZXUFgxubmsbMhXmMTU2XluExhx+o9OD6eKELZxeBXY/97oPBeZLwC+t6cH280IWzFoDryC74NiEYXNXY4H5pt3owONDGyKGJ0ukokg8d8fp1yzBdP6a/XwdeARhC9nRv2WBwfg0gHVvcL2E6Bh8eLTXYFMmHjnj9cbIGzDLfL/KOjsHalvahEmddQK4ju7e/bDC4IjKaii3lMTjQxuFT5xajv4q6LorkQ0e8/jhZZw7KfL/IOzrOO9jSPlTiLQCL0HnRexPcPbKzvvgWSVMmf7vVg5Hb+0svWure6pklP1pmWQuvaTKyylwkXzoWfeu4hdYrAEOYCgZXRIarvs4sE71Mnkya/LoHG115i8otkwdV6apqsHa1r4R4F5AhbA4G52q4gCwTvUye6mjyh+jKW9PCi7jaV0K8BWAIm4PBueouypr15bkEkmZydTT5Q3TlrWnhRVw/X+AtAEPoDgb3xOgUDo6fWQzVLLvo7Hro57x94K7P5GzBtRm8LK6fL/AWgCF0BYMLZcxenM/1pYpYAHXcAhfF9ZlcEarwV5edwbvuS3eV2isAWxuWrmBwoYzulSuWDGiyi851HyBdcz2UwWYlbnPa6kztXUC2mvZVhIKIDmayLifXTd06U9Q9Z5N7Jp5mm9LWJGpvAdh6wbxIulSHggjT1rd5nfLFXlstrjpTdPaswspRVb/xNLtmgdWlnddeAZhqWHmdMildVYWCCD8zfGhCSIaKfHvUU+UkR1X9uu5SrEs7r70LCDCzg6WMSZvkppnozGLfgWPLXDVpO3iK7OwJ09S3eR2GD00UllEEb8rLI9peq3TPqapfm1yKZcaHeDm4uksu1wIgop8Q0VkiejHybAMRPUNErwQ/10f+tpeIThDRcSK6SVfC84iaaCa0dRnLI+207snOeaxd1VX67gBgqQUQpm1salpIRhFkLa66mNYy2Dy7jA7cdamnMuUdb+c211kWRSyAfwbwfQAPR57dA+BZZr6fiO4J/v8tItoO4FYAOwB8AMC/ENG1zPyO2mTnE138dWVWmjTzjgYGqyoYnOxpTqD8jMjWRfsqcaG91qmeVJS3C3WWRK4FwMy/BnAu9vhmAA8Fvz8E4EuR548x80VmPgXgBICdapIqRtTH6MoCU9LMu93qwb27t1caDE5FmOiyMyLXfcMqSGqvtllGdaonFeODK2NMnLKLwFcx82kACH6GewivBjAZ+dxU8GwZRDRERKNENNrpdEomIx0XK8SWYHAqQkEMDrSx+7pNmJm7JDRouVJvVQ/IJlwMqiJzeuxF9S4gSnjGSR9k5mFm7mfm/larlSvYthmQDkRm7yIDvSoZIrRbPVizugtPHT0jdXGIrVQ9IJuYcbvq11aBi22yDGUVwOtEtAkAgp9ng+dTAHojn9sM4LXyyXsP1aF+baRuweCKDlouDjRVD8gmZty2uHkmOrPY88gLuOuRP9Ta4jJBWQWwH8Adwe93APhF5PmtRLSSiK4BsBXAYbkkLpDVGOtSWWkDclLeRWfvKmSIUnTQqmKgUT1JqLMLJCwrAErzWLYORg5N4KmjZ3Dg6OlaW1wmKLIN9FEAvwWwjYimiOhOAPcD+DwRvQLg88H/wczjAB4HcAzAQQB7VO0Akgn16wpV3/xlSzjoKgZT2Xt2m0QVl8WIMDjQxt9etxG7r9tUqI+rUPaibdJVL0TuNlBmvi3lT59L+fx9AO6TSZQoNh0qkSEtNk/SljvRC2GKykg7eOY6svfs2oqOA0i6tjSWldtu9eAHX/t44c9XvUV1ojOLwYdHcbJzvrLvVEWtQ0G4ppXTLBmRUBCyMsKDZ1ta3UtkuFaWcWTv2bUVHbN1XRZZVW6zqj0CaX3GBWobCsJlrRwnaVbft3kdnjpyGn2b1wnJ2HfgGEY2XI52qyf14Nlbc/Pg2OatOhz8qYulGMXVA0g6qbqe0yLwylJFeInaWgAuamWRffnDhyYwc2Eew7HPZslYu6pr0QUCpB88S9q+6eo1kXWnzovRrqCrDqrY3FJbBRCagSO39zvTOUR25QwNtLF2VReGCrh6QhkzF+aXKMS0z4q4nDz247r7rqlU4cqqrQJwcWYksitnbGoaMxfmMTY1LSTj3t3bF8tEJKBcHS2ApgyMddkm3TSqGMNquwbgImn+XBWB3HTdQOYydVjXKEKZdQJXwxt7xLDWAmjK7CxKmsa34XxAHS0AGRNbtH2abM9lZpLeapAnWuey9a+r/VirAHwDfA8bgsHVcQ1AxsQWbZ+utWebD1e6MjmM1rls/etqP9a6gKo2W202eZPKQvQgWFkZvRsux0jk1rB4fdhcbjoRbZ+ubde0ectsla47mfadVOdl619X+7HWAqjabHVthlZVKIiwXJ4YnUqUYVuY4qoQbZ8ubEqwoVyL4ModyNE6j/5eppx1tR9rLYAyyGhJm2doVYaCiMsIy2Nm7lLirEvF7WGiNGXxtmpcKVcX70COIlPOqi1uKyyAi/PvKpl5yGhJm2doIj55FXv7k+4PvqW/V2vkUBFs9k+7jC/X5egYF2TKWbXFbYUCeGP2olPul6oR2ZWjYm+/ibsDRJA1p6O44vaoAtsmQXWtG5lyVq2krVAAV/Ss9DOPDJIqXVcwuDQZuu8OKEvajKjo4OHa2k+T8HWzHNVK2goFsLLrfVbNPFxAdOYtu7df1AKoavaWpvCKDh5Vuz3qOqvVgXdJ6ccKBeDJJmkwSxuQVeztT5IhagFUNXtLmxEVHTyqdnv4WW1xbHNJ1ZFa7QLSicn97ln7iU2HghBNR1XYuo/d5t1mnubhLYAYaSa6yZmbyEwoz1XTt3ldrgtCRSgI02sDtuJntcVx3V3mQvq9AoiRNtDb5o+MpjPa0PJcNcMFjqfLuovSZLhG2Q6suuO7MJDowHV3mQvp9wogRtpAb9vMLem07sihiVwLYCiSPxG/vsj5gjQZKgeyKgbFsh1Ydcd3YSDRQVWTLl1tybZJYxJ+DSCGrb7jOEmndQcH2pg893bmGsDY1PRi/sJb0+J+fZE1gLTnSTJUnjSt4tRqWX+9aj9/U9cNquqLutqSC2OJVwAlMLEgHP/O+D2keYO6bEA5nYvOZahiUCzbgVV3fBcGEpdpqoIFauYCqspXasIkj3+nzXcHiMoog20uOY+7NLkt1UoBVDUwm/DtFf1OXXv7dd4/0NRFTo/HNLVSAFUNzGVmDLKDXNHvVB0MLk+uChlNXeT0vIdM/2jSBEJ1XmulAGw25aoa5GwJBmd7QDmPXTTpLg8ZVOfVLwJXRFULTbpO9+Yt9u47cAwjGy5fVL4qbjGToak3lblKXe/yUI3qvEpZAET0JyI6SkRjRDQaPNtARM8Q0SvBz/UiMsuYOC6YgFVZJ2nfozMY3NpVXTjZOZ87KxGxAGTrtEmzQhtJqr+sOg13toUHG0Ww2fJXjY3RQD/LzH3M3B/8/x4AzzLzVgDPBv8vTJmO6zt7PjqDwc1cmMeWVveSWYmuReeiisGFQzh1Jqn+8vqp78fVo8MFdDOAG4LfHwLwHIBvFX25jInTJBOwLEX35U90ZvHa9Bw+uH61sLsoOivR5XIqemhHx95571YqjkidZr3j0YusBcAAniaiF4hoKHh2FTOfBoDgp5BTN8nEyZv1NckELEvRGfHIoQk8//IbePXNOeXB4HSdL6jz+Q/bef74WXzugefw/PGzuZ/NCxDo+3H1yCqATzLz9QC+AGAPEX266ItENEREo0Q02ul0Mj/rO548SZ0rqVwHB9q44doWejMsgCg23D9Q5/MfNpGkaPcdOLa4CSCKSJ16zCGlAJj5teDnWQA/B7ATwOtEtAkAgp+JUwNmHmbmfmbub7Vamd/jSsercjFaxXcldch2qweb1q3CZEELQFdAORXnC1RTpxlq2H6eP3628GJt0qA+NNDG2lVdGJK4gtRjjtIKgIi6iWhN+DuAGwG8CGA/gDuCj90B4BeyiXSl41Vpqaj4LpFBVuR8gQoZKs4XuLA7rAg68hG2n30HjhVerE2qk7GpacxcmMfY1PSSz6q4U0IG2TKrS9vJQ2YR+CoAPyeiUM5PmfkgEf0ewONEdCeAVwF8VT6ZbizAVbmIpeK70mRUHQzO9IIxYHf70pGPsPx27di4JKJs+KzKOtWBbITPKqLN2kBpC4CZJ5j53wT/djDzfcHz/8vMn2PmrcHPcyoS6sI6QJWWiorvsjmgnM40JM3ubIlTlPR9Ii7QovkI6753w+XLZNhQp7LIuo1dcTvL4kwoiKZUSBpVDkRVB4OrcsFYRMZEZxaDD48WnnioqKMHn34Zjx6exINPv7z4TETZi8ZbqjJIYJVrALITJFfczrJYpQDyTgo2oULSqNICSlK2ooOsrIw0hS8ysMjKCO9WiB9yU3FvdJoMBi/5KSpDV1mkyQjfj981LVKnHnNYpQBccPOYwnTn0eUWUCFD1xmFwYE2dl+3CR/euGbJ+yKzXlFlcUt/L7a0unFLf+8SGXc98gfseeSFXLdVNA1FLBJV7rfhWFr8lk83sEoBmB7kbKZKC6hKN0uaDF3uIhEZ7VYP1qzuwlNHzxi/R+HA0dOF0hF9P/rdut1vQ7G+a/K8hqc4VimAqga5pmzxKkuSIhadvSe5BlTs7Zc9X6BChq58ZMnoXb8aN1zbyq2T6PvRetRdFuFd02HfNXleo+6oHL+sUgBV4Wci2SQp4jTrLG9vf9Q1oGJvf13OKIjKmHxzDpvWrVoWbykrH9F6zEtHEUWtqyw8Yqgcvxp5H4APOiVOWnC1vH3gQwNtjE1NL4b6de3CepH97CruQDBVFsNB3ZgoizxsPqNhgmjZ3i8pq5EWQNN3FFVBkmtAdrE3LVKpiAwgedY7ee5tHD51DpPn3s6VoWvhe6IziweePo635uZLyyibjrgPv6gMmTQUdWV4i30pKsevRioAjzqq3NufFqm07EJr1D2VFtRMRz7SZIwcmsBTR8/gwNHTlS86RxV1URkTndnFcitTFkUHdr85RB+NcQF5M1IPusJJpL0/M3cJBFLiZtm1470wCH2b1+G+X760LKhZmXyElsXgQFvIRTI40Mbp6Qs42Zl1wuWUdk4iTUb8/aKuWB13O3gWcMYC8FcE2kmaORruoZ+Zu5RbZ0VdC+1WD+6+cRvWrO7KfT9JRtiGQjdP74bLF9OeFtSsjIujyMJ3EqKRWLPSoNNdFMpIOyeRJiP+vnfFmscZBSA7gNfBjHRp+2rSHnrToSCyImCqdLNE/eki+aiyLFTISDsnIZoPjzmcUQCyA3iVsw1dA7VrVky8znTvRS8yY71tZy+GBtpa9/YXWfg2XRaqZCQtqIvmwwVcmnyJ4IwCcMlc1DVQu2bFxOtMZGDRsRc9TM/Y1HRle/tF8lFWxhOjU4uDk6gFIFOe0WdFQkG4bAG4NvkqSmMWgatE1zkD1xfDqlww1ilDZKG1ijMKsxfnceDo6cXnJsoiet5DNB8u4HLas3DGAnAJl6yVKslaMC5q2ahwLajez25aRvfKFbkhH0TSUSYN8VAQSXXqcr9wOe1ZeAXgMU7RC+sBdcHgetevxunpC5k+6zQZVezLF5FxS39vbsgHXeWZprzrOmDWDa8APFaSNrA8MTq16PcOKbPQOvnmHJ57uZMbvljVImlRn7ysDJ2LznWb1Xu8AvBYStrAknRhSpmF1nh0Td2LpEVdPVkyoou9aTJ0Ljq7MNjXdbeOLvwisMcp7r5xG9auvmzZCdroz7znoQXwqa1XLPFZJ322ygXjPBnRxd7vfOWjlS86u0DopgLqfZm7KrwCiGAyXIQPVVGMpJ1QabujRCOYJpE0QKYNpkXfLyvjwxvXLDtXIZuOpLJwebeZy8pLhHC8oK73r5SR4xVABJOzBz9zqY6kAS6t/JMGlLTBNEmGqAWQJCOahujkIEmGSD7SysJl6pafNMJ6XtG9/ioZOVYogIvz7y4GzzI5+zU5e2jKzMVWRAZIEZeT6BkFkRl5WkC5ovnwuEtYv9994M3XZeQQM+d/SjOb/moHr/y77+G2nb2+kXpqRWiq79rxXuRRE5OcuIvRuxzdIq2+iOgFZu4vK9eKXUBX9Kx0KsSBx1OUcOZ9cPyM0VAC8T38dQ1tUFd01ZcVLqCVXe+Tmvn72Yx+ZMo4vOmKQPiPN17byDqqwsWXVUfx76/a5ej7qBy66ssKCyCk7B5eP5vRj2gZR+sy7aarJlHFHvqsOop/f9V7+n0flUNXfVlhAYSU3QnjF1D1I1rG0bocHEi+ycujFpv7gc1pM41J60jbIjAR7QLwXwCsAPBfmfn+tM/29/fz6OioNxNrRB3rso55yqOJea6avU8ewaOHJ0ttgpFdBNZiARDRCgA/APB5AFMAfk9E+5n5WNZ7fqua28QHi7rVZdyqacLA6M+n6MekdaRrDWAngBPMPMHMfwHwGICbNX2XxxJM+3l1x4GJBkMznVcZRMopHgCubBn7GD3pmIyxpMUFRER/B2AXM/+H4P9fB/BvmfmuyGeGAAwF//0IgBeVJ8QergDwhulEaOQKAG9Q1/tXruhef9U75998nef/crHqRHT9q6s++L7Va1rvzr3Vmf9/r7+qUPSy+jOdVxkSyqlw+yxbxhrrJo+6971tzLym7Mu6FoEp4dkSTcPMwwCGAYCIRmX8WLbj8+c2Pn/uUue8AQv5k3lflwtoCkBv5P+bAbym6bs8Ho/HUwJdCuD3ALYS0TVE9H4AtwLYr+m7PB6Px1MCLS4gZp4norsA/A8sbAP9CTOPZ7wyrCMdFuHz5zY+f+5S57wBkvmzIhicx+PxeKrHqlAQHo/H46kOrwA8Ho+noRhXAES0i4iOE9EJIrrHdHpUQER/IqKjRDQWbtMiog1E9AwRvRL8XG86nUUgop8Q0VkiejHyLDUvRLQ3qMvjRHSTmVQXJyV/3yaiPwf1N0ZEX4z8zbX89RLRr4joJSIaJ6JvBM9rUYcZ+XO+DoloFREdJqI/Bnn7x+C5urpjZmP/sLBAfBJAG8D7AfwRwHaTaVKUrz8BuCL27HsA7gl+vwfAd02ns2BePg3gegAv5uUFwPagDlcCuCao2xWm81Aif98G8J8SPuti/jYBuD74fQ2Al4N81KIOM/LnfB1i4TxVT/D7ZQB+B+ATKuvOtAXQpJARNwN4KPj9IQBfMpeU4jDzrwGciz1Oy8vNAB5j5ovMfArACSzUsbWk5C8NF/N3mpn/EPz+FoCXAFyNmtRhRv7ScCZ/vEAYO+Oy4B9DYd2ZVgBXA5iM/H8K2ZXnCgzgaSJ6IQh5AQBXMfNpYKHRArgy9W37SctLnerzLiI6EriIQhPb6fwR0YcAfAwLM8na1WEsf0AN6pCIVhDRGICzAJ5hZqV1Z1oB5IaMcJRPMvP1AL4AYA8Rfdp0giqiLvX5IwBbAPQBOA3ggeC5s/kjoh4APwPwTWaeyfpowjPr85iQv1rUITO/w8x9WIimsJOIPpLxceG8mVYAtQwZwcyvBT/PAvg5Fsyw14loEwAEP8+aS6E0aXmpRX0y8+tBx3sXwAjeM6OdzB8RXYaFwfERZn4yeFybOkzKX93qkJmnATwHYBcU1p1pBVC7kBFE1E1Ea8LfAdyIhUin+wHcEXzsDgC/MJNCJaTlZT+AW4loJRFdA2ArgMMG0idF2LkCvoz3ItU6lz8iIgA/BvASMz8Y+VMt6jAtf3WoQyJqEdG64PfVAP4GwP+GyrqzYKX7i1hYuT8J4B9Mp0dBftpYWIn/I4DxME8A/jWAZwG8EvzcYDqtBfPzKBZM6EtYmGHcmZUXAP8Q1OVxAF8wnf6S+ftvAI4COBJ0qk0O5+9TWHADHAEwFvz7Yl3qMCN/ztchgI8C+F9BHl4EcG/wXFnd+VAQHo/H01BMu4A8Ho/HYwivADwej6eheAXg8Xg8DcUrAI/H42koXgF4PB5PQ/EKwOPxeBqKVwAej8fTUP4/+lfTvIcERkIAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "utilities.plot_raster_dot(input_list[10][1])\n",
    "utilities.plot_raster_dot(output_list[10][1])"
   ]
  }
 ],
 "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": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}