Train_model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sys\n",
    "sys.path.append(\"../../Chapter_3/\")\n",
    "\n",
    "import pickle\n",
    "\n",
    "import torch\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "\n",
    "import fastai\n",
    "from fastai.basics import Learner, DataLoaders\n",
    "from fastai.vision.all import *\n",
    "from fastai.callback.all import *\n",
    "from fastai.torch_core import trainable_params\n",
    "\n",
    "# from Utils.model_datasets import RPIDataset\n",
    "from Utils.train import set_seed, MainModelDenseInputSizeCalculator, main_model_splitter\n",
    "from Utils.model_utils import LossLogger\n",
    "from Utils.model_datasets import EncodedRPIDataset\n",
    "from Utils.model_blocks import *\n",
    "from Utils.train import BlockInputShapeReformatter, MainModelDenseInputSizeCalculator\n",
    "\n",
    "import pandas as pd\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib.path import Path\n",
    "import matplotlib.patches as patches\n",
    "import seaborn as sbn\n",
    "import numpy as np\n",
    "\n",
    "from sklearn.metrics import accuracy_score, balanced_accuracy_score, f1_score, roc_auc_score, matthews_corrcoef\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "NNI_DIR = \"/data/Chapter_3/nni\"\n",
    "\n",
    "models = ('Dense','StackedRes','DenseSAE','Attention','Reccurent')\n",
    "\n",
    "model_eval = {model : pd.read_csv(f\"{NNI_DIR}/experiment_results/Main_{model}.csv\", index_col = 0).dropna().sort_values('final_metric')\n",
    "              for model in models}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def determine_block_name_from_params(params):\n",
    "    \"\"\"\n",
    "    Searches for a parameter exclusive to a different block type\n",
    "    within the given parameters, and returns the corresponding\n",
    "    block name. \n",
    "    \"\"\"\n",
    "    \n",
    "    param_names = params.keys()\n",
    "    if 'pooling_type' in param_names:\n",
    "        return 'ConvPool'\n",
    "    elif 'reccurent_type' in param_names:\n",
    "        return 'Reccurent'\n",
    "    elif 'layer_count' in param_names:\n",
    "        return 'Attention'\n",
    "    elif 'hidden_sizes' in param_names:\n",
    "        return 'StackedRes'\n",
    "    elif 'hidden_layer_sizes' in param_names:\n",
    "        return 'Dense'\n",
    "    elif 'sae_type' in param_names:\n",
    "        return 'SAE'\n",
    "    \n",
    "class BlockInializer:\n",
    "    \"\"\"\n",
    "    Used to initialize MainModel blocks from the parameters and \n",
    "    \"\"\"\n",
    "    \n",
    "    def __init__(self, seed):\n",
    "        self.seed = seed\n",
    "    \n",
    "    def _reformat_block_act(self, block_params):\n",
    "        preprocessed_block_str = re.sub(r\"'act': 'nn.(.+?)\\(\\)'\", r\"'act': nn.\\1()\", block_params)\n",
    "        preprocessed_block_str = re.sub(r\"'act_cls': 'nn.(.+?)\\'\", r\"'act_cls': nn.\\1\", preprocessed_block_str)\n",
    "        return preprocessed_block_str\n",
    "    \n",
    "    def _reformat_list_str(self, block_params):\n",
    "        preprocessed_block_str = re.sub(r\"'\\[(.*)\\]'\", r\"[\\1]\", block_params)\n",
    "        return preprocessed_block_str\n",
    "    \n",
    "    def generate_block_str(self, block_name, input_size, block_params):\n",
    "        return f\"{block_name}Block({input_size}, **{block_params})\"\n",
    "    \n",
    "    def _reformat_sae_block(self, input_size, preprocessed_block_str):\n",
    "        return preprocessed_block_str.replace(f\"{input_size}, \",\"\")\n",
    "    \n",
    "    def initialize_block(self, block_name, input_size, block_params):\n",
    "        block_str = self.generate_block_str(block_name, input_size, block_params)\n",
    "        preprocessed_block_str = self._reformat_block_act(block_str)\n",
    "        preprocessed_block_str = self._reformat_list_str(preprocessed_block_str)\n",
    "        \n",
    "        if block_name == 'SAE':\n",
    "            preprocessed_block_str = self._reformat_sae_block(input_size, preprocessed_block_str)\n",
    "        \n",
    "        set_seed(self.seed)\n",
    "        print(preprocessed_block_str)\n",
    "        return eval(preprocessed_block_str)\n",
    "\n",
    "def swap_seq_input_order(data : list) -> None:\n",
    "    seq1, seq2 = data[0], data[1]\n",
    "    data[0] = seq2\n",
    "    data[1] = seq1\n",
    "\n",
    "def initiate_lr_option(lr_option, lr_min, lr_steep):\n",
    "    \"having min first is intentional. It results in better performance\"\n",
    "    options = {'slice' : slice(lr_min, lr_steep), 'min' : lr_min}\n",
    "    return options[lr_option]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'batchsize': 32,\n",
       " 'dense_hidden_layers': '[3000, 2000, 1000, 10]',\n",
       " 'p': 0.2,\n",
       " 'act': 'None',\n",
       " 'bn': False,\n",
       " 'lin_first': False,\n",
       " 'lr_max_arg': 'slice',\n",
       " 'rna_hidden_sizes': '[70, 40, 10]',\n",
       " 'rna_act_cls': 'nn.Sigmoid',\n",
       " 'rna_bn_1st': False,\n",
       " 'prot_hidden_sizes': '[55, 10]',\n",
       " 'prot_act_cls': 'nn.ReLU',\n",
       " 'prot_bn_1st': True,\n",
       " 'final_metric': 0.0482180081307888}"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_eval['StackedRes'].iloc[0,:].to_dict()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'StackedRes'"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "SEED = 255\n",
    "\n",
    "PARAMS = model_eval['StackedRes'].iloc[0,:].to_dict()\n",
    "PARAMS\n",
    "\n",
    "_ = PARAMS.pop(\"final_metric\")\n",
    "LR_OPTION = PARAMS.pop(\"lr_max_arg\")\n",
    "BATCHSIZE = PARAMS.pop(\"batchsize\")\n",
    "MAIN_ACT = eval(PARAMS.pop(\"act\"))\n",
    "\n",
    "main_model_params = {key : value for key,value in PARAMS.items() if 'rna' not in key and 'prot' not in key}\n",
    "rna_block_params = {key.replace(\"rna_\",\"\") : value for key,value in PARAMS.items() if 'rna' in key}\n",
    "prot_block_params = {key.replace(\"prot_\",\"\") : value for key,value in PARAMS.items() if 'prot' in key}\n",
    "\n",
    "rna_block_name = determine_block_name_from_params(rna_block_params)\n",
    "prot_block_name = determine_block_name_from_params(prot_block_params)\n",
    "\n",
    "# TEMPORARY: ASSUMES THAT PROT AND RNA BLOCKS ARE IDENTICAL\n",
    "block_name = prot_block_name\n",
    "block_name"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total training data: 24294\n",
      "Total positive interactions: 12147.0\n",
      "Total negative interactions: 12147.0\n"
     ]
    }
   ],
   "source": [
    "#####################################################\n",
    "# Load params\n",
    "######################################################\n",
    "\n",
    "with open(\"../Data/TrainingData/FullEmbeddedTrainingData.pickle\",'rb') as handle:\n",
    "    data = pickle.load(handle)\n",
    "    print(f\"Total training data: {len(data[0])}\")\n",
    "    print(f\"Total positive interactions: {sum(data[-1])}\")\n",
    "    print(f\"Total negative interactions: {len(data[-1]) - sum(data[-1])}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tqdm.notebook import tqdm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4f5d9f6382b64cecbfe880fe0908458f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/5 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DenseBlock(1024, **{'hidden_layer_sizes': [840, 680, 520, 360, 200], 'p': 0.8, 'act': nn.Sigmoid(), 'bn': False, 'lin_first': False})\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [600, 200], 'p': 0.2, 'act': nn.ReLU(), 'bn': False, 'lin_first': True})\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [840, 680, 520, 360, 200], 'p': 0.8, 'act': nn.Sigmoid(), 'bn': False, 'lin_first': False})\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [600, 200], 'p': 0.2, 'act': nn.ReLU(), 'bn': False, 'lin_first': True})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.406051</td>\n",
       "      <td>0.222645</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.312130</td>\n",
       "      <td>0.164714</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.287990</td>\n",
       "      <td>0.140308</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.276178</td>\n",
       "      <td>0.128403</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.255821</td>\n",
       "      <td>0.116298</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.252889</td>\n",
       "      <td>0.111330</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.243031</td>\n",
       "      <td>0.102817</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.232752</td>\n",
       "      <td>0.098659</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.234854</td>\n",
       "      <td>0.093567</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.237825</td>\n",
       "      <td>0.090115</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.223106</td>\n",
       "      <td>0.087874</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.231914</td>\n",
       "      <td>0.085436</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.212779</td>\n",
       "      <td>0.079385</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.219061</td>\n",
       "      <td>0.081142</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.214488</td>\n",
       "      <td>0.081301</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.210078</td>\n",
       "      <td>0.076644</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.206977</td>\n",
       "      <td>0.078792</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.204005</td>\n",
       "      <td>0.072983</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.221692</td>\n",
       "      <td>0.074299</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.214572</td>\n",
       "      <td>0.071770</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.199621</td>\n",
       "      <td>0.072830</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.203078</td>\n",
       "      <td>0.074410</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.207634</td>\n",
       "      <td>0.072964</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.197485</td>\n",
       "      <td>0.067457</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.200716</td>\n",
       "      <td>0.065328</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.192429</td>\n",
       "      <td>0.064955</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.196755</td>\n",
       "      <td>0.063738</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.195553</td>\n",
       "      <td>0.064047</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.191849</td>\n",
       "      <td>0.063473</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.193304</td>\n",
       "      <td>0.061714</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.200106</td>\n",
       "      <td>0.062963</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.193908</td>\n",
       "      <td>0.063924</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.196313</td>\n",
       "      <td>0.063119</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.181259</td>\n",
       "      <td>0.060961</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.190408</td>\n",
       "      <td>0.057304</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.188102</td>\n",
       "      <td>0.059260</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.187590</td>\n",
       "      <td>0.057273</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.181630</td>\n",
       "      <td>0.055988</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.179201</td>\n",
       "      <td>0.057087</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.185366</td>\n",
       "      <td>0.057456</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.182054</td>\n",
       "      <td>0.053321</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.179320</td>\n",
       "      <td>0.054169</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.177739</td>\n",
       "      <td>0.060175</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.187395</td>\n",
       "      <td>0.059498</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.179224</td>\n",
       "      <td>0.053905</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.190198</td>\n",
       "      <td>0.052192</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>46</td>\n",
       "      <td>0.177125</td>\n",
       "      <td>0.054054</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>47</td>\n",
       "      <td>0.184305</td>\n",
       "      <td>0.053684</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>48</td>\n",
       "      <td>0.178106</td>\n",
       "      <td>0.050292</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>49</td>\n",
       "      <td>0.174757</td>\n",
       "      <td>0.051104</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>50</td>\n",
       "      <td>0.178510</td>\n",
       "      <td>0.050505</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>51</td>\n",
       "      <td>0.174030</td>\n",
       "      <td>0.052699</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>52</td>\n",
       "      <td>0.175094</td>\n",
       "      <td>0.054233</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>53</td>\n",
       "      <td>0.174012</td>\n",
       "      <td>0.052298</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>54</td>\n",
       "      <td>0.171772</td>\n",
       "      <td>0.055670</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>55</td>\n",
       "      <td>0.179324</td>\n",
       "      <td>0.051130</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>56</td>\n",
       "      <td>0.171667</td>\n",
       "      <td>0.051415</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>57</td>\n",
       "      <td>0.178889</td>\n",
       "      <td>0.058252</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>58</td>\n",
       "      <td>0.177635</td>\n",
       "      <td>0.049190</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>59</td>\n",
       "      <td>0.174761</td>\n",
       "      <td>0.054884</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>60</td>\n",
       "      <td>0.171352</td>\n",
       "      <td>0.053145</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>61</td>\n",
       "      <td>0.177087</td>\n",
       "      <td>0.050299</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>62</td>\n",
       "      <td>0.168053</td>\n",
       "      <td>0.057543</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>63</td>\n",
       "      <td>0.177309</td>\n",
       "      <td>0.052632</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>64</td>\n",
       "      <td>0.167107</td>\n",
       "      <td>0.053520</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>65</td>\n",
       "      <td>0.168778</td>\n",
       "      <td>0.051011</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>66</td>\n",
       "      <td>0.176846</td>\n",
       "      <td>0.051993</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>67</td>\n",
       "      <td>0.171928</td>\n",
       "      <td>0.056492</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>68</td>\n",
       "      <td>0.166138</td>\n",
       "      <td>0.053883</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 58: early stopping\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [840, 680, 520, 360, 200], 'p': 0.8, 'act': nn.Sigmoid(), 'bn': False, 'lin_first': False})\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [600, 200], 'p': 0.2, 'act': nn.ReLU(), 'bn': False, 'lin_first': True})\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [840, 680, 520, 360, 200], 'p': 0.8, 'act': nn.Sigmoid(), 'bn': False, 'lin_first': False})\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [600, 200], 'p': 0.2, 'act': nn.ReLU(), 'bn': False, 'lin_first': True})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.409303</td>\n",
       "      <td>0.243432</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.309674</td>\n",
       "      <td>0.179973</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.281425</td>\n",
       "      <td>0.154879</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.265318</td>\n",
       "      <td>0.143242</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.247379</td>\n",
       "      <td>0.129397</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.241300</td>\n",
       "      <td>0.123463</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.235132</td>\n",
       "      <td>0.113876</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.227006</td>\n",
       "      <td>0.110389</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.236065</td>\n",
       "      <td>0.107135</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.226747</td>\n",
       "      <td>0.099871</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.217747</td>\n",
       "      <td>0.098672</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.227035</td>\n",
       "      <td>0.094438</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.212536</td>\n",
       "      <td>0.090145</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.207474</td>\n",
       "      <td>0.091666</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.220846</td>\n",
       "      <td>0.091205</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.208105</td>\n",
       "      <td>0.085256</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.204857</td>\n",
       "      <td>0.089489</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.202659</td>\n",
       "      <td>0.080854</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.203179</td>\n",
       "      <td>0.083477</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.196556</td>\n",
       "      <td>0.081513</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.191982</td>\n",
       "      <td>0.078122</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.189991</td>\n",
       "      <td>0.079021</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.189176</td>\n",
       "      <td>0.074685</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.199827</td>\n",
       "      <td>0.076795</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.200616</td>\n",
       "      <td>0.071703</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.189096</td>\n",
       "      <td>0.075925</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.189215</td>\n",
       "      <td>0.071053</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.186152</td>\n",
       "      <td>0.070849</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.188715</td>\n",
       "      <td>0.071342</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.191029</td>\n",
       "      <td>0.072125</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.186768</td>\n",
       "      <td>0.069259</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.187679</td>\n",
       "      <td>0.069357</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.185604</td>\n",
       "      <td>0.067987</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.181516</td>\n",
       "      <td>0.065292</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.184673</td>\n",
       "      <td>0.068090</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.185095</td>\n",
       "      <td>0.066507</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.195881</td>\n",
       "      <td>0.063363</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.180796</td>\n",
       "      <td>0.067000</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.192018</td>\n",
       "      <td>0.065478</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.175880</td>\n",
       "      <td>0.069206</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.177351</td>\n",
       "      <td>0.066314</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.178324</td>\n",
       "      <td>0.068862</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.178781</td>\n",
       "      <td>0.064344</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.180230</td>\n",
       "      <td>0.070601</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.182856</td>\n",
       "      <td>0.065965</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.182302</td>\n",
       "      <td>0.064344</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>46</td>\n",
       "      <td>0.182144</td>\n",
       "      <td>0.062915</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>47</td>\n",
       "      <td>0.172468</td>\n",
       "      <td>0.071168</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>48</td>\n",
       "      <td>0.173723</td>\n",
       "      <td>0.061407</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>49</td>\n",
       "      <td>0.174257</td>\n",
       "      <td>0.063171</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>50</td>\n",
       "      <td>0.177139</td>\n",
       "      <td>0.062953</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>51</td>\n",
       "      <td>0.167508</td>\n",
       "      <td>0.064808</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>52</td>\n",
       "      <td>0.173455</td>\n",
       "      <td>0.077571</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>53</td>\n",
       "      <td>0.179791</td>\n",
       "      <td>0.066873</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>54</td>\n",
       "      <td>0.173596</td>\n",
       "      <td>0.064840</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>55</td>\n",
       "      <td>0.178029</td>\n",
       "      <td>0.071420</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>56</td>\n",
       "      <td>0.171634</td>\n",
       "      <td>0.068347</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>57</td>\n",
       "      <td>0.174461</td>\n",
       "      <td>0.068272</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>58</td>\n",
       "      <td>0.175477</td>\n",
       "      <td>0.070195</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 48: early stopping\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [840, 680, 520, 360, 200], 'p': 0.8, 'act': nn.Sigmoid(), 'bn': False, 'lin_first': False})\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [600, 200], 'p': 0.2, 'act': nn.ReLU(), 'bn': False, 'lin_first': True})\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [840, 680, 520, 360, 200], 'p': 0.8, 'act': nn.Sigmoid(), 'bn': False, 'lin_first': False})\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [600, 200], 'p': 0.2, 'act': nn.ReLU(), 'bn': False, 'lin_first': True})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.447924</td>\n",
       "      <td>0.288354</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.319493</td>\n",
       "      <td>0.202795</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.291767</td>\n",
       "      <td>0.174863</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.269474</td>\n",
       "      <td>0.157224</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.252847</td>\n",
       "      <td>0.141362</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.250511</td>\n",
       "      <td>0.130995</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.243915</td>\n",
       "      <td>0.123961</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.236388</td>\n",
       "      <td>0.113881</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.231694</td>\n",
       "      <td>0.112867</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.230255</td>\n",
       "      <td>0.105937</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.219362</td>\n",
       "      <td>0.102689</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.224318</td>\n",
       "      <td>0.100981</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.217377</td>\n",
       "      <td>0.098204</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.220332</td>\n",
       "      <td>0.091588</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.213760</td>\n",
       "      <td>0.092589</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.207208</td>\n",
       "      <td>0.085899</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.205463</td>\n",
       "      <td>0.088882</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.210176</td>\n",
       "      <td>0.084424</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.215849</td>\n",
       "      <td>0.085823</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.201816</td>\n",
       "      <td>0.086572</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.191428</td>\n",
       "      <td>0.080888</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.197666</td>\n",
       "      <td>0.078593</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.200553</td>\n",
       "      <td>0.078280</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.200897</td>\n",
       "      <td>0.079766</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.197020</td>\n",
       "      <td>0.073180</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.192603</td>\n",
       "      <td>0.078342</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.197832</td>\n",
       "      <td>0.074519</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.186349</td>\n",
       "      <td>0.073136</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.193958</td>\n",
       "      <td>0.075294</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.189293</td>\n",
       "      <td>0.077014</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.195956</td>\n",
       "      <td>0.074080</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.181777</td>\n",
       "      <td>0.072160</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.192317</td>\n",
       "      <td>0.075321</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.195991</td>\n",
       "      <td>0.067330</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.189236</td>\n",
       "      <td>0.072967</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.185015</td>\n",
       "      <td>0.067517</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.185973</td>\n",
       "      <td>0.068547</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.183314</td>\n",
       "      <td>0.066490</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.186909</td>\n",
       "      <td>0.071617</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.180909</td>\n",
       "      <td>0.066597</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.173597</td>\n",
       "      <td>0.074135</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.180571</td>\n",
       "      <td>0.071557</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.181609</td>\n",
       "      <td>0.067587</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.175384</td>\n",
       "      <td>0.069663</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.176083</td>\n",
       "      <td>0.070670</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.190920</td>\n",
       "      <td>0.069644</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>46</td>\n",
       "      <td>0.171973</td>\n",
       "      <td>0.074328</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>47</td>\n",
       "      <td>0.180767</td>\n",
       "      <td>0.068648</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 37: early stopping\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [840, 680, 520, 360, 200], 'p': 0.8, 'act': nn.Sigmoid(), 'bn': False, 'lin_first': False})\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [600, 200], 'p': 0.2, 'act': nn.ReLU(), 'bn': False, 'lin_first': True})\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [840, 680, 520, 360, 200], 'p': 0.8, 'act': nn.Sigmoid(), 'bn': False, 'lin_first': False})\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [600, 200], 'p': 0.2, 'act': nn.ReLU(), 'bn': False, 'lin_first': True})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.447750</td>\n",
       "      <td>0.270469</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.314872</td>\n",
       "      <td>0.181194</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.287777</td>\n",
       "      <td>0.149473</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.276555</td>\n",
       "      <td>0.135006</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.249321</td>\n",
       "      <td>0.116500</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.250488</td>\n",
       "      <td>0.107299</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.247413</td>\n",
       "      <td>0.099983</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.241191</td>\n",
       "      <td>0.095534</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.236598</td>\n",
       "      <td>0.088214</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.233648</td>\n",
       "      <td>0.083995</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.225368</td>\n",
       "      <td>0.080580</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.228628</td>\n",
       "      <td>0.076904</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.222897</td>\n",
       "      <td>0.071988</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.216830</td>\n",
       "      <td>0.070144</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.223414</td>\n",
       "      <td>0.068679</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.216201</td>\n",
       "      <td>0.064854</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.206396</td>\n",
       "      <td>0.065301</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.211237</td>\n",
       "      <td>0.060374</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.207265</td>\n",
       "      <td>0.060897</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.205415</td>\n",
       "      <td>0.056537</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.196665</td>\n",
       "      <td>0.055818</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.199848</td>\n",
       "      <td>0.055644</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.196136</td>\n",
       "      <td>0.055367</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.204686</td>\n",
       "      <td>0.055696</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.200124</td>\n",
       "      <td>0.052017</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.193886</td>\n",
       "      <td>0.052198</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.198445</td>\n",
       "      <td>0.050200</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.190723</td>\n",
       "      <td>0.048582</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.192805</td>\n",
       "      <td>0.048040</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.190650</td>\n",
       "      <td>0.049105</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.192990</td>\n",
       "      <td>0.047254</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.191900</td>\n",
       "      <td>0.043887</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.199858</td>\n",
       "      <td>0.046553</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.178389</td>\n",
       "      <td>0.040389</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.198856</td>\n",
       "      <td>0.046347</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.186618</td>\n",
       "      <td>0.043803</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.185961</td>\n",
       "      <td>0.042336</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.182736</td>\n",
       "      <td>0.042311</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.186553</td>\n",
       "      <td>0.040952</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.183757</td>\n",
       "      <td>0.043520</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.178047</td>\n",
       "      <td>0.041141</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.183325</td>\n",
       "      <td>0.040398</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.176410</td>\n",
       "      <td>0.040664</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.180373</td>\n",
       "      <td>0.042935</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 33: early stopping\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [840, 680, 520, 360, 200], 'p': 0.8, 'act': nn.Sigmoid(), 'bn': False, 'lin_first': False})\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [600, 200], 'p': 0.2, 'act': nn.ReLU(), 'bn': False, 'lin_first': True})\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [840, 680, 520, 360, 200], 'p': 0.8, 'act': nn.Sigmoid(), 'bn': False, 'lin_first': False})\n",
      "DenseBlock(1024, **{'hidden_layer_sizes': [600, 200], 'p': 0.2, 'act': nn.ReLU(), 'bn': False, 'lin_first': True})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.418819</td>\n",
       "      <td>0.256459</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.317042</td>\n",
       "      <td>0.189797</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.288619</td>\n",
       "      <td>0.163689</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.271018</td>\n",
       "      <td>0.148757</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.253303</td>\n",
       "      <td>0.131935</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.252742</td>\n",
       "      <td>0.124247</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.245838</td>\n",
       "      <td>0.117276</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.245682</td>\n",
       "      <td>0.112643</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.237380</td>\n",
       "      <td>0.104234</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.231595</td>\n",
       "      <td>0.102895</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.222701</td>\n",
       "      <td>0.097569</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.232123</td>\n",
       "      <td>0.094023</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.217612</td>\n",
       "      <td>0.091843</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.227500</td>\n",
       "      <td>0.087463</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.220624</td>\n",
       "      <td>0.087919</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.210763</td>\n",
       "      <td>0.086078</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.205785</td>\n",
       "      <td>0.077681</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.209040</td>\n",
       "      <td>0.078433</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.214936</td>\n",
       "      <td>0.078079</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.196743</td>\n",
       "      <td>0.071626</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.195009</td>\n",
       "      <td>0.073674</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.201069</td>\n",
       "      <td>0.072404</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.196083</td>\n",
       "      <td>0.073715</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.196890</td>\n",
       "      <td>0.067072</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.213641</td>\n",
       "      <td>0.069983</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.196037</td>\n",
       "      <td>0.068072</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.191151</td>\n",
       "      <td>0.067419</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.187820</td>\n",
       "      <td>0.062307</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.190127</td>\n",
       "      <td>0.065334</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.187589</td>\n",
       "      <td>0.063171</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.187738</td>\n",
       "      <td>0.064715</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.186587</td>\n",
       "      <td>0.064427</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.193038</td>\n",
       "      <td>0.060335</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.182454</td>\n",
       "      <td>0.064403</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.183576</td>\n",
       "      <td>0.061144</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.191048</td>\n",
       "      <td>0.061859</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.186419</td>\n",
       "      <td>0.058213</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.182411</td>\n",
       "      <td>0.058865</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.186555</td>\n",
       "      <td>0.055938</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.182015</td>\n",
       "      <td>0.062703</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.182850</td>\n",
       "      <td>0.059452</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.179796</td>\n",
       "      <td>0.058055</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.176317</td>\n",
       "      <td>0.054063</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.180803</td>\n",
       "      <td>0.053403</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.181285</td>\n",
       "      <td>0.054047</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.183992</td>\n",
       "      <td>0.049130</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>46</td>\n",
       "      <td>0.174746</td>\n",
       "      <td>0.050515</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>47</td>\n",
       "      <td>0.177257</td>\n",
       "      <td>0.055337</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>48</td>\n",
       "      <td>0.172662</td>\n",
       "      <td>0.049091</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>49</td>\n",
       "      <td>0.185943</td>\n",
       "      <td>0.053312</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>50</td>\n",
       "      <td>0.173926</td>\n",
       "      <td>0.057890</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>51</td>\n",
       "      <td>0.168417</td>\n",
       "      <td>0.050279</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>52</td>\n",
       "      <td>0.175717</td>\n",
       "      <td>0.052609</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>53</td>\n",
       "      <td>0.177564</td>\n",
       "      <td>0.052862</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>54</td>\n",
       "      <td>0.174828</td>\n",
       "      <td>0.055554</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>55</td>\n",
       "      <td>0.176536</td>\n",
       "      <td>0.053057</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>56</td>\n",
       "      <td>0.164998</td>\n",
       "      <td>0.056374</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>57</td>\n",
       "      <td>0.176417</td>\n",
       "      <td>0.052095</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>58</td>\n",
       "      <td>0.168665</td>\n",
       "      <td>0.058808</td>\n",
       "      <td>00:03</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 48: early stopping\n",
      "StackedResBlock(1024, **{'hidden_sizes': [70, 40, 10], 'act_cls': nn.Sigmoid, 'bn_1st': False})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [55, 10], 'act_cls': nn.ReLU, 'bn_1st': True})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [70, 40, 10], 'act_cls': nn.Sigmoid, 'bn_1st': False})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [55, 10], 'act_cls': nn.ReLU, 'bn_1st': True})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.288335</td>\n",
       "      <td>0.228136</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.238510</td>\n",
       "      <td>0.174065</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.200082</td>\n",
       "      <td>0.136850</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.172269</td>\n",
       "      <td>0.127613</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.150293</td>\n",
       "      <td>0.129133</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.138957</td>\n",
       "      <td>0.122051</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.122444</td>\n",
       "      <td>0.107041</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.113492</td>\n",
       "      <td>0.106169</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.110688</td>\n",
       "      <td>0.103089</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.109668</td>\n",
       "      <td>0.101200</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.100360</td>\n",
       "      <td>0.094326</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.132260</td>\n",
       "      <td>0.099685</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.082241</td>\n",
       "      <td>0.094406</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.091674</td>\n",
       "      <td>0.082796</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.086509</td>\n",
       "      <td>0.082839</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.079508</td>\n",
       "      <td>0.081153</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.075884</td>\n",
       "      <td>0.079899</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.067461</td>\n",
       "      <td>0.089940</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.075199</td>\n",
       "      <td>0.078676</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.086784</td>\n",
       "      <td>0.083221</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.072073</td>\n",
       "      <td>0.079693</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.110439</td>\n",
       "      <td>0.070958</td>\n",
       "      <td>00:11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.063923</td>\n",
       "      <td>0.067114</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.061827</td>\n",
       "      <td>0.083435</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.065963</td>\n",
       "      <td>0.065455</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.063148</td>\n",
       "      <td>0.083298</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.059882</td>\n",
       "      <td>0.069273</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.063856</td>\n",
       "      <td>0.072591</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.069660</td>\n",
       "      <td>0.066598</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.063915</td>\n",
       "      <td>0.066215</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.074391</td>\n",
       "      <td>0.068785</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.070817</td>\n",
       "      <td>0.061327</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.047580</td>\n",
       "      <td>0.063234</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.063258</td>\n",
       "      <td>0.066551</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.069497</td>\n",
       "      <td>0.065654</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.046728</td>\n",
       "      <td>0.059780</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.060724</td>\n",
       "      <td>0.062364</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.048437</td>\n",
       "      <td>0.069715</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.044490</td>\n",
       "      <td>0.068074</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.052301</td>\n",
       "      <td>0.064572</td>\n",
       "      <td>00:11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.049149</td>\n",
       "      <td>0.055413</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.043242</td>\n",
       "      <td>0.061983</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.048896</td>\n",
       "      <td>0.054664</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.045181</td>\n",
       "      <td>0.067228</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.043871</td>\n",
       "      <td>0.061960</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.047812</td>\n",
       "      <td>0.063233</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>46</td>\n",
       "      <td>0.049591</td>\n",
       "      <td>0.066228</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>47</td>\n",
       "      <td>0.043493</td>\n",
       "      <td>0.064128</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>48</td>\n",
       "      <td>0.052758</td>\n",
       "      <td>0.063868</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>49</td>\n",
       "      <td>0.045892</td>\n",
       "      <td>0.059111</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>50</td>\n",
       "      <td>0.043599</td>\n",
       "      <td>0.056694</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>51</td>\n",
       "      <td>0.098821</td>\n",
       "      <td>0.059795</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>52</td>\n",
       "      <td>0.047002</td>\n",
       "      <td>0.063306</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 42: early stopping\n",
      "StackedResBlock(1024, **{'hidden_sizes': [70, 40, 10], 'act_cls': nn.Sigmoid, 'bn_1st': False})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [55, 10], 'act_cls': nn.ReLU, 'bn_1st': True})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [70, 40, 10], 'act_cls': nn.Sigmoid, 'bn_1st': False})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [55, 10], 'act_cls': nn.ReLU, 'bn_1st': True})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.288752</td>\n",
       "      <td>0.256233</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.223898</td>\n",
       "      <td>0.193489</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.198566</td>\n",
       "      <td>0.164352</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.161777</td>\n",
       "      <td>0.158365</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.147728</td>\n",
       "      <td>0.142887</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.139026</td>\n",
       "      <td>0.143048</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.133233</td>\n",
       "      <td>0.117530</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.122960</td>\n",
       "      <td>0.117590</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.137823</td>\n",
       "      <td>0.116460</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.116009</td>\n",
       "      <td>0.114026</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.108502</td>\n",
       "      <td>0.100958</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.118410</td>\n",
       "      <td>0.112372</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.085431</td>\n",
       "      <td>0.100463</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.080041</td>\n",
       "      <td>0.102382</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.090720</td>\n",
       "      <td>0.093790</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.085624</td>\n",
       "      <td>0.094184</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.073126</td>\n",
       "      <td>0.093310</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.087970</td>\n",
       "      <td>0.094584</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.067477</td>\n",
       "      <td>0.081463</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.068056</td>\n",
       "      <td>0.088510</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.065670</td>\n",
       "      <td>0.097329</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.073422</td>\n",
       "      <td>0.079910</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.067912</td>\n",
       "      <td>0.079800</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.064783</td>\n",
       "      <td>0.089911</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.058940</td>\n",
       "      <td>0.077470</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.056156</td>\n",
       "      <td>0.097953</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.067934</td>\n",
       "      <td>0.080676</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.065738</td>\n",
       "      <td>0.088860</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.057860</td>\n",
       "      <td>0.082781</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.060202</td>\n",
       "      <td>0.082342</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.059290</td>\n",
       "      <td>0.082202</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.074045</td>\n",
       "      <td>0.083162</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.059446</td>\n",
       "      <td>0.082002</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.065776</td>\n",
       "      <td>0.078448</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.051192</td>\n",
       "      <td>0.091478</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 24: early stopping\n",
      "StackedResBlock(1024, **{'hidden_sizes': [70, 40, 10], 'act_cls': nn.Sigmoid, 'bn_1st': False})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [55, 10], 'act_cls': nn.ReLU, 'bn_1st': True})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [70, 40, 10], 'act_cls': nn.Sigmoid, 'bn_1st': False})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [55, 10], 'act_cls': nn.ReLU, 'bn_1st': True})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.318260</td>\n",
       "      <td>0.318073</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.248517</td>\n",
       "      <td>0.231344</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.206271</td>\n",
       "      <td>0.195101</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.189572</td>\n",
       "      <td>0.188517</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.158900</td>\n",
       "      <td>0.179161</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.169747</td>\n",
       "      <td>0.169656</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.160723</td>\n",
       "      <td>0.142989</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.153712</td>\n",
       "      <td>0.139089</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.122964</td>\n",
       "      <td>0.140915</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.118091</td>\n",
       "      <td>0.136729</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.105174</td>\n",
       "      <td>0.123842</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.129587</td>\n",
       "      <td>0.136091</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.108847</td>\n",
       "      <td>0.124171</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.106966</td>\n",
       "      <td>0.109176</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.108791</td>\n",
       "      <td>0.110172</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.099389</td>\n",
       "      <td>0.108475</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.096329</td>\n",
       "      <td>0.110134</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.093117</td>\n",
       "      <td>0.119233</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.095744</td>\n",
       "      <td>0.098671</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.101637</td>\n",
       "      <td>0.111966</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.077562</td>\n",
       "      <td>0.108898</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.079818</td>\n",
       "      <td>0.090631</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.082472</td>\n",
       "      <td>0.095578</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.075961</td>\n",
       "      <td>0.113754</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.083421</td>\n",
       "      <td>0.087269</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.064937</td>\n",
       "      <td>0.115311</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.086016</td>\n",
       "      <td>0.092525</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.081277</td>\n",
       "      <td>0.105247</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.068124</td>\n",
       "      <td>0.099168</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.074649</td>\n",
       "      <td>0.090170</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.071421</td>\n",
       "      <td>0.097511</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.071051</td>\n",
       "      <td>0.089856</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.054157</td>\n",
       "      <td>0.094205</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.072962</td>\n",
       "      <td>0.089633</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.067637</td>\n",
       "      <td>0.100417</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 24: early stopping\n",
      "StackedResBlock(1024, **{'hidden_sizes': [70, 40, 10], 'act_cls': nn.Sigmoid, 'bn_1st': False})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [55, 10], 'act_cls': nn.ReLU, 'bn_1st': True})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [70, 40, 10], 'act_cls': nn.Sigmoid, 'bn_1st': False})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [55, 10], 'act_cls': nn.ReLU, 'bn_1st': True})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.374126</td>\n",
       "      <td>0.335641</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.290812</td>\n",
       "      <td>0.250828</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.264581</td>\n",
       "      <td>0.199947</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.228100</td>\n",
       "      <td>0.185948</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.213976</td>\n",
       "      <td>0.159424</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.207459</td>\n",
       "      <td>0.159123</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.190196</td>\n",
       "      <td>0.134225</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.169641</td>\n",
       "      <td>0.128002</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.162134</td>\n",
       "      <td>0.118540</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.169948</td>\n",
       "      <td>0.112905</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.144531</td>\n",
       "      <td>0.102146</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.148792</td>\n",
       "      <td>0.105593</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.121739</td>\n",
       "      <td>0.092678</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.127829</td>\n",
       "      <td>0.082636</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.134040</td>\n",
       "      <td>0.087881</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.135080</td>\n",
       "      <td>0.079126</td>\n",
       "      <td>00:11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.107248</td>\n",
       "      <td>0.079237</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.123097</td>\n",
       "      <td>0.084601</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.110078</td>\n",
       "      <td>0.071055</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.114382</td>\n",
       "      <td>0.075186</td>\n",
       "      <td>00:11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.099134</td>\n",
       "      <td>0.074645</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.106007</td>\n",
       "      <td>0.073907</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.095793</td>\n",
       "      <td>0.063334</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.104328</td>\n",
       "      <td>0.075137</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.106412</td>\n",
       "      <td>0.055344</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.082673</td>\n",
       "      <td>0.078616</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.086457</td>\n",
       "      <td>0.064647</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.104368</td>\n",
       "      <td>0.067916</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.084863</td>\n",
       "      <td>0.057476</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.083385</td>\n",
       "      <td>0.058555</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.091184</td>\n",
       "      <td>0.063327</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.097232</td>\n",
       "      <td>0.052124</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.071901</td>\n",
       "      <td>0.054062</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.072058</td>\n",
       "      <td>0.052416</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.078921</td>\n",
       "      <td>0.056215</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.064133</td>\n",
       "      <td>0.051104</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.077255</td>\n",
       "      <td>0.050935</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.072639</td>\n",
       "      <td>0.051196</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.075187</td>\n",
       "      <td>0.052120</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.062847</td>\n",
       "      <td>0.060341</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.063512</td>\n",
       "      <td>0.048415</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.055075</td>\n",
       "      <td>0.057241</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.059045</td>\n",
       "      <td>0.044097</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.067571</td>\n",
       "      <td>0.047131</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.064164</td>\n",
       "      <td>0.043285</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.056492</td>\n",
       "      <td>0.046530</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>46</td>\n",
       "      <td>0.060964</td>\n",
       "      <td>0.044880</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>47</td>\n",
       "      <td>0.055004</td>\n",
       "      <td>0.049480</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>48</td>\n",
       "      <td>0.052883</td>\n",
       "      <td>0.056075</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>49</td>\n",
       "      <td>0.048138</td>\n",
       "      <td>0.052834</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>50</td>\n",
       "      <td>0.046484</td>\n",
       "      <td>0.049004</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>51</td>\n",
       "      <td>0.048533</td>\n",
       "      <td>0.044674</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>52</td>\n",
       "      <td>0.052122</td>\n",
       "      <td>0.053890</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>53</td>\n",
       "      <td>0.052193</td>\n",
       "      <td>0.052666</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>54</td>\n",
       "      <td>0.058223</td>\n",
       "      <td>0.046991</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 44: early stopping\n",
      "StackedResBlock(1024, **{'hidden_sizes': [70, 40, 10], 'act_cls': nn.Sigmoid, 'bn_1st': False})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [55, 10], 'act_cls': nn.ReLU, 'bn_1st': True})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [70, 40, 10], 'act_cls': nn.Sigmoid, 'bn_1st': False})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [55, 10], 'act_cls': nn.ReLU, 'bn_1st': True})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.292560</td>\n",
       "      <td>0.267943</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.239080</td>\n",
       "      <td>0.207295</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.205257</td>\n",
       "      <td>0.167165</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.174641</td>\n",
       "      <td>0.152607</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.162289</td>\n",
       "      <td>0.138544</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.168355</td>\n",
       "      <td>0.150460</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.148335</td>\n",
       "      <td>0.114709</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.142111</td>\n",
       "      <td>0.111405</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.123491</td>\n",
       "      <td>0.115366</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.125714</td>\n",
       "      <td>0.103980</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.107328</td>\n",
       "      <td>0.094866</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.118034</td>\n",
       "      <td>0.097671</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.094367</td>\n",
       "      <td>0.098978</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.108340</td>\n",
       "      <td>0.087817</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.102402</td>\n",
       "      <td>0.084871</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.087353</td>\n",
       "      <td>0.084562</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.083697</td>\n",
       "      <td>0.098718</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.099054</td>\n",
       "      <td>0.083635</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.091516</td>\n",
       "      <td>0.081588</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.076096</td>\n",
       "      <td>0.078249</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.067034</td>\n",
       "      <td>0.080951</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.092052</td>\n",
       "      <td>0.077327</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.077872</td>\n",
       "      <td>0.082872</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.079798</td>\n",
       "      <td>0.079487</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.077204</td>\n",
       "      <td>0.075956</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.065152</td>\n",
       "      <td>0.069763</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.063997</td>\n",
       "      <td>0.078877</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.064588</td>\n",
       "      <td>0.067177</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.071889</td>\n",
       "      <td>0.077581</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.055826</td>\n",
       "      <td>0.069823</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.054375</td>\n",
       "      <td>0.074138</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.051911</td>\n",
       "      <td>0.077887</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.050719</td>\n",
       "      <td>0.069653</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.085355</td>\n",
       "      <td>0.073823</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.055862</td>\n",
       "      <td>0.065722</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.059515</td>\n",
       "      <td>0.069407</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.069927</td>\n",
       "      <td>0.059429</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.050692</td>\n",
       "      <td>0.068077</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.040298</td>\n",
       "      <td>0.070708</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.065960</td>\n",
       "      <td>0.079781</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.040295</td>\n",
       "      <td>0.078622</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.052616</td>\n",
       "      <td>0.066028</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.049249</td>\n",
       "      <td>0.059674</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.054041</td>\n",
       "      <td>0.063086</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.059544</td>\n",
       "      <td>0.059179</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.064501</td>\n",
       "      <td>0.060330</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>46</td>\n",
       "      <td>0.053287</td>\n",
       "      <td>0.054549</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>47</td>\n",
       "      <td>0.047674</td>\n",
       "      <td>0.066697</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>48</td>\n",
       "      <td>0.030188</td>\n",
       "      <td>0.082962</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>49</td>\n",
       "      <td>0.060154</td>\n",
       "      <td>0.067681</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>50</td>\n",
       "      <td>0.047742</td>\n",
       "      <td>0.073864</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>51</td>\n",
       "      <td>0.045936</td>\n",
       "      <td>0.062219</td>\n",
       "      <td>00:11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>52</td>\n",
       "      <td>0.039162</td>\n",
       "      <td>0.062218</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>53</td>\n",
       "      <td>0.042010</td>\n",
       "      <td>0.065432</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>54</td>\n",
       "      <td>0.029799</td>\n",
       "      <td>0.080002</td>\n",
       "      <td>00:11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>55</td>\n",
       "      <td>0.040418</td>\n",
       "      <td>0.058411</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>56</td>\n",
       "      <td>0.032120</td>\n",
       "      <td>0.064906</td>\n",
       "      <td>00:10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 46: early stopping\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/rnavec_encoded_aggregated_u_t_swapped_rnas_DenseSAE.pt'})\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/ProtBERT_encoded_110_swissprot_DenseSAE.pt'})\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/rnavec_encoded_aggregated_u_t_swapped_rnas_DenseSAE.pt'})\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/ProtBERT_encoded_110_swissprot_DenseSAE.pt'})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.693112</td>\n",
       "      <td>0.693086</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.655587</td>\n",
       "      <td>0.602333</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.537114</td>\n",
       "      <td>0.499385</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.508215</td>\n",
       "      <td>0.477505</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.488435</td>\n",
       "      <td>0.465755</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.485050</td>\n",
       "      <td>0.457301</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.482641</td>\n",
       "      <td>0.451205</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.475256</td>\n",
       "      <td>0.446708</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.470949</td>\n",
       "      <td>0.443822</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.471461</td>\n",
       "      <td>0.441848</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.464937</td>\n",
       "      <td>0.437283</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.467813</td>\n",
       "      <td>0.435358</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.452966</td>\n",
       "      <td>0.433092</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.455248</td>\n",
       "      <td>0.432586</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.456205</td>\n",
       "      <td>0.430927</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.447572</td>\n",
       "      <td>0.426581</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.449028</td>\n",
       "      <td>0.425691</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.443916</td>\n",
       "      <td>0.424076</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.445676</td>\n",
       "      <td>0.420992</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.437814</td>\n",
       "      <td>0.421485</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.439570</td>\n",
       "      <td>0.417199</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.436348</td>\n",
       "      <td>0.418118</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.432915</td>\n",
       "      <td>0.415590</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.429300</td>\n",
       "      <td>0.415880</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.430150</td>\n",
       "      <td>0.413106</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.422543</td>\n",
       "      <td>0.413720</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.421248</td>\n",
       "      <td>0.409232</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.419383</td>\n",
       "      <td>0.406991</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.417714</td>\n",
       "      <td>0.408754</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.413548</td>\n",
       "      <td>0.407384</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.410814</td>\n",
       "      <td>0.414444</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.405745</td>\n",
       "      <td>0.406336</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.410318</td>\n",
       "      <td>0.407467</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.395323</td>\n",
       "      <td>0.404732</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.403820</td>\n",
       "      <td>0.401474</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.399119</td>\n",
       "      <td>0.400220</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.395711</td>\n",
       "      <td>0.395051</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.390581</td>\n",
       "      <td>0.396398</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.392163</td>\n",
       "      <td>0.389942</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.388313</td>\n",
       "      <td>0.393237</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.379796</td>\n",
       "      <td>0.383370</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.370410</td>\n",
       "      <td>0.380944</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.368618</td>\n",
       "      <td>0.369144</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.354884</td>\n",
       "      <td>0.358166</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.331069</td>\n",
       "      <td>0.343918</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.319834</td>\n",
       "      <td>0.310635</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>46</td>\n",
       "      <td>0.292909</td>\n",
       "      <td>0.277586</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>47</td>\n",
       "      <td>0.268966</td>\n",
       "      <td>0.246092</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>48</td>\n",
       "      <td>0.246685</td>\n",
       "      <td>0.227426</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>49</td>\n",
       "      <td>0.220549</td>\n",
       "      <td>0.199245</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>50</td>\n",
       "      <td>0.214429</td>\n",
       "      <td>0.192478</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>51</td>\n",
       "      <td>0.200110</td>\n",
       "      <td>0.179150</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>52</td>\n",
       "      <td>0.190626</td>\n",
       "      <td>0.177072</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>53</td>\n",
       "      <td>0.180327</td>\n",
       "      <td>0.163574</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>54</td>\n",
       "      <td>0.167609</td>\n",
       "      <td>0.153714</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>55</td>\n",
       "      <td>0.170453</td>\n",
       "      <td>0.149524</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>56</td>\n",
       "      <td>0.160558</td>\n",
       "      <td>0.155530</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>57</td>\n",
       "      <td>0.160415</td>\n",
       "      <td>0.132351</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>58</td>\n",
       "      <td>0.146073</td>\n",
       "      <td>0.134994</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>59</td>\n",
       "      <td>0.144352</td>\n",
       "      <td>0.130587</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>60</td>\n",
       "      <td>0.143381</td>\n",
       "      <td>0.134815</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>61</td>\n",
       "      <td>0.139176</td>\n",
       "      <td>0.130982</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>62</td>\n",
       "      <td>0.136925</td>\n",
       "      <td>0.131535</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>63</td>\n",
       "      <td>0.137040</td>\n",
       "      <td>0.127149</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>64</td>\n",
       "      <td>0.124174</td>\n",
       "      <td>0.116102</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>65</td>\n",
       "      <td>0.126220</td>\n",
       "      <td>0.111884</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>66</td>\n",
       "      <td>0.122966</td>\n",
       "      <td>0.109510</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>67</td>\n",
       "      <td>0.131604</td>\n",
       "      <td>0.120291</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>68</td>\n",
       "      <td>0.117818</td>\n",
       "      <td>0.100442</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>69</td>\n",
       "      <td>0.110085</td>\n",
       "      <td>0.110702</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>70</td>\n",
       "      <td>0.116673</td>\n",
       "      <td>0.110472</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>71</td>\n",
       "      <td>0.113654</td>\n",
       "      <td>0.109908</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>72</td>\n",
       "      <td>0.117457</td>\n",
       "      <td>0.098515</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>73</td>\n",
       "      <td>0.108003</td>\n",
       "      <td>0.097654</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>74</td>\n",
       "      <td>0.107035</td>\n",
       "      <td>0.106367</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>75</td>\n",
       "      <td>0.105179</td>\n",
       "      <td>0.113630</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>76</td>\n",
       "      <td>0.097135</td>\n",
       "      <td>0.099390</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>77</td>\n",
       "      <td>0.098859</td>\n",
       "      <td>0.097488</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>78</td>\n",
       "      <td>0.093207</td>\n",
       "      <td>0.093644</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>79</td>\n",
       "      <td>0.111304</td>\n",
       "      <td>0.105752</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>80</td>\n",
       "      <td>0.098510</td>\n",
       "      <td>0.097729</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>81</td>\n",
       "      <td>0.093297</td>\n",
       "      <td>0.097808</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>82</td>\n",
       "      <td>0.094289</td>\n",
       "      <td>0.095224</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>83</td>\n",
       "      <td>0.093360</td>\n",
       "      <td>0.084061</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>84</td>\n",
       "      <td>0.093740</td>\n",
       "      <td>0.083341</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>85</td>\n",
       "      <td>0.088990</td>\n",
       "      <td>0.095898</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>86</td>\n",
       "      <td>0.082277</td>\n",
       "      <td>0.086089</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>87</td>\n",
       "      <td>0.088413</td>\n",
       "      <td>0.088433</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>88</td>\n",
       "      <td>0.092525</td>\n",
       "      <td>0.093204</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>89</td>\n",
       "      <td>0.080738</td>\n",
       "      <td>0.082743</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>90</td>\n",
       "      <td>0.075048</td>\n",
       "      <td>0.092847</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>91</td>\n",
       "      <td>0.079704</td>\n",
       "      <td>0.087081</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>92</td>\n",
       "      <td>0.081729</td>\n",
       "      <td>0.087003</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>93</td>\n",
       "      <td>0.075703</td>\n",
       "      <td>0.080640</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>94</td>\n",
       "      <td>0.072478</td>\n",
       "      <td>0.078665</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>95</td>\n",
       "      <td>0.078542</td>\n",
       "      <td>0.095519</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>96</td>\n",
       "      <td>0.082319</td>\n",
       "      <td>0.087759</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>97</td>\n",
       "      <td>0.071677</td>\n",
       "      <td>0.079357</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>98</td>\n",
       "      <td>0.066225</td>\n",
       "      <td>0.087943</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>99</td>\n",
       "      <td>0.074273</td>\n",
       "      <td>0.089354</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>100</td>\n",
       "      <td>0.062849</td>\n",
       "      <td>0.091098</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>101</td>\n",
       "      <td>0.063755</td>\n",
       "      <td>0.087423</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>102</td>\n",
       "      <td>0.062362</td>\n",
       "      <td>0.082446</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>103</td>\n",
       "      <td>0.063045</td>\n",
       "      <td>0.081525</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>104</td>\n",
       "      <td>0.064987</td>\n",
       "      <td>0.085083</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 94: early stopping\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/rnavec_encoded_aggregated_u_t_swapped_rnas_DenseSAE.pt'})\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/ProtBERT_encoded_110_swissprot_DenseSAE.pt'})\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/rnavec_encoded_aggregated_u_t_swapped_rnas_DenseSAE.pt'})\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/ProtBERT_encoded_110_swissprot_DenseSAE.pt'})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.693103</td>\n",
       "      <td>0.693072</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.631615</td>\n",
       "      <td>0.564241</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.516449</td>\n",
       "      <td>0.498000</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.497104</td>\n",
       "      <td>0.482480</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.478943</td>\n",
       "      <td>0.474994</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.477880</td>\n",
       "      <td>0.470020</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.473321</td>\n",
       "      <td>0.465794</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.470231</td>\n",
       "      <td>0.462119</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.470216</td>\n",
       "      <td>0.457795</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.465285</td>\n",
       "      <td>0.455282</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.458957</td>\n",
       "      <td>0.452803</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.460340</td>\n",
       "      <td>0.449596</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.450406</td>\n",
       "      <td>0.448606</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.447725</td>\n",
       "      <td>0.445815</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.449755</td>\n",
       "      <td>0.445333</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.441864</td>\n",
       "      <td>0.442012</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.444501</td>\n",
       "      <td>0.436952</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.441179</td>\n",
       "      <td>0.434112</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.437894</td>\n",
       "      <td>0.432146</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.431408</td>\n",
       "      <td>0.432522</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.428358</td>\n",
       "      <td>0.430357</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.425137</td>\n",
       "      <td>0.426572</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.422690</td>\n",
       "      <td>0.422742</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.421818</td>\n",
       "      <td>0.428501</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.421684</td>\n",
       "      <td>0.424475</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.415703</td>\n",
       "      <td>0.426337</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.414799</td>\n",
       "      <td>0.422904</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.412692</td>\n",
       "      <td>0.414340</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.406698</td>\n",
       "      <td>0.417469</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.400484</td>\n",
       "      <td>0.414411</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.403352</td>\n",
       "      <td>0.417247</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.398443</td>\n",
       "      <td>0.410350</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.397904</td>\n",
       "      <td>0.406595</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.388058</td>\n",
       "      <td>0.402086</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.390088</td>\n",
       "      <td>0.404120</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.383903</td>\n",
       "      <td>0.400045</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.377734</td>\n",
       "      <td>0.390077</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.367685</td>\n",
       "      <td>0.397124</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.365344</td>\n",
       "      <td>0.377798</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.346579</td>\n",
       "      <td>0.354784</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.323911</td>\n",
       "      <td>0.328794</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.294435</td>\n",
       "      <td>0.294859</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.268199</td>\n",
       "      <td>0.268037</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.249211</td>\n",
       "      <td>0.235165</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.230462</td>\n",
       "      <td>0.235741</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.212160</td>\n",
       "      <td>0.224735</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>46</td>\n",
       "      <td>0.200298</td>\n",
       "      <td>0.193209</td>\n",
       "      <td>00:01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>47</td>\n",
       "      <td>0.188578</td>\n",
       "      <td>0.184212</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>48</td>\n",
       "      <td>0.177031</td>\n",
       "      <td>0.166066</td>\n",
       "      <td>00:01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>49</td>\n",
       "      <td>0.170793</td>\n",
       "      <td>0.182381</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>50</td>\n",
       "      <td>0.162965</td>\n",
       "      <td>0.171148</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>51</td>\n",
       "      <td>0.154956</td>\n",
       "      <td>0.162314</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>52</td>\n",
       "      <td>0.154985</td>\n",
       "      <td>0.151823</td>\n",
       "      <td>00:01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>53</td>\n",
       "      <td>0.152914</td>\n",
       "      <td>0.135664</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>54</td>\n",
       "      <td>0.141901</td>\n",
       "      <td>0.157678</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>55</td>\n",
       "      <td>0.139879</td>\n",
       "      <td>0.153463</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>56</td>\n",
       "      <td>0.136438</td>\n",
       "      <td>0.136227</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>57</td>\n",
       "      <td>0.129838</td>\n",
       "      <td>0.138012</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>58</td>\n",
       "      <td>0.132748</td>\n",
       "      <td>0.132336</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>59</td>\n",
       "      <td>0.132068</td>\n",
       "      <td>0.144315</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>60</td>\n",
       "      <td>0.125233</td>\n",
       "      <td>0.127751</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>61</td>\n",
       "      <td>0.123755</td>\n",
       "      <td>0.135078</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>62</td>\n",
       "      <td>0.129106</td>\n",
       "      <td>0.124096</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>63</td>\n",
       "      <td>0.121956</td>\n",
       "      <td>0.139770</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>64</td>\n",
       "      <td>0.115544</td>\n",
       "      <td>0.143275</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>65</td>\n",
       "      <td>0.115878</td>\n",
       "      <td>0.116428</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>66</td>\n",
       "      <td>0.113553</td>\n",
       "      <td>0.123438</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>67</td>\n",
       "      <td>0.110518</td>\n",
       "      <td>0.132433</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>68</td>\n",
       "      <td>0.107670</td>\n",
       "      <td>0.133207</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>69</td>\n",
       "      <td>0.107987</td>\n",
       "      <td>0.125809</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>70</td>\n",
       "      <td>0.106061</td>\n",
       "      <td>0.132067</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>71</td>\n",
       "      <td>0.097812</td>\n",
       "      <td>0.137051</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>72</td>\n",
       "      <td>0.100551</td>\n",
       "      <td>0.120129</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>73</td>\n",
       "      <td>0.107083</td>\n",
       "      <td>0.118423</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>74</td>\n",
       "      <td>0.100237</td>\n",
       "      <td>0.103637</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>75</td>\n",
       "      <td>0.096903</td>\n",
       "      <td>0.124319</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>76</td>\n",
       "      <td>0.088250</td>\n",
       "      <td>0.118015</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>77</td>\n",
       "      <td>0.096283</td>\n",
       "      <td>0.124201</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>78</td>\n",
       "      <td>0.086269</td>\n",
       "      <td>0.124418</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>79</td>\n",
       "      <td>0.087990</td>\n",
       "      <td>0.117072</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>80</td>\n",
       "      <td>0.083522</td>\n",
       "      <td>0.125124</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>81</td>\n",
       "      <td>0.091367</td>\n",
       "      <td>0.113481</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>82</td>\n",
       "      <td>0.089450</td>\n",
       "      <td>0.130109</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>83</td>\n",
       "      <td>0.074095</td>\n",
       "      <td>0.134585</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>84</td>\n",
       "      <td>0.082017</td>\n",
       "      <td>0.108544</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 74: early stopping\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/rnavec_encoded_aggregated_u_t_swapped_rnas_DenseSAE.pt'})\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/ProtBERT_encoded_110_swissprot_DenseSAE.pt'})\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/rnavec_encoded_aggregated_u_t_swapped_rnas_DenseSAE.pt'})\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/ProtBERT_encoded_110_swissprot_DenseSAE.pt'})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.597723</td>\n",
       "      <td>0.507039</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.492582</td>\n",
       "      <td>0.479586</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.475345</td>\n",
       "      <td>0.468781</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.466629</td>\n",
       "      <td>0.467041</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.452770</td>\n",
       "      <td>0.457176</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.450972</td>\n",
       "      <td>0.449427</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.445112</td>\n",
       "      <td>0.453900</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.439974</td>\n",
       "      <td>0.441794</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.430573</td>\n",
       "      <td>0.437932</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.421257</td>\n",
       "      <td>0.430411</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.417050</td>\n",
       "      <td>0.417384</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.401538</td>\n",
       "      <td>0.400895</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.360303</td>\n",
       "      <td>0.365853</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.318724</td>\n",
       "      <td>0.303149</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.271329</td>\n",
       "      <td>0.258646</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.235898</td>\n",
       "      <td>0.232999</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.218527</td>\n",
       "      <td>0.216366</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.196296</td>\n",
       "      <td>0.198764</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.188350</td>\n",
       "      <td>0.177874</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.176182</td>\n",
       "      <td>0.179349</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.161295</td>\n",
       "      <td>0.174725</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.161992</td>\n",
       "      <td>0.162514</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.156496</td>\n",
       "      <td>0.163616</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.152064</td>\n",
       "      <td>0.157001</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.146973</td>\n",
       "      <td>0.160974</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.144022</td>\n",
       "      <td>0.159407</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.145284</td>\n",
       "      <td>0.155281</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.138162</td>\n",
       "      <td>0.148872</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.134388</td>\n",
       "      <td>0.153719</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.135806</td>\n",
       "      <td>0.143335</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.134330</td>\n",
       "      <td>0.171764</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.126706</td>\n",
       "      <td>0.135433</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.124338</td>\n",
       "      <td>0.151602</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.127067</td>\n",
       "      <td>0.138429</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.124800</td>\n",
       "      <td>0.137589</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.124880</td>\n",
       "      <td>0.163140</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.116237</td>\n",
       "      <td>0.151582</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.111099</td>\n",
       "      <td>0.128918</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.107201</td>\n",
       "      <td>0.141038</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.112232</td>\n",
       "      <td>0.145635</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.107266</td>\n",
       "      <td>0.142557</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.102123</td>\n",
       "      <td>0.136932</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.110673</td>\n",
       "      <td>0.148197</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.094464</td>\n",
       "      <td>0.150687</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.100842</td>\n",
       "      <td>0.126628</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.096602</td>\n",
       "      <td>0.131704</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>46</td>\n",
       "      <td>0.093727</td>\n",
       "      <td>0.151402</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>47</td>\n",
       "      <td>0.093672</td>\n",
       "      <td>0.144771</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>48</td>\n",
       "      <td>0.096105</td>\n",
       "      <td>0.118489</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>49</td>\n",
       "      <td>0.086226</td>\n",
       "      <td>0.131837</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>50</td>\n",
       "      <td>0.095578</td>\n",
       "      <td>0.133509</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>51</td>\n",
       "      <td>0.082214</td>\n",
       "      <td>0.124841</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>52</td>\n",
       "      <td>0.085630</td>\n",
       "      <td>0.149963</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>53</td>\n",
       "      <td>0.080898</td>\n",
       "      <td>0.117076</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>54</td>\n",
       "      <td>0.083928</td>\n",
       "      <td>0.123552</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>55</td>\n",
       "      <td>0.078326</td>\n",
       "      <td>0.125546</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>56</td>\n",
       "      <td>0.078081</td>\n",
       "      <td>0.123635</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>57</td>\n",
       "      <td>0.077094</td>\n",
       "      <td>0.146594</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>58</td>\n",
       "      <td>0.075974</td>\n",
       "      <td>0.116681</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>59</td>\n",
       "      <td>0.073027</td>\n",
       "      <td>0.106136</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>60</td>\n",
       "      <td>0.066386</td>\n",
       "      <td>0.145163</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>61</td>\n",
       "      <td>0.070714</td>\n",
       "      <td>0.150904</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>62</td>\n",
       "      <td>0.064111</td>\n",
       "      <td>0.123711</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>63</td>\n",
       "      <td>0.067454</td>\n",
       "      <td>0.121797</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>64</td>\n",
       "      <td>0.063583</td>\n",
       "      <td>0.136952</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>65</td>\n",
       "      <td>0.067271</td>\n",
       "      <td>0.123518</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>66</td>\n",
       "      <td>0.062777</td>\n",
       "      <td>0.178195</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>67</td>\n",
       "      <td>0.067786</td>\n",
       "      <td>0.122643</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>68</td>\n",
       "      <td>0.055045</td>\n",
       "      <td>0.146015</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>69</td>\n",
       "      <td>0.057658</td>\n",
       "      <td>0.212908</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 59: early stopping\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/rnavec_encoded_aggregated_u_t_swapped_rnas_DenseSAE.pt'})\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/ProtBERT_encoded_110_swissprot_DenseSAE.pt'})\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/rnavec_encoded_aggregated_u_t_swapped_rnas_DenseSAE.pt'})\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/ProtBERT_encoded_110_swissprot_DenseSAE.pt'})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.587273</td>\n",
       "      <td>0.493277</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.491992</td>\n",
       "      <td>0.463152</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.471957</td>\n",
       "      <td>0.452394</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.470493</td>\n",
       "      <td>0.446791</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.451068</td>\n",
       "      <td>0.440245</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.451722</td>\n",
       "      <td>0.430708</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.449109</td>\n",
       "      <td>0.428618</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.443093</td>\n",
       "      <td>0.427056</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.436171</td>\n",
       "      <td>0.425076</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.429577</td>\n",
       "      <td>0.415628</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.416162</td>\n",
       "      <td>0.408136</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.402834</td>\n",
       "      <td>0.393414</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.380705</td>\n",
       "      <td>0.359972</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.340087</td>\n",
       "      <td>0.312928</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.295166</td>\n",
       "      <td>0.259519</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.253611</td>\n",
       "      <td>0.223831</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.230341</td>\n",
       "      <td>0.206414</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.212341</td>\n",
       "      <td>0.176617</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.195477</td>\n",
       "      <td>0.157031</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.188014</td>\n",
       "      <td>0.152186</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.179078</td>\n",
       "      <td>0.149226</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.177555</td>\n",
       "      <td>0.140882</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.170252</td>\n",
       "      <td>0.135013</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.160496</td>\n",
       "      <td>0.123868</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.153505</td>\n",
       "      <td>0.118536</td>\n",
       "      <td>00:01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.154211</td>\n",
       "      <td>0.119901</td>\n",
       "      <td>00:01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.152723</td>\n",
       "      <td>0.110648</td>\n",
       "      <td>00:01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.144288</td>\n",
       "      <td>0.112922</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.147850</td>\n",
       "      <td>0.112097</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.138213</td>\n",
       "      <td>0.103665</td>\n",
       "      <td>00:01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.138281</td>\n",
       "      <td>0.100359</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.137815</td>\n",
       "      <td>0.098115</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.130163</td>\n",
       "      <td>0.107816</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.127878</td>\n",
       "      <td>0.094983</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.128549</td>\n",
       "      <td>0.089434</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.115303</td>\n",
       "      <td>0.094927</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.120497</td>\n",
       "      <td>0.080841</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.116131</td>\n",
       "      <td>0.093449</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.117194</td>\n",
       "      <td>0.092027</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.106892</td>\n",
       "      <td>0.084275</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.112504</td>\n",
       "      <td>0.072239</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.100374</td>\n",
       "      <td>0.087151</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.100857</td>\n",
       "      <td>0.080813</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.098908</td>\n",
       "      <td>0.093346</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.099694</td>\n",
       "      <td>0.076229</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.101948</td>\n",
       "      <td>0.077857</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>46</td>\n",
       "      <td>0.100830</td>\n",
       "      <td>0.077552</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>47</td>\n",
       "      <td>0.087990</td>\n",
       "      <td>0.080390</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>48</td>\n",
       "      <td>0.085170</td>\n",
       "      <td>0.079120</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>49</td>\n",
       "      <td>0.086110</td>\n",
       "      <td>0.065998</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>50</td>\n",
       "      <td>0.086295</td>\n",
       "      <td>0.074578</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>51</td>\n",
       "      <td>0.080441</td>\n",
       "      <td>0.063544</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>52</td>\n",
       "      <td>0.090178</td>\n",
       "      <td>0.076520</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>53</td>\n",
       "      <td>0.078713</td>\n",
       "      <td>0.060420</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>54</td>\n",
       "      <td>0.080822</td>\n",
       "      <td>0.060999</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>55</td>\n",
       "      <td>0.094385</td>\n",
       "      <td>0.071494</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>56</td>\n",
       "      <td>0.082837</td>\n",
       "      <td>0.064950</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>57</td>\n",
       "      <td>0.080044</td>\n",
       "      <td>0.068272</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>58</td>\n",
       "      <td>0.073713</td>\n",
       "      <td>0.059809</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>59</td>\n",
       "      <td>0.079529</td>\n",
       "      <td>0.057483</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>60</td>\n",
       "      <td>0.075151</td>\n",
       "      <td>0.066198</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>61</td>\n",
       "      <td>0.068269</td>\n",
       "      <td>0.062778</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>62</td>\n",
       "      <td>0.073785</td>\n",
       "      <td>0.064495</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>63</td>\n",
       "      <td>0.071196</td>\n",
       "      <td>0.072973</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>64</td>\n",
       "      <td>0.068108</td>\n",
       "      <td>0.071621</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>65</td>\n",
       "      <td>0.067888</td>\n",
       "      <td>0.060724</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>66</td>\n",
       "      <td>0.064617</td>\n",
       "      <td>0.051361</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>67</td>\n",
       "      <td>0.068447</td>\n",
       "      <td>0.056613</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>68</td>\n",
       "      <td>0.064913</td>\n",
       "      <td>0.068651</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>69</td>\n",
       "      <td>0.064681</td>\n",
       "      <td>0.064958</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>70</td>\n",
       "      <td>0.066551</td>\n",
       "      <td>0.056086</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>71</td>\n",
       "      <td>0.063980</td>\n",
       "      <td>0.050868</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>72</td>\n",
       "      <td>0.056323</td>\n",
       "      <td>0.065880</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>73</td>\n",
       "      <td>0.060539</td>\n",
       "      <td>0.062332</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>74</td>\n",
       "      <td>0.063643</td>\n",
       "      <td>0.045861</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>75</td>\n",
       "      <td>0.063455</td>\n",
       "      <td>0.063566</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>76</td>\n",
       "      <td>0.068177</td>\n",
       "      <td>0.053254</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>77</td>\n",
       "      <td>0.059552</td>\n",
       "      <td>0.048621</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>78</td>\n",
       "      <td>0.058996</td>\n",
       "      <td>0.056119</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>79</td>\n",
       "      <td>0.064578</td>\n",
       "      <td>0.050817</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>80</td>\n",
       "      <td>0.061418</td>\n",
       "      <td>0.052045</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>81</td>\n",
       "      <td>0.058727</td>\n",
       "      <td>0.075169</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>82</td>\n",
       "      <td>0.062401</td>\n",
       "      <td>0.051022</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>83</td>\n",
       "      <td>0.060991</td>\n",
       "      <td>0.049404</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>84</td>\n",
       "      <td>0.060179</td>\n",
       "      <td>0.057948</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 74: early stopping\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/rnavec_encoded_aggregated_u_t_swapped_rnas_DenseSAE.pt'})\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/ProtBERT_encoded_110_swissprot_DenseSAE.pt'})\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/rnavec_encoded_aggregated_u_t_swapped_rnas_DenseSAE.pt'})\n",
      "SAEBlock(**{'sae_type': 'Dense', 'sae_model_path': '/data/Chapter_3/Notebooks/ProtBERT_encoded_110_swissprot_DenseSAE.pt'})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.595141</td>\n",
       "      <td>0.503792</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.500857</td>\n",
       "      <td>0.479116</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.476829</td>\n",
       "      <td>0.465272</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.468313</td>\n",
       "      <td>0.456534</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.450989</td>\n",
       "      <td>0.453781</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.452083</td>\n",
       "      <td>0.446729</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.446496</td>\n",
       "      <td>0.440042</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.440470</td>\n",
       "      <td>0.433584</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.430805</td>\n",
       "      <td>0.429339</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.426979</td>\n",
       "      <td>0.426638</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.415234</td>\n",
       "      <td>0.415613</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.415108</td>\n",
       "      <td>0.406797</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.398140</td>\n",
       "      <td>0.409899</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.383024</td>\n",
       "      <td>0.385286</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.357154</td>\n",
       "      <td>0.347789</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.304048</td>\n",
       "      <td>0.294300</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.265451</td>\n",
       "      <td>0.248480</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.241221</td>\n",
       "      <td>0.220769</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.216925</td>\n",
       "      <td>0.191054</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.197574</td>\n",
       "      <td>0.185415</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.179336</td>\n",
       "      <td>0.185525</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.176883</td>\n",
       "      <td>0.165827</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.168043</td>\n",
       "      <td>0.159689</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.159028</td>\n",
       "      <td>0.161904</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.158709</td>\n",
       "      <td>0.160371</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.155673</td>\n",
       "      <td>0.165629</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.149877</td>\n",
       "      <td>0.156540</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.134958</td>\n",
       "      <td>0.144633</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.144739</td>\n",
       "      <td>0.150301</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.136291</td>\n",
       "      <td>0.139781</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.135104</td>\n",
       "      <td>0.145680</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.132490</td>\n",
       "      <td>0.144845</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.124976</td>\n",
       "      <td>0.135536</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.124313</td>\n",
       "      <td>0.140288</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.119725</td>\n",
       "      <td>0.131550</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.122648</td>\n",
       "      <td>0.138254</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.113131</td>\n",
       "      <td>0.125698</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.112870</td>\n",
       "      <td>0.137381</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.107838</td>\n",
       "      <td>0.132828</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.106108</td>\n",
       "      <td>0.128436</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.101586</td>\n",
       "      <td>0.128954</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.109528</td>\n",
       "      <td>0.126038</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.102086</td>\n",
       "      <td>0.145707</td>\n",
       "      <td>00:01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.105698</td>\n",
       "      <td>0.117617</td>\n",
       "      <td>00:01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.103419</td>\n",
       "      <td>0.106546</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.092959</td>\n",
       "      <td>0.114339</td>\n",
       "      <td>00:01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>46</td>\n",
       "      <td>0.092388</td>\n",
       "      <td>0.120238</td>\n",
       "      <td>00:01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>47</td>\n",
       "      <td>0.094669</td>\n",
       "      <td>0.106641</td>\n",
       "      <td>00:01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>48</td>\n",
       "      <td>0.086652</td>\n",
       "      <td>0.112197</td>\n",
       "      <td>00:01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>49</td>\n",
       "      <td>0.084063</td>\n",
       "      <td>0.105490</td>\n",
       "      <td>00:01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>50</td>\n",
       "      <td>0.088854</td>\n",
       "      <td>0.095166</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>51</td>\n",
       "      <td>0.081302</td>\n",
       "      <td>0.096279</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>52</td>\n",
       "      <td>0.078598</td>\n",
       "      <td>0.119440</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>53</td>\n",
       "      <td>0.082530</td>\n",
       "      <td>0.119206</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>54</td>\n",
       "      <td>0.074105</td>\n",
       "      <td>0.114332</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>55</td>\n",
       "      <td>0.079982</td>\n",
       "      <td>0.104847</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>56</td>\n",
       "      <td>0.082319</td>\n",
       "      <td>0.101255</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>57</td>\n",
       "      <td>0.070035</td>\n",
       "      <td>0.125756</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>58</td>\n",
       "      <td>0.068184</td>\n",
       "      <td>0.099260</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>59</td>\n",
       "      <td>0.076892</td>\n",
       "      <td>0.102378</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>60</td>\n",
       "      <td>0.066230</td>\n",
       "      <td>0.119809</td>\n",
       "      <td>00:02</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 50: early stopping\n",
      "AttentionBlock(1024, **{'layer_count': 32})\n",
      "AttentionBlock(1024, **{'layer_count': 8})\n",
      "AttentionBlock(1024, **{'layer_count': 32})\n",
      "AttentionBlock(1024, **{'layer_count': 8})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.313070</td>\n",
       "      <td>0.244215</td>\n",
       "      <td>00:37</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.242603</td>\n",
       "      <td>0.188821</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.216145</td>\n",
       "      <td>0.165229</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.205099</td>\n",
       "      <td>0.155765</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.181534</td>\n",
       "      <td>0.149744</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.179737</td>\n",
       "      <td>0.138496</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.167059</td>\n",
       "      <td>0.132900</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.153367</td>\n",
       "      <td>0.126550</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.160104</td>\n",
       "      <td>0.124236</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.158187</td>\n",
       "      <td>0.118287</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.147288</td>\n",
       "      <td>0.115443</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.161171</td>\n",
       "      <td>0.122323</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.132954</td>\n",
       "      <td>0.107892</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.139679</td>\n",
       "      <td>0.106672</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.130614</td>\n",
       "      <td>0.104527</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.133512</td>\n",
       "      <td>0.105475</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.136713</td>\n",
       "      <td>0.107608</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.130197</td>\n",
       "      <td>0.103006</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.127933</td>\n",
       "      <td>0.103784</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.131592</td>\n",
       "      <td>0.102938</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.129045</td>\n",
       "      <td>0.107201</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.131730</td>\n",
       "      <td>0.098418</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.128609</td>\n",
       "      <td>0.099460</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.128485</td>\n",
       "      <td>0.097896</td>\n",
       "      <td>00:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.132517</td>\n",
       "      <td>0.099161</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.122164</td>\n",
       "      <td>0.097845</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.127242</td>\n",
       "      <td>0.099044</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.133609</td>\n",
       "      <td>0.098162</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.134173</td>\n",
       "      <td>0.104952</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.128413</td>\n",
       "      <td>0.101959</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.126307</td>\n",
       "      <td>0.099952</td>\n",
       "      <td>00:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.128619</td>\n",
       "      <td>0.099072</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.116585</td>\n",
       "      <td>0.102796</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.112482</td>\n",
       "      <td>0.091927</td>\n",
       "      <td>00:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.129473</td>\n",
       "      <td>0.095538</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.108350</td>\n",
       "      <td>0.113156</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.117119</td>\n",
       "      <td>0.094174</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.124539</td>\n",
       "      <td>0.098584</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.116604</td>\n",
       "      <td>0.101917</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.120816</td>\n",
       "      <td>0.097479</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.116427</td>\n",
       "      <td>0.100158</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.114653</td>\n",
       "      <td>0.092458</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.109349</td>\n",
       "      <td>0.093673</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.113833</td>\n",
       "      <td>0.098835</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 33: early stopping\n",
      "AttentionBlock(1024, **{'layer_count': 32})\n",
      "AttentionBlock(1024, **{'layer_count': 8})\n",
      "AttentionBlock(1024, **{'layer_count': 32})\n",
      "AttentionBlock(1024, **{'layer_count': 8})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.255938</td>\n",
       "      <td>0.235745</td>\n",
       "      <td>00:38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.211102</td>\n",
       "      <td>0.198602</td>\n",
       "      <td>00:26</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.188265</td>\n",
       "      <td>0.179093</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.172812</td>\n",
       "      <td>0.169751</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.155574</td>\n",
       "      <td>0.167045</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.160113</td>\n",
       "      <td>0.154855</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.145446</td>\n",
       "      <td>0.148207</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.142339</td>\n",
       "      <td>0.155472</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.145855</td>\n",
       "      <td>0.144865</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.149092</td>\n",
       "      <td>0.140773</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.131254</td>\n",
       "      <td>0.131627</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.141866</td>\n",
       "      <td>0.135902</td>\n",
       "      <td>00:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.129976</td>\n",
       "      <td>0.125896</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.118517</td>\n",
       "      <td>0.127013</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.121719</td>\n",
       "      <td>0.136762</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.128756</td>\n",
       "      <td>0.127035</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.116534</td>\n",
       "      <td>0.128621</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.135380</td>\n",
       "      <td>0.124818</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.113812</td>\n",
       "      <td>0.121961</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.129952</td>\n",
       "      <td>0.135151</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.119383</td>\n",
       "      <td>0.119863</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.108485</td>\n",
       "      <td>0.133810</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.109896</td>\n",
       "      <td>0.120512</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.122604</td>\n",
       "      <td>0.116485</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.117331</td>\n",
       "      <td>0.127240</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.111161</td>\n",
       "      <td>0.126107</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.120490</td>\n",
       "      <td>0.122628</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.117835</td>\n",
       "      <td>0.119900</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.110983</td>\n",
       "      <td>0.136438</td>\n",
       "      <td>00:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.119373</td>\n",
       "      <td>0.112057</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.113574</td>\n",
       "      <td>0.122433</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.112498</td>\n",
       "      <td>0.130366</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.122973</td>\n",
       "      <td>0.115098</td>\n",
       "      <td>00:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.107394</td>\n",
       "      <td>0.118027</td>\n",
       "      <td>00:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.114312</td>\n",
       "      <td>0.129127</td>\n",
       "      <td>00:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.112163</td>\n",
       "      <td>0.108649</td>\n",
       "      <td>00:34</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.119040</td>\n",
       "      <td>0.127912</td>\n",
       "      <td>00:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.097601</td>\n",
       "      <td>0.116165</td>\n",
       "      <td>00:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.116573</td>\n",
       "      <td>0.131858</td>\n",
       "      <td>00:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.108918</td>\n",
       "      <td>0.126613</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.107158</td>\n",
       "      <td>0.114951</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.108709</td>\n",
       "      <td>0.109612</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.113774</td>\n",
       "      <td>0.110531</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.105399</td>\n",
       "      <td>0.119493</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.110256</td>\n",
       "      <td>0.113051</td>\n",
       "      <td>00:34</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.100260</td>\n",
       "      <td>0.111233</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 35: early stopping\n",
      "AttentionBlock(1024, **{'layer_count': 32})\n",
      "AttentionBlock(1024, **{'layer_count': 8})\n",
      "AttentionBlock(1024, **{'layer_count': 32})\n",
      "AttentionBlock(1024, **{'layer_count': 8})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.320053</td>\n",
       "      <td>0.294385</td>\n",
       "      <td>00:40</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.240293</td>\n",
       "      <td>0.235289</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.219667</td>\n",
       "      <td>0.213289</td>\n",
       "      <td>00:29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.193856</td>\n",
       "      <td>0.205017</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.181628</td>\n",
       "      <td>0.191321</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.185137</td>\n",
       "      <td>0.177533</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.169827</td>\n",
       "      <td>0.170682</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.168786</td>\n",
       "      <td>0.163973</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.156136</td>\n",
       "      <td>0.159839</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.155345</td>\n",
       "      <td>0.155737</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.141654</td>\n",
       "      <td>0.151279</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.150011</td>\n",
       "      <td>0.155783</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.145340</td>\n",
       "      <td>0.144544</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.136604</td>\n",
       "      <td>0.137085</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.145505</td>\n",
       "      <td>0.140173</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.132047</td>\n",
       "      <td>0.129010</td>\n",
       "      <td>00:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.140544</td>\n",
       "      <td>0.134055</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.133888</td>\n",
       "      <td>0.131691</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.141134</td>\n",
       "      <td>0.130044</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.138636</td>\n",
       "      <td>0.136203</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.124002</td>\n",
       "      <td>0.132025</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.136352</td>\n",
       "      <td>0.123448</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.131194</td>\n",
       "      <td>0.124839</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.120500</td>\n",
       "      <td>0.133906</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.131664</td>\n",
       "      <td>0.123657</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.120433</td>\n",
       "      <td>0.123099</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.124678</td>\n",
       "      <td>0.119995</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.119794</td>\n",
       "      <td>0.124263</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.125255</td>\n",
       "      <td>0.119656</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.127944</td>\n",
       "      <td>0.117692</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.119504</td>\n",
       "      <td>0.124966</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.124807</td>\n",
       "      <td>0.115272</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.110751</td>\n",
       "      <td>0.122114</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.124938</td>\n",
       "      <td>0.120388</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.119696</td>\n",
       "      <td>0.116383</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.110533</td>\n",
       "      <td>0.127333</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.117716</td>\n",
       "      <td>0.122128</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.115129</td>\n",
       "      <td>0.113593</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.122303</td>\n",
       "      <td>0.125601</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.111042</td>\n",
       "      <td>0.116702</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.105723</td>\n",
       "      <td>0.129748</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.113897</td>\n",
       "      <td>0.124704</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.119934</td>\n",
       "      <td>0.115529</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.114386</td>\n",
       "      <td>0.128259</td>\n",
       "      <td>00:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.110567</td>\n",
       "      <td>0.115940</td>\n",
       "      <td>00:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.111910</td>\n",
       "      <td>0.115518</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>46</td>\n",
       "      <td>0.107672</td>\n",
       "      <td>0.115398</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>47</td>\n",
       "      <td>0.105190</td>\n",
       "      <td>0.124481</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 37: early stopping\n",
      "AttentionBlock(1024, **{'layer_count': 32})\n",
      "AttentionBlock(1024, **{'layer_count': 8})\n",
      "AttentionBlock(1024, **{'layer_count': 32})\n",
      "AttentionBlock(1024, **{'layer_count': 8})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.319853</td>\n",
       "      <td>0.264279</td>\n",
       "      <td>00:41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.244580</td>\n",
       "      <td>0.204719</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.221282</td>\n",
       "      <td>0.185323</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.208146</td>\n",
       "      <td>0.165694</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.186466</td>\n",
       "      <td>0.153196</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.180642</td>\n",
       "      <td>0.146195</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.180904</td>\n",
       "      <td>0.138208</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.169070</td>\n",
       "      <td>0.134202</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.166780</td>\n",
       "      <td>0.128487</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.170465</td>\n",
       "      <td>0.122771</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.146255</td>\n",
       "      <td>0.117475</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.160799</td>\n",
       "      <td>0.116917</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.151580</td>\n",
       "      <td>0.110026</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.136821</td>\n",
       "      <td>0.109204</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.158173</td>\n",
       "      <td>0.105494</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.142688</td>\n",
       "      <td>0.104097</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.132488</td>\n",
       "      <td>0.103274</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.138089</td>\n",
       "      <td>0.105742</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.131059</td>\n",
       "      <td>0.099888</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.137320</td>\n",
       "      <td>0.099973</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.130689</td>\n",
       "      <td>0.101856</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.135094</td>\n",
       "      <td>0.097590</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.124599</td>\n",
       "      <td>0.104055</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.130307</td>\n",
       "      <td>0.095512</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.136306</td>\n",
       "      <td>0.098009</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.125639</td>\n",
       "      <td>0.097065</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.128819</td>\n",
       "      <td>0.092314</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.130747</td>\n",
       "      <td>0.091784</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.130261</td>\n",
       "      <td>0.105865</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.121263</td>\n",
       "      <td>0.097914</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.125855</td>\n",
       "      <td>0.092484</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.125479</td>\n",
       "      <td>0.088765</td>\n",
       "      <td>00:28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.118135</td>\n",
       "      <td>0.098099</td>\n",
       "      <td>00:26</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.119960</td>\n",
       "      <td>0.084978</td>\n",
       "      <td>00:27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.119981</td>\n",
       "      <td>0.093714</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.113474</td>\n",
       "      <td>0.086981</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.130051</td>\n",
       "      <td>0.095541</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.125269</td>\n",
       "      <td>0.086367</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.125315</td>\n",
       "      <td>0.097073</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.115236</td>\n",
       "      <td>0.095046</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.113944</td>\n",
       "      <td>0.086849</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.108720</td>\n",
       "      <td>0.087239</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.117014</td>\n",
       "      <td>0.097145</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.117296</td>\n",
       "      <td>0.085263</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 33: early stopping\n",
      "AttentionBlock(1024, **{'layer_count': 32})\n",
      "AttentionBlock(1024, **{'layer_count': 8})\n",
      "AttentionBlock(1024, **{'layer_count': 32})\n",
      "AttentionBlock(1024, **{'layer_count': 8})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.268310</td>\n",
       "      <td>0.241909</td>\n",
       "      <td>00:42</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.227702</td>\n",
       "      <td>0.204178</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.199268</td>\n",
       "      <td>0.190493</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.181408</td>\n",
       "      <td>0.174537</td>\n",
       "      <td>00:34</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.161746</td>\n",
       "      <td>0.162546</td>\n",
       "      <td>00:33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.168975</td>\n",
       "      <td>0.156567</td>\n",
       "      <td>00:32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.157533</td>\n",
       "      <td>0.148621</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.155293</td>\n",
       "      <td>0.139982</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.145127</td>\n",
       "      <td>0.140040</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.140383</td>\n",
       "      <td>0.137346</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.135134</td>\n",
       "      <td>0.133884</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.143824</td>\n",
       "      <td>0.128633</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.132886</td>\n",
       "      <td>0.129696</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.130281</td>\n",
       "      <td>0.127956</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.130384</td>\n",
       "      <td>0.120465</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.123927</td>\n",
       "      <td>0.124902</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.124886</td>\n",
       "      <td>0.123555</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.142570</td>\n",
       "      <td>0.120427</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.130506</td>\n",
       "      <td>0.119365</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.124531</td>\n",
       "      <td>0.122702</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.110629</td>\n",
       "      <td>0.124329</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.132662</td>\n",
       "      <td>0.137189</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.119363</td>\n",
       "      <td>0.118454</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.121466</td>\n",
       "      <td>0.130375</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.123308</td>\n",
       "      <td>0.125640</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.117091</td>\n",
       "      <td>0.113754</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.120937</td>\n",
       "      <td>0.128181</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.118121</td>\n",
       "      <td>0.134405</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.125697</td>\n",
       "      <td>0.119725</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.122025</td>\n",
       "      <td>0.116254</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.117169</td>\n",
       "      <td>0.128345</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.107767</td>\n",
       "      <td>0.117295</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.109701</td>\n",
       "      <td>0.115514</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.122500</td>\n",
       "      <td>0.113621</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.113619</td>\n",
       "      <td>0.116413</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.121674</td>\n",
       "      <td>0.125386</td>\n",
       "      <td>00:30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.113347</td>\n",
       "      <td>0.113682</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.114618</td>\n",
       "      <td>0.112076</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.104317</td>\n",
       "      <td>0.112304</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.121468</td>\n",
       "      <td>0.120126</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.102606</td>\n",
       "      <td>0.120279</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.124879</td>\n",
       "      <td>0.115610</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.105352</td>\n",
       "      <td>0.111171</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.117484</td>\n",
       "      <td>0.113191</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.113652</td>\n",
       "      <td>0.112564</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.119143</td>\n",
       "      <td>0.103409</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>46</td>\n",
       "      <td>0.118799</td>\n",
       "      <td>0.101023</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>47</td>\n",
       "      <td>0.115289</td>\n",
       "      <td>0.116836</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>48</td>\n",
       "      <td>0.101421</td>\n",
       "      <td>0.113090</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>49</td>\n",
       "      <td>0.105062</td>\n",
       "      <td>0.113833</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>50</td>\n",
       "      <td>0.112654</td>\n",
       "      <td>0.114876</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>51</td>\n",
       "      <td>0.104883</td>\n",
       "      <td>0.109748</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>52</td>\n",
       "      <td>0.104985</td>\n",
       "      <td>0.106545</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>53</td>\n",
       "      <td>0.105932</td>\n",
       "      <td>0.107047</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>54</td>\n",
       "      <td>0.103812</td>\n",
       "      <td>0.113058</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>55</td>\n",
       "      <td>0.098119</td>\n",
       "      <td>0.112536</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>56</td>\n",
       "      <td>0.108831</td>\n",
       "      <td>0.114904</td>\n",
       "      <td>00:31</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 46: early stopping\n",
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.4, 'num_layers': 2, 'hidden_size': 32})\n",
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.7000000000000001, 'num_layers': 1, 'hidden_size': 8})\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/bobby/anaconda3/lib/python3.8/site-packages/torch/nn/modules/rnn.py:60: UserWarning: dropout option adds dropout after all but last recurrent layer, so non-zero dropout expects num_layers greater than 1, but got dropout=0.7000000000000001 and num_layers=1\n",
      "  warnings.warn(\"dropout option adds dropout after all but last \"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.4, 'num_layers': 2, 'hidden_size': 32})\n",
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.7000000000000001, 'num_layers': 1, 'hidden_size': 8})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/bobby/anaconda3/lib/python3.8/site-packages/fastai/callback/schedule.py:207: RuntimeWarning: More than 20 figures have been opened. Figures created through the pyplot interface (`matplotlib.pyplot.figure`) are retained until explicitly closed and may consume too much memory. (To control this warning, see the rcParam `figure.max_open_warning`).\n",
      "  fig, ax = plt.subplots(1,1)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.291685</td>\n",
       "      <td>0.213042</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.225838</td>\n",
       "      <td>0.170587</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.212556</td>\n",
       "      <td>0.146657</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.190649</td>\n",
       "      <td>0.139308</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.164973</td>\n",
       "      <td>0.128893</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.151940</td>\n",
       "      <td>0.118569</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.139798</td>\n",
       "      <td>0.118466</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.141963</td>\n",
       "      <td>0.109096</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.132526</td>\n",
       "      <td>0.104969</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.132946</td>\n",
       "      <td>0.096877</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.108304</td>\n",
       "      <td>0.102638</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.128075</td>\n",
       "      <td>0.099431</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.095508</td>\n",
       "      <td>0.096375</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.100177</td>\n",
       "      <td>0.090059</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.098663</td>\n",
       "      <td>0.097403</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.096253</td>\n",
       "      <td>0.092888</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.095929</td>\n",
       "      <td>0.089746</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.096378</td>\n",
       "      <td>0.089938</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.094394</td>\n",
       "      <td>0.097958</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.103623</td>\n",
       "      <td>0.092693</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.095475</td>\n",
       "      <td>0.088284</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.091639</td>\n",
       "      <td>0.087368</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.093421</td>\n",
       "      <td>0.085674</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.085078</td>\n",
       "      <td>0.095698</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.092438</td>\n",
       "      <td>0.086637</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.082000</td>\n",
       "      <td>0.081785</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.078250</td>\n",
       "      <td>0.088983</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.085699</td>\n",
       "      <td>0.091392</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.083497</td>\n",
       "      <td>0.085946</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.086492</td>\n",
       "      <td>0.080400</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.085953</td>\n",
       "      <td>0.090451</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.082713</td>\n",
       "      <td>0.088891</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.073169</td>\n",
       "      <td>0.093319</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.090128</td>\n",
       "      <td>0.096420</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.087300</td>\n",
       "      <td>0.087498</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.072014</td>\n",
       "      <td>0.084803</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.075554</td>\n",
       "      <td>0.085090</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.082521</td>\n",
       "      <td>0.090742</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.069674</td>\n",
       "      <td>0.092272</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.078340</td>\n",
       "      <td>0.101583</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 29: early stopping\n",
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.4, 'num_layers': 2, 'hidden_size': 32})\n",
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.7000000000000001, 'num_layers': 1, 'hidden_size': 8})\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/bobby/anaconda3/lib/python3.8/site-packages/torch/nn/modules/rnn.py:60: UserWarning: dropout option adds dropout after all but last recurrent layer, so non-zero dropout expects num_layers greater than 1, but got dropout=0.7000000000000001 and num_layers=1\n",
      "  warnings.warn(\"dropout option adds dropout after all but last \"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.4, 'num_layers': 2, 'hidden_size': 32})\n",
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.7000000000000001, 'num_layers': 1, 'hidden_size': 8})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.352705</td>\n",
       "      <td>0.291999</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.263170</td>\n",
       "      <td>0.230361</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.225657</td>\n",
       "      <td>0.199368</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.206409</td>\n",
       "      <td>0.186998</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.182932</td>\n",
       "      <td>0.171610</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.179353</td>\n",
       "      <td>0.165347</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.166348</td>\n",
       "      <td>0.157048</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.162955</td>\n",
       "      <td>0.152516</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.167852</td>\n",
       "      <td>0.144899</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.163963</td>\n",
       "      <td>0.140931</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.140281</td>\n",
       "      <td>0.138453</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.152126</td>\n",
       "      <td>0.133394</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.130068</td>\n",
       "      <td>0.120568</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.116856</td>\n",
       "      <td>0.122851</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.123049</td>\n",
       "      <td>0.123951</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.126545</td>\n",
       "      <td>0.121892</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.105824</td>\n",
       "      <td>0.120200</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.135814</td>\n",
       "      <td>0.114526</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.097380</td>\n",
       "      <td>0.113661</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.108950</td>\n",
       "      <td>0.112986</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.101063</td>\n",
       "      <td>0.110940</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.092890</td>\n",
       "      <td>0.107832</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.095235</td>\n",
       "      <td>0.105396</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.101816</td>\n",
       "      <td>0.103445</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.101834</td>\n",
       "      <td>0.096399</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.086924</td>\n",
       "      <td>0.107932</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.093379</td>\n",
       "      <td>0.102395</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.100296</td>\n",
       "      <td>0.101033</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.092519</td>\n",
       "      <td>0.100990</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.089740</td>\n",
       "      <td>0.099026</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.084543</td>\n",
       "      <td>0.101616</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.089721</td>\n",
       "      <td>0.104841</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.092530</td>\n",
       "      <td>0.094471</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.087215</td>\n",
       "      <td>0.092987</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.083414</td>\n",
       "      <td>0.109190</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.082652</td>\n",
       "      <td>0.087272</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.083094</td>\n",
       "      <td>0.092994</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.074785</td>\n",
       "      <td>0.109315</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.082091</td>\n",
       "      <td>0.104384</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.076688</td>\n",
       "      <td>0.102214</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.077722</td>\n",
       "      <td>0.094442</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.081059</td>\n",
       "      <td>0.088061</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.081718</td>\n",
       "      <td>0.104263</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.075200</td>\n",
       "      <td>0.108696</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.075921</td>\n",
       "      <td>0.098243</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.071839</td>\n",
       "      <td>0.088321</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 35: early stopping\n",
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.4, 'num_layers': 2, 'hidden_size': 32})\n",
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.7000000000000001, 'num_layers': 1, 'hidden_size': 8})\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/bobby/anaconda3/lib/python3.8/site-packages/torch/nn/modules/rnn.py:60: UserWarning: dropout option adds dropout after all but last recurrent layer, so non-zero dropout expects num_layers greater than 1, but got dropout=0.7000000000000001 and num_layers=1\n",
      "  warnings.warn(\"dropout option adds dropout after all but last \"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.4, 'num_layers': 2, 'hidden_size': 32})\n",
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.7000000000000001, 'num_layers': 1, 'hidden_size': 8})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.348715</td>\n",
       "      <td>0.293049</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.251428</td>\n",
       "      <td>0.230690</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.225728</td>\n",
       "      <td>0.201927</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.203486</td>\n",
       "      <td>0.199510</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.186316</td>\n",
       "      <td>0.184586</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.186518</td>\n",
       "      <td>0.163486</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.178319</td>\n",
       "      <td>0.157128</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.161327</td>\n",
       "      <td>0.151673</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.144006</td>\n",
       "      <td>0.144981</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.144833</td>\n",
       "      <td>0.134488</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.133798</td>\n",
       "      <td>0.134238</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.141670</td>\n",
       "      <td>0.136203</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.127883</td>\n",
       "      <td>0.129170</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.122871</td>\n",
       "      <td>0.120383</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.131029</td>\n",
       "      <td>0.126003</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.112831</td>\n",
       "      <td>0.115303</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.120942</td>\n",
       "      <td>0.118887</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.121415</td>\n",
       "      <td>0.111201</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.117941</td>\n",
       "      <td>0.111146</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.115249</td>\n",
       "      <td>0.123283</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.098979</td>\n",
       "      <td>0.110750</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.108637</td>\n",
       "      <td>0.099574</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.099176</td>\n",
       "      <td>0.104240</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.099221</td>\n",
       "      <td>0.106970</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.097296</td>\n",
       "      <td>0.101344</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.083991</td>\n",
       "      <td>0.105749</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.100850</td>\n",
       "      <td>0.117395</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.086134</td>\n",
       "      <td>0.104799</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.093190</td>\n",
       "      <td>0.100954</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.092094</td>\n",
       "      <td>0.105093</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.092099</td>\n",
       "      <td>0.111914</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.097222</td>\n",
       "      <td>0.101821</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 21: early stopping\n",
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.4, 'num_layers': 2, 'hidden_size': 32})\n",
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.7000000000000001, 'num_layers': 1, 'hidden_size': 8})\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/bobby/anaconda3/lib/python3.8/site-packages/torch/nn/modules/rnn.py:60: UserWarning: dropout option adds dropout after all but last recurrent layer, so non-zero dropout expects num_layers greater than 1, but got dropout=0.7000000000000001 and num_layers=1\n",
      "  warnings.warn(\"dropout option adds dropout after all but last \"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.4, 'num_layers': 2, 'hidden_size': 32})\n",
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.7000000000000001, 'num_layers': 1, 'hidden_size': 8})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.296836</td>\n",
       "      <td>0.242093</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.237828</td>\n",
       "      <td>0.186836</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.212195</td>\n",
       "      <td>0.158790</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.192988</td>\n",
       "      <td>0.136545</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.169311</td>\n",
       "      <td>0.122960</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.160687</td>\n",
       "      <td>0.118723</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.155101</td>\n",
       "      <td>0.113242</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.143536</td>\n",
       "      <td>0.102120</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.142632</td>\n",
       "      <td>0.095718</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.138403</td>\n",
       "      <td>0.103397</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.124176</td>\n",
       "      <td>0.086582</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.128279</td>\n",
       "      <td>0.084058</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.121108</td>\n",
       "      <td>0.081251</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.103534</td>\n",
       "      <td>0.076791</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.116224</td>\n",
       "      <td>0.073780</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.110133</td>\n",
       "      <td>0.080290</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.098006</td>\n",
       "      <td>0.074261</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.106197</td>\n",
       "      <td>0.073988</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.102630</td>\n",
       "      <td>0.072601</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.097911</td>\n",
       "      <td>0.068163</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.102042</td>\n",
       "      <td>0.070535</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.100818</td>\n",
       "      <td>0.071889</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.085768</td>\n",
       "      <td>0.070801</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.098189</td>\n",
       "      <td>0.066825</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.091986</td>\n",
       "      <td>0.070333</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.084894</td>\n",
       "      <td>0.064653</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.093891</td>\n",
       "      <td>0.061653</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.084334</td>\n",
       "      <td>0.061543</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.087748</td>\n",
       "      <td>0.067053</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.089246</td>\n",
       "      <td>0.065731</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.091209</td>\n",
       "      <td>0.060295</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.086681</td>\n",
       "      <td>0.053935</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.074998</td>\n",
       "      <td>0.062427</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.073165</td>\n",
       "      <td>0.059883</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.079223</td>\n",
       "      <td>0.071311</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.074164</td>\n",
       "      <td>0.054875</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.086133</td>\n",
       "      <td>0.048712</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.079654</td>\n",
       "      <td>0.060868</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.076698</td>\n",
       "      <td>0.067838</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.069713</td>\n",
       "      <td>0.070164</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.073630</td>\n",
       "      <td>0.055201</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.062657</td>\n",
       "      <td>0.058347</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.067507</td>\n",
       "      <td>0.054411</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.073676</td>\n",
       "      <td>0.053794</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.074951</td>\n",
       "      <td>0.056949</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.069512</td>\n",
       "      <td>0.053462</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>46</td>\n",
       "      <td>0.071542</td>\n",
       "      <td>0.057310</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 36: early stopping\n",
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.4, 'num_layers': 2, 'hidden_size': 32})\n",
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.7000000000000001, 'num_layers': 1, 'hidden_size': 8})\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/bobby/anaconda3/lib/python3.8/site-packages/torch/nn/modules/rnn.py:60: UserWarning: dropout option adds dropout after all but last recurrent layer, so non-zero dropout expects num_layers greater than 1, but got dropout=0.7000000000000001 and num_layers=1\n",
      "  warnings.warn(\"dropout option adds dropout after all but last \"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.4, 'num_layers': 2, 'hidden_size': 32})\n",
      "ReccurentBlock(1024, **{'reccurent_type': 'GRU', 'bidirectional': True, 'dropout': 0.7000000000000001, 'num_layers': 1, 'hidden_size': 8})\n"
     ]
    },
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.289563</td>\n",
       "      <td>0.245677</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.238526</td>\n",
       "      <td>0.202271</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.206238</td>\n",
       "      <td>0.183816</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.184973</td>\n",
       "      <td>0.168940</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.164252</td>\n",
       "      <td>0.159646</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.163229</td>\n",
       "      <td>0.148352</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.151879</td>\n",
       "      <td>0.144554</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.146399</td>\n",
       "      <td>0.132782</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.139104</td>\n",
       "      <td>0.130388</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.129364</td>\n",
       "      <td>0.125341</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.116944</td>\n",
       "      <td>0.119294</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.134865</td>\n",
       "      <td>0.112589</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.114740</td>\n",
       "      <td>0.115634</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.110078</td>\n",
       "      <td>0.112188</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.116068</td>\n",
       "      <td>0.104110</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.106385</td>\n",
       "      <td>0.104266</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.105124</td>\n",
       "      <td>0.105343</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.120521</td>\n",
       "      <td>0.106859</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.107424</td>\n",
       "      <td>0.096904</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.095875</td>\n",
       "      <td>0.095705</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.082776</td>\n",
       "      <td>0.109855</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.101365</td>\n",
       "      <td>0.094721</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.089293</td>\n",
       "      <td>0.108453</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.084880</td>\n",
       "      <td>0.102450</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.093144</td>\n",
       "      <td>0.096593</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.092022</td>\n",
       "      <td>0.096339</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.088286</td>\n",
       "      <td>0.097838</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.095733</td>\n",
       "      <td>0.089195</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.106305</td>\n",
       "      <td>0.090445</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.084125</td>\n",
       "      <td>0.087667</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.080290</td>\n",
       "      <td>0.092751</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.070487</td>\n",
       "      <td>0.087290</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.083114</td>\n",
       "      <td>0.101241</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.084205</td>\n",
       "      <td>0.083043</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.081732</td>\n",
       "      <td>0.086430</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.077359</td>\n",
       "      <td>0.086915</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.085833</td>\n",
       "      <td>0.083922</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.074327</td>\n",
       "      <td>0.079679</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.065963</td>\n",
       "      <td>0.085316</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.080899</td>\n",
       "      <td>0.100281</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>40</td>\n",
       "      <td>0.062992</td>\n",
       "      <td>0.105842</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>41</td>\n",
       "      <td>0.078141</td>\n",
       "      <td>0.079808</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>42</td>\n",
       "      <td>0.066107</td>\n",
       "      <td>0.097787</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>43</td>\n",
       "      <td>0.088125</td>\n",
       "      <td>0.071313</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>44</td>\n",
       "      <td>0.072154</td>\n",
       "      <td>0.079197</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>45</td>\n",
       "      <td>0.075092</td>\n",
       "      <td>0.079953</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>46</td>\n",
       "      <td>0.075941</td>\n",
       "      <td>0.064341</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>47</td>\n",
       "      <td>0.075311</td>\n",
       "      <td>0.081737</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>48</td>\n",
       "      <td>0.056881</td>\n",
       "      <td>0.088542</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>49</td>\n",
       "      <td>0.068667</td>\n",
       "      <td>0.081750</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>50</td>\n",
       "      <td>0.067933</td>\n",
       "      <td>0.101603</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>51</td>\n",
       "      <td>0.067761</td>\n",
       "      <td>0.077192</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>52</td>\n",
       "      <td>0.068642</td>\n",
       "      <td>0.074159</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>53</td>\n",
       "      <td>0.070546</td>\n",
       "      <td>0.075747</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>54</td>\n",
       "      <td>0.062687</td>\n",
       "      <td>0.066894</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>55</td>\n",
       "      <td>0.062752</td>\n",
       "      <td>0.075407</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>56</td>\n",
       "      <td>0.060842</td>\n",
       "      <td>0.093171</td>\n",
       "      <td>00:08</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 46: early stopping\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEKCAYAAAAIO8L1AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAAn3ElEQVR4nO3deZgU5bn38e/dPRvDDPuAMOwIChIRJKi4kUVFY9TgcjAmwWg0eqKJMXqi7/GoIcmbnJwYl8SomING36AhagwqiUvcFSODiAIKjIgyLLIKDELPdPf9/tEFtGMDjU5P9fT8PtfV11Q9VdV1T9H0Pc9S9Zi7IyIi0lQk7ABERCQ/KUGIiEhGShAiIpKREoSIiGSkBCEiIhkpQYiISEZFYQfQXLp16+b9+/cPOwwRkVZlzpw569y9KtO2gkkQ/fv3p6amJuwwRERaFTN7b3fb1MQkIiIZKUGIiEhGShAiIpKREoSIiGSkBCEiIhkpQYiISEZKECIirdi767byytL1OXlvJQgRkVYonkhy+3PvMP6m57nm4fkkk80/t0/B3CgnItJWLFy5mR8/+AZvrtjE8cN68NPThhOJWLOfRwlCRKSViMUT/O7pWm579h06lRfz+3NGceLw/TBr/uQAShAiIq3C26s3c8m0udSuqWfCqGr+6yvD6Ny+JKfnVIIQEWkFbn5qCWs2b+fub3+ecQd0b5Fz5rST2szGm9kiM6s1s6sybO9nZv80szfM7Fkz6522bZKZLQlek3IZp4hIvquPxRnUvaLFkgPkMEGYWRS4FTgRGAacbWbDmuz2a+Aedz8YmAz8Iji2C3AdcBgwBrjOzDrnKlYRkXwXa0xSWtSyA09zebYxQK27L3X3BuB+4NQm+wwDng6Wn0nbfgLwpLtvcPeNwJPA+BzGKiKS12LxBKVF0RY9Zy4TRDWwPG29LihLNw+YECx/Dag0s65ZHisi0mbE4oVVg8jGFcCxZjYXOBZYASSyPdjMLjSzGjOrWbt2ba5iFBEJXSyepLS4cGoQK4A+aeu9g7Kd3H2lu09w95HAfwZlH2ZzbLDvFHcf7e6jq6oyzpgnIlIQGgqsBjEbGGxmA8ysBJgIzEjfwcy6mdmOGK4GpgbLjwPHm1nnoHP6+KBMRKRNSvVBFEiCcPc4cAmpL/a3gOnuvsDMJpvZKcFu44BFZrYY6AH8PDh2A/BTUklmNjA5KBMRaZNSo5hatokppzfKuftMYGaTsmvTlh8AHtjNsVPZVaMQEWnTUn0QBVKDEBGR5pFMOg2JJCVRJQgREUnTkEgCqAYhIiIfF2sMEkQB3SgnIiLNIBZP3R5WMKOYRESkecTiO2oQShAiIpJmZ4IooDupRUSkGaiJSUREMlITk4iIZKRRTCIiktGOJqYS1SBERCSdmphERCSjHQmiTHdSi4hIuoa4+iBERCQDDXMVEZGMNIpJREQy2nUntWoQIiKSZucwV80HISIi6WLxJMVRIxKxFj2vEoSISJ4LYz5qUIIQEcl7sXiixUcwgRKEiEjei8WThZcgzGy8mS0ys1ozuyrD9r5m9oyZzTWzN8zspKC8v5ltM7PXg9ftuYxTRCSfNcSTLT4XBEBRrt7YzKLArcBxQB0w28xmuPvCtN2uAaa7+21mNgyYCfQPtr3j7ofkKj4RkdaiEJuYxgC17r7U3RuA+4FTm+zjQIdguSOwMofxiIi0SoXYxFQNLE9brwvK0l0PfMPM6kjVHi5N2zYgaHp6zsyOznQCM7vQzGrMrGbt2rXNGLqISP5oq6OYzgbudvfewEnAvWYWAVYBfd19JHA5MM3MOjQ92N2nuPtodx9dVVXVooGLiLSUWDzR4ndRQ24TxAqgT9p676As3fnAdAB3nwWUAd3cPebu64PyOcA7wJAcxioikrdi8WSL30UNuU0Qs4HBZjbAzEqAicCMJvu8D3wJwMyGkkoQa82sKujkxswGAoOBpTmMVUQkb8XiyVBqEDkbxeTucTO7BHgciAJT3X2BmU0Gatx9BvAj4E4z+yGpDutz3d3N7Bhgspk1AkngInffkKtYRUTyWWoUUwENcwVw95mkOp/Ty65NW14IHJnhuAeBB3MZm4hIa9FQgKOYRESkGRTiMFcREWkGscZw7qRWghARyWPuXpB3UouIyGcUTzpJb/n5qEEJQkQkr+2cbrQN3kktIiJ7EGsMphtVDUJERNLtqkEoQYiISJqGHQmiwJ7FJCIin5H6IEREJKNYPNUHoSYmERH5GNUgREQko1ij+iBERCQDNTGJiEhGO5qYdB+EiIh8zK4ahPogREQkTYNulBMRkUx0J7WIiGS0axSTmphERCSNRjGJiEhGsXiSiEFRxFr83DlNEGY23swWmVmtmV2VYXtfM3vGzOaa2RtmdlLatquD4xaZ2Qm5jFNEJF+l5qOOYtbyCaIoV29sZlHgVuA4oA6YbWYz3H1h2m7XANPd/TYzGwbMBPoHyxOBg4BewFNmNsTdE7mKV0QkH8UaE6HcRQ25rUGMAWrdfam7NwD3A6c22ceBDsFyR2BlsHwqcL+7x9z9XaA2eD8RkTYlFk9SEi28BFENLE9brwvK0l0PfMPM6kjVHi7dh2MxswvNrMbMatauXdtccYuI5I1YPFmQNYhsnA3c7e69gZOAe80s65jcfYq7j3b30VVVVTkLUkQkLA1BH0QYctYHAawA+qSt9w7K0p0PjAdw91lmVgZ0y/JYEZGCF4snQhniCrmtQcwGBpvZADMrIdXpPKPJPu8DXwIws6FAGbA22G+imZWa2QBgMPBqDmMVEclLqVFM4SSInNUg3D1uZpcAjwNRYKq7LzCzyUCNu88AfgTcaWY/JNVhfa67O7DAzKYDC4E48D2NYBKRtijWWJhNTLj7TFKdz+ll16YtLwSO3M2xPwd+nsv4RETyXSyeoHP7klDOHXYntYiI7EGYTUxKECIieSwW4igmJQgRkTwWa0yEMpscKEGIiOS1hoSamEREJIMwRzEpQYiI5LG2/KgNERHZjWTS1cQkIiKf1JDYMR+1mphERCTNzvmoVYMQEZF0O+ejVh+EiIiki8XVxCQiIhnsSBC6UU5ERD5mZxOTEoSIiKTb1cSkBCEiIml2jWJSH4SIiKTRKCYREclITUwiIpJRqxjmambtzSwSLA8xs1PMrDi3oYmItG2xxtYxiul5oMzMqoEngG8Cd+cqKBERSX8WU34nCHP3j4AJwO/d/UzgoL0eZDbezBaZWa2ZXZVh+41m9nrwWmxmH6ZtS6Rtm5FlnCIiBSPsUUxFWe5nZnYEcA5wflC2x4jNLArcChwH1AGzzWyGuy/csY+7/zBt/0uBkWlvsc3dD8kyPhGRgrOzDyLPRzFdBlwN/NXdF5jZQOCZvRwzBqh196Xu3gDcD5y6h/3PBu7LMh4RkYK3Y5hrSTScBJFVDcLdnwOeAwg6q9e5+/f3clg1sDxtvQ44LNOOZtYPGAA8nVZcZmY1QBz4pbs/nOG4C4ELAfr27ZvNryIi0mrE4klKohEiEQvl/NmOYppmZh3MrD0wH1hoZlc2YxwTgQfcPZFW1s/dRwNfB24ys0FND3L3Ke4+2t1HV1VVNWM4IiLhS81HHd7dCNmeeZi7bwZOA/5O6q/9b+7lmBVAn7T13kFZJhNp0rzk7iuCn0uBZ/l4/4SISMGLxROh9T9A9gmiOLjv4TRghrs3Ar6XY2YDg81sgJmVkEoCnxiNZGYHAp2BWWllnc2sNFjuBhwJLGx6rIhIIYvFk6GNYILsE8QdwDKgPfB80GeweU8HuHscuAR4HHgLmB50cE82s1PSdp0I3O/u6QlnKFBjZvNIdYb/Mn30k4hIW5BKEOHVILLtpL4FuCWt6D0z+0IWx80EZjYpu7bJ+vUZjnsZ+Fw2sYmIFKqGeCK0yYIg+07qjmb2GzOrCV43kKpNiIhIjoRdg8j2zFOBLcBZwWszcFeughIRkR2jmMLrg8j2TupB7n562vpPzOz1HMQjIiKBWDxB+9Jsv6abX7Y1iG1mdtSOFTM7EtiWm5BERATCb2LKNjVdBNxjZh2D9Y3ApNyEJCIiEP4w12xHMc0DRphZh2B9s5ldBryRw9hERNq0WDzRKjqpgVRiCO6oBrg8B/G0OHfnodfq2LStMexQREQ+JtaYbBV3UmcSztOjmtnSdVu58oE3uObh+Xz8Xj0RkXA1JFrHndSZFMS36aCqCi4/bgiPzFvJA3Pqwg5HRGSnWGMy1Bvl9tgHYWZbyJwIDGiXk4hCcNGxg3hhyVqum7GA0f27MKCb7gEUkXC5e373Qbh7pbt3yPCqdPfwBuc2s2jEuPHfDqGkKML375tLQzCLk4hIWOJJJ+nhzUcNn62JqaD07NiO/z79YN5csYkbnlgUdjgi0sbtnG60lfZBFJwTDtqPcw7ryx3PL+WFJWvDDkdE2rBYY2r+tNY6iqkgXfOVYQzuXsHl0+exvj4Wdjgi0kbtqkEoQeSNdiVRbjl7JJu2NXLlA29o6KuIhEJNTHlqaM8O/J8TD+Tpt9fw1d+9yNQX32WdahMi0oIa8qAGUTAjkZrbpLH9iUYj/Hn2+0x+dCE/n/kWxw6pYsKoar48tAdlxeFldREpfLF4qg8ib++DaMvMjG8e3o9vHt6PxR9s4aHXVvDw3BU8/fYaKsuK+MGXBvPtIwcQjTT/DeXujll+3qjemEiybN1Wlqyppz4W5/ABXenbtTzssEQKTj40MSlBZGFIj0quOvFArjzhAF5Zup47X1jKzx57i8feXMX/nHEw+3ev/MznWLN5OzPfXMWjb6zi9eUf8sUDuzNpbH/GDurarMkimXRWfLgNd4hEUveARCz1cnc2b4+zeXsjm7c1siVYXrM5Ru2aehZ/sIV3120lnvx4v0zfLuUcNbgbR+/fjbGDutGxvLjZ4hVpq2KNQYIIcRSTEsQ+iEaMI/fvxthBXfnb6yu5/pEFnHTzi/zgy4P57jEDKYru2z/kuvoYf5+/mkfnreTVZRtwhwP3q+TM0b15fMEHPLHwA/bvXsGkI/oxYVTvzzRxyJbtjTw4p457Zr3H0nVb9+lYM+jXpZz9u1fy5WE9GNKjgsHdKykpivBy7TperF3H3+auYNq/3idiMKBbe9qVRCmORiiJRigpSv0sLY7QrriI9qVRykuKaF8Spby0iJKiCMmkk/TUjUHuqeWSaIRulaV0q0i9qipK6dCu6FMlzMZEkq2xOFu2p171sTjtS6MMqqpQc6HkpR1NTGH2QVguR+mY2XjgZiAK/MHdf9lk+43AF4LVcqC7u3cKtk0Crgm2/czd/7inc40ePdpramqaMfq9W7slxrV/m8/f56/mc9Ud+dUZB9OnSzlrt8TSXttZWx9jfX0D6+obWL81xoatDayvb6A+FgdgUFV7Tj64F18d0XNnbWR7Y4JH31jFH19expsrNlFZWsRpI6sZ2bcTg7tXMqh7e8pL9p4watfUc8+sZTw4p46tDQkO6dOJ00dV066kKPWFnHQSwU/M6FBWRId2xXQoK6Zju6LUz/LivVZzGxNJXl/+IS8sWcfi1VtoTCRpSCRpiKd+NiaSbG9Msq0hwdaGOB/FEjQk9v2O9ZJohIqyIoxU4gLDbNeTI1OVm/REk/qPtr0x87kiBv27teeAHpUcsF8lB/SopFendrQvLaKiNJXM2pcUEcmiKXFbQ4J19THWb21gfX2MomiELuUldG5fTJf2JbQrjuZt06Hkn5lvruLf//Qaj192DAfs99lbKXbHzOa4++iM23KVIMwsCiwGjgPqgNnA2e6+cDf7XwqMdPfzzKwLUAOMJvUsqDnAoe6+cXfnCyNB7DDzzVVc+7f5rKtvyLg9GjG6ti+hS/sSulWU0rUitdy9soxjh1QxtGflbr843J25yz/knpeXMXP+6o89BqR353YM7l5Bv67tiZiRdCex8y9x5731H/HyO+spiUY4eURPJh3RnxF9OuXiEnwqjYkkHzUkaIgniVjqOpkZEYOIGdsbE6yrb2BdfYx19UHCrY+xNRbHPfXB8CAhECzvON6C9zBSnXyVZcVUlqW+9FM/i9m0rZFFqzez6IMtLP6gnmXrt7K7/w7lJVHKiqMURYziaISiqO1c/ihIDB81JPb4+5YUpRJGj45l9Oncjj5dyunTuZw+XdrRt0s51Z3a7XMtVArXX+fW8cM/z+PZK8bRP4fPh9tTgshlE9MYoNbdlwZB3A+cCmRMEMDZwHXB8gnAk+6+ITj2SWA8cF8O4/3UTvpcTw4f2JX7Xn2foohRVVlKVWUp3SvLqKospVO74qz+As3EzBjVtzOj+nbmfxJJ3lv/EbVrtrDkg3qWrEm9apal8mYkYju/aCNmVJQVccXxQ5g4pi/dKkqb81duFsXRCB3b7f4LsX1pEV0rSjmA3P319JWDe+5c3taQoHZNPWu2bKc+lmqG2hqLUx9LsDUWJxZPEE84jQknnkwGy0nKS6J0DRJ/t/a7/gBIurNhayMbtzaw4aOG1M+tDazatJ036jbxj/mrP9afUxKN0L9bOft3r2D/7pXs372Cwd0rGFjVPtSOSglHofdBVAPL09brgMMy7Whm/YABwNN7OLY6BzE2my7tS/jeF/bP6TmKo5Hgy6OC8cNzeqo2qV1JlM/17gh03Ou+zSGRdFZt2sbyDdtYvvEjlq7dSu2aLSxcuZl/zF/NjtxRHDWG9KhkeK+ODK/uwEHVHRm6XwfalShpFLIdTbAaxQQTgQfcfc919CbM7ELgQoC+ffvmIi6RnIlGjN6dy+nduZwj6PqxbdsbE7wbDCdeuHIzC1Zu4omFq/lzTervpojB/t0rGF7dkeG9OvK53h0Z2rMDFZ9hIIPkl501iAK9D2IF0CdtvXdQlslE4HtNjh3X5Nhnmx7k7lOAKZDqg/j0oYrkl7LiKEN7dmBozw6cMqIXkOqPWrlpO/NXbGL+ik0sWLmZF5as46HXUv+tLBhBdkjvTozs24mRfTtzwH6VFKtfo1Uq9BvlZgODzWwAqS/8icDXm+5kZgcCnYFZacWPA//XzDoH68cDV+cwVpG8Z2ZUd2pHdad2nHDQfjvL12zezvyVm3izbjNvrtjE80vW8dDcVNIoK45wcPWuhHFov85UVeZff5R8UiwYvFGUg5txs5WzBOHucTO7hNSXfRSY6u4LzGwyUOPuM4JdJwL3e9pwKnffYGY/JZVkACbv6LAWkY/r3qGML3Yo44sH9gBSNY0VH25j7vsfpl7LN3LXS8u44/mlQOrGxkP7dWZUv86M7teZA/fb/Sg6CU8snpqPOsx/m5w2WLr7TGBmk7Jrm6xfv5tjpwJTcxacSIEy29W38dWgeSoWTzB/xWZee28jc97byAtL1vHXoJbRr2s5px1SzYRR1fTrqul280WsMRHqCCbIn05qEcmh0qIoh/ZLNTFdQKqWsXzDNl5Zup6/zVvBLU8v4eZ/LuHQfp2ZMKqakz/XS49MCVmqBqEEISItzMzo27Wcvl3LOevzfVi1aRsPz13JQ6/V8Z9/nc9PZizkG4f34wdfGqxEEZIdTUxhUoIQEXp2bMfF4wZx0bEDmb9iM/e+soy7Xn6Xh+bWcdmXBnPO4f00GqqFxeKJ0GsQ+hcXkZ3MjM/17sivzhjBY5cezbCeHbj+kYWMv+l5nn77A82w2IIa4snQ+yCUIEQko2G9OvCn7xzGnd8aTdLhvLtr+NbUV6ldsyXs0NqEfGhiUoIQkd0yM44b1oPHLzuGa08exrzlH3LizS/w3/94m48a4mGHV9BijUlKQm7WU4IQkb0qKYpw3lEDePqKcZx6SDW3PfsOx/3meZ5YsDrs0ApWLB7+MFclCBHJWreKUn595gimf/cIKkqLuPDeOXznj7NZvuGjsEMrOPkwzFUJQkT22ZgBXXj0+0fxnycN5eV31nPcjc/xwJy6sMMqKOqDEJFWqzga4YJjBvLPHx3Lof06c8Vf5nH9jAU0foqZAuWTYo0a5ioirVzPju3447fHcP5RA7j75WV8639fZcPWzLMrSvZiGuYqIoWgKBrhv04exg1njmDO+xs55XcvsnDl5rDDatUa1MQkIoXk9EN785fvHkE84Uy47SUembcy7JBaLXVSi0jBGdGnEzMuPZLhvTpy6X1z+ckjC3ZOfiPZSSadhkQy1MmCQAlCRHKge2UZ0y44nHPH9ueul5Zxxm2zeG/91rDDajXyYT5qUIIQkRwpKYpw/SkHccc3D+W99Vs5+ZYXeeyNVWGH1Srkw3zUoAQhIjl2wkH7MfMHR7N/jwq+N+01rnn4TbY3qslpT3Y0yWkUk4gUvN6dy5n+3SP47jED+X+vvM9pt77E68s/DDusvBWLq4lJRNqQ4miEq08ayl3nfp519Q2cdutLXP7n11m9aXvYoeWdnTUINTGJSFvyhQO78+yV47h43CAefWMVX/j1s/z2n0vU7JRmVw1CCUJE2piK0iJ+PP5Anrr8WI4dUsUNTy7mSzc8x2NvrNKkRKQliOICbmIys/FmtsjMas3sqt3sc5aZLTSzBWY2La08YWavB68ZuYxTRMLRt2s5t3/zUO674HA6tCvme9Ne45Jpc9n0UWPYoYUqX0Yx5WxOajOLArcCxwF1wGwzm+HuC9P2GQxcDRzp7hvNrHvaW2xz90NyFZ+I5I8jBnXl0UuPYsrzS7nhiUXMfX8jN00cyZgBXcIOLRQ7+iAK+Ua5MUCtuy919wbgfuDUJvtcANzq7hsB3H1NDuMRkTwWjRgXjxvEgxePpaQowsQps/jNk4uJt8Gnw7aFPohqYHnael1Qlm4IMMTMXjKzV8xsfNq2MjOrCcpPy3QCM7sw2Kdm7dq1zRq8iIRjRJ9OPPr9o5kwqje3/HMJZ90xq81NSKRhrilFwGBgHHA2cKeZdQq29XP30cDXgZvMbFDTg919iruPdvfRVVVVLRSyiORaRWkRvz5zBLecPZIlH9Rz0s0vcP+r77eZDuxYY+EPc10B9Elb7x2UpasDZrh7o7u/CywmlTBw9xXBz6XAs8DIHMYqInnolBG9mPmDozmougNXPfQm3/jff7WJ2sSuUUyFmyBmA4PNbICZlQATgaajkR4mVXvAzLqRanJaamadzaw0rfxIYCEi0ub06VLOtO8czs+/Npx5yzdx/I3Pc9dL75JMFm5toqHQm5jcPQ5cAjwOvAVMd/cFZjbZzE4JdnscWG9mC4FngCvdfT0wFKgxs3lB+S/TRz+JSNsSiRjnHNaPJ354DIcP7MJPHlnImXfMonZNfdih5US+dFJbobTpjR492mtqasIOQ0RyzN3569wVTH50IdsaEvz+nFF8aWiPsMNqVjc9tZibnlrCu784CTPL6bnMbE7Q3/sJYXdSi4jsEzNjwqjePPHDYxjSo5Lv3juHv79ZWI8Rj8VTkwXlOjnsjRKEiLRK3SvL+NMFhzGiTycuuW8uD89tOgam9Yo1JimNhv/1HH4EIiKfUoeyYu45bwyf79+ZH05/nT/Pfj/skJpFLJ4IfQQTKEGISCvXvrSIu789hqMHV/HjB9/knlnLwg7pM4vFk6GPYAIlCBEpAGXFUe781qEcN6wH1/5tAbc/906rvqkulSDC/3oOPwIRkWZQWhTl9+eM4isH9+SXf3+br/7uRZ55e02rTBQN8UToD+oDJQgRKSDF0Qi3TBzJDWeOYNO2Rr5992zOuH0WL7+zLuzQ9kksngx9LghQghCRAhONGKcf2punfzSOn39tOCs2buPrd/6Lc/7wCnPf3xh2eFmJNaqJSUQkZ4qjEc45rB/PXjmOa08exqLVWzj9tpd5+u0Pwg5tr2LxhBKEiEiulRVHOe+oATx75Rc4qFdHLpk2l/krNoUd1h6pk1pEpAVVlBbxh0mj6dSumPP/OJtVm7aFHdJuaZiriEgL69GhjP899/NsjSU47+4a6mPxsEPKSE1MIiIhGNqzA7eeM4rFH2zh0mmv5eWUprHGpO6kFhEJw7FDqph86kE8s2gtP3lkYd7dK5EvTUxFYQcgIhKGcw7rx3vrP2LK80vp17Wc7xw9MOyQdmrIk05qJQgRabOuGn8g76//iJ899hYrPtzG1ScODf0OZndXH4SISNgiEePmsw/h3LH9ueulZZxx+8u8vz7cOa/jSSfp6E5qEZGwlRZFuf6Ug7j9G6N4d91WvnLLC6FOQJQv042CEoSICADjh/dk5vePZmBVey7+02tc+7f5bG9MtHgcseCcYTd1gRKEiMhOfbqU85eLxvKdowZwz6z3OPP2WXyweXuLxtBmahBmNt7MFplZrZldtZt9zjKzhWa2wMympZVPMrMlwWtSLuMUEdmhpCjCNScP485vjWbp2nq+dutLLFq9pcXOvytBFHAfhJlFgVuBE4FhwNlmNqzJPoOBq4Ej3f0g4LKgvAtwHXAYMAa4zsw65ypWEZGmjhvWg+kXHUHCnTNue5mXalvmkeGxeKqJqdBrEGOAWndf6u4NwP3AqU32uQC41d03Arj7mqD8BOBJd98QbHsSGJ/DWEVEPuGgXh35678fSa9O7Zg09VUemFOX83M27KhBFPid1NXA8rT1uqAs3RBgiJm9ZGavmNn4fTgWM7vQzGrMrGbt2rXNGLqISEqvTu34y8VHcNjALlzxl3nc/NSSnN553SaamLJUBAwGxgFnA3eaWadsD3b3Ke4+2t1HV1VV5SZCEWnzOpQVc9e5Yzh9VG9ufGoxP5o+j03bGnNyrlhj2+ikXgH0SVvvHZSlqwNmuHuju78LLCaVMLI5VkSkxZQURfj1mQdz2ZcH8/DrK/jir5/lgTl1JJPNW5vY1QdR2DWI2cBgMxtgZiXARGBGk30eJlV7wMy6kWpyWgo8DhxvZp2DzunjgzIRkdCYGZd9eQgzLjmKvl3LueIv8zjrjlksXLm52c4Rawt9EO4eBy4h9cX+FjDd3ReY2WQzOyXY7XFgvZktBJ4BrnT39e6+AfgpqSQzG5gclImIhG54dUcevGgsvzr9YJau28rJv32B62csaJZmpx01iJJo+Akipw/rc/eZwMwmZdemLTtwefBqeuxUYGou4xMR+bQiEeOsz/fh+IN6cMMTi/njrGU8+sYqrv3qML56cE/M7FO9784+iEKuQYiItAWdykv46WnDeeSSo+jVqYzv3zeXc++azfINn+6hfxrFJCJSYIZXp+6ZuPbkYdQs28BxNz7HHc+9Q+M+zljX0FYetSEi0pZEI8Z5Rw3gycuP5aj9q/jF39/mlN+9xGvvb8z6PdrKndQiIm1Sr07t+MOk0dz+jUPZsDXGhN+/zNlTXuGZRWv2epNdLJ4kGjGKCr2TWkSkLRs/fD+OGtyNaf96j6kvLuPbd81mSI8KLjh6IKceUp3xkd6xPJluFFSDEBHJqYrSIi48ZhDP/8cX+M1ZI4iYceUDb3D0r57mln8uoXZN/cf2jzXmx3SjoBqEiEiLKCmKMGFUb742spoXlqxjyvNL+c2Ti/nNk4sZ3L2C8cP3Y/zw/djemMyLEUygBCEi0qLMjGOGVHHMkCpWbdrGEws+4O/zV3HrM7X89ulaIga9O5eHHSagBCEiEpqeHdsxaWx/Jo3tz/r6GE8u/IB/LFjN4O4VYYcGKEGIiOSFrhWlTBzTl4lj+oYdyk750RMiIiJ5RwlCREQyUoIQEZGMlCBERCQjJQgREclICUJERDJSghARkYyUIEREJCPb26NnWwszWwu8l1bUEdi0D+vdgHU5CK3peZrzmD3tt7ttmcr35Vrl6jrtLrbmOCZX16lpmT5Tuy9rS5+pPW3Px2vVz92rMm5x94J8AVP2cb2mJeJozmP2tN/utmUq35drlavrlMtrlavrlOHa6DOVxWcow3UrqM9Ua71WmV6F3MT0yD6ut1QczXnMnvbb3bZM5YV+rXJ1nZqWtfbrtLf99JnKfp/WeK0+oWCamD4rM6tx99Fhx5HvdJ2yp2uVHV2n7LX0tSrkGsS+mhJ2AK2ErlP2dK2yo+uUvRa9VqpBiIhIRqpBiIhIRkoQIiKSkRKEiIhkpASxF2Z2tJndbmZ/MLOXw44nn5lZxMx+bma/NbNJYceTz8xsnJm9EHy2xoUdTz4zs/ZmVmNmJ4cdSz4zs6HB5+kBM7u4Od6zoBOEmU01szVmNr9J+XgzW2RmtWZ21Z7ew91fcPeLgEeBP+Yy3jA1x7UCTgV6A41AXa5iDVszXSsH6oEyCvRaNdN1AvgxMD03UeaHZvqueiv4rjoLOLJZ4irkUUxmdgyp/4T3uPvwoCwKLAaOI/UfczZwNhAFftHkLc5z9zXBcdOB8919SwuF36Ka41oFr43ufoeZPeDuZ7RU/C2pma7VOndPmlkP4Dfufk5Lxd9Smuk6jQC6kkqk69z90ZaJvmU113eVmZ0CXAzc6+7TPmtcRZ/1DfKZuz9vZv2bFI8Bat19KYCZ3Q+c6u6/ADJWYc2sL7CpUJMDNM+1MrM6oCFYTeQw3FA11+cqsBEozUmgIWumz9Q4oD0wDNhmZjPdPZnLuMPQXJ8pd58BzDCzxwAliE+hGlietl4HHLaXY84H7spZRPlrX6/VQ8Bvzexo4PlcBpaH9ulamdkE4ASgE/C7nEaWX/bpOrn7fwKY2bkEta6cRpdf9vUzNQ6YQOoPjpnNEUBbTBD7zN2vCzuG1sDdPyKVTGUv3P0hUglVsuDud4cdQ75z92eBZ5vzPQu6k3o3VgB90tZ7B2XySbpW2dO1yo6uU/ZCv1ZtMUHMBgab2QAzKwEmAjNCjilf6VplT9cqO7pO2Qv9WhV0gjCz+4BZwAFmVmdm57t7HLgEeBx4C5ju7gvCjDMf6FplT9cqO7pO2cvXa1XQw1xFROTTK+gahIiIfHpKECIikpEShIiIZKQEISIiGSlBiIhIRkoQIiKSkRKEFDQzq2/h8zXLnCHBfBGbzOx1M3vbzH6dxTGnmdmw5ji/CChBiOwTM9vj88vcfWwznu4Fdz8EGAmcbGZ7e8b/aaSeeirSLJQgpM0xs0Fm9g8zmxPM6nZgUP5VM/uXmc01s6eCuRows+vN7F4zewm4N1ifambPmtlSM/t+2nvXBz/HBdsfCGoAfzIzC7adFJTNMbNbzGyPcxy4+zbgdVJP98TMLjCz2WY2z8weNLNyMxsLnAL8T1DrGLS731MkW0oQ0hZNAS5190OBK4DfB+UvAoe7+0jgfuA/0o4ZBnzZ3c8O1g8k9bjuMcB1Zlac4TwjgcuCYwcCR5pZGXAHcGJw/qq9BWtmnYHB7HqE+kPu/nl3H0HqEQznu/vLpJ7Tc6W7H+Lu7+zh9xTJih73LW2KmVUAY4G/BH/Qw64Je3oDfzaznkAJ8G7aoTOCv+R3eMzdY0DMzNYAPfjk1KGvuntdcN7Xgf6kZg1b6u473vs+4MLdhHu0mc0jlRxucvfVQflwM/sZqbkkKkg9q2dffk+RrChBSFsTAT4M2vab+i2p6T9nBJOvXJ+2bWuTfWNpywky/1/KZp89ecHdTzazAcArZjbd3V8H7gZOc/d5wUQ64zIcu6ffUyQramKSNsXdNwPvmtmZAJYyItjckV3P25+UoxAWAQPTppf8t70dENQ2fgn8OCiqBFYFzVrpc1lvCbbt7fcUyYoShBS68uDxyTtel5P6Uj0/aL5ZAJwa7Hs9qSaZOcC6XAQTNFP9O/CP4DxbgE1ZHHo7cEyQWP4L+BfwEvB22j73A1cGneyD2P3vKZIVPe5bpIWZWYW71wejmm4Flrj7jWHHJdKUahAiLe+CoNN6AalmrTvCDUckM9UgREQkI9UgREQkIyUIERHJSAlCREQyUoIQEZGMlCBERCQjJQgREcno/wNmpSaJkTAz5AAAAABJRU5ErkJggg==\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"
    },
    {
     "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"
    },
    {
     "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"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZIAAAEKCAYAAAA4t9PUAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAAtxUlEQVR4nO3deXxcZdn/8c+VrWmSJmmbdF+S0nQHWhpKS1kKsqstIAJVERDBDRXcl8dHf+4+Pioi+CAg4sIqKlZBCkIFaWlpCi003Ztuabok6ZKl2XP9/sgUQ+iSZGZ6Jpnv+/WaF5kz58xcd0PyzX3uc+7b3B0REZHuSgi6ABER6dkUJCIiEhYFiYiIhEVBIiIiYVGQiIhIWBQkIiISlqSgC4iUnJwcz8vLC7oMEZEeZcWKFRXunhvOe/SaIMnLy6OoqCjoMkREehQz2xbue+jUloiIhEVBIiIiYYlqkJjZJWa23sw2mdlXjvD6z8xsZeixwcwOtHvtejPbGHpcH806RUSk+6I2RmJmicDdwIVAKbDczBa4+5rD+7j77e32/zQwLfT1AOCbQCHgwIrQsfujVa+IiHRPNHskM4BN7l7i7o3Ao8C8Y+w/H3gk9PXFwHPuvi8UHs8Bl0SxVhER6aZoBslwYEe756Whbe9gZqOBfOCFrhxrZreYWZGZFZWXl0ekaBER6ZpYGWy/FnjC3Vu6cpC73+vuhe5emJsb1mXQEoby6gZe2VxJeXVD0KWISACieR/JTmBku+cjQtuO5FrgUx2OndPh2H9FsDYJw7PFu1myuZL1u6vZsKeaytpGANJSErn57DHccs4Y0vv0mluUROQ4ovnTvhwoMLN82oLhWuADHXcyswlAf+CVdpsXAt83s/6h5xcBX41irdJJdz6/kZ8+t4G+yYmMG9KPCyYOZtyQfuTnpPGnFTv5+fMbeWjZdm6/sIBrCkeSlBgrnV4RiZaoBYm7N5vZrbSFQiLwgLsXm9m3gSJ3XxDa9VrgUW+3VKO77zOz79AWRgDfdvd90apVjs/d+dk/N3Ln8xu58rTh/M/7TnlHSJw/YTAf3b6f7z+9lq//ZTUPvLyFH73vFArzBkSsjpZWx4CEBIvYe4pIeKy3LLVbWFjomiIlOtydnzy7gbsWbeLqwhH84MpTSDzGL3J357k1e/jOU2uoa2zh2dvPZUB6Src//+ChJv66aiePLd9BcVnVW9sTDBLMyMnow0fPzueDZ4ymb0pitz9HJB6Z2Qp3LwzrPRQkcizuzo+eWc89L25m/oyRfO/ykzvdG1i3u4r3/uJlLpkylF/Mn9alz21tdZaWVPJY0Q7+sXo3jc2tTBqaybsmDiIxwWhtdVodWt1ZueMASzZXkpORwsfPPUmBItIFkQgSjYjKUbW2Ot9/ei33v7yFD80cxbfnTunSKaUJQzL5zPkF/OS5Dbz75CFcMmVop46rb2rhE39YwaL15WSmJnHt6SO5unAkU4ZnHfWYV7fs4+fPb+C7T63lnhc389Gzx/CeU4Yyon9ap+sVke5Rj0SOqLahmc8/vopnindzw5l5fPO9kzDr+rhEU0srl9+9mD1V9Tx3+7n0P84prvqmFm75/Qr+vbGcr182kQ/NHE1qcud7F4cDZfGmSgAmDs3kwomDeNfEwZw8PEtjKyId6NRWOwqSyNmx7xA3/66IDXuq+dplE7nprPxuhchha3dVMfeul7l0ylDuPMYprvYh8qMrT+Hq00cedd/j2Vxew/Nr9/DPNXsp2raPVofh2X2578OFTBqW2e33FeltFCTtKEgi45XNlXzyoRW0tDq/+MBpnDsuMjd6Hr5s+J4PTeeSKUPe8XokQ6Sj/bWNLFq/lx8vXE9jcyuP3jKTgsH9Ivb+7T/nmwuKSUwwLp48mHPG5ZKWorPHEtsUJO0oSMLT0NzCI8u2892n1jJ6YBr3X386+TnpEXv/ppZW5t21mL3VDTx3+zlvO8UVzRBpr6S8hmvuXYoBj39sFnkRbN/e6nquu/9VtlTWkpaSyIFDTfRJSuDsghwumjyEKcOyGJiRQv+0FFKSdG+NxA4FSTsKkq6rbWjmX+vLWVi8mxfW7aWmoZnzJwzijmunkpmaHPHPW1PWdoorNTmRxASjuaWV5lanudVpdY9qiBy2YU811967lNSkBB772CxGDgh/MH7ngTo+eN9S9lY3cP+HC5mRP4BXt+7j2eI9PLdmDzsP1L1t/36pSQxMT6FPUiKO46Grz5y202/nFORy7vhcCgZlhHVKUaQzFCTtKEg6p7mllefX7eWPRaW8tLGcxuZWBqancOGkwVw8ZQjnFuRGdUD62eLdvLihnKQEIykxIfRfozBvAOeNHxS1z22vuOwg8+9dSlZaMo9/bBZDs/p2+722VNTyofuXUVXfxIM3zmD66P5ve93dWbOriu2Vh6isbWRf6FFZ20hjcwuGYUbbA2Pj3mo27KkBYGhWKueOy+Wq6SMielOnSHsKknYUJMdWdqCOR5fv4LHl29lT1cCgfn149ylDuWTyEArzBhzzBsPeaOWOA3zo/mUMzEhh3tThjB6QRl5OGqMHpjMwPaVTPYH1u6v50K+X0dLq/O4jM455eXJXlB2o46UN5by4oZyXN1ZQ39zC/defHrHxKpH2FCTtKEiOrKKmga/86U1eWLcHB84pyOUDZ4ziXRMGxf08WMu37uPLT7zB1spaWtv9GKSnJDIgI4XsvilkpyWT2TeZzNQkahpaqKxpeKtHsa+2kYHpKTz00TOiMngPcLCuifn3LqWkoobf33QGp6tnIhGmIGlHQfJOra3ODQ8uZ1lJJR89O59rTx8VkTGB3qaxuZXS/YfYVnmIrZW1bN93iAOHmjhwqJGDdU0cqGuiqq6ZjD6JDMzow4D0FAamp5Dbrw9XF46M+r9pRU0DV//qFcqrGnjklpkR6/mIgILkbRQk7/SbxVv4f39bw3fmTea6WXlBlyNhKDtQx/vveYW6phYe/9hMxg5q6wG5O6tKD/K3VWWUVzdw1fQRnDU2RzdeSqcpSNpRkLzdut1VzL1rMWeNzeHX1xfq6p9eYEtFLe+/5xWSEowfXXUKr2yu5Kk3y9ixr47kRCOjTxL7DzWRn5POdTNH877pI8jqG/mr76R3UZC0oyD5j/qmFubdtZjK2gaeue0ccjL6BF2SRMjaXVVc86tXqKpvJjHBOGtsDu85ZSgXTR5CanICz6zezW+XbOW17Qfom5zI/Bmj+Pq7J8bdxRTSeZq0UY7of55Zz/o91fzmxtMVIr3MxKGZPP7xWbxRepALJg5+x/T886YOZ97U4bxZepAHFm/hgcVbyE5L5jPvKgioYokHCpJe5sUN5TyweAs3nJl3wu7LkBNrwpBMJgw59nxhJ4/I4qdXn0qrO3f8cwNn5A/gjDEDT1CFEm/i+/rPXqaypoEv/HEV4wZn8JVLJwRdjgTMzPjeFSczakAan310JftrG4MuSXopBUkv4e58+U9vcPBQE3dcM61LU69L75XRJ4m7PnAalbVtf2T0ljFRiS0KkihYWLybh5dtp7G5tcvHtrY6NQ3NXT7uD8u288+1e/nypRM0Tbq8zZThWXz10ok8v24vDyzeGnQ50gtpjCTC3J2v/+VNKmoauefFzXzh4vG85+Shnbqu/7Xt+/nan99k3e5qJgzpx8wxA5k5ZgAz8gcec83zTXur+e7f13B2QQ43npkXwdZIb3Hj7DyWbK7gh/9Yy4y8AZw8Qjc1SuSoRxJh6/dUU1HTyAfOGEVaSiKfeeR15t79Mi9vrDjqMVX1TXzjydW87/+WcLCuiVvPG0tuvz48tnwHH//Da5z2nee44peLKS47+I5jG5pb+PQjK0nvk8RP3n+qbkSTIzIzfnzVqeRk9OHWR16jvqkl6JKkF1GPJMIOL/F663ljGZKZyl9X7eR/F27gQ79exoQh/ZgyPItJQzOZNCyTiUMzeWVzBd9cUEx5dQM3nJnH5y8aT0aftm9LY3Mrb+48yNKSSh5cspXL717MbReM42PnjHlrnqwfP7Oetbuq+PX1hQzKTA2s3RL7+qen8IMrT+aG3yxnwaoyri6M7pT9Ej90Q2KE3fTgckoqaln0hTlvbWtobuHhZdt5Yd1e1u6qoqLm7VfPTBqayQ/fdzKnjMg+6vvur23kv55czVNv7mLaqGx+evVUSvcf4rpfv8p1M0fzncunRKlF0pu4Oxff8RLJiQn8/dNnacYD0Q2JsaappZWlJZVcPm3427b3SUrkxtn53Dg7H2hbTW9NWRXFZVVkpyVzTeHI487E2z89hbs+MI2LVg3mG0+u5rKf/5u+KYmMHZTB1y6bGLU2Se9iZnx4Vh7/9eRqXtu+n+mjNZuwhE9jJBH0RukBahtbmD0255j7DeqXypzxg/jUeWP54BmjOz2du5kxb+pwnr39XE7PH0BdYws/v3YqfVN0qa903hXThtMvNYkHl2wLuhTpJdQjiaDFmyoxg1lRvoN4SFYqv73xdOqaWkhL0bdQuia9TxJXF47kt0u2sufdExmssTUJk3okEbR4UwWTh2XS/xiX6kaKmSlEpNuumzmaFnceXrY96FKkF1CQRMihxmZe276f2Scd+7SWSCzIy0lnzrhcHn61ezfOirSnIImQ5Vv309TinHmc8RGRWHH9mXmUVzfwj9W7gi5FejgFSYQs2VRBcqJxel7/oEsR6ZRzCnLJz0nnt0u2Bl2K9HAKkghZvLmC00b117iF9BgJCcZ1M0fz2vYDvFn6zlkTRDpLQRIB+2sbKS6rOu5lvyKx5qrCEaSlJPKgeiUSBgVJBLxSUok7zB6rhYOkZ8lMTeZ9p43gb2+UUV7dEHQ50kNFNUjM7BIzW29mm8zsK0fZ52ozW2NmxWb2cLvt/xPattbM7rQYnsth8aYK0lMSjznFiUis+shZ+bS0Oncv2hR0KdJDRS1IzCwRuBu4FJgEzDezSR32KQC+Csx298nAbaHtZwKzgVOAKcDpwLnRqjVcSzZXMnPMQJI7eYe6SCzJz0nnmtNH8tCybWyrrA26HOmBovmbbwawyd1L3L0ReBSY12Gfm4G73X0/gLvvDW13IBVIAfoAycCeKNbabTsP1LGlolaX/UqPdtu7CkhKSOAnz24IuhTpgaIZJMOBHe2el4a2tTcOGGdmi81sqZldAuDurwCLgF2hx0J3XxvFWrtt8aa2dUY0PiI92aDMVG46K58Fq8pYvVNXcEnXBH0uJgkoAOYA84H7zCzbzMYCE4ERtIXP+WZ2dseDzewWMysys6Ly8vITWPZ/LNlUQU5GCuMH9wvk80Ui5ZZzx9A/LZkfPbMu6FKkh4lmkOwE2q+cMyK0rb1SYIG7N7n7FmADbcFyBbDU3WvcvQb4BzCr4we4+73uXujuhbm5uVFpxLG0tjovb6pk1kk5WtdBerzM1GRuPb+Af2+s4N8bg/nDTHqmaAbJcqDAzPLNLAW4FljQYZ8naeuNYGY5tJ3qKgG2A+eaWZKZJdM20B5zp7aKy6qoqGngvPEnPsREouFDM0cxPLsvP3pmHa2tvWPRO4m+qAWJuzcDtwILaQuBx9292My+bWZzQ7stBCrNbA1tYyJfdPdK4AlgM/AmsApY5e5/i1at3fXCur2YwbnjFCTSO/RJSuQLF49j9c4q/v7m2+fgcnfqGrXWu7yTltoNw+V3LwbgyU/NPqGfKxJNra3Ou3/xMtX1Tcw9dRjbKg+xtbKWbZWHqGlo5rcfmaE/nnqRSCy1G/Rge49VWdPAqtIDnDd+UNCliERUQoLx1UsnULq/jntfKmHNripy+/XhqukjGJCewuPLdxz/TSSuaIbBbnpxQznucP4EBYn0PueMy+X1b1xIv9Skty0F7e48unwH1fVN9EtNDrBCiSXqkXTTovXl5GT0YfKwzKBLEYmK/ukpbwsRgLlTh9PQ3Mpza2Ly/mAJiIKkG5pbWnlx/V7mjM8lIUGX/Ur8OG1UNiP69+WvK8uCLkViiIKkG17fcYCq+mad1pK4Y2bMPXUYL2+qoKJGswVLGwVJN7ywbi9JCcZZBZpfS+LPvKnDaWl1nn5TS/RKGwVJNyxat5fCvP5karBR4tD4If2YMKQfC3R6S0IUJF2062Ad63ZX67JfiWvvPXUYRdv2s2PfoaBLkRigIOmiReva5iDS+IjEs7mnDgPgb2+oVyIKki5btH4vw7P7MnZQRtCliARm5IA0po/ur9NbAihIuqShuYXFmyo4f8IgzfYrcW/e1GGs213N+t3VQZciAVOQdMGrW/ZxqLGF8yZoniGRy04eSmKCsWBVx9UhJN4oSLpg0bpy+iQlMGuMLvsVycnow+yxOfx1ZRm9ZfJX6R4FSRf8a/1eZp00kL4piUGXIhIT5p06jNL9dby2fX/QpUiAFCSddOBQIyUVtZyRr7XZRQ67aPJg0lISeeRVzQgczxQknbSmrApAkzSKtNMvNZn3nTaCBSvLNGVKHFOQdFKxgkTkiK4/czSNLa08+ur2oEuRgChIOmnNriqGZKYyMKNP0KWIxJSxg/pxdkEOv1+6jaaW1qDLkQAoSDqpuOygeiMiR3Hj7Dz2VDXwzOrdQZciAVCQdEJ9Uwuby2sVJCJHMWfcIEYPTOO3S7YGXYoEQEHSCet2V9PS6kwalhV0KSIxKSHBuH5WHkXb9vNm6cGgy5ETTEHSCcVlbT8Y6pGIHN1VhSNIS0nkQfVK4o6CpBOKy6rITE1iRP++QZciErMyU5O5avoI/rZKlwLHGwVJJxSXVTFpWKYmahQ5jg/PyqOxpZVHlulS4HiiIDmO5pZW1u2qYrLGR0SOa+ygDM4Zl8sflulS4HiiIDmOLRW1NDS3anxEpJNuPLPtUmCt6R4/FCTH8Z872tUjEemMc8flMn5wP3723AYam9UriQcKkuMoLjtISlICJ+WmB12KSI+QkGB85dIJbK08xCOaNiUuKEiOo7isiglD+pGUqH8qkc6aMz6XWWMG8vPnN1Jd3xR0ORJl+u14DO5OcVmVxkdEusjM+OplE9hX28ivXiwJuhyJMgXJMew8UMfBuibd0S7SDaeMyGbuqcO4/+USdh+sD7ociSIFyTFo6niR8Hzx4vG0tDo/e25D0KVIFClIjqG4rIoEg4lDFCQi3TFyQBofnpXHH1fsYP3u6qDLkShRkBzDmrIqxuRmaI12kTDcet5Y0vsk8aNn1gVdikRJVIPEzC4xs/VmtsnMvnKUfa42szVmVmxmD7fbPsrMnjWztaHX86JZ65GsKTvIpKHqjYiEo396CreeN5YX1u1lyeaKoMuRKIhakJhZInA3cCkwCZhvZpM67FMAfBWY7e6Tgdvavfw74MfuPhGYAeyNVq1Hsr+2kbKD9RofEYmA68/MY2hWKv+7cD3uHnQ5EmHR7JHMADa5e4m7NwKPAvM67HMzcLe77wdw970AocBJcvfnQttr3P1QFGt9B93RLhI5qcmJ3Hr+WF7bfoAXN5QHXY5EWDSDZDiwo93z0tC29sYB48xssZktNbNL2m0/YGZ/NrPXzezHoR7O25jZLWZWZGZF5eWR/Z9Ta5CIRNb7p49kRP++/Oy5DeqV9DJBD7YnAQXAHGA+cJ+ZZYe2nw18ATgdGAPc0PFgd7/X3QvdvTA3NzeihRWXVTEsK5X+6SkRfV+ReJWSlMBnzi9gVelBXlh3Qs9US5RFM0h2AiPbPR8R2tZeKbDA3ZvcfQuwgbZgKQVWhk6LNQNPAqdFsdZ3KC47qBsRRSLsitOGM3pgGj9Vr6RXiWaQLAcKzCzfzFKAa4EFHfZ5krbeCGaWQ9sprZLQsdlmdribcT6wJoq1vmX1zoPc+JtX2Vxey/TR/U/ER4rEjeTEtl5JcVkVC4v3BF2OREjUgiTUk7gVWAisBR5392Iz+7aZzQ3tthCoNLM1wCLgi+5e6e4ttJ3Wet7M3gQMuC9atQJsLq/hUw+/xnt+8TKvbT/Aly+ZwEfOyovmR4rEpXlThzEmJ507/rmB1lb1SnoD6y3dy8LCQi8qKurycRU1DfzPM+t4YkUpqcmJ3HRWPh89ewxZfZOjUKWIAPx15U4+++hK7v7Aabz7lKFBlxPXzGyFuxeG8x5BD7YHLjkhgRfWlXPDmfm89KXz+PxF4xUiIlH2nlOGMXZQBnf8cwMt6pX0eHEfJFlpybz85fP47/dOIiejT9DliMSFxATjtgsK2Li3hr+/URZ0ORKmTgWJmaWbWULo63FmNtfMes2f7anJmktL5ES7bMpQJgzpx8//uZHmFi3J25N1tkfyEpBqZsOBZ4HrgAejVZSI9H4JCcZtF4yjpKKWv7ze8c4A6Uk6GyQWmqLkSuCX7v5+YHL0yhKReHDx5MFMGZ7JnS9spEm9kh6r00FiZrOADwJPhbbpfJCIhMXM+PyF49mxr44/FpUGXY50U2eD5DbaZun9S+hekDG03fchIhKWOeNzmTYqm7te2EhDc0vQ5Ug3dCpI3P1Fd5/r7j8KDbpXuPtnolybiMSBw72SsoP1PPrqjuMfIDGns1dtPWxmmWaWDqwG1pjZF6NbmojEi9ljB3JG/gDuWrSJukb1Snqazp7amuTuVcDlwD+AfNqu3BIRCZuZ8fmLxlNe3cAflm4Luhzpos4GSXLovpHLCc3WC+h2VBGJmBn5Azi7IIf/e3EztQ3NQZcjXdDZIPkVsBVIB14ys9FAVbSKEpH49LkLx7GvtpHfvrI16FKkCzo72H6nuw9398u8zTbgvCjXJiJxZtqo/pye159nVu8OuhTpgs4OtmeZ2U8PL2trZj+hrXciIhJRhXkDWFNWRX2TBt17is6e2noAqAauDj2qgN9EqygRiV/TRmbT3OoUlx0MuhTppKRO7neSu7+v3fP/Z2Yro1CPiMS5qaOyAXh9+wGmjx4QbDHSKZ3tkdSZ2VmHn5jZbKAuOiWJSDwb1C+V4dl9eX3HgaBLkU7qbI/k48DvzCwr9Hw/cH10ShKReDd1VDYrtx8IugzppM5etbXK3U8FTgFOcfdpwPlRrUxE4ta0kdnsPFDH3qr6oEuRTujSConuXhW6wx3gc1GoR0SEaYfHSXR6q0cIZ6ldi1gVIiLtTB6WRVKCsVJB0iOEEySaIkVEoiI1OZFJwzI1TtJDHHOw3cyqOXJgGNA3KhWJiABTR2bzpxWltLQ6iQk6ARLLjtkjcfd+7p55hEc/d+/sFV8iIl02bVQ2tY0tbNxbHXQpchzhnNoSEYmaqSP7A203JkpsU5CISEzKG5hGdlqyxkl6AAWJiMQkM2PqyGxe37E/6FLkOBQkIhKzpo7MZuPeGqrrm4IuRY5BQSIiMWvaqP64wxulmgk4lilIRCRmTR2RDaAbE2OcgkREYlZWWjJjctN5fbvGSWKZgkREYtrUkdms3HEAd02mEasUJCIS06aNzKaippHS/VoCKVZFNUjM7BIzW29mm8zsK0fZ52ozW2NmxWb2cIfXMs2s1MzuimadIhK7po0K3ZiocZKYFbUgMbNE4G7gUmASMN/MJnXYpwD4KjDb3ScDt3V4m+8AL0WrRhGJfeOH9KNPUoJuTIxh0eyRzAA2uXuJuzcCjwLzOuxzM3C3u+8HcPe9h18ws+nAYODZKNYoIjEuOTGBk4dnsVI3JsasaAbJcGBHu+eloW3tjQPGmdliM1tqZpcAmFkC8BPgC1GsT0R6iCnDs1i7q5qWVg24x6KgB9uTgAJgDjAfuM/MsoFPAk+7e+mxDjazW8ysyMyKysvLo12riARkyvAs6ppa2FJRE3QpcgTRnAp+JzCy3fMRoW3tlQLL3L0J2GJmG2gLllnA2Wb2SSADSDGzGnd/24C9u98L3AtQWFioP1VEeqnJwzIBKC6rYuygfgFXIx1Fs0eyHCgws3wzSwGuBRZ02OdJ2nojmFkObae6Stz9g+4+yt3zaDu99buOISIi8WPsoAxSkhJYvVNTpcSiqAWJuzcDtwILgbXA4+5ebGbfNrO5od0WApVmtgZYBHzR3SujVZOI9EzJiQlMGNKP4rKqoEuRI4jqKofu/jTwdIdt/93uawc+F3oc7T0eBB6MToUi0lNMHpbJU2/swt0x09K7sSTowXYRkU6ZPCyLqvpm3eEegxQkItIjtB9wl9iiIBGRHmHCkEwSDIrLNOAeaxQkItIj9E1JZOygDPVIYpCCRER6jMnDstQjiUEKEhHpMSYPy2RPVQPl1Q1BlyLtKEhEpMeYPCwL0DhJrFGQiEiPMUlXbsUkBYmI9BhZfZMZNSBNPZIYoyARkR5l8rBM9UhijIJERHqUKcOz2FZ5iKr6pqBLkRAFiYj0KIfHSdaoVxIzFCQi0qNMCV25pSnlY4eCRER6lNx+fRjUr496JDFEQSIiPc6U4VkacI8hChIR6XEmD8tkU3kN9U0tQZciKEhEpAeaPCyTllZn3e7qoEsRFCQi0gNN1oB7TFGQiEiPM6J/X7L6JmucJEYoSESkxzEzJg/LZNG6vazccSDocuKegkREeqTPvquAVneu+OVivvHkag7W6U73oChIRKRHOmPMQJ7//LlcPyuPh5Zt410/eZG/rtyJuwddWtyx3vKPXlhY6EVFRUGXISIBWL3zIF/7y5u8UXqQCUP6cVJuBsOyUxma1Zdh2amcOjKboVl9gy4zJpnZCncvDOs9FCQi0hu0tDoPv7qdf7y5i10H6yk7UEdDcysA/VKTePlL55OVlhxwlZG3eudB+qYkclJuRreOj0SQJIVzsIhIrEhMMK6bOZrrZo4GwN3Zf6iJ17bt56O/K+KJ10q56az8gKuMrJqGZj718GskJybw7G3nkJBggdShMRIR6ZXMjAHpKVwwaTDTRmXz0LJtvW785FsLitmx7xDfv+LkwEIEFCQiEgc+eMZoSsprWVqyL+hSIubvb5TxxIpSPnXeWGbkDwi0FgWJiPR67zllKFl9k3lo2bagS4mIsgN1fO3PbzJ1ZDafeVdB0OUoSESk90tNTuSq6SNYWLyb8uqGoMsJS0urc/tjK2lpdX5+7VSSE4P/NR58BSIiJ8AHzhhFU4vzeNGOoEsJy69e2syyLfv41tzJjB6YHnQ5gIJEROLESbkZzBozkEde3U5La88cdF+14wA/fXYD7z55KFdNHxF0OW9RkIhI3PjgzFGU7q/jpY3lQZfSLd97ei05GX34/hUnYxbcVVodKUhEJG5cNGkIORl9eGhpzxt0d3fW7qriwkmDY+7GSgWJiMSNlKQErjl9BC+s28vOA3VBl9MlFTWNVNc3MyY3NsZF2otqkJjZJWa23sw2mdlXjrLP1Wa2xsyKzezh0LapZvZKaNsbZnZNNOsUkfhx7emjcOCxV7cHXUqXlJTXAJCfE0dBYmaJwN3ApcAkYL6ZTeqwTwHwVWC2u08Gbgu9dAj4cGjbJcAdZpYdrVpFJH6MHJDGnHG5PLp8B42hubh6gi0VtQDdnlMrmqLZI5kBbHL3EndvBB4F5nXY52bgbnffD+Due0P/3eDuG0NflwF7gdwo1ioiceSG2fnsrW7gjyt6zqXAJRW1pCQlMCw79mYxjmaQDAfaf5dKQ9vaGweMM7PFZrbUzC7p+CZmNgNIATYf4bVbzKzIzIrKy3vmVRgicuKdU5DDtFHZ3PXCJhqaW4Iup1NKymvIG5hGYoBzah1N0IPtSUABMAeYD9zX/hSWmQ0Ffg/c6O7v6IO6+73uXujuhbm56rCISOeYGZ+7cBy7Dtbz+PKe0SspqahlTE7sndaC6AbJTmBku+cjQtvaKwUWuHuTu28BNtAWLJhZJvAU8HV3XxrFOkUkDp01NofC0f25e9Fm6ptiu1fS1NLK9spDMXnFFkQ3SJYDBWaWb2YpwLXAgg77PElbbwQzy6HtVFdJaP+/AL9z9yeiWKOIxCkz4/YLx7G7qp7HYrxXsmPfIZpbPSav2IIoBom7NwO3AguBtcDj7l5sZt82s7mh3RYClWa2BlgEfNHdK4GrgXOAG8xsZegxNVq1ikh8OvOkgczIH8DdizbFdK/k8BVbY2Lwii2I8gqJ7v408HSHbf/d7msHPhd6tN/nD8AfolmbiIiZcfsF45h/31IeXradj8ToCool5Ycv/Y2zHomISE8w66SBzBwzgP97MXbHSkoqauiflkx2WkrQpRyRgkRE4t7tF4yjvLqBP8ToHFwl5bUxe1oLFCQiIpwxZiCzxw7knhc3U13fFHQ579B26W9sntYCBYmICABfungC+w818d9/LQ66lLeprm+ivLqB/BgdHwEFiYgIAKeOzOYz5xfwl9d38uTrHW95C87hgfZYvRkRFCQiIm/51HknUTi6P//15Gq2Vx4Kuhyg/WSN6pGIiMS8pMQE7rh2Kmbw2cdep7kl+NmBS8prSDAYNTAt6FKOSkEiItLOiP5pfO+Kk3l9+wHufH5j0OWwuaKWEf3T6JOUGHQpR6UgERHpYO6pw7hq+gjuWrSJZSWVgdaypbw2ZufYOiyqd7aLiPRU35o7meVb93HrI69zRv4AADz0WmpSIl+7bAIDM/pEtYbWVmdLRS0zxwyM6ueES0EiInIEGX2SuGv+aXztL2+yZlcVAIdXAtlSUcvAjBS+dtnEqNawu6qeuqaWmL70FxQkIiJHdfKILP726bPesf2zj77OH5Zu45NzTorqtCVvXbEVwzcjgsZIRES67BNzTuJQYwu/XRLdKVVKymuA2J319zAFiYhIF00YkskFEwfx4JItHGpsjtrnbC6vJS0lkcGZ0R2LCZeCRESkGz4xZyz7DzXxyKvRWxSrpKKW/Jx0zGJvnfb2FCQiIt0wfXR/zsgfwH0vldDYHJ0bF7dU1MT8aS1QkIiIdNsnzxvL7qp6/vJ6acTfu76phdL9dTG7vG57ChIRkW46pyCHycMyuefFElpa/fgHAHur62nqxNQr2yoP4R7bc2wdpiAREekmM+OTc8aypaKWZ1bvPu7+L2+sYPYPX2D+vUvZV9t4zH23VISu2IrhWX8PU5CIiIThkilDGJOTzi//tQn3o/dK3ig9wMd+X8Sw7L68sfMgV/5y8Vv3iRzJ5tD08Xk5sTtZ42EKEhGRMCQmGB8/9ySKy6r48cL1Rxx431JRy42/WU7/9BQe/9gsHrn5DKrqm7nyl4sp2rrviO9bUl7LoH596JeaHO0mhE1BIiISpitOG86V04bzy39t5r2/eJlVOw689dqeqnqu+/UyHPjdR2YwODOV6aMH8OdPnEl2WgofuH8Zf3+jDHensqaBoq37eHz5DpZv3RfzkzUeZsfqivUkhYWFXlRUFHQZIhLHXli3h6/9eTV7q+v56Nlj+OhZ+Xz4gVfZse8Qj9wyk1NGZL9t//21jdzy+yKWb91PVt9kDtb9Z7345ETjSxdP4OZzxkS1ZjNb4e6FYb2HgkREJHKq6pv4wdPreOTV7SQntt1I+JsbZnBWQc4R969vauHO5zdysK6JMbkZjMlJZ0xuOsOz+5KUGP2TRgqSdhQkIhJLlmyq4KfPbeAjZ+Vz2clDgy7nqCIRJJr9V0QkCs4cm8OZY4/cC+ltNNguIiJhUZCIiEhYFCQiIhIWBYmIiIRFQSIiImFRkIiISFgUJCIiEhYFiYiIhKXX3NluZuXANiALONjupWM9b/91DlARoXI6fmY4+x7t9SNtV9uPvu1o/xaRbPvR6ujuvvred257T2v78fbv6mvhtn+0u+d2ruyjcPde9QDu7ezzDl8XRauGcPY92utH2q62H33b0f4tItn2WGp/PH/vY73tx9u/q6/FQvt746mtv3XhecfXolVDOPse7fUjbVfbj77teP82kRIr7Y/n732st/14+3f1tcDb32tObYXLzIo8zInLeiq1PT7bDvHd/nhuO0S2/b2xR9Jd9wZdQIDU9vgVz+2P57ZDBNuvHomIiIRFPRIREQmLgkRERMKiIBERkbAoSI7DzM42s3vM7H4zWxJ0PSeamSWY2ffM7Bdmdn3Q9ZxIZjbHzP4d+v7PCbqeIJhZupkVmdl7gq7lRDKziaHv+xNm9omg6znRzOxyM7vPzB4zs4uOt3+vDhIze8DM9prZ6g7bLzGz9Wa2ycy+cqz3cPd/u/vHgb8Dv41mvZEWifYD84ARQBNQGq1aIy1CbXegBkilB7UdItZ+gC8Dj0enyuiI0M/92tDP/dXA7GjWG2kRav+T7n4z8HHgmuN+Zm++asvMzqHtF8Hv3H1KaFsisAG4kLZfDsuB+UAi8IMOb/ERd98bOu5x4CZ3rz5B5YctEu0PPfa7+6/M7Al3v+pE1R+OCLW9wt1bzWww8FN3/+CJqj9cEWr/qcBA2oK0wt3/fmKqD0+kfu7NbC7wCeD37v7wiao/XBH+vfcT4CF3f+1Yn5kU0RbEGHd/yczyOmyeAWxy9xIAM3sUmOfuPwCO2H03s1HAwZ4UIhCZ9ptZKdAYetoSxXIjKlLf+5D9QJ+oFBolEfrezwHSgUlAnZk97e6t0aw7EiL1vXf3BcACM3sK6DFBEqHvvQE/BP5xvBCBXh4kRzEc2NHueSlwxnGOuQn4TdQqOrG62v4/A78ws7OBl6JZ2AnQpbab2ZXAxUA2cFdUKzsxutR+d/86gJndQKh3FtXqoqur3/s5wJW0/QHxdDQLO0G6+nP/aeACIMvMxrr7Pcd683gMki5z928GXUNQ3P0QbUEad9z9z7QFaVxz9weDruFEc/d/Af8KuIzAuPudwJ2d3b9XD7YfxU5gZLvnI0Lb4kU8tz+e2w7x3f54bjtEuf3xGCTLgQIzyzezFOBaYEHANZ1I8dz+eG47xHf747ntEOX29+ogMbNHgFeA8WZWamY3uXszcCuwEFgLPO7uxUHWGS3x3P54bjvEd/vjue0QTPt79eW/IiISfb26RyIiItGnIBERkbAoSEREJCwKEhERCYuCREREwqIgERGRsChIpFczs5oT/HkRWbPG2tZCOWhmK81snZn9byeOudzMJkXi80W6QkEi0gVmdsz56dz9zAh+3L/dfSowDXiPmR1vXYzLaZupV+SEUpBI3DGzk8zsGTNbYW0rIE4IbX+vmS0zs9fN7J+hdUgws2+Z2e/NbDHw+9DzB8zsX2ZWYmafaffeNaH/zgm9/kSoR/FQaGpuzOyy0LYVZnanmR1znQ93rwNW0jaDK2Z2s5ktN7NVZvYnM0szszOBucCPQ72Yk47WTpFIU5BIPLoX+LS7Twe+APwytP1lYKa7TwMeBb7U7phJwAXuPj/0fAJtU8zPAL5pZslH+JxpwG2hY8cAs80sFfgVcGno83OPV6yZ9QcK+M80/n9299Pd/VTapru4yd2X0DZ30hfdfaq7bz5GO0UiStPIS1wxswzgTOCPoQ4C/GfRqhHAY2Y2FEgBtrQ7dEGoZ3DYU+7eADSY2V5gMO9cjvdVdy8Nfe5KII+2letK3P3wez8C3HKUcs82s1W0hcgd7r47tH2KmX2XtnVSMmibP6kr7RSJKAWJxJsE4EBo7KGjX9C2pO6C0MJG32r3Wm2HfRvafd3CkX+WOrPPsfzb3d9jZvnAUjN73N1XAg8Cl7v7qtCiU3OOcOyx2ikSUTq1JXHF3auALWb2fmhbUtTMTg29nMV/1mi4PkolrAfGtFsK9ZrjHRDqvfwQ+HJoUz9gV+h0Wvt15KtDrx2vnSIRpSCR3i4tNJX24cfnaPvle1PotFExMC+077doOxW0AqiIRjGh02OfBJ4JfU41cLATh94DnBMKoG8Ay4DFwLp2+zwKfDF0scBJHL2dIhGlaeRFTjAzy3D3mtBVXHcDG939Z0HXJdJd6pGInHg3hwbfi2k7nfarYMsRCY96JCIiEhb1SEREJCwKEhERCYuCREREwqIgERGRsChIREQkLAoSEREJy/8HNcwLvWaXE0YAAAAASUVORK5CYII=\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"
    },
    {
     "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"
    },
    {
     "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"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEKCAYAAAAIO8L1AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAAdZUlEQVR4nO3deZhddZ3n8fentoSEbJBiS0ISNiECIU4aFwRpFwSGAcRRyWg/+AyPtHSDo7RO49JCY9s40yo90iBgNzqigoiiUaOAoywqahJIkASDIQgkLFZYspHkbt/5455KbopTlVt176lbOfV5PU89uWf5nfP9pW6d7/0t9xxFBGZmZn21tToAMzMbmZwgzMwslROEmZmlcoIwM7NUThBmZpbKCcLMzFJ1tDqAZpk6dWrMmjWr1WGYme1Rli5duj4iutO25SZBzJo1iyVLlrQ6DDOzPYqkJ/rb5i4mMzNL5QRhZmapnCDMzCyVE4SZmaVygjAzs1ROEGZmlsoJwswsA89t3MaGl4utDqMhThBmZk1WKFV4xzW/4uxrf8W2YrnV4QyZE4SZWZN9f9k6nt6wjcfXb+ELd65qdThD5gRhZtZElUpw3T2PMefAiSw4/mD+45ePs/SJF1sd1pA4QZiZNdGdK59jTc8WLjz5UD5x+pEcOGkvPnbb8j2yq8kJwsysSSKCL9/zGDP3HcdpRx/AhLGdfO6dx7CmZwtX/ezRVoc3aE4QZmZNcv+a51n+1EtccNIhdLRXL68nHt7NguNn8JV717DsqZdaG+AgOUGYmTXJdfesYereY3jna6bvsv4Tpx/FARPH8tHvLGfz9lKLohs8JwgzsyZ4eN0G7n20h/PfOJuxne27bJswtpN/eddcHl+/hQu/sZRCqdKiKAfHCcLMrAmuu+cxJozp4L2vOzh1+wmHTeXKc47hvj+u59LvPkREDHOEg5ebBwaZmbVKz6btLPr9M1xw0qFMHNvZ737vnj+D5zZs4wt3Pcp+E8dy6WlHDmOUg+cEYWbWoOe3bKcSMHf6pN3ue9GbD+PZjdu47p7HOGDiGN5/wuxhiHBonCDMzBpULFW7izrbd99rL4krzjqaP2/azj/+aCXTpozjbXP2zzrEIfEYhJlZgwrl6qBzZ0d9l9T2NnH1gnkcM20Sl3x7GY+v35JleEPmBGFm1qBib4JoV91lxna2c+17X0N7u7jwG0t5uTDypr86QZiZNag3QXTV0cVUa/qUcfyfc+ex6rlNfPL2h0fczCYnCDOzBu1sQQz+kvqmI7r58FuO4PYH1/GN3zzR7NAa4gRhZtagQjJI3TGILqZaF7/5MP7yVd1c8aOVPPDkyLnzqxOEmVmDhtrF1KutTVz1nuM4YNJYLvrmAyPmSXROEGZmDSpVht7F1GvyuC6uXvAa/rxpO5+4/fcjYjzCCcLMrEE7vgdR5zTX/hw3YzKXnHIEP/79M3xnydpmhNYQJwgzswYVhjDNtT9/fdKhvP6Qfbls4Qoe69nc8PEa4QRhZtagRscgarUn4xFjOtv4H7c8yPZS655El2mCkHSqpFWSVku6NGX7TEn/T9JDku6WNL1m23mS/pj8nJdlnGZmjWhkmmuaAyaN5X+/81geXreRL9zZuifRZZYgJLUD1wCnAXOABZLm9Nnt88DXI+JY4ArgyqTsPsBlwGuB44HLJE3JKlYzs0YUy/Xfi6lep7z6AN73uoO54d41/Hr1+qYddzCybEEcD6yOiDURUQBuAc7qs88c4OfJ61/UbH87cFdEvBARLwJ3AadmGKuZ2ZD1PgCoGWMQtT55+hwOmTqej932EBu3Df/U1ywTxDTgqZrltcm6WsuBc5LX7wAmSNq3zrJmZiNCsVyhs11IzU0Qe3W18/l3z+WZDVv5zA9XNvXY9Wj1IPVHgTdJehB4E7AOqHtERtIFkpZIWtLT05NVjGZmAyqWK3S0ZXM5fc3BU/ibkw/jO0vXctfK5zI5R3+yTBDrgBk1y9OTdTtExNMRcU5EzAM+max7qZ6yyb43RMT8iJjf3d3d5PDNzOpTLEfTu5dqfegthzPnwIl8/HsP8fzm7Zmdp68sE8Ri4HBJsyV1AecCC2t3kDRVUm8MHwduTF7fAZwiaUoyOH1Kss7MbMQplCt0NfgluYF0dbTxxffMZePW0rDe9TWzGkVECbiI6oX9EeDWiFgh6QpJZya7nQyskvQosD/w2aTsC8BnqCaZxcAVyTozsxGnVK40dQZTmiMPmMglpxzBT1c8yx0rhqerKdNHjkbEImBRn3Wfrnl9G3BbP2VvZGeLwsxsxKp2MWU/pPuBEw/hpvuf4NuLn+TUow/I/HytHqQ2M9vjFZJZTFlrbxNnzD2Q+/64nhe3FDI/nxOEmVmDiqXsu5h6nTn3IEqVYNHDz2R+LicIM7MGFTMepK4158CJHNo9nh8sezrzczlBmJk1aLjGIAAkcebcaSz+0ws8s2FrpudygjAza9BwjUH0OvO4g4iAHy3PtpvJCcLMrEHFYZjmWmv21PEcO30SC5dn283kBGFm1qDhThBQHaz+/boNrMnwoUJOEGZmDSqWsr3VRpozjj0IiUxbEU4QZmYNKlaGvwVxwKSxHD9rHxYufzqzW284QZiZNahYrjTlcaODdeZxB7GmZwsrnt6YyfGdIMzMGlTtYhr+y+npRx9IR5sy62ZygjAza1CxXKGzY3jHIACmjO/ipCO6uffRbJ6Hk+nN+szMRoNCC2Yx9brynGOYMq4rk2M7QZiZNahVYxAA+08cm9mx3cVkZtag4bzVxnDKX43MzIZRuRKUK04QZmbWR7FcAWjJIHXWnCDMzBqwI0G05e9ymr8amZkNo2K5+i3m4b7VxnBwgjAza0BpRxdT/i6n+auRmdkwKvQmCA9Sm5lZrd4uplZ9DyJL+auRmdkwKroFYWZmaQql3gThQWozM6tR9CC1mZml8RiEmZml8hiEmZml2jnN1WMQZmZWo1hyC8LMzFLsvNVG/i6n+auRmdkwKlXcxTQkkk6VtErSakmXpmw/WNIvJD0o6SFJpyfrZ0naKmlZ8nNdlnGamQ1VIcddTJk9clRSO3AN8DZgLbBY0sKIWFmz26eAWyPiy5LmAIuAWcm2xyLiuKziMzNrhh3TXP09iEE5HlgdEWsiogDcApzVZ58AJiavJwFPZxiPmVnTeZrr0EwDnqpZXpusq3U58D5Ja6m2Hi6u2TY76Xq6R9KJaSeQdIGkJZKW9PT0NDF0M7P6FD3NNTMLgK9FxHTgdOAmSW3AM8DBETEPuAT4lqSJfQtHxA0RMT8i5nd3dw9r4GZm4Nt9D9U6YEbN8vRkXa3zgVsBIuJ+YCwwNSK2R8TzyfqlwGPAERnGamY2JMWSp7kOxWLgcEmzJXUB5wIL++zzJPAWAElHUU0QPZK6k0FuJB0CHA6syTBWM7MhKZYrtLeJ9rb8dTFlNospIkqSLgLuANqBGyNihaQrgCURsRD4O+Arkj5CdcD6/RERkk4CrpBUBCrAByPihaxiNTMbqmK5ksvxB8gwQQBExCKqg8+16z5d83olcEJKue8C380yNjOzZiiUK3S25a97CVo/SG1mtkcrliu5fBYEOEGYmTWkVI7cdjE5QZiZNaBQruRyBhM4QZiZNaRYjlw+TQ6cIMzMGlIsuQVhZmYpqoPUHoMwM7M+PAZhZmapik4QZmaWxoPUZmaWKs+32nCCMDNrQKFUocMtCDMz66tUcReTmZmlcBeTmZml8hflzMwsVaEcvpurmZm9UrFc8RiEmZm9kscgzMwslb9JbWZmrxARFMvhBGFmZrsqlgOArtE8SC1pvKS25PURks6U1JltaGZmI1uxXAGgo210j0HcC4yVNA24E/gr4GtZBWVmtifoTRCjvYtJEfEycA5wbUS8C3h1dmGZmY18vV1Mo/17EJL0euC9wI+Tde3ZhGRmtmfobUF0jfJprh8GPg7cHhErJB0C/CKzqMzM9gB572LqqGeniLgHuAcgGaxeHxEfyjIwM7ORLu8Jot5ZTN+SNFHSeOBhYKWkj2UbmpnZyFYoJWMQozlBAHMiYiNwNvATYDbVmUxmZqPWjjGIjtE9BtGZfO/hbGBhRBSByCwqM7M9gLuYqq4H/gSMB+6VNBPYmFVQZmZ7goITBETElyJiWkScHlVPAH+5u3KSTpW0StJqSZembD9Y0i8kPSjpIUmn12z7eFJulaS3D6pWZmbDYMf3IEZzgpA0SdIXJS1Jfr5AtTUxUJl24BrgNGAOsEDSnD67fQq4NSLmAecC1yZl5yTLrwZOBa5NjmdmNmIUS70tiNE9BnEjsAl4d/KzEfjqbsocD6yOiDURUQBuAc7qs08AE5PXk4Cnk9dnAbdExPaIeBxYnRzPzGzEyPsYRF3fgwAOjYh31iz/o6RluykzDXiqZnkt8No++1wO3CnpYqotkrfWlP1Nn7LT6ozVzGxYFCvuYgLYKumNvQuSTgC2NuH8C4CvRcR04HTgpt67xtZD0gW93V49PT1NCMfMrH69XUx5feRovS2IDwJflzQpWX4ROG83ZdYBM2qWpyfrap1PdYyBiLhf0lhgap1liYgbgBsA5s+f72m3ZjasdnQxjebvQUTE8oiYCxwLHJsMKr95N8UWA4dLmi2pi+qg88I++zwJvAVA0lHAWKAn2e9cSWMkzQYOB35XZ53MzIZF3scgBlWriNiYfKMa4JLd7FsCLgLuAB6hOltphaQrJJ2Z7PZ3wAckLQduBt6fTKNdAdwKrAR+CvxtRJQHE6uZWdYKOZ/mWm8XU5rdtqkiYhGwqM+6T9e8Xgmc0E/ZzwKfbSA+M7NM7bzddz4TRCO1cp+/mY1qef8exIAtCEmbSE8EAvbKJCIzsz1EsVxBgvacPpN6wAQREROGKxAzsz1NoRx0trch5TNB5LPjzMxsGBTLFTpz2noAJwgzsyErlit0duT3MprfmpmZZayYdDHlVX5rZmaWsWK5ktspruAEYWY2ZMVyJbdTXMEJwsxsyKoJIr+X0fzWzMwsY4WSxyDMzCyFZzGZmVmq6iC1xyDMzKwPj0GYmVmqgr8HYWZmaYolT3M1M7MU7mIyM7NUpYq7mMzMLEWh5BaEmZmlKJYrdHV4DMLMzPrwGISZmaXy7b7NzCxVwS0IMzPrKyJ8qw0zM3ulciWIwC0IMzPbVbEcAL6bq5mZ7apQrgDQ0eYuJjMzq1FKEkSXWxBmZlZrRxeTxyDMzKxWMWlBOEGYmdkuCjsShMcgzMysRm8LosstiKGRdKqkVZJWS7o0ZftVkpYlP49KeqlmW7lm28Is4zQzG6xiKf9jEB1ZHVhSO3AN8DZgLbBY0sKIWNm7T0R8pGb/i4F5NYfYGhHHZRWfmVkjdnQxeRbTkBwPrI6INRFRAG4Bzhpg/wXAzRnGY2bWNEWPQTRkGvBUzfLaZN0rSJoJzAZ+XrN6rKQlkn4j6ex+yl2Q7LOkp6enSWGbme2exyCGz7nAbRFRrlk3MyLmA/8N+FdJh/YtFBE3RMT8iJjf3d09XLGamXmaa4PWATNqlqcn69KcS5/upYhYl/y7BribXccnzMxaqpAMUne4i2lIFgOHS5otqYtqEnjFbCRJRwJTgPtr1k2RNCZ5PRU4AVjZt6yZWauUKvnvYspsFlNElCRdBNwBtAM3RsQKSVcASyKiN1mcC9wSEVFT/CjgekkVqknsc7Wzn8zMWm00dDFlliAAImIRsKjPuk/3Wb48pdyvgWOyjM3MrBE7vgfhaa5mZlbLt9owM7NUnuZqZmapRsMYRH5rZmaWIT8PwszMUhVKHoMwM7MUxXKFznYhOUGYmVmNYrlCR1u+L6H5rp2ZWUaK5ch19xI4QZiZDUmxXKErx1+SAycIM7MhqY5B5PsSmu/amZllpNrFlO9LaL5rZ2aWkUIyiynPnCDMzIagWHIXk5mZpfAgtZmZpfIYhJmZpfIYhJmZpfI0VzMzS+UEYWZmqYol32rDzMxSFCtuQZiZWYpiuZLrx42CE4SZ2ZBUu5jyfQnNd+3MzDJSLFfo7PAYhJmZ9VHwLCYzM0vjMQgzM0vlW22YmdkrlCtBueIEYWZmfRTLFQAPUpuZ2a52JIi2fF9C8107M7MMlMoB4FttmJnZrnZ2MeX7Eppp7SSdKmmVpNWSLk3ZfpWkZcnPo5Jeqtl2nqQ/Jj/nZRmnmdlgFHoTRM4HqTuyOrCkduAa4G3AWmCxpIURsbJ3n4j4SM3+FwPzktf7AJcB84EAliZlX8wqXjOzehWTLiZ/D2LojgdWR8SaiCgAtwBnDbD/AuDm5PXbgbsi4oUkKdwFnJphrGZmdSuOkhZElrWbBjxVs7w2WfcKkmYCs4GfD6aspAskLZG0pKenpylBm5ntTqHUmyA8SD0czgVui4jyYApFxA0RMT8i5nd3d2cUmpnZrjxI3bh1wIya5enJujTnsrN7abBlzcyGlccgGrcYOFzSbEldVJPAwr47SToSmALcX7P6DuAUSVMkTQFOSdaZmbXcaBmDyGwWU0SUJF1E9cLeDtwYESskXQEsiYjeZHEucEtERE3ZFyR9hmqSAbgiIl7IIs6thTI33LuGNkFbm2iTaG+DNlX7FiUhoE3V120Ckn8jqlOs2Bk6SSF6eyZrKvWK7RIIoT7dmLWLQe954hWHqd1n1xUpO/WJr+95+tml/+2orv12F0ra+dLrWOeBdneeAWvdvPM0w+5iHfJxB3i/1V+m+r7t/ZtobxPjujoYP6advcd0sPeYDqZPGUdXTrtgdk5zzfcYRGYJAiAiFgGL+qz7dJ/ly/speyNwY2bBJbYUSlz1s0ezPo3ZqNPV3sZRB05g7ozJHDt9Mq+dvQ8z9hnX6rCaolhyC2JU2Hd8F2v++XTKEVQiqFSgEkE5ovpJNvn0XgmIqH6+rCTbBNCnFVD7aZ/YuR2qn8J2bI+drYNaaZ9gd3xa27kidZ9dlvv5YBM7Qhvgk/Ku4fe3eUitg4Fi2l2ZRj+r1ds2GAmfCbNqx9TzfttdGdj5N9D7fi6Wg5cLJTZvL/Hy9jIbtxVZ9ewmlj31Et9dupav3/8EAPNnTuEdr5nGfz7mQCaP62pGlVqiVOm91YYTRK5J1Ytv24i4LJjlT7kSPNazmZ898hy3P7COT97+MP+4cCVvnbMf/3DGHA6ctFerQxy0oruYzMwa194mjth/AkfsP4EL33QoK57eyPcfXMfNv3uS3675JVcvmMcbDpva6jAHpTBKupjyXTszG1EkcfS0SXzqjDn84KI3MmV8F+/7j9/y5bsfI+rtsxwBdkxzzekgfK98187MRqzD9tubH/ztCZx2zIH8r5/+gb++aSmPPrdpR/fNSOZprmZmGRs/poN/WzCPeTMmc+VP/sCdK5+jq72NQ7rH86oDJrDfhDFsLZZ5uVBma6F6o4U3HdHNaUcfyKRxnQ2du1SusGFrkRdfLrBle/XYvZMjKgHPbdzGMy9t5ZkN23hmwzYO329v/ur1M5k8rmvUjEFoT2rWDWT+/PmxZMmSVodhZkP0xPNbeODJF1n17GZWPbuRVc9u4oWXC4zr6mCvznb26mpna6HMupe20tkuTn7Vfpw59yCmT9mLlwtltmwvsbVYZvP2Epu3ldiyvcSm7SU2bSuxaVuRTdtKbEz+fenlIhu2FuuKq6ujje69x7Dupa2M72rnva+bSbFc4au/+hN/+MypjO1sz/h/JluSlkbE/LRtbkGY2Ygwc9/xzNx3/ID7RAS/X7eBHyx7mh8uf5q7Vj7X774SO760N3FsJxPGdrDfhLEc2t3B5L06mTyui33GdzF5XCd7j+lIjl+dYtwm6J4whoMm78W+47uQxCPPbOTLdz/Gv9+3hmSWa+67mNyCMLM9UrkSPPDki2zeXmJ8VwfjutoZ19XO+CQpjOtqR7v7Es4QPPH8Fq6/dw2btpW4esG8ph9/uLkFYWa5094m/mLWPsN+3pn7juef33HMsJ+3FfLdPjIzsyFzgjAzs1ROEGZmlsoJwszMUjlBmJlZKicIMzNL5QRhZmapnCDMzCxVbr5JLakHeKJm1SRgwyBeTwXWNxBC7bGGsl+96wdaTnvd6nr1t63V9eovrsHsU0+9+q7ze3Ho/F7M5r04MyK6U48aEbn8AW4Y5OslzTrfUPard/1Ay2mvW12v/ra1ul711q3RetVTl5TXfi9mUK96f2ej8b3Y30+eu5h+OMjXzTzfUPard/1Ay1nUrdF69bet1fWq91iN1qvvupFSr4H283ux/+W8vhdT5aaLqVGSlkQ/N6zak7lee5681s312vPkuQUxWDe0OoCMuF57nrzWzfXaw7gFYWZmqdyCMDOzVE4QZmaWygnCzMxSOUHshqQTJV0n6d8l/brV8TSTpDZJn5V0taTzWh1Ps0g6WdJ9ye/t5FbH00ySxktaIumMVsfSTJKOSn5ft0m6sNXxNIuksyV9RdK3JZ3S6ngGK9cJQtKNkv4s6eE+60+VtErSakmXDnSMiLgvIj4I/Aj4v1nGOxjNqBtwFjAdKAJrs4p1MJpUrwA2A2PJV70A/h64NZsoh6ZJf2ePJH9n7wZOyDLeejWpXt+PiA8AHwTek2W8Wcj1LCZJJ1G9UHw9Io5O1rUDjwJvo3rxWAwsANqBK/sc4r9HxJ+TcrcC50fEpmEKf0DNqFvy82JEXC/ptoj4r8MVf3+aVK/1EVGRtD/wxYh473DF358m1WsusC/VxLc+In40PNEPrFl/Z5LOBC4EboqIbw1X/P1p8vXjC8A3I+KBYQq/KTpaHUCWIuJeSbP6rD4eWB0RawAk3QKcFRFXAqnNdkkHAxtGSnKA5tRN0lqgkCyWMwy3bs36nSVeBMZkEuggNen3dTIwHpgDbJW0KCIqWcZdj2b9ziJiIbBQ0o+BlieIJv3OBHwO+Mmelhwg5wmiH9OAp2qW1wKv3U2Z84GvZhZR8wy2bt8DrpZ0InBvloE1aFD1knQO8HZgMvBvmUbWmEHVKyI+CSDp/SStpEyja8xgf2cnA+dQTeiLsgysQYP9G7sYeCswSdJhEXFdlsE122hMEIMWEZe1OoYsRMTLVJNfrkTE96gmv1yKiK+1OoZmi4i7gbtbHEbTRcSXgC+1Oo6hyvUgdT/WATNqlqcn6/Igr3VzvfY8ea1bXuuVajQmiMXA4ZJmS+oCzgUWtjimZslr3VyvPU9e65bXeqXKdYKQdDNwP/AqSWslnR8RJeAi4A7gEeDWiFjRyjiHIq91c732rHpBfuuW13oNRq6nuZqZ2dDlugVhZmZD5wRhZmapnCDMzCyVE4SZmaVygjAzs1ROEGZmlsoJwnJN0uZhPl9Tnhmi6jMtNkhaJukPkj5fR5mzJc1pxvnNwAnCbFAkDXj/soh4QxNPd19EHAfMA86QtLvnJJxN9U6vZk3hBGGjjqRDJf1U0lJVnzx3ZLL+v0j6raQHJf0seZ4Eki6XdJOkXwE3Jcs3Srpb0hpJH6o59ubk35OT7bclLYBvJrd+RtLpybqlkr4kacDnOkTEVmAZ1TuJIukDkhZLWi7pu5LGSXoDcCbwL0mr49D+6mlWLycIG41uAC6OiP8EfBS4Nln/S+B1ETEPuAX4nzVl5gBvjYgFyfKRVG8pfjxwmaTOlPPMAz6clD0EOEHSWOB64LTk/N27C1bSFOBwdt6S/XsR8RcRMZfq7R7Oj4hfU70n0Mci4riIeGyAeprVxbf7tlFF0t7AG4DvJB/oYedDhaYD35Z0INAFPF5TdGHySb7XjyNiO7Bd0p+B/Xnl401/FxFrk/MuA2ZRfULZmojoPfbNwAX9hHuipOVUk8O/RsSzyfqjJf0T1edd7E31vkCDqadZXZwgbLRpA15K+vb7uprqI0oXJg+wubxm25Y++26veV0m/W+pnn0Gcl9EnCFpNvAbSbdGxDLga8DZEbE8eXjQySllB6qnWV3cxWSjSkRsBB6X9C6oPhJS0txk8yR23tv/vIxCWAUcUvMoy90+yD5pbXwO+Ptk1QTgmaRbq/Z525uSbburp1ldnCAs78Ylt2ru/bmE6kX1/KT7ZgVwVrLv5VS7ZJYC67MIJumm+hvgp8l5NgEb6ih6HXBSklj+Afgt8CvgDzX73AJ8LBlkP5T+62lWF9/u22yYSdo7IjYns5quAf4YEVe1Oi6zvtyCMBt+H0gGrVdQ7da6vrXhmKVzC8LMzFK5BWFmZqmcIMzMLJUThJmZpXKCMDOzVE4QZmaWygnCzMxS/X9+xx8ytjxSCAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAY4AAAENCAYAAAAYIIIKAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAAo2ElEQVR4nO3de3xc9Xnn8c+j0V22JVm+32Rj7BiHix0EBJwUkg2p07LBZFMumzS0SSEpTdKUlg1st2mWNm3Tbpu+SEkTJ0toUgghhFAnhUC2DZcABptgAzbB+IJtWQbLlixbt5FG8+wf54w0liV7ZOvMTd/36zUvz5w5c87zs+R5/LubuyMiIpKpklwHICIihUWJQ0RExkSJQ0RExkSJQ0RExkSJQ0RExkSJQ0RExiTSxGFmq83sNTPbbma3jvD+V8xsU/jYZmaH09673sxeDx/Xpx0/38xeDq95h5lZlGUQEZFjWVTzOMwsBmwDLgeagQ3Ade6+dZTzPwOsdPePm9lUYCPQBDjwAnC+u7eb2fPAZ4HngIeBO9z9kUgKISIix4myxnEhsN3dd7p7H3AfcOUJzr8O+F74/NeBn7l7m7u3Az8DVpvZbGCKu6/3ION9B1gTWQlEROQ4pRFeey6wN+11M3DRSCeaWSOwCPjPE3x2bvhoHuH4SNe8EbgRoKam5vxly5aNvQQiIhPYCy+8cNDdpw8/HmXiGItrgQfcfWC8Lujua4G1AE1NTb5x48bxurSIyIRgZrtHOh5lU9U+YH7a63nhsZFcy1Az1Yk+uy98nsk1RUQkAlEmjg3AEjNbZGblBMlh3fCTzGwZUA88m3b4UeD9ZlZvZvXA+4FH3X0/cMTM3hmOpvoY8G8RlkFERIaJrKnK3RNm9mmCJBAD7nL3LWZ2O7DR3VNJ5FrgPk8b3uXubWb2FwTJB+B2d28Ln98E3A1UAY+EDxERyZLIhuPmE/VxiIiMnZm94O5Nw49r5riIiIyJEoeIiIyJEoeISBHa39HDY1vepDOeGPdrK3GIiBSh53e1ceN3X+CtI73jfm0lDhGRItQVD+ZT15SP/+BZJQ4RkSLUFTZR1VTExv3aShwiIkWoqy9IHNWqcYiISCa64gmqymLESsZ/yyIlDhGRItTVN0BNRTSLgyhxiIgUoa54IpL+DVDiEBEpSl3xgUhGVIESh4hIUeqKJ5ikpioREclUV1+CajVViYhIpoI+DtU4REQkQ0Efh2ocIiKSoa4+1ThERCRD7h40VWlUlYiIZKK3P0nSUY1DREQyk1qnapJGVYmISCZSK+NGscAhKHGIiBSdwb041FQlIiKZSDVVaa0qERHJyNAmTqpxiIhIBqLcNhaUOEREik6U28aCEoeISNEZGo6rGoeIiGRAw3FFRGRMuvoGKI+VUF4azVe8EoeISJHpike3FwcocYiIFJ3OCBc4BCUOEZGi0x0fiGxEFShxiIgUnSj34gAlDhGRotMVT0Q2FBciThxmttrMXjOz7WZ26yjnXG1mW81si5ndm3b8y2b2Svi4Ju343Wa2y8w2hY8VUZZBRKTQdMUHqI5o21iAyFKSmcWAO4HLgWZgg5mtc/etaecsAW4DVrl7u5nNCI//JvAOYAVQATxuZo+4+5Hwo7e4+wNRxS4iUsgKuanqQmC7u+909z7gPuDKYefcANzp7u0A7n4gPL4ceNLdE+7eBbwErI4wVhGRohHltrEQbeKYC+xNe90cHku3FFhqZk+b2XozSyWHzcBqM6s2s2nAe4D5aZ/7kpm9ZGZfMbOKqAogIlKIuuIDBVvjyEQpsAS4DLgO+KaZ1bn7Y8DDwDPA94BngYHwM7cBy4ALgKnA50e6sJndaGYbzWxja2trpIUQEckXfYkkfQNJaiLs44gycezj2FrCvPBYumZgnbv3u/suYBtBIsHdv+TuK9z9csDC93D3/R6IA98maBI7jruvdfcmd2+aPn36uBZMRCRfdfdFuxcHRJs4NgBLzGyRmZUD1wLrhp3zEEFtg7BJaimw08xiZtYQHj8XOBd4LHw9O/zTgDXAKxGWQUSkoHT1BY0zUQ7HjezK7p4ws08DjwIx4C5332JmtwMb3X1d+N77zWwrQVPULe5+yMwqgaeC3MAR4KPunggvfY+ZTSeohWwCPhVVGURECs3gyrgRzhyPLiUB7v4wQV9F+rEvpD134ObwkX5OL8HIqpGu+d7xj1REpDh0RrxtLOS+c1xERMZRd8TbxoISh4hIUemMeNtYUOIQESkq3RFvGwtKHCIiRSXqbWNBiUNEpKhkYziuEoeISBHpiicoMagsi+7rXYlDRKSIpLaNDefBRUKJQ0SkiHTHByKd/AcRTwAsdPc9v4eWjl5mTalkVm0FM6dUMmNyJdXlMSrLYsRKosvoIiKnojPivThAieOEntjWyk+3vIn7yO+XxYzKshhTKsuYWlNOXXXwZ3mshN5Ekt7+AXr7B+hLJCkvLaGiNEZFWQmVpTGS7sQTA8T7k8QTSZLuVJXFqCyPUVUWPEpjRmmJESspobTEGMxTYRV0IJmku2+Anr4BuvuCeyXdcQcH3J1U6OnHEgPOQNJJJJMMOJSWGOWxEspLg0fpsIRoBiVmxErCh1l4raF7mDEYa1nMKAljTI9h8HowWI1OuoeP4HpmwfslZqliHvP3P/xq7jCQ9KHHKD+s1D0tfGFYWK6h50N/tZb+15wWQ/h368fHMTzO4OyTS31mpOtZWhzpfyepGFPHzY49N/3aHsZ8zHXDc0rMhq5J8PtVUmKDP++Rrm0W/PxT55aknptRUhL8vqb+HdRXl9MwqZxZUyojbTaRY3VHvG0sKHGc0D9/9HwSA0laO+Ps7+jlrY5eWjvj9PQN0NufpDcRfGkf6e2nvauPtu5+9rR105dIUlkWCx8llMdK6IwnONjZRzxMJrGYBYmktISK0hLMjI6efnr6g2TS0z9AYiBJIunBYyB5TBKA4EuvuryUqvIYNWEtKP3LxdK+XAAI/5GXlZQQKzEqyoJ20IFkkr5Eku7uxGASS5d0SIZfyqkv6MEv4vBeqS/w/gFnIBnEfcw5qdjT4nd3YiWpLy8bvE4q2STdBz+X/sUz/CtoMKGFj+HvDya5tC/+1P1T94Gh5Jr+d5xupKR23DnHfSbzL8zjE9WxcboHsQ79HfqwmMNypMVhaQkg/bqp/0QkU9cc7c8RYhirCxbW8xdrzmbZrClj/7CMWdTbxoISx0mVxkqYXVvF7NqqXIciknMeJpWBVE0xGSTkVALqTyRp7+6nvbuPQ519vHGoi288sYPfvOMXXH/xQv7o8iVMrizLdTGKWmc8wZy6ykjvocQhIhlL1QxLjqtbDWmYdOymnNc0zefvHnuNbz+zix+/1MJfX3UO71s+M+pQJ6zuvkSkk/9Ao6pEJGL1NeX81VXn8NBNq2ioKecP73txcD0lGX+dEW8bC0ocIpIl582v40tXnUNX3wDrNrXkOpyi1d2XiHTbWFDiEJEseseCOpbNmsw9z+0e7MyX8ZNMOt19qnGISBExMz7yzka2tBxhc3NHrsMpOt390a9TBUocIpJla1bMobo8xr3P7c51KEUnG9vGghKHiGTZ5Moyrlwxh3WbW+jo6c91OEUlNehANQ4RKTr//cJGevuT/OiXzbkOpaikto3VcFwRKTrnzKvlvHm13PPcHnWSj6NsbBsLShwikiMfuaiR1w90snF3e65DKRqpbWNrVOMQkWJ0xXmzmVxRyj3r1Uk+XoZqHEocIlKEqstL+dA75vLwy29ysDOe63CKQncWto0FJQ4RyaGPXbKQRDLJ136+I9ehFAUNxxWRord4+iSubprPd9e/wd627lyHU/AGm6rUxyEixexz71tKiRn/8LNtuQ6l4HX3DVBZVhL57qRKHCKSU7NqK/n4uxbx0KZ9bGnRMiSnozMLu/+BEoeI5IFPXbqYKZVl/O1PX8t1KAWtOx79fuOgxCEieaC2qow/eM9intjWyjM7DuY6nILVGR+IfNY4KHGISJ742MULmVNbyZcf+ZVmk5+irniCSRGPqAIlDhHJE5VlMf7o8qVsbu7gp6+8metwClI2to0FJQ4RySMfesc8ptaU88S21lyHUpDUOS4iE06sxFg0rYbdhzSn41R09w1QHfG2sRBx4jCz1Wb2mpltN7NbRznnajPbamZbzOzetONfNrNXwsc1accXmdlz4TW/b2blUZZBRLKrcWo1uw915TqMgtRZ6KOqzCwG3Al8AFgOXGdmy4edswS4DVjl7m8HPhce/03gHcAK4CLgT8xsSvixLwNfcfczgXbgE1GVQUSyb0FDNfuP9BJPDOQ6lILiHuw3XuhNVRcC2919p7v3AfcBVw475wbgTndvB3D3A+Hx5cCT7p5w9y7gJWC1mRnwXuCB8Lx/AdZEWAYRybLGhmrcYW9bT65DKSjxRJKBpEe+ThVEmzjmAnvTXjeHx9ItBZaa2dNmtt7MVofHNxMkimozmwa8B5gPNACH3T1xgmsCYGY3mtlGM9vY2qqONpFCsWBqDQB72tRcNRbZ2jYWIPo7nPz+S4DLgHnAk2Z2jrs/ZmYXAM8ArcCzwJjqre6+FlgL0NTUpEHhIgWisaEaQB3kY5StbWMh2hrHPoJaQsq88Fi6ZmCdu/e7+y5gG0Eiwd2/5O4r3P1ywML3DgF1ZlZ6gmuKSAFrqClnUkWpEscYDdU4CrupagOwJBwFVQ5cC6wbds5DBLUNwiappcBOM4uZWUN4/FzgXOAxD6aT/hz4cPj564F/i7AMIpJlZsYCjawas9S2sQVd4wj7IT4NPAq8Ctzv7lvM7HYz+2B42qPAITPbSpAQbnH3Q0AZ8FR4fC3w0bR+jc8DN5vZdoI+j/8bVRlEJDcaG6rZrf05xiRb28ZCxH0c7v4w8PCwY19Ie+7AzeEj/ZxegpFVI11zJ8GILREpUgsaqvmPVw8wkPTI95YoFl3x7GwbC5o5LiJ5qHFqDX0DSd480pvrUApG12BTVWH3cYiInJKhkVXq58hUVxaH4ypxiEje0ZDcsevuC4fjFvioKhGRUzK7toqymClxjEFnPEFZzKgoVeIQkQkoVmLMr6/W7PEx6I5nZy8OUOIQkTy1oKFaNY4x6IxnZ4FDUOIQkTzVOLWaPYe6tY1shrriiayMqAIlDhHJUwsaajgaT9DW1ZfrUApCS0cPs2ors3IvJQ4RyUsLUyOrNIP8pNydHQc6WTx9Ulbul1HiMLMaMysJny81sw+aWVm0oYnIRJYakrtH/Rwn9daROF19AyyeXpOV+2Va43gSqDSzucBjwG8Dd0cVlIjIvPpqzDSXIxM7WjsB8qvGAZi7dwMfAr7m7r8FvD26sERkoqssizFrSiW7NST3pAYTx4w8SxxmdjHwEeDfw2PZ6b4XkQkrWF5dNY6T2dnaxaSKUmZMrsjK/TJNHJ8DbgN+FC6NfgbBMugiIpFp1FyOjOxo7eSM6TWYZWcl4Yxmi7j7E8ATAGEn+UF3/2yUgYmINDbUcLCzma54Iiv7TBSqHQc6ueiMhqzdL9NRVfea2RQzqwFeAbaa2S3RhiYiE93gyCoNyR1VVzxBS0dv1kZUQeZNVcvd/QiwBngEWEQwskpEJDKNU4MvQzVXjW7XwWDwQLZGVEHmiaMsnLexBljn7v2A1gEQkUgt0L4cJ5XtEVWQeeL4BvAGUAM8aWaNwJGoghIRAaitKqOuukyzx09gR2sXJTbUrJcNmXaO3wHckXZot5m9J5qQRESGNDbUaPb4Cexo7WT+1Oqs7MORkmnneK2Z/YOZbQwff09Q+xARiVTj1GpNAjyBbK5RlZJpU9VdwFHg6vBxBPh2VEGJiKTMra/izY5eBpLqVh1uIOnsOtiV1RFVkGFTFbDY3f9b2uv/bWabIohHROQYc2or6R9wDnbGmTklO8uGF4qWwz3EE8m8rXH0mNm7Ui/MbBXQE01IIiJD5tRVAcGXpBxrew5GVEHmNY5PAd8xs9rwdTtwfTQhiYgMmV0bJI79Hb2szHEs+WZna/bncEDmo6o2A+eZ2ZTw9REz+xzwUoSxiYgwVzWOUe1o7aS+uoypNeVZve+YdgB09yPhDHKAmyOIR0TkGFOqSqkuj9FyuDfXoeSdHQc6OSPLtQ04va1js7MMo4hMaGbG7NpK9neoxjHcjtbsj6iC00scGhsnIlkxp65KTVXDdHT3c7AznvX+DThJH4eZHWXkBGFAVSQRiYgMM6e2il+9eTTXYeSVHQezu11suhMmDnefnK1ARERGM7uuktajceKJgawurZHPBkdUZXkoLpxeU5WISFak5nK81RHPcST5Y0drJ2UxY3599ht/lDhEJO/NCedytKiDfNCOA500NtRQGsv+13ikdzSz1Wb2mpltN7NbRznnajPbamZbzOzetON/Gx571czusHAzXTN7PLzmpvAxI8oyiEjuzakLlhpRB/mQHa2dORlRBZnPHB8zM4sBdwKXA83ABjNb5+5b085ZAtwGrHL39lQSMLNLgFXAueGpvwAuBR4PX3/E3TdGFbuI5Jf02eMSJI3dh7p5/9tn5eT+UdY4LgS2u/tOd+8D7gOuHHbODcCd7t4O4O4HwuMOVALlQAVQBrwVYawikseqymPUV5exTzUONu89zG99/Vnqqsv48PnzchJDlIljLrA37XVzeCzdUmCpmT1tZuvNbDWAuz8L/BzYHz4edfdX0z737bCZ6s9STVjDmdmNqf1DWltbx6tMIpIjc+qq2D/BE8dTr7dy3TfXU1MR44FPXZKTobiQ+87xUmAJcBlwHfBNM6szszOBs4B5BMnmvWb27vAzH3H3c4B3h4/fHunC7r7W3ZvcvWn69OkRF0NEoja7tmpCN1Wt29zCx+/ewIKp1fzwU5ewcFru9tKLMnHsA+anvZ4XHkvXDKxz93533wVsI0gkVwHr3b3T3TuBR4CLAdx9X/jnUeBegiYxESlyc+oqJ2xT1fqdh/jD+15k5YJ6vv/Ji5mR431JokwcG4AlZrbIzMqBa4F1w855iKC2gZlNI2i62gnsAS41s1IzKyPoGH81fD0tPL8MuAJ4JcIyiEiemFNXxdHeBJ3xRK5Dybrnd7XhDnf9zgXUVpXlOpzoEoe7J4BPA48CrwL3u/sWM7vdzD4YnvYocMjMthL0adzi7oeAB4AdwMvAZmCzu/+YoKP8UTN7CdhEUIP5ZlRlEJH8Mbs2+F/2ROzn2NvWzYzJFUyqiGwg7JhEGoW7Pww8POzYF9KeO8Hy7DcPO2cA+OQI1+sCzo8kWBHJa6l9OfYd7mHJzIm1GtKetm4WTK3OdRiDct05LiKSkdl1o8/leG7nIZrbu7MdUtY0t/cwX4lDRGRsZk6uoMSOnz3e2z/Ax+56njv+4/UcRRatvkSSlo6enKxJNRolDhEpCKWxEmZOqTxuJ8Bndx4inkiy62BXjiKLVsvhHtxRjUNE5FSMtBPgE68FE3x3HyrOpqq9YROcEoeIyCmYPcJOgE9sCxLHgaNxuvuKb6junrYgcahzXETkFMytC2aPBwMyYfehLnYd7OL8xnpg6Eu2mOxt66EsZszM8aS/dEocIlIwZtdWEk8kaevqA4ZqG9dfshAozuaqve3dzK2rIlYy4rJ8OaHEISIFI7UTYKqD/PHXWlnYUM2vLZkGwJ5iTBxt3XnVvwFKHCJSQNJ3AuztH+CZHQe5dOl06qrLmVJZyu624htZpcQhInIaZqftBLjhjTZ6+5Nc9rZgE9CF02qKrqnqaG8/7d39edUxDhEvOSIiMp4aasopLy1hf0cvze09lJeW8M4zGoBg1NFLzR05jnB87W0LRpDNr8+vxKEah4gUDDNjTm0lLYd7eGJbKxctmkpVeQyAxoZq9h3uoX8gmeMox8/QHI78mTUOShwiUmBm11bxy93tbD/QOdhMBdDYUMNA0tnXXjyr5+7NwzkcoMQhIgVmTl0VLeFCh5cuHdrdszH8ct1dRHM59rZ1M7miNC/24EinxCEiBWVO2EE+r76KxdOHtk9tbAie7zlUPCOr9oQjqszyZw4HKHGISIFJzeW4dOn0Y75QZ0yuoKK0pKhGVu1t78m7/g1Q4hCRApNqknrvshnHHC8pMRobqnmjSBKHuwdzOPJsRBVoOK6IFJiLFzdwz+9dxCWLG457b8HUGvYUySTA1qNx4okkCxryL3GoxiEiBcXMWHXmtBHb/RsbqtnT1j24CGIhGxyKm4c1DiUOESkajQ3V9PYnOXA0nutQTtvg5D/1cYiIRCc1suqNItgNMLVE/DzVOEREolNMczn2tnUzY3IFlWWxXIdyHCUOESkac+uDfSuKYXn1PW3deTdjPEWJQ0SKRlmshLl1VUVR42hu78m75dRTlDhEpKg0NlSzu8Bnj/clkuzv6GF+ff51jIMSh4gUmQVTqwt+9njL4R6SjmocIiLZ0NhQTUdPP4e7+3IdyikbWk5diUNEJHKpIbmFXOvYk6fLqacocYhIUWlsKPwhuXvbeiiLGTOnVOY6lBEpcYhIUUn9L72Ql1ff29bN3LpgaHE+UuIQkaJSXV7K9MkVBdtUtaO1k5+9+hbvaKzPdSijUuIQkaKzsKEwR1Ylk86tP3yJytISbl29LNfhjEqJQ0SKzoKpNbxRgE1V312/mw1vtPNnVyxnRp72b0DEicPMVpvZa2a23cxuHeWcq81sq5ltMbN7047/bXjsVTO7w8I1lM3sfDN7Obzm4HERkZSzZk/mwNE4B4725jqUjDW3d/Pln/6KX1s6nQ+fPy/X4ZxQZInDzGLAncAHgOXAdWa2fNg5S4DbgFXu/nbgc+HxS4BVwLnA2cAFwKXhx/4ZuAFYEj5WR1UGESlMK+bXAbBpz+GcxjGaZPLY/ULcndsefBkD/uqqs/Nuj/HhotwB8EJgu7vvBDCz+4Arga1p59wA3Onu7QDufiA87kAlUA4YUAa8ZWazgSnuvj685neANcAjEZZDRArM2XNrKS0xXtx7mPe/fVZOYwm2gO3huV2HeG5XG8/vamN/Rw/nzavj4sUNXHxGA7sOdfHU6we5/cq35+Uy6sNFmTjmAnvTXjcDFw07ZymAmT0NxIAvuvtP3f1ZM/s5sJ8gcfyTu79qZk3hddKvOXekm5vZjcCNAAsWLBiH4ohIoagsi7F8zpS8qHH88Q828+Av9wFQX13GhYum8r6zZvLCnnbu/Pl2vvqf2wG4cOFUPnpRYy5DzViu9xwvJWhuugyYBzxpZucA04CzwmMAPzOzdwM9mV7Y3dcCawGampoKfx9JERmTFfPr+OELzQwkPafzIZ7f1cZFi6byF2vO5szpkyhJi+Vobz8b3mhj094Orm6ad8x7+SzKzvF9wPy01/PCY+magXXu3u/uu4BtBInkKmC9u3e6eydBU9TF4efnneSaIiKsXFBHV98Arx84mrMYBpLOmx29nN9Yz9KZk49LDJMry3jvspncfPnSgmiiSokycWwAlpjZIjMrB64F1g075yGC2gZmNo2g6WonsAe41MxKzayMoGP8VXffDxwxs3eGo6k+BvxbhGUQkQK1Yn4wge7FHDZXtR6Nk0g6c+ryc3n0UxVZ4nD3BPBp4FHgVeB+d99iZreb2QfD0x4FDpnZVuDnwC3ufgh4ANgBvAxsBja7+4/Dz9wEfAvYHp6jjnEROc7Chmrqqst4cU97zmJo6Qha1+cWWeKItI/D3R8GHh527Atpzx24OXyknzMAfHKUa24kGKIrIjIqM2PF/Do27T2csxhaDgeJY3Zd/k7mOxWaOS4iRWvl/HpeP9DJ0d7+nNw/lTjUVCUiUiBWLqjDHV5q7sjJ/VsO9zK5opQplWU5uX9UlDhEpGidF84gz1U/x77DPUVX2wAlDhEpYrVVZSyeXpOzfo79HT3MKbL+Dcj9BEARkUitmF/P468dwN0zWgPqwNFent1xiGe2H2LL/g7+9DeWc/HihlO6d8vhXs6dV3dKn81nShwiUtRWLqjjh79sZm9bDwsaRp5k1xlP8K/rd/PgL5vZ9lYnAFMqS6koi/HJ727kwZsu4cwZk8d0356+Adq6+opuKC4ocYhIkVu5oA6AF/e2H5c4Orr7ufuZN7jr6V109PRz4aKpfH71Mlad2cDb59TScriHq772NL979wZ+dNMqpk2qyPi+qTkcaqoSESkwb5s5maqyGC/uOcyVK4I1UXv7B/jnx3dw1y92cTSe4H1nzeQz7z1zsDM9Zf7Uar51/QVc841nufE7G7n3hndSWRbL6L6DQ3FrVeMQESkopbESzplXO9hB/syOg/zPB1/mjUPd/MY5s/j0e5awfM6UUT+/Yn4d/3jNCm6695f88Q8289VrV2a0GGGxzuEAJQ4RmQBWzq/j20+/wZ/8YDMPvNBMY0M19/zeRaw6c1pGn//AObO57QPL+KuHf8X0SRV84YrlJ00eLYd7MYNZtWqqEhEpOCsX1PGNJ5M89OI+brpsMZ/9L0sybnJKueHdZ3DgSJxv/WIX+zt6+MdrVlJVPvo1Wg73MGNyBWWx4pv1oMQhIkXvsrfN4A/es5grzp3DWbNHb5Y6ETPjf12xnLn1Vdz+k61cu/ZZvnl9EzMmj1yjaOkozsl/oAmAIjIBVJbFuOXXl51y0kj3u6sWsfa3m9j2VidX3fkMr7058n4fLYd7lThERCRw+fKZ3P/Ji+kfSHLt2mfp7R845n13p+VwT1HO4QAlDhGRU3LOvFr+cs3ZtHf38/K+YxdRbOvqI55IMqcIO8ZBiUNE5JRdsHAqEOwrnq7lcC8As1XjEBGRdPU15SyZMYmNbxybOPYdLs6d/1KUOERETkPTwqls3N3OQNIHjxXz5D9Q4hAROS0XLKznaG+CbW8Nja5qOdxDZVkJ9dXFtYFTihKHiMhpSPVzpDdXpeZwZLKMeyFS4hAROQ3z6quYNaWS598Y2mWw5XBvUS5umKLEISJyGsyMpoX1bNjVhnvQz9FyuDh3/ktR4hAROU0XLprKm0d6aW7vIZ4Y4MDReNF2jIMSh4jIaWtqDPs5drfxVkccKN4RVaBFDkVETtvbZk1mckUpG95oZ9aUIGEU6xwOUI1DROS0xUqM88N+jv0dxT2HA5Q4RETGxQULp/L6gU62tBwBYHaRrlMFShwiIuOiqbEegJ+81EJDTfmYN4oqJEocIiLj4Lz5dZTHSnjrSHGPqAIlDhGRcVFZFuOcebUART2HA5Q4RETGTdPCoLlKNQ4REcnIheG6VcW83AgocYiIjJuLzmjgXWdO411LpuU6lEhFmjjMbLWZvWZm283s1lHOudrMtprZFjO7Nzz2HjPblPboNbM14Xt3m9mutPdWRFkGEZFMTaoo5V9/7yLOmj0l16FEKrKZ42YWA+4ELgeagQ1mts7dt6adswS4DVjl7u1mNgPA3X8OrAjPmQpsBx5Lu/wt7v5AVLGLiMjooqxxXAhsd/ed7t4H3AdcOeycG4A73b0dwN0PjHCdDwOPuHt3hLGKiEiGokwcc4G9aa+bw2PplgJLzexpM1tvZqtHuM61wPeGHfuSmb1kZl8xs4qRbm5mN5rZRjPb2NraeqplEBGRYXLdOV4KLAEuA64Dvmlmdak3zWw2cA7waNpnbgOWARcAU4HPj3Rhd1/r7k3u3jR9+vRIghcRmYiiTBz7gPlpr+eFx9I1A+vcvd/ddwHbCBJJytXAj9y9P3XA3fd7IA58m6BJTEREsiTKxLEBWGJmi8ysnKDJad2wcx4iqG1gZtMImq52pr1/HcOaqcJaCBZs5rsGeGX8QxcRkdFENqrK3RNm9mmCZqYYcJe7bzGz24GN7r4ufO/9ZrYVGCAYLXUIwMwWEtRYnhh26XvMbDpgwCbgU1GVQUREjmepPXKLWVNTk2/cuDHXYYiIFBQze8Hdm447PhESh5m1ArvDl7VAR9rb6a9Hej4NOHgatx9+v1M5b6T3TlSO4a+jKNdocY3lnEzKNfxYJs9z/TPL9Lh+F4vndzH9dT79Lo72XqY/s0Z3P350kbtPqAewdrTXIz0naFYbt/udynkjvXeicmSjXJmW7XTLlUlZRnie059Zpsf1u1g8v4vDypM3v4vj8TMb6ZHr4bi58OMTvB7t+Xje71TOG+m9E5Vj+OsoypXptU63XMOP5Uu5TnRepsf1uzh+cv27mP46n34XR3tvLD+z40yIpqrTYWYbfYQ2vkJXrOWC4i2bylV4irVsE7HGMVZrcx1ARIq1XFC8ZVO5Ck9Rlk01DhERGRPVOEREZEyUOEREZEyUOEREZEyUOEREZEyUOE6Dmb3bzL5uZt8ys2dyHc94MbMSM/uSmX3VzK7PdTzjxcwuM7Onwp/ZZbmOZ7yZWU24B80VuY5lvJjZWeHP6wEz+/1cxzOezGyNmX3TzL5vZu/PdTxjMWETh5ndZWYHzOyVYcdPuk96irs/5e6fAn4C/EuU8WZqPMpFsFPjPKCfYOn7nBuncjnQCVSSJ+WCcSsbBHvT3B9NlGM3Tv/GXg3/jV0NrIoy3rEYp7I95O43ECzUek2U8Y63CTsc18x+jeBL5DvufnZ4LEawJ8jgPukES7vHgL8edomPe7jVrZndD3zC3Y9mKfxRjUe5wke7u3/DzB5w9w9nK/7RjFO5Drp70sxmAv/g7h/JVvwnMk5lOw9oIEiKB939J9mJfnTj9W/MzD4I/D7wXXe/N1vxn8g4f3/8PXCPu/8yS+GftsiWVc937v5kuHR7usF90gHM7D7gSnf/a2DE6r+ZLQA68iFpwPiUy8yagb7w5UCE4WZsvH5eoXZgxC2Hc2GcfmaXATXAcqDHzB5292SUcZ/MeP3MPNiCYZ2Z/TuQF4ljnH5mBvwN8EghJQ2YwIljFCPtk37RST7zCYKdCPPZWMv1IPBVM3s38GSUgZ2mMZXLzD4E/DpQB/xTpJGdvjGVzd3/FMDMfoewZhVpdKdurD+zy4APEST6h6MMbByM9d/ZZ4D3AbVmdqa7fz3K4MaTEsdpcvc/z3UM483duwkSYlFx9wcJkmLRcve7cx3DeHL3x4HHcxxGJNz9DuCOXMdxKiZs5/goMtknvRCpXIWnWMtWrOWC4i7bMZQ4jpXJPumFSOUqPMVatmItFxR32Y4xYROHmX0PeBZ4m5k1m9kn3D0BpPZJfxW439235DLOsVK5CqtcULxlK9ZyQXGXLRMTdjiuiIicmglb4xARkVOjxCEiImOixCEiImOixCEiImOixCEiImOixCEiImOixCETkpl1Zvl+47JfiwV7inSY2SYz+5WZ/Z8MPrPGzJaPx/1FQIlDZFyY2QnXfXP3S8bxdk+5+wpgJXCFmZ1sn4o1BKvmiowLJQ6RkJktNrOfmtkLFuwUuCw8/l/N7Dkze9HM/l+4nwdm9kUz+66ZPQ18N3x9l5k9bmY7zeyzadfuDP+8LHz/gbDGcE+4vDZm9hvhsRfM7A4zO+GeGu7eA2wiWJUVM7vBzDaY2WYz+6GZVZvZJcAHgb8LaymLRyunSKaUOESGrAU+4+7nA38CfC08/gvgne6+ErgP+B9pn1kOvM/drwtfLyNYuv1C4M/NrGyE+6wEPhd+9gxglZlVAt8APhDef/rJgjWzemAJQ0vfP+juF7j7eQRLXnzC3Z8hWC/pFndf4e47TlBOkYxoWXURwMwmAZcAPwgrADC02dM84PtmNhsoB3alfXRd+D//lH939zgQN7MDwEyO36b2eXdvDu+7CVhIsJvcTndPXft7wI2jhPtuM9tMkDT+0d3fDI+fbWZ/SbDfyCSCNZPGUk6RjChxiARKgMNh38FwXyXYanZduLHQF9Pe6xp2bjzt+QAj/xvL5JwTecrdrzCzRcB6M7vf3TcBdwNr3H1zuKHTZSN89kTlFMmImqpEAHc/Auwys9+CYFtPMzsvfLuWoX0Vro8ohNeAM9K2I73mZB8Iayd/A3w+PDQZ2B82j6Xvp340fO9k5RTJiBKHTFTV4XLYqcfNBF+2nwibgbYAV4bnfpGgaecF4GAUwYTNXTcBPw3vcxToyOCjXwd+LUw4fwY8BzwN/CrtnPuAW8LO/cWMXk6RjGhZdZE8YWaT3L0zHGV1J/C6u38l13GJDKcah0j+uCHsLN9C0Dz2jdyGIzIy1ThERGRMVOMQEZExUeIQEZExUeIQEZExUeIQEZExUeIQEZEx+f98kjZC8mDzUAAAAABJRU5ErkJggg==\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"
    },
    {
     "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"
    },
    {
     "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"
    },
    {
     "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"
    },
    {
     "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": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEKCAYAAAAIO8L1AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAAuRUlEQVR4nO3deXxV1bn/8c+TeYKEIQRIGAIkYkQGCaBSEK0Dtk5tHUBta+uV20E7Xb21t7O30+/a3rZWWqe2VlulatVLKy22VsUJJSgig0CYJIAkzEkg8/P745zYEE9CgLNzMnzfr9d5kb322ns/y5g8WXvtvZa5OyIiIq3FxToAERHpmpQgREQkIiUIERGJSAlCREQiUoIQEZGIlCBERCSihFgHEC0DBw70kSNHxjoMEZFuZfny5bvdPTvSvh6TIEaOHElJSUmswxAR6VbMbGtb+3SLSUREIlKCEBGRiJQgREQkIiUIERGJSAlCREQiUoIQEZGIAk0QZjbbzNaZWamZ3Rph/0/NbEX4s97M9rfYN9zMnjaztWa2xsxGBhGju/O3Ve+yt7ouiNOLiHRbgSUIM4sH5gMXAkXAXDMralnH3b/s7hPdfSLwC+DxFrsfAG5395OBqUB5EHFu2XOIz/1hOXc/vzGI04uIdFtB9iCmAqXuvsnd64AFwKXt1J8LPAwQTiQJ7v53AHevcvdDQQSZPzCdyybm8rtXtlB+sCaIS4iIdEtBJohcYFuL7bJw2fuY2QggH/hnuKgQ2G9mj5vZG2Z2e7hH0vq4eWZWYmYlFRUVxx3oF88toL7R+eVz6kWIiDTrKoPUc4DH3L0xvJ0AzABuBqYAo4DrWh/k7ve4e7G7F2dnR5xKpENGDEjnyuI8Hnr1HbbvP3zc5xER6UmCTBDbgWEttvPCZZHMIXx7KawMWBG+PdUAPAmcFkSQzW46pwCAXzyzIcjLiIh0G0EmiGVAgZnlm1kSoSSwsHUlMxsL9ANeaXVslpk1dwvOAdYEGCtDs1K5etpwHl1expbd1UFeSkSkWwgsQYT/8r8RWAysBR5x99VmdpuZXdKi6hxggbt7i2MbCd1eesbM3gIMuDeoWJt97uzRJMXH8bN/rA/6UiIiXZ61+L3crRUXF3s0pvv+0V/f5u4lG1n8pZkU5vSJQmQiIl2XmS139+JI+7rKIHWXccOMfNzhmbWBvHYhItJtKEG0MiAjmczURLbvD+S1CxGRbkMJIoLcrFR27NdLcyLSuylBRJDbL5Xt+/Q+hIj0bkoQEeRmpbJ9/2F6ygC+iMjxUIKIIDcrlaraBg7WNMQ6FBGRmFGCiCC3XyqAbjOJSK+mBBFBblY4QWheJhHpxZQgIhjanCD26VFXEem9lCAiGJiRRHJCHDsO6FFXEem9lCAiMLPQk0wagxCRXkwJog1Ds1Ip0xiEiPRiShBtCL1NrQQhIr2XEkQbcvulUlFZS01949Eri4j0QEoQbWh+kmmnBqpFpJdSgmhD87sQus0kIr2VEkQb8vQ2tYj0coEmCDObbWbrzKzUzG6NsP+nZrYi/FlvZvtb7e9rZmVmdmeQcUYyODMFM/Qkk4j0WglBndjM4oH5wHlAGbDMzBa6+5rmOu7+5Rb1bwImtTrNfwNLgoqxPYnxceT0SdEtJhHptYLsQUwFSt19k7vXAQuAS9upPxd4uHnDzCYDOcDTAcbYLq0LISK9WZAJIhfY1mK7LFz2PmY2AsgH/hnejgN+Atzc3gXMbJ6ZlZhZSUVFRVSCbql5XQgRkd6oqwxSzwEec/fmlw4+Byxy97L2DnL3e9y92N2Ls7Ozox7U0KxUdh44TFOTFg4Skd4nsDEIYDswrMV2XrgskjnA51tsnwHMMLPPARlAkplVufv7BrqDlNsvlfpGp6Kqlpy+KZ15aRGRmAsyQSwDCswsn1BimANc3bqSmY0F+gGvNJe5+zUt9l8HFHd2cgDIC78LUbbvsBKEiPQ6gd1icvcG4EZgMbAWeMTdV5vZbWZ2SYuqc4AF3gUXgB6qhYNEpBcLsgeBuy8CFrUq+1ar7e8c5Rz3A/dHObQOaV56VI+6ikhv1FUGqbukjOQEMlMT9airiPRKShBHMVSPuopIL6UEcRRaF0JEeisliKPI09vUItJLKUEcRW5WKpW1DRw4XB/rUEREOpUSxFEM1boQItJLKUEcRa7WhRCRXkoJ4iiGhRPEi6W7YxyJiEjnUoI4igEZyVwzbTj3v7yF59dHf8ZYEZGuSgmiA755URGFORn8xyMrKK+sOWLf2+8e5Op7l/Lqpj0xik5EJBhKEB2QkhjPnVefRmVNA//xyJvvTf/9l5U7+Mj8l3l54x7ufLY0xlGKiERXoHMx9SSFOX341sVFfP2JVdy1ZCMHDzdw1/MbmTyiH2MH9+Gh195hx/7D7z31JCLS3akHcQyunjqcC8cN5n/+to67nt/ItacP5+EbTuczZ43GHf60vN31jUREuhX1II6BmfGjj46nrqGJC04ZzJVTQushDeufxpmjB/Do8jI+f/YY4uIsxpGKiJw49SCOUWZaIr++bsp7yaHZlcXDeGfvIV7dvDdGkYmIRJcSRJRccMpg+iQn8GjJtliHIiISFUoQUZKaFM/FE4eyaNVOKms0b5OIdH+BJggzm21m68ys1Mzet6a0mf3UzFaEP+vNbH+4fKKZvWJmq81spZldFWSc0XJl8TBq6pv4y8qd75Xtrqrl5//YwPxnS/nLyh2s2n6AqtqGGEYpItIxgQ1Sm1k8MB84DygDlpnZQndf01zH3b/cov5NwKTw5iHgE+6+wcyGAsvNbLG77w8q3miYkJdJYU4Gj5Rs47KJufz6xU3c9fwmqusaaL3i9kXjh/CNDxcxODMlNsGKiBxFkE8xTQVK3X0TgJktAC4F1rRRfy7wbQB3X99c6O47zKwcyAb2BxjvCTMzrpg8jO8vWsvM25+lorKW84ty+OqFYxmSmcKW3YfYuqeaFdv2c//LW/jn2+V84YMFfHp6PkkJutsnIl1LkAkiF2g5YlsGTItU0cxGAPnAPyPsmwokARsj7JsHzAMYPnz4iUccBR85LZc7ntlAblYq868+jan5/d/bVzS0L0VD+3LhqUO49vQRfPfPa/jRX9/mkZJtTMjLoqa+kdqGJprcuemcMUwe0b+dK4mIBMu89b2PaJ3Y7HJgtrv/W3j748A0d78xQt2vAnnuflOr8iHAc8An3X1pe9crLi72kpKSaIV/Qg7XNZKSGIfZ0d+H+Ofbu/jJ0+uprGkgOSGO5MQ4du6vITE+jsVfmklmWmInRCwivZWZLXf34kj7guxBbAdaviyQFy6LZA7w+ZYFZtYXeAr4+tGSQ1eTmhTf4brnjM3hnLE5R5StLNvPR375Mt/582p+etXEKEcnItIxQd74XgYUmFm+mSURSgILW1cys7FAP+CVFmVJwBPAA+7+WIAxdknj87K48ewxPPHGdv62aufRDxARCUBgCcLdG4AbgcXAWuARd19tZreZ2SUtqs4BFviR97quBGYC17V4DHZiULF2RTeeM4ZxuX35rydWsbuqtsPH1dQ38mjJNnYe0Ap4InJiAhuD6GxdaQwiWtbvquSiX7zIrMJs7v745HbHNBqbnD+9XsZP/76enQdqGJ2dzuOfna4xDBFpV3tjEHq2sgsrzOnDzecX8vSaXfz6xc1t1nv27XJm/2wJ//nYSgb1TeG7l5zCO3sP8dk/LKe+sakTIxaRnkSzuXZx139gFMu37uN7T63lYE0DXz634L2eRF1DEz9YtJb7X97CqIHp/Oqa05g9bjBmRkZyAv/x6Jt844lV/Ohjp753zOodB1jw2jZGDkzn8tPy1MMQkTYpQXRx8XHG/KtP47+eeIs7ntnAvuo6vnPJKeyuquXzf3idkq37+PT0fL72obEkxv+rQ/ixyXls3l3Nnc+WMio7nSn5/Zn/z1KeebucpIQ46hqauH3x21w6IZdrTx/BqXmZMWyliHRFShDdQEJ8HP/vY+Ppl57E3c9vomzfId7afpBDdQ38Yu4kLp4wNOJxXzmvkM17qvnhX98GoF9aIv9xXiGfOHMkZfsO8ful7/DkG9v5Y8k2zivK4fuXjWNQX039ISIhGqTuZu5ZspEfLHqbUdnp3H3tZApy+rRbv6a+ke/+eTWjszOYO3U46clH/k1wsKaeB1/Zyh3PbCA5IY5vXlTE5ZPzOvSSXySNTc7/rdjO/kP19EtPpF9aEv3SkhgzKON9127vHAYRF15qbHJeKt3NyAHpDB+Qdlwxisi/tDdIrQTRDa3ZcZARA9I6/Au3Izbvruarj63ktS17mVmYzXlFoZf3DDCDsYP7MCEvi4T4tp9r2HngMF/+4wqWbnr/okkJccb4vExOHzWAM0YPYGp+f5ITjnyhsL6xid+9vIWf/n09AzKSmTt1OFcU5zEwI5m6hiaefGM7v3p+I5t3V5MYb3zijJF84ZyC942jlFfW4A4D0pMixtvY5MQZx50ERXoSJQjpkKYm5/evbuVHf32bQ3WN79vfJzmBM0YPYEZhNpOGZTFmUAYpiaFf8n99aye3Pv4W9Y1NfOfiUzivKId9h+rYd6iePVW1rNi2n6Wb9rCy7AANTU5maiKXThzK5ZPzODU3k5Kt+/jGE6tYt6uSswqzqalv5NXNe0mMN84+aRCrth9gx4EaxuX25YYZo1i6aQ9/XLaNvqmJfOGcAgZkJPHKxj0s3bSHLXsOAaHE1i8tiYEZSTQ5VNbUU1nTwKG6RhLijL6pifRNSSAzNZEzxwzkU2eO1C026XWUIOSYHK5rpKq2ASf0/0ZDo7Ni235e2FDBkvW72b4/9BJenMGIAelkZyTz2pa9TMjL5GdzJpE/ML3Nc1fXNvDq5j08+cYOFq9+l9qGJob1T2Xb3sPkZqXy7YuLOK8oBzOjtLySh17dxv+t2M6o7HQ+f/YYzirMfu8v/7U7D/L9p9byYuluAPqkJDAtfwCnj+pPSmI8u6tqqagMfRLijT7JiWSkJJCRnEBDUxMHDtdz4HADFZU1vLZ5LwlxcVw2aSjzZo4iOSGeVzbu4ZVNe3ht816qakNzZaUkxpOcEMcpQ/vy4fFDmVEw8L0kKdIdKUFI1Lg7W/ccYs3Og6x7t5L1uyrZvLuac0/O4YvnFhzxJNXRHDhcz19W7uBvq95lfF4mnz97DGlJx3bbzN15/Z39JCfEcfKQvsRHGLfoiC27q7nvxU08WlJGbcO/3h0ZkJ7E6aMHMDA9idqGJmrqG6mua+S1zXs5cLiejOQEzj15EJ+ans+EYVnHdW2RWFKCEOmgPVW1PLq8jJSEOM4cM5CCQRkRxyrqG5t4eeMeFq3cyd9Wv8vBmnquO3MkN59/UlTHhkSCpgQhEqDKmnpuX7yOB5duZWhmKt+7bBxnjx0U67BEOkQJQqQTLN+6l6/+6S1Ky6uYOCyLswqzmVmYzcRhWcd960skaEoQIp2ktqGR+1/awqJV77KybD/u0DclgYEZyVTXNXCotpFD9Y2clNOHyyfncdmkXPqnJ8U6bOnFlCBEYmBfdR0vlu7mpdLdVNc1kp4UT1pSAkkJcby8cTcryw6QGG98cGwOs8cNZmp+f4ZmpcY6bOlllCBEuqC33z3IYyVlPLliO7ur6gDI65fK1JH9yclMocmdpianyUOPFMfFGQlxRnxcHKfmZnL2Sdntvrgo0hFKECJdWGOT8/a7B3lt815e27yXZVtCj9CaGfFmxBk40NAUShgNTaGf2UF9krmiOI85U4YzrL+mHZHjowQh0oM0NDbx7LoKHn7tHZ5bV06Tw4XjBnPrhWMZMaDtlxRFIonZgkFmNtvM1plZqZndGmH/T1ssKbrezPa32PdJM9sQ/nwyyDhFupOE+DjOK8rhN9dN4cWvnsNN54zhuXUVnPe/S/jhorUcrKmPdYjSQwTWgzCzeGA9cB5QBiwD5rr7mjbq3wRMcvdPm1l/oAQoJtS7Xg5Mdvd9bV1PPQjpzXYdrOF//raOP71exoD0JM4YPYD6xibqG536xiZSE+PJ7pP83mda/gDGDMqIddjSBbTXgwjylc+pQKm7bwoHsQC4FIiYIIC5wLfDX18A/N3d94aP/TswG3g4wHhFuq2cvin85MoJfPLMEdy+eB2rdxwkKT6OxAQjMT6O8oO1lGzdx97q0GC4GZxflMNnZ41hoqYIkTYEmSBygW0ttsuAaZEqmtkIIB/4ZzvH5gYQo0iPMj4viwevj/hjBoSmCHn3QA2Plmzj/pe3sHj1Ls4cPYCLJwxl5IB08gemk9M3WVOhC9B1VpSbAzzm7u+fY7odZjYPmAcwfPjwIOIS6VES4+MY1j+Nr5x/EvPOGs1Dr27lvhc287XH33qvTmpiPH1SEnBCkyG6w8UThvKti4oiLuIkPVeQCWI7MKzFdl64LJI5wOdbHTur1bHPtT7I3e8B7oHQGMTxhyrS+2QkJzBv5miu/8Aoduw/zJY91WzZXc3m3Yc4VNeAhR+x3Vtdx/0vb6GusYnvXzZOvYteJMgEsQwoMLN8Qr/w5wBXt65kZmOBfsArLYoXAz8ws37h7fOBrwUYq0ivFR9nDOufxrD+acwoyH7ffnfn9sXr+OVzG0mKj+PbFxcpSfQSgSUId28wsxsJ/bKPB37j7qvN7DagxN0XhqvOARZ4i8ep3H2vmf03oSQDcFvzgLWIdC4z45YLTqKuoYn7XtxMckIct144VkmiF9CLciLSIe7Ot/5vNQ8u3cq5Jw/igyfn8IExA/UWdzcXq8dcRaQHMTO+e8kpZKYm8ujybfxjbTkAIwakcdWUYcybMUpzQ/Uw6kGIyDFzdzZWVPHCht08s7acF0t3c2puJj++YgInDe4T6/DkGGguJhEJ1KK3dvLNJ1dRWdPAF88t4N9nqjfRXcRsLiYR6R0+dOoQnv7yTM4ryuH2xeu44YESGpt6xh+fvZkShIhExYCMZOZfcxrfubiIZ9dV8MNFa2MdkpygDg1Sm1k6cNjdm8ysEBgL/NXdNW2kiBzhuun5bNlziPte3ExhTh+unDLs6AdJl9TRHsQSIMXMcoGngY8D9wcVlIh0b9/48Ml8YMxAvv7kWyzboleYuquOJghz90PAR4FfuvsVwCnBhSUi3VlCfBzzrz6NvH5pfObB5WzbeyjWIclx6HCCMLMzgGuAp8Jl8cGEJCI9QWZaIvd9spi6xiau/90yLWTUDXU0QXyJ0FxIT4SnyxgFPBtYVCLSI4zOzuCuayezqaKaz/5+OXUNTbEOSY5BhxKEuz/v7pe4+/8zszhgt7t/IeDYRKQHmD5mID/62HheKt3D1x5/i57y7lVv0KEEYWYPmVnf8NNMq4A1ZnZLsKGJSE9x+eQ8vnRuAX96vYyfP7Mh1uFIB3X0FlORux8ELgP+Smj1t48HFZSI9Dxf/GABl0/O42f/2MAjJduOfoDEXEcn60s0s0RCCeJOd683M/UTRaTDzIwffORUdh2s4dY/rSQ5IY5LJ2ol4a6soz2Iu4EtQDqwJLyG9MGgghKRnikpIY57Pl7MlJH9+fIfV/DUyp2xDkna0dFB6jvcPdfdP+QhW4GzA45NRHqg1KR4fnPdFCaP6McXFrzB31YpSXRVHR2kzjSz/zWzkvDnJ4R6EyIixyw9OYHffmoqE/IyufGhN/jHml2xDkki6Ogtpt8AlcCV4c9B4LdHO8jMZpvZOjMrNbNb26hzpZmtMbPVZvZQi/L/CZetNbM7TOsbivQoGckJ3P/pqRQN7ctND7/Bqu0HYh2StNLRBDHa3b/t7pvCn+8Co9o7wMzigfnAhUARMNfMilrVKSD0At50dz+F0At5mNmZwHRgPDAOmAKc1eFWiUi30Dcl9LZ1v7RE/u13Jew6WBPrkKSFjiaIw2b2geYNM5sOHD7KMVOB0nBCqQMWAJe2qnMDMN/d9wG4e3m43IEUIAlIBhIB9UFFeqBBfVK475NTOFhTzw0PlHC4rjHWIUlYRxPEZ4D5ZrbFzLYAdwL/fpRjcoGWDzuXhctaKgQKzewlM1tqZrMB3P0VQlN57Ax/Fru7JpcX6aGKhvbl53Mm8db2A9z82Js0abGhLqGjTzG96e4TCN3yGe/uk4BzonD9BKAAmAXMBe41sywzGwOcDOQRSirnmNmM1geb2bzmgfOKiooohCMisXJeUQ63zh7LUyt38qvnN8Y6HOEYV5Rz94PhN6oBvnKU6tuBliuF5IXLWioDFrp7vbtvBtYTShgfAZa6e5W7VxF6e/uMCPHc4+7F7l6cnZ19LE0RkS5o3sxRXDhuMHf+s5SKytpYh9PrnciSo0d7qmgZUGBm+WaWBMwBFraq8ySh3gNmNpDQLadNwDvAWWaWEH6D+yxAt5hEejgz45YLTqK2oZFfPadeRKydSIJo9yahuzcANwKLCf1yfyQ8VfhtZnZJuNpiYI+ZrSE05nCLu+8BHgM2Am8BbwJvuvufTyBWEekmRmVn8LHT8vj9q1vZeeBoz8JIkKy9qXfNrJLIicCAVHfv6FxOgSsuLvaSkpJYhyEiUbBt7yHO+clzXDVlGN+77NRYh9Ojmdlydy+OtK/dHoS793H3vhE+fbpSchCRnmVY/zSuLB7GH5dt03KlMXQit5hERAJz4zljMDPuaLF+RFOT8/o7+9ixX7eeOoN6ASLSJQ3JTOXaaSP43Stb+ODJg1i+dR9/WbmTnQdqGDu4D3/94gw0A0+w1IMQkS7rs7NGkxQfx2d+/zr3v7yFU4b25RNnjODtdyt5dl350U8gJ0Q9CBHpsrL7JPOra0+jorKW84sGk5mWSH1jE8+sLeeXz27knLE5sQ6xR1MPQkS6tFknDeKK4mFkpiUCkBgfxw0z8inZuo9lW/bGOLqeTQlCRLqdq6YMZ0B6Er98tjTWofRoShAi0u2kJsXzqekjeXZdBWt2aPXjoChBiEi39PHTR5KeFK+J/QKkBCEi3VJmWiLXnj6Cp1buYOue6liH0yMpQYhIt3X9B/JJiIvj3hc2xTqUHkkJQkS6rUF9U/jQqYP585s7qW9sinU4PY4ShIh0axeNH8qBw/W8VLo71qH0OEoQItKtzSgcSJ+UBP6ycmesQ+lxlCBEpFtLTojn/KLBLF79LnUNus0UTUoQItLtXTR+CJU1DbywQWvTR5MShIh0e9PHDCQzNZGndJspqpQgRKTbS0qI44JTcnh6zS5q6htjHU6PEWiCMLPZZrbOzErN7NY26lxpZmvMbLWZPdSifLiZPW1ma8P7RwYZq4h0bxeNH0pVbQNL1us2U7QEliDMLB6YD1wIFAFzzayoVZ0C4GvAdHc/BfhSi90PALe7+8nAVECTv4tIm84YPYB+aYl6mimKguxBTAVK3X2Tu9cBC4BLW9W5AZjv7vsA3L0cIJxIEtz97+HyKnfXwrQi0qbE+DhmjxvCP9bqNlO0BJkgcoFtLbbLwmUtFQKFZvaSmS01s9ktyveb2eNm9oaZ3R7ukRzBzOaZWYmZlVRUqFsp0ttdNH4Ih+oaefZt3XCIhlgPUicABcAsYC5wr5llhctnADcDU4BRwHWtD3b3e9y92N2Ls7OzOylkEemqpuX3Z2BGEgvf3BHrUHqEIBPEdmBYi+28cFlLZcBCd693983AekIJowxYEb491QA8CZwWYKwi0gMkxMdx6cRc/rF2FxWVtbEOp9sLMkEsAwrMLN/MkoA5wMJWdZ4k1HvAzAYSurW0KXxslpk1dwvOAdYEGKuI9BBzpw6nvtF5dPm2o1eWdgWWIMJ/+d8ILAbWAo+4+2ozu83MLglXWwzsMbM1wLPALe6+x90bCd1eesbM3gIMuDeoWEWk5xgzKINp+f1Z8No2mpo81uF0a+beM/4DFhcXe0lJSazDEJEuYOGbO/jCw2/wwKenMrNQ45PtMbPl7l4caV+sB6lFRKLuglNy6J+exEOvvhPrULo1JQgR6XGSE+K5YnIef1+7i10Ha2IdTrelBCEiPdLcqcNpbHIeWXbkYPXWPdWUK2l0iBKEiPRIIwem84ExA1mwbBuNTU5FZS1fe/wtzv7xc1z/uxJ6yvhrkJQgRKTHunracLbvP8zNj77JrNuf5dGSbUwe0Y+3th/gzbIDsQ6vy1OCEJEe67yiHLL7JPPEG9s5c8xAnv7yTH5z3RTSkuL5w9KtsQ6vy0uIdQAiIkFJjI/jt9dNoaa+keKR/d8rv2xSLo+/XsY3PlxEZlpiDCPs2tSDEJEebVxu5hHJAeDqqcOpqW/iT6+XxSiq7kEJQkR6nXG5mUwclsUfXt2qwep2KEGISK907ekj2FhRzaub98Y6lC5LCUJEeqWLxg+hb0oCv9dgdZuUIESkV0pJjOfyycNYvPpdTQ3eBiUIEem1rjk9NDX4b1/a3G3HIrbtPcTe6rpAzq0EISK91ujsDM49OYdfPreRK+56hWVbut94xDeeXMW1970ayLmVIESkV/vVtafxg4+cyjt7D3HFXa9w/f3L2LCrMtZhdVhpeRUFORmBnFsJQkR6tcT4OK6eNpznbzmb/5x9Eq9t2cuH73iR+17Y1OUXHDpU18D2/YcZk60EISISmNSkeD43awzP3TyLmYXZfO+ptXziN6/x7oGuO/PrxvJqILSKXhACTRBmNtvM1plZqZnd2kadK81sjZmtNrOHWu3ra2ZlZnZnkHGKiDQbkJHMvZ+YzA8/eirLt+7jgp8t4Zm1u2IdVkSlFaFbYd0uQZhZPDAfuBAoAuaaWVGrOgXA14Dp7n4K8KVWp/lvYElQMYqIRGJmzJ06nEVfnMGw/ql89vevs3xr1xvALi2vIiHOGDEgPZDzB9mDmAqUuvsmd68DFgCXtqpzAzDf3fcBuHt58w4zmwzkAE8HGKOISJvyB6bz++unMTQrhXkPLGfb3kOxDukIG3ZVMWJAGkkJwfwqDzJB5AItl3IqC5e1VAgUmtlLZrbUzGYDmFkc8BPg5vYuYGbzzKzEzEoqKiqiGLqISEhWWhK/vm4K9Y1NXP+7ZVTW1Mc6pPeUVlQFdnsJYj9InQAUALOAucC9ZpYFfA5Y5O7tTrXo7ve4e7G7F2dnZwcdq4j0UqOzM/jVtZPZWFHNTQ+/QUNjU6xDoq6hia17DlEwqE9g1wgyQWwHhrXYzguXtVQGLHT3enffDKwnlDDOAG40sy3Aj4FPmNmPAoxVRKRd08cM5LZLT+G5dRX8z+J1sQ6HrXuqaWzybtuDWAYUmFm+mSUBc4CFreo8Saj3gJkNJHTLaZO7X+Puw919JKHbTA+4e8SnoEREOss100ZwzbTh3PvCppgPWm8orwKCe4IJAkwQ7t4A3AgsBtYCj7j7ajO7zcwuCVdbDOwxszXAs8At7r4nqJhERE7Uf33oZIZmpvKfj62kpr4xZnGUhhPEqOxgnmCCgMcg3H2Ruxe6+2h3/3647FvuvjD8tbv7V9y9yN1PdfcFEc5xv7vfGGScIiIdlZ6cwA8+eiobK6qZ/2xpzOIoLa8ir18qaUnBrRwd60FqEZFu56zCbD52Wh6/em4ja3YcjEkMG8qDfYIJlCBERI7LNy86may0RL76p5Wd/lRTY5OzqaIqsDmYmilBiIgch6y0JG67dBxvbT/AfS9u7tRrb993mNqGpsBmcW2mBCEicpwuHDeYc8YO4q7nN3K4rvMGrIOeg6mZEoSIyHEyM/595ij2H6rnyRWtX/MKzoZd4Udcs4N7SQ6UIERETsjU/P4UDenbqcuWlpZXMTAjmcy0xECvowQhInICzIzrpo9k/a4qXt7YOa9xlVZUURDw7SVQghAROWGXTBjKgPQkfvtS8IPV7k7pruAfcQUlCBGRE5aSGM/V04bzzNvlbN1THei1yitrqaxtUIIQEekurj19BPFm3P/ylkCv0zzFhm4xiYh0Ezl9U/jw+CE8WlIW6JoRpZ0wSV8zJQgRkSj51PR8qmobeGx5u0vZnJAN5ZX0SUkgu09yYNdopgQhIhIlE4dlcdrwLO5dsimwmV7XvVtJYU4fzCyQ87ekBCEiEkU3X3ASOw7U8MArW6J+7tqGRt4sO8CkYVlRP3ckShAiIlF05uiBzDopmzv/Wcr+Q3VRPfeq7Qepa2iieGS/qJ63LUoQIiJRduuFY6msbYj6ehHNq9hNHtE/qudtixKEiEiUjR3cl8tPy+N3L29l295DUTtvyZZ9jBiQ1ikD1BBwgjCz2Wa2zsxKzSzimtJmdqWZrTGz1Wb2ULhsopm9Ei5baWZXBRmniEi0feX8Qszgf/++/piOa2zyiOtLuDvLt+6juJN6DxBggjCzeGA+cCFQBMw1s6JWdQqArwHT3f0U4EvhXYeAT4TLZgM/M7OsoGIVEYm2IZmpfPoD+TzxxnZWbT/Q4eNuevh1/u2BkveVb95dzZ7quk4bf4BgexBTgVJ33+TudcAC4NJWdW4A5rv7PgB3Lw//u97dN4S/3gGUA9kBxioiEnWfnTWafmmJ3L54XYfq19Q38szacp5bV/G+W1MlW/cBUDyiZySIXGBbi+2ycFlLhUChmb1kZkvNbHbrk5jZVCAJ2Bhh3zwzKzGzkoqKiiiGLiJy4vqmJHLDzFE8v76iQ72IVzfvpbYhdHtp4Zs7jti3fMs+MlMTGR3wMqMtxXqQOgEoAGYBc4F7W95KMrMhwIPAp9z9fTfl3P0edy929+LsbHUwRKTruWbaCDKSE7h7yaaj1l2yvoKkhDgm5GXy+OtlR6wvUbJ1L5NH9CMuLvgX5JoFmSC2A8NabOeFy1oqAxa6e727bwbWE0oYmFlf4Cng6+6+NMA4RUQCk5mayDWnD+eplTuOOtPrkvUVTMvvz1VThrOxoppV2w8CsLe6jo0V1UzuxNtLEGyCWAYUmFm+mSUBc4CFreo8Saj3gJkNJHTLaVO4/hPAA+7+WIAxiogE7vrp+STExXHvC233InbsP8yG8ipmFAzkw6cOISk+jifeCP1NvTwG4w8QYIJw9wbgRmAxsBZ4xN1Xm9ltZnZJuNpiYI+ZrQGeBW5x9z3AlcBM4DozWxH+TAwqVhGRIA3qm8LHJufySEkZFZW1Eeu8uGE3ADMLs8lMS+ScsYNY+OYOGhqbKNm6l8R4Y0InTbHRLNAxCHdf5O6F7j7a3b8fLvuWuy8Mf+3u/hV3L3L3U919Qbj89+6e6O4TW3xWBBmriEiQ5s0cTX1jU5urzj2/oYKcvsmclNMHgMsm5bK7qpaXNu5h+ZZ9jMvNJCUxvjNDjvkgtYhIr5A/MJ0Lxw3mwaVb37deRGOT8+KG3cwoyH5vltazx2aTmZrIH5e9w8qyA51+ewmUIEREOs1nzhpNZU0Df3j1nSPKV5bt58DhemYW/utpzOSEeD48fgiL3nqXusamTpt/qSUlCBGRTjI+L4uZhdn8/B8bjngvYsn63ZjBjDEDj6j/kUn/enWss59gAiUIEZFO9eMrxpOVlsgND5RQXlkDwJINFYzPzaRfetIRdScP70dev1RGduIEfS0pQYiIdKJBfVK49xPF7D9Uz7wHllNeWcOKbfuZUfD+l33j4oyfz5nE7VdMiEGkShAiIp1uXG4mP71qAiu27WfOPUtpbPIjxh9amjyiH1NGdv74AyhBiIjExOxxQ7j5/EI2VVSTkZzApOFZsQ7pfRJiHYCISG/1+bPHsLuqjozkBBLju97f60oQIiIxYmZ855JTYh1Gm7peyhIRkS5BCUJERCJSghARkYiUIEREJCIlCBERiUgJQkREIlKCEBGRiJQgREQkInP3WMcQFWZWAewHDrQozmyxHenrlmUDgd3HefmW5znWOpHKW5epHccm1u1o+bXaoXa0Vd5V2jHC3SNPBOXuPeYD3NPWdqSvW5WVROu6x1InUrna0b3b0apNaofa0e3a0fzpabeY/tzOdqSvW9eP1nWPpU6kcrXjxMS6HR2N4WjUDrWjva+PV4fP0WNuMZ0oMytx9+JYx3Gi1I6uRe3oWtSOY9PTehAn4p5YBxAlakfXonZ0LWrHMVAPQkREIlIPQkREIlKCEBGRiJQgREQkIiWIozCzGWZ2l5ndZ2Yvxzqe42VmcWb2fTP7hZl9MtbxnAgzm2VmL4S/L7NiHc/xMrN0Mysxs4tiHcuJMLOTw9+Lx8zss7GO53iZ2WVmdq+Z/dHMzo91PMfLzEaZ2a/N7LETPVePThBm9hszKzezVa3KZ5vZOjMrNbNb2zuHu7/g7p8B/gL8Lsh42xKNdgCXAnlAPVAWVKxHE6W2OFAFpBCDtkSpDQBfBR4JJsqOidLPyNrwz8iVwPQg421LlNrxpLvfAHwGuCrIeNsSpXZscvfroxJPT36KycxmEvpF8oC7jwuXxQPrgfMI/XJZBswF4oEftjrFp929PHzcI8D17l7ZSeG/JxrtCH/2ufvdZvaYu1/eWfG3FKW27Hb3JjPLAf7X3a/prPjD8UajDROAAYSS3G53/0vnRH+kaP2MmNklwGeBB939oc6Kv1mUf9Z/AvzB3V/vpPDfE+V2nPDPecKJHNzVufsSMxvZqngqUOrumwDMbAFwqbv/EIjY1Tez4cCBWCQHiE47zKwMqAtvNgYYbrui9T0J2wckBxJoO6L0/ZgFpANFwGEzW+TuTUHGHUm0vh/uvhBYaGZPAZ2eIKL0PTHgR8BfY5EcIOo/HyesRyeINuQC21pslwHTjnLM9cBvA4vo+BxrOx4HfmFmM4AlQQZ2HI6pLWb2UeACIAu4M9DIOu6Y2uDuXwcws+sI94gCje7YHOv3YxbwUULJelGQgR2jY/0ZuQk4F8g0szHufleQwR2DY/1+DAC+D0wys6+FE8lx6Y0J4pi5+7djHcOJcvdDhBJdt+fujxNKeN2eu98f6xhOlLs/BzwX4zBOmLvfAdwR6zhOlLvvITSOcsJ69CB1G7YDw1ps54XLupue0g7oGW3pCW1o1lPaonacoN6YIJYBBWaWb2ZJwBxgYYxjOh49pR3QM9rSE9rQrKe0Re04Ucc7p3h3+AAPAzv516Od14fLP0ToqYCNwNdjHWdvaUdPaUtPaENPa4vaEcynRz/mKiIix6833mISEZEOUIIQEZGIlCBERCQiJQgREYlICUJERCJSghARkYiUIKRHM7OqTr5eVNYMsdCaFwfMbIWZvW1mP+7AMZeZWVE0ri8CShAix8TM2p2/zN3PjOLlXnD3icAk4CIzO9paC5cRmh1WJCqUIKTXMbPRZvY3M1tuoZXpxobLLzazV83sDTP7R3i9CczsO2b2oJm9BDwY3v6NmT1nZpvM7Astzl0V/ndWeP9j4R7AH8LTSWNmHwqXLTezO8ys3bUg3P0wsILQrJ6Y2Q1mtszM3jSzP5lZmpmdCVwC3B7udYxuq50iHaUEIb3RPcBN7j4ZuBn4Zbj8ReB0d58ELAD+s8UxRcC57j43vD2W0JTjU4Fvm1lihOtMAr4UPnYUMN3MUoC7gQvD188+WrBm1g8o4F/TtD/u7lPcfQKwltB0DC8Tmp/nFnef6O4b22mnSIdoum/pVcwsAzgTeDT8Bz38a9GhPOCPZjYESAI2tzh0Yfgv+WZPuXstUGtm5UAO71/+9DV3LwtfdwUwktBqYZvcvfncDwPz2gh3hpm9SSg5/Mzd3w2XjzOz7xFaDyMDWHyM7RTpECUI6W3igP3he/ut/YLQEqYLw4vgfKfFvupWdWtbfN1I5J+ljtRpzwvufpGZ5QNLzewRd18B3A9c5u5vhhccmhXh2PbaKdIhusUkvYq7HwQ2m9kVEFpm0swmhHdn8q959j8ZUAjrgFEtlpW86mgHhHsbPwK+Gi7qA+wM39ZquR53ZXjf0dop0iFKENLTpZlZWYvPVwj9Ur0+fPtmNXBpuO53CN2SWQ7sDiKY8G2qzwF/C1+nEjjQgUPvAmaGE8s3gVeBl4C3W9RZANwSHmQfTdvtFOkQTfct0snMLMPdq8JPNc0HNrj7T2Mdl0hr6kGIdL4bwoPWqwnd1ro7tuGIRKYehIiIRKQehIiIRKQEISIiESlBiIhIREoQIiISkRKEiIhEpAQhIiIR/X+tIEFUncph/QAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEKCAYAAAAIO8L1AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAAuE0lEQVR4nO3deXhV5bn+8e+TGQgJQ8KUQJgCgsgYURGROlS0zlYFrcPRak+tHY+2tj2/tse2p572WFstVrGD1WNLnaUVRa1WkUEJyDwGZAhjGJIAIWR6fn/sDd2ETUhIdnayc3+uK1ezhr3W8zYmN+9613qXuTsiIiK1xUW7ABERaZkUECIiEpYCQkREwlJAiIhIWAoIEREJSwEhIiJhJUS7gKaSkZHhffv2jXYZIiKtysKFC3e7e2a4bTETEH379iU/Pz/aZYiItCpmtulE23SJSUREwlJAiIhIWAoIEREJSwEhIiJhKSBERCQsBYSIiITV5gOivLKaf6zayZa9ZdEuRUSkRWnzAXHwcBV3/imff6zaGe1SRERalDYfEF06JJEUH8f20vJolyIi0qK0+YAwM7qnJ7OzRAEhIhKqzQcEQM+0dmxXQIiIHEMBAXRPT2GnLjGJiBxDAQH0TE9he0k57h7tUkREWgwFBNA9LYXDVTWUHKqMdikiIi2GAoJADwLQOISISAgFBIEeBMAOBYSIyFERDQgzm2Rma8yswMweCLP9ETNbHPxaa2bFtbanmVmhmf0mknUe6UHs0EC1iMhREXujnJnFA1OBi4FCYIGZzXD3lUf2cfdvhuz/VWBUrcP8GPggUjUekdkxGTNdYhIRCRXJHsRYoMDdN7h7BTAduKqO/acAfzmyYGZjgO7AWxGsEYDE+DgyU/WwnIhIqEgGRBawJWS5MLjuOGaWA/QD3g0uxwEPA/fVdQIzu9vM8s0sv6ioqFHF9kxP0XQbIiIhWsog9WTgRXevDi7fA8x098K6PuTu09w9z93zMjMzG1VA97QU9SBEREJEbAwC2Ar0DlnODq4LZzLwlZDlc4DzzOweIBVIMrMD7n7cQHdT6ZmewvwNeyJ1eBGRVieSAbEAyDWzfgSCYTJwU+2dzOw0oDMw78g6d785ZPvtQF4kwwEC022UlldRVlFF+6RI/t8iItI6ROwSk7tXAfcCs4BVwPPuvsLMHjSzK0N2nQxM9yjPc3H0VlddZhIRASLbg8DdZwIza637Qa3lH53kGE8DTzdxaccJfViuf2ZqpE8nItLitZRB6qjrmd4O0MNyIiJHKCCCeqRpPiYRkVAKiKB2SfGkt0vUeyFERIIUECF6pKWoByEiEqSACNFDb5YTETlKARFCPQgRkX9RQITokZ7C7gOHqayuiXYpIiJRp4AI0SM9BXfYtf9wtEsREYk6BUSIHkefpj4U5UpERKJPARHiX9NtqAchIqKACPGvh+XUgxARUUCESG+XSEpinG51FRFBAXEMM9OtriIiQQqIWvSwnIhIgAKiFvUgREQCFBC19Ehvx87Scmpqovr+IhGRqFNA1NIjLZnKamdvWUW0SxERiSq9fLmWHsEXBxXuO8TyrSXMWLyN+Rv2MO3WPIZlpUe5OhGR5qOAqOXI09Q3PjmPw1U1pKUkcOBwFW+v3KmAEJE2RQFRy8BuqQztmUa/jA5cNbIX5w/O5Jqpc1m0eV+0SxMRaVYKiFpSkxOY+fXzjlk3Jqczr3yyleoaJz7OolSZiEjz0iB1PYzO6cSBw1Ws3bk/2qWIiDQbBUQ9jOnTBYCFm3SZSUTajogGhJlNMrM1ZlZgZg+E2f6ImS0Ofq01s+Lg+pFmNs/MVpjZUjO7MZJ1nkzvLu3ISE1mkQJCRNqQiI1BmFk8MBW4GCgEFpjZDHdfeWQfd/9myP5fBUYFF8uAW919nZn1Ahaa2Sx3L45UvXUxM8bkdGKhBqpFpA2JZA9iLFDg7hvcvQKYDlxVx/5TgL8AuPtad18X/H4bsAvIjGCtJzUmpzOb9pRRpLfNiUgbEcmAyAK2hCwXBtcdx8xygH7Au2G2jQWSgPVhtt1tZvlmll9UVNQkRZ/ImJzOALrdVUTajJYySD0ZeNHdq0NXmllP4Fng39y9pvaH3H2au+e5e15mZmQ7GKf3SicpPk7jECLSZkQyILYCvUOWs4PrwplM8PLSEWaWBrwOfN/d50ekwgZISYxnWFbacT2IyuoaPlhbRFX1cfklItKqRTIgFgC5ZtbPzJIIhMCM2juZ2WlAZ2BeyLok4BXgGXd/MYI1NsiYnM4sKSyhoupfYfDwW2u59Q8f89t/HncFTESkVYtYQLh7FXAvMAtYBTzv7ivM7EEzuzJk18nAdHcPnV/7BmACcHvIbbAjI1VrfY3J6UxFVQ0rtpUA8MnmfUz7YD0dkxN49N11rN5RGuUKRUSaTkTHINx9prsPcvcB7v7T4LofuPuMkH1+5O4P1Prc/7l7oruPDPlaHMla62N0n8BA9cJN+yivrOa+F5bQIy2F1792Hmkpidz3whIqdalJRGJESxmkbhW6paXQu0s7Fm3exyNvr2V90UEeum44fbq25ydXD2P51lKefF+XmkQkNmiyvgYa06cz76zaRVlFFVPG9mbCoMDdU5ee0ZPPDe/Jr/+xjouH9mBwj45RrlREpHHUg2igMTmdOXC4ip7p7fjeZUOO2fbglaeTlpLIt55fzKLN+3Rnk4i0aupBNNCEQZlkpCbx888Pp2NK4jHbuqYm89/XnsE9zy3i2sfn0jE5gbP6d+GiId25Ia83cZoqXERaETv25qHWKy8vz/Pz86NdBgB7Dhxm3oY9zCnYw5yC3WzeW8ZPrh7GF87OiXZpIiLHMLOF7p4Xbpt6EBHQNTWZy4f34vLhvXB3bvn9x/xs5irOH5RJ7y7to12eiEi9aAwiwsyMh647AzPjOy8t5WQ9trdX7uT7ryyjuubUenZb9pbxyeZ91Jzi50VEjlAPohlkd27P9y4bwvdeWcZzH20+4aWm/I17+cpzi6iormFMTmeuHZ1d73NUVNXwxPvr+c27BVRU19AzPYXPndGTK0b0Ynh2OmYa/xCRhlEPoplMGdub8QMz+NnMVWzZW3bc9o27D3LXM/lkdW7HaT068qt31tX7obvFW4q58jcf8su313Lx6d15+PoRnN4rjT/N28hVU+fwhd9/RHll9ckPJCISQgHRTI5cagL4jxeWsKHowNFtxWUV3PH0AgD+ePuZfHvSYDbvLeOF/MI6j7llbxnffXkZ1z4+h31lFTx1ax5TbxrNdWOy+d1tZ5L//Yv5z88NYe76Pdzz3CI95S0iDaK7mJrZC/lb+PZLS3GHIT3TuHx4Tz5YW8Qnm4t57q6zOLNvF9yd6347l+0l5bx330RSEuOPOUbBrv08/t56XluyjXgzJo/tzX2XDCat1m23Rzz30Sa+/8pyrhzRi0duHEl8FG+3rayuITH+1P5dUl3jFOw6wKDuqfW+ZObuurwmUgfdxdSCXJ/Xm/G5GbyxbAevL9vOL2atAeDXk0dyZt8uQKC3cd8lg7npqY947qPN3Dm+HwD7yyv5yd9X8fzCLaQkxHP7uL7cdV5/eqSn1HnOm8/KoeRQJT9/cw1p7RL48VXDIvJH0915b80uivYfpsahxp3Kqho27ilj7c79rN15gOKyCr532RDuCLapIX40YwXPzt/ERUO68eOrh9Ezvd0J912xrYT/eXMNq7eX8siNIzl3YEZjmibSJqkHEWXbig+xo7T86ESAoW7+3XxWb9/PB9/+DMu3lvAfLyxhW/Ehvnhef/79/AF06ZDUoHP97I1VPPn+BqaM7c0Dlw4hvV34HsepOFRRzXdeWsqMJduO29YhKZ6B3TsyqFsq20vK+bBgNw9fP4LrxtR/EH76x5t54OVlnJebwYKNe0mIi+M7l57GzWP7HPMA4uY9ZTz89hpeW7yNTu0T6dw+iU17DnLfJYP58vkD1JsQqaWuHoQCogVbtHkf1z4+l5G9O7GksJg+XdrzyxtGMCanyykdz9156I3VTJu9gS7tk7j/ksFcn9e70Zectuwt40vPLmTVjlLu++xgrh2dhWHEGcTFGV07JB39w3y4qpo7nl7A/A17eeILY7h4aPeTHn/hpr1Mnjafs/t35el/G8vWfYf43ivL+LBgN7ndUklNSeBQRTWHKqvZVnyI+DjjjnP78aXzB5AQF7i9+O9Lt3Px0O48fMOIE16KE2mLFBCt2Bf/tIB3Vu3iprP68P3LhtAhufFXBZdvLeFHM1aQv2kfw7LSuH1cP3K7pdI/swMdUxKprK5hzY79LC0sYeX2Es4dkMGlZ/QMe6y5Bbv5yp8XUV3j/HrKKD4zuNtJz3/gcBU3PzWfVTv288wdYzm7f1eKyyrYtKeM3QcOM7RX2tHLRztLy7n8sQ9pnxTPa185l07tA70md+fFhYW8tKiQxPg42iXG0y4pnh7pKfzbuH7HXHZzd/44ZyP/PXMVnTskMen0Hlw4pBvnDOhKckJ82BpF2goFRCtWWl7J5j1lDMtKb9LjujszlmzjZzNXs6O0/Oj6zI7JlB6q5HDwrXlJCXFUVNXw+THZ/NeVpx8NqOKyCn4+aw1/+XgzAzNTmXZrHv0yOtT7/HsPVnDDk/Mo3FdGUnwcpeVVx2zP6dqes/p1YdX2/awvOsAr95zb6BlyF27ay5Pvb2D2ut0cqqymQ1I8Fw3tzpcnDuC0HmnH7LujpJynZm9g056D9MvowIDMVPpnpnJ6r7QmCWmRlkIBISdUWV3Dpj1lrC86wPqiA2woOkhaSiIj+3RiZHYnenZK4bF/rOM37xWQ07UDv7pxJGt27OehN1dTcqiS28f15ZsXDyL1FP5obi85xMNvrSUlMY6cLh3o07U9XToksWRLMR99upePP91LaXklj980+oQ9mFNRXlnNvPV7eHvVTmYs3saBw1Vccnp3vnpBLl06JPHE++uZ/vEWatzpl9GBzXvLjgZmp/aJ3PuZgXzh7Jzj7i4TaY0UENJoH23Ywzf+upjtJYHeRl5OZ3589TCG9Ew7ySdPXU2NU3yossGD8Q1RXFbBH+ds5A9zPmV/eRUJwfGY6/OyuWfiQHp3aU9NjbO1+BDrdu3nT3M38f7aIrI6teNbFw/i6lFZUb1tWKSxFBDSJIrLKvjVO+sYlpXOtaOyYmr68tLySp6dt4nisgpuG9eX7M4nnlRxbsFuHnpzNUsLS+jdpR3XjsrmutHZ9OmqiRil9VFAiDQxd+fN5Tt47qPNzFm/G3c4s29nrhqZxWdP7063jnU/myLSUiggRCJoW/EhXl28lZcWFrK+6CBmcGZOFy4Z1oMxOZ3plZ5CRmry0R5XZXUNuw8cZlfpYSqC058YYAYdkhPo1C6JTu0TNcYhzUIBIdIM3J21Ow/wxvLtvLl8B6t37D+6LSHO6J6WQkUwHOrza5eSGEd6u0TSUhJJa5dIp3aJjBuYwRUjeqqHIk1GASESBZuDU4xsLznEtpJydpSUk5wQR7e0FHqkpdCtYzIpifE4gd/BGoeDh6vYV1ZBcVklxWUVlB6qorS8ktLySnaUlLO+6CBxBucOzODqkVlcMaIXSQmac1NOXdTmYjKzScCvgXjgd+7+UK3tjwCfCS62B7q5e6fgttuA/wxu+4m7/ymStYo0tT5d2zf5wHXBrv28tngbry7eyn+8sIQ/zPmUR24cyaDujXtGRCSciPUgzCweWAtcDBQCC4Ap7r7yBPt/FRjl7neYWRcgH8gDHFgIjHH3fSc6n3oQ0pa4O7NW7OD7ryxn/+Eqvn3JYO44t19M3VkmzaOuHkQk+6ZjgQJ33+DuFcB04Ko69p8C/CX4/SXA2+6+NxgKbwOTIlirSKtiZkwa1pNZ35zA+YMy+cnrq5g8bT5/XbCZ1TtKqdK7P6QJRPISUxawJWS5EDgr3I5mlgP0A96t47NZYT53N3A3QJ8+fRpfsUgrk5GazLRbxvDiwkJ+9sZqvvPSMgDaJcYzPDudG8/szeXDNU4hp6alTCozGXjR3Rv0Xkx3nwZMg8AlpkgUJtLSmRnX5/XmutHZbNxzkKWFJSwpLGb2ut186/kl/PzNNdwxvi+Tx/bRTLbSIJEMiK1A75Dl7OC6cCYDX6n12Ym1PvvPJqxNJObExRn9g5MKXj0qC3fnn2uLeOqDDfz3zNU8+o8CPj8mm9vH9aVvAyZWlLYrkoPUCQQGqS8k8Ad/AXCTu6+otd9pwJtAPw8WExykXgiMDu62iMAg9d4TnU+D1CIntnxrCb+bvYHXl22nqsa5YHA37hjfj3EDuuolSm1cVAap3b0KuBeYBawCnnf3FWb2oJldGbLrZGC6hyRVMAh+TCBUFgAP1hUOIlK3YVnp/GryKOZ85wK++pmBLN5SzM2/+4hrfzuXD9YWESvPQ0nT0oNyIm1QeWU1Ly0qZOq7BWwrKScvpzNfvyiX8QMz1KNoY/QktYiEdbiqmufzA0Gxo7Scwd07cuu4HK4emaUXI7URCggRqVN5ZTUzFm/j6bkbWbm9lI4pCVw3OpvPj8nm9F5p6lXEMAWEiNSLu7No8z7+NHcTbyzfTmW1MyCzA1ePzOLqUVn07qJ3XsQaBYSINFhxWQUzl+3g1cVb+fjTvSTGG1+7IJcvTxxAQrwevIsVCggRaZTCfWX8z5tr+NuSbQzPTufh60eQqwkCY0K05mISkRiR3bk9j00ZxdSbRlO47xCfe/RDnnh/PdU1sfEPTAlPASEi9fa54T1565sTuOC0bjz0xmqu++1cCnbtP/kHpVVSQIhIg2SkJvPbL4zm0Smj2LTnIJcFexOaQTb2KCBEpMHMjCtH9OKtb57PBYODvYkn5lGw60C0S5MmVK+AMLMOZhYX/H6QmV1pZpoWUqSNy+wY6E08FuxNfO7R2fzhw0+p0dhETKhvD+IDIMXMsoC3gFuApyNVlIi0HmbGFSN68dY3JzB+YAYP/n0lN/1uPlv2lkW7NGmk+gaEuXsZcC3wuLtfD5weubJEpLXp1jGF392Wx8+vG87yraV8/om5lByqjHZZ0gj1DggzOwe4GXg9uC4+MiWJSGtlZtxwZm/+ctfZ7D5Qwc9mrop2SdII9Q2IbwDfBV4JTtndH3gvYlWJSKt2RnY6d53Xn+kLtjCnYHe0y5FTVK+AcPf33f1Kd/+f4GD1bnf/WoRrE5FW7BsX5dIvowPffXkZhyoa9DZhaSHqexfTn80szcw6AMuBlWZ2f2RLE5HWLCUxnp9dewab95bxy7fXRLscOQX1vcQ01N1LgauBN4B+BO5kEhE5obP7d+Wms/rw+w8/ZfGW4miXIw1U34BIDD73cDUww90rAd3oLCIn9cClp9GtYwr3vbCEsoqqaJcjDVDfgHgS2Ah0AD4wsxygNFJFiUjsSEtJ5H+vH8H6ogP84LUV0S5HGqC+g9SPunuWu1/mAZuAz0S4NhGJEeNzM/jqBbm8uLCQF/K3RLscqaf6DlKnm9kvzSw/+PUwgd6EiEi9fP3CXM7p35X/99py1u7UDLCtQX0vMf0B2A/cEPwqBf4YqaJEJPbExxm/njKS1ORE7nlukcYjWoH6BsQAd/+hu28Ifv0X0D+ShYlI7OnWMYVHJ49kQ9EBHvzbymiXIydR34A4ZGbjjyyY2bnAoZN9yMwmmdkaMyswswdOsM8NZrbSzFaY2Z9D1v88uG6VmT1qZlbPWkWkBRs3MIM7x/fjr/lb9LKhFq6+AfHvwFQz22hmG4HfAF+q6wNmFg9MBS4FhgJTzGxorX1yCUzhca67n05gSg/MbBxwLjAcGAacCZxfz1pFpIX78sSBtE+M55F31kW7FKlDfe9iWuLuIwj8wR7u7qOAC07ysbFAQfCSVAUwHbiq1j53AVPdfV/wPLuOnBJIAZKAZCAR2FmfWkWk5evSIYk7xvfj9aXbWb1Dd8y3VA16o5y7lwafqAb41kl2zwJC72crDK4LNQgYZGZzzGy+mU0KnmcegckAtwe/Zrm7poUUiSFfHN+fjskJ/Opt9SJaqsa8crQpxgQSgFxgIjAFeMrMOpnZQGAIkE0gVC4ws/OOK8Ds7iO33hYVFTVBOSLSXNLbJ3Lnef14c8UOlm8tiXY5EkZjAuJkU21sBXqHLGcH14UqJDh1h7t/CqwlEBjXAPPd/YC7HyAw/9M5xxXgPs3d89w9LzMz81TbISJRcsf4fqS3S+RX76yNdikSRp0BYWb7zaw0zNd+oNdJjr0AyDWzfmaWBEwGZtTa51UCvQfMLIPAJacNwGbgfDNLCM4BdT6gS0wiMSYtJZG7J/TnnVW7NJlfC1RnQLh7R3dPC/PV0d0TTvLZKuBeYBaBP+7PB1829KCZXRncbRawx8xWEhhzuN/d9wAvAuuBZcASYIm7/61RLRWRFum2cX3p0iGJx/6hsYiWps4/8o3l7jOBmbXW/SDkeycw2P2tWvtUc5LbaEUkNqQmJ/CFs3N47N11bNlbRu8u7aNdkgQ1ZgxCRKRJ3DS2D3Fm/N9Hm6JdioRQQIhI1PVIT+GzQ7vz/IItlFfq9aQthQJCRFqEW87JYV9ZJa8v3R7tUiRIASEiLcI5/bsysFsqz8zXZaaWQgEhIi2CmXHL2Tks2VLM0sLiaJcjKCBEpAW5ZnQW7ZPieXaeehEtgQJCRFqMtJRErhmVxYwl2yguq4h2OW2eAkJEWpRbzsnhcFUNL+QXRruUNk8BISItymk90ji7fxemzd6g15JGmQJCRFqc+y8ZTNH+w/xxzsZol9KmKSBEpMUZk9OFi4d254l/rmfvQY1FRIsCQkRapG9fMpiDFVVMfa8g2qW0WQoIEWmRcrt35PNjsnl23iYK95VFu5w2SQEhIi3WNy4aBAa/fFsvFIoGBYSItFi9OrXj38b15ZVPtrJ6R2m0y2lzFBAi0qJ9eeIAUpMTePy99dEupc1RQIhIi9apfRJXjujF2yt36rmIZqaAEJEW7/LhvThUWc27q3dFu5Q2RQEhIi3e2H5d6NYxmb8t2RbtUtoUBYSItHjxccZlZ/TkvTVF7C+vjHY5bYYCQkRahStG9KKiqoa3V+6MdilthgJCRFqF0X06kdWpnS4zNSMFhIi0CmbG5cN7Mnvdbr0roplENCDMbJKZrTGzAjN74AT73GBmK81shZn9OWR9HzN7y8xWBbf3jWStItLyXT68F1U1zpvLd0S7lDYhYgFhZvHAVOBSYCgwxcyG1tonF/gucK67nw58I2TzM8Av3H0IMBbQ/W0ibdywrDT6dm3P35duj3YpbUIkexBjgQJ33+DuFcB04Kpa+9wFTHX3fQDuvgsgGCQJ7v52cP0Bd9dsXSJtnJlxxYhezF2/m6L9h6NdTsyLZEBkAVtClguD60INAgaZ2Rwzm29mk0LWF5vZy2b2iZn9ItgjEZE27vLhvahxeGO5ehGRFu1B6gQgF5gITAGeMrNOwfXnAfcBZwL9gdtrf9jM7jazfDPLLyoqaqaSRSSaBvfoyNCeaTw7bxM1NR7tcmJaJANiK9A7ZDk7uC5UITDD3Svd/VNgLYHAKAQWBy9PVQGvAqNrn8Ddp7l7nrvnZWZmRqINItIC3T2hP+t2HeC9NRqajKRIBsQCINfM+plZEjAZmFFrn1cJ9B4wswwCl5Y2BD/bycyO/NW/AFgZwVpFpBX53PCeZHVqx5Pvb4h2KTEtYgER/Jf/vcAsYBXwvLuvMLMHzezK4G6zgD1mthJ4D7jf3fe4ezWBy0v/MLNlgAFPRapWEWldEuPjuHN8Pz7euJeFm/ZFu5yYZe6xcQ0vLy/P8/Pzo12GiDSTg4erGPfQu5zdvwtP3pIX7XJaLTNb6O5h/w+M9iC1iMgp6ZCcwK3n5PDWyp2sLzoQ7XJikgJCRFqt28b1JTE+jt/N1lhEJCggRKTVykhN5vox2by0cCu79pdHu5yYo4AQkVbtrvP6U1VTw5/mbox2KTFHASEirVrfjA5ccFp3/rpgC4erqqNdTkxRQIhIq3fLOTnsPlChWV6bmAJCRFq98wZmkNO1Pf83f1O0S4kpCggRafXi4owvnJXDgo37WLW9NNrlxAwFhIjEhM+PySY5IU69iCakgBCRmNC5QxJXjOjFK59sZX95ZbTLiQkKCBGJGbecnUNZRTUvL6o9cbScCgWEiMSMEb07MTw7nWfnbyJW5pmLJgWEiMSUL5ydQ8GuA8zbsCfapbR6CggRiSlXDO9Fx5QEXsgvjHYprZ4CQkRiSrukeC4f3pM3l+/g4OGqaJfTqikgRCTmXDMqm0OV1cxaoSerG0MBISIxJy+nM9md2/HKJ7qbqTEUECISc+LijGtGZTGnYDc7SzUN+KlSQIhITLpmVBY1Dq8tVi/iVCkgRCQm9c9MZUTvTnporhEUECISs64dlcXqHfs1gd8pUkCISMy6YkQvEuLsuMFqPWVdPwoIEYlZXTokMXFwJq8t3kpZRRWvL93OF/+0gEH/+YbGJuohogFhZpPMbI2ZFZjZAyfY5wYzW2lmK8zsz7W2pZlZoZn9JpJ1ikjsumZUNjtLDzP6x2/zlT8vYtnWEjJSk/n1O+uormn9PYk9Bw5HrB0JETkqYGbxwFTgYqAQWGBmM9x9Zcg+ucB3gXPdfZ+Zdat1mB8DH0SqRhGJfRcO6cb4gRn0SE/hmlFZnN2/K28s3869f/6EWSt2cNkZPaNdYqM88PIydpaW89pXzsXMmvTYEQsIYCxQ4O4bAMxsOnAVsDJkn7uAqe6+D8Dddx3ZYGZjgO7Am0BeBOsUkRiWkhjP/33xrGPWXTqsJ/0y1vLbf67n0mE9mvwPa3PZXnKIf6zayZfOHxCRNkTyElMWsCVkuTC4LtQgYJCZzTGz+WY2CcDM4oCHgfsiWJ+ItFHxccaXJvRn2dYSPizYHe1yTtlfF2zBgSln9onI8aM9SJ0A5AITgSnAU2bWCbgHmOnudU7HaGZ3m1m+meUXFRVFulYRiSHXjM6ie1oyj7+3PtqlnJKq6hqmf7yF83Iz6dO1fUTOEcmA2Ar0DlnODq4LVQjMcPdKd/8UWEsgMM4B7jWzjcD/Area2UO1T+Du09w9z93zMjMzI9EGEYlRyQnxfHF8f+Zt2MMnm/dFu5wGe29NETtKy7lpbGR6DxDZgFgA5JpZPzNLAiYDM2rt8yqB3gNmlkHgktMGd7/Z3fu4e18Cl5mecfewd0GJiJyqKWf1Ib1dIr/9Z+vrRTz30Sa6dUzmwiG17+1pOhELCHevAu4FZgGrgOfdfYWZPWhmVwZ3mwXsMbOVwHvA/e6u10CJSLNITU7gtnF9eWvlTgp27Y92OfW2ZW8Z768tYvKZvUmMj9y/8yM6BuHuM919kLsPcPefBtf9wN1nBL93d/+Wuw919zPcfXqYYzzt7vdGsk4RabtuPScHM/j70u3RLqXe/rpgCwbcGMHLSxD9QWoRkajKSE1meHYnPljbOm50qayu4a/5W5g4uBtZndpF9FwKCBFp887PzWDxlmJKyiqjXcpJvbNyJ0X7D3PzWZHtPYACQkSECYMyqXGYs77lPxPx0qKt9ExPYeLgyA1OH6GAEJE2b2TvTnRMSWjxl5kqq2uYt343F5zWjfi4yD/9rYAQkTYvIT6Ocwdk8MHaoiafCvzN5Tu45fcfsWt/4199+snmYg5WVHNebvM896WAEBEhcJlpW0k564sONNkxl2wp5uvTP2H2ut3c9cxCyiurG3W82euKiDM4Z0DXJqqwbgoIERFgwqAMAN5f2zTjEDtKyrnrmXwyUpP5n+vOYGlhMf/x/BJqGjE19+x1uxnZuxPp7RKbpMaTUUCIiADZndvTP7NDk4xDlFdWc/ez+Rw8XMXvb8/jxjP78L1Lh/D6su08/PaaUzpmSVklSwuLGd9Ml5cgstN9i4i0KhNyM5m+YDPlldWkJMaf0jHcnftfXMqyrSVMuyWP03qkAfDF8/qxYfdBpr63nk7tkjitZ0cOVVRzqLKanuntGNuvS53Hnbt+NzUOE3IzTqmuU6GAEBEJOn9QJk/P3ciCjXsbPBC8ZW8Zry3eymuLt7Fu1wG+PWkwFw/tfnS7mfHgVaezZW8ZP5256pjPJsQZ8793IRmpySc8/gfrdpOanMCI3p0aVFdjKCBERILO6t+FpIQ4PlhbdDQg3lu9i5cWFfK9y4bQK8yTy5v3lPGt5xeTvykwI+yZfTvz8+uGc31e9nH7JsbH8fvb81i4aR9J8XGkJMaz92AFt/7hY2Ys3sYd4/uFrcvdmb2uiHMGdI3o3Eu1KSBERILaJyUwtm8XPli7m5vOOsiP/76Sd1cHXnRZVe08ccuYY/Z3d777ylLW7NjPtycN5soRvcjuXPe7GZIT4hk34NjLRGdkpfPyJ4UnDIhNe8oo3HeIuyf0b0TrGk6D1CIiISYMymDNzv189pH3+fjTvXz/siF8/cJc3lyx47gB7NeXbWdOwR7unzSYeyYOPGk4nMi1o7NYvrWUNTvCzyg7O/jWu+Z6/uEIBYSISIjPDu1Bx+QErh6Zxbv3nc9dE/pzz2cG0Ldre340YwUVVTUAHDxcxU/+vorTe6Vx81k5jTrnFSN6kRBnvPxJ+Jdozl5bRFandvSN0JvjTkQBISISom9GB5b91yX84voRdOuYAgQuC/3wytPZsPsgv//wUwAefXcdO0rL+fHVwxo97UVGajITB2fy6idbqa71nERVdQ3z1u9hwqAMzCI/vUYoBYSISD18ZnA3LhrSncfeXcfsdUX8fvan3JjXm9F9OjfJ8a8bnc3O0sPMKTj2Qb0lhcXsP1zF+IHN/1plBYSISD398IqhVNc4t/9xAR2SE/j2pMFNduwLhnQjLSWBlxf96zKTu/O3Jdsxg3HNNL1GKAWEiEg99e7Sni9PHEB1jXP/JYPpWsdzCw2VnBDPFSN68eaKHRw4XMWhimrue2EpT8/dyOXDe9G5Q1KTnau+dJuriEgD3PuZgYwbkEFeTtNcWgp17ehsnvtoM0++v563V+5kzc79fP3CXL52YW6Tn6s+FBAiIg2QEB930mkxTtXoPp3o27U9j71bQKf2ifzx9jOb5cVAJ6KAEBFpIcyMr1+Uy+tLt/NfVw2L+DunT0YBISLSglwzKptrRh0/TUc0aJBaRETCimhAmNkkM1tjZgVm9sAJ9rnBzFaa2Qoz+3Nw3Ugzmxdct9TMboxknSIicryIXWIys3hgKnAxUAgsMLMZ7r4yZJ9c4LvAue6+z8yOjMaUAbe6+zoz6wUsNLNZ7l4cqXpFRORYkexBjAUK3H2Du1cA04Grau1zFzDV3fcBuPuu4P+udfd1we+3AbuA5n+MUESkDYtkQGQBW0KWC4PrQg0CBpnZHDObb2aTah/EzMYCScD6iFUqIiLHifZdTAlALjARyAY+MLMzjlxKMrOewLPAbe5eU/vDZnY3cDdAnz59mqlkEZG2IZI9iK1A75Dl7OC6UIXADHevdPdPgbUEAgMzSwNeB77v7vPDncDdp7l7nrvnZWbqCpSISFOKZEAsAHLNrJ+ZJQGTgRm19nmVQO8BM8sgcMlpQ3D/V4Bn3P3FCNYoIiInYO5+8r1O9eBmlwG/AuKBP7j7T83sQSDf3WdYYHLzh4FJQDXwU3efbmZfAP4IrAg53O3uvriOcxUBxUBJyOr0kOVw34euywCOnWe3/kKP05Dt4dbXXneyNoR+H8k21LVPU7QjFtoQ+n00/nsKt60hy7H0s9Dv9slrPCLH3cNfgnH3mPkCpp1oOdz3tdblN9V567s93PqGtqFWeyLWhki3Ixba0FztqGt7XTXXt02x8LPQ73bDfhYn+oq1J6n/VsdyuO9r799U563v9nDrG9qG+py/PupzjEi2IxbaUN8aTuZU/3sKt60hy7H0s9DvdhMcI6KXmFoTM8t397xo19EYakPLEQvtiIU2QGy0I1ptiLUeRGNMi3YBTUBtaDlioR2x0AaIjXZEpQ3qQYiISFjqQYiISFgKCBERCUsBISIiYSkgTsLMzjOzJ8zsd2Y2N9r1nCozizOzn5rZY2Z2W7TrORVmNtHMZgd/HhOjXU9jmFkHM8s3s8ujXcupMLMhwZ/Di2b25WjXc6rM7Goze8rM/mpmn412PafCzPqb2e/NrMlnnYjpgDCzP5jZLjNbXmv9SV9kdIS7z3b3fwf+DvwpkvWeSFO0g8BU69lAJYE5sJpVE7XBgQNAClFoAzRZOwC+AzwfmSrr1kS/F6uCvxc3AOdGst4TaaJ2vOrudwH/DjT7i8maqA0b3P3OiNQXy3cxmdkEAn9QnnH3YcF18QQmBTz6IiNgCoHpQH5W6xB3ePAdFWb2PHCnu+9vpvKPaop2BL/2ufuTZvaiu3++ueoP1tsUbdjt7jVm1h34pbvf3Fz1H9FE7RgBdCUQdLvd/e/NU31AU/1emNmVwJeBZ939z81V/xFN/Pv9MPCcuy9qpvIJnrcp29Dkv9fRnu47otz9AzPrW2v10RcZAZjZdOAqd/8ZELa7b2Z9gJJohAM0TTvMrBCoCC5WR7DcsJrqZxG0D0iOSKEn0UQ/i4lAB2AocMjMZnqY6ewjpal+Fu4+A5hhZq8DzR4QTfSzMOAh4I3mDgdo8t+LJhfTAXEC4V5kdNZJPnMngckDW5KGtuNl4DEzOw/4IJKFNUCD2mBm1wKXAJ2A30S0soZpUDvc/fsAZnY7wV5RRKurn4b+LCYC1xII6pmRLKyBGvp78VXgIiDdzAa6+xORLK6eGvqz6Ar8FBhlZt8NBkmTaIsB0WDu/sNo19BY7l5GIOhaLXd/mUDQxQR3fzraNZwqd/8n8M8ol9Fo7v4o8Gi062gMd99DYAylycX0IPUJ1OdFRq1BLLQjFtoAsdGOWGgDxEY7Wkwb2mJA1OdFRq1BLLQjFtoAsdGOWGgDxEY7Wk4bTnWe8NbwBfwF2M6/bu28M7j+MgJ3Cawn8ErTqNca6+2IhTbESjtioQ2x0o6W3oaYvs1VREROXVu8xCQiIvWggBARkbAUECIiEpYCQkREwlJAiIhIWAoIEREJSwEhMc3MDjTz+ZrknSEWePdFiZktNrPVZva/9fjM1WY2tCnOLwIKCJEGMbM65y9z93FNeLrZ7j4SGAVcbmYne+/C1QRmiBVpEgoIaXPMbICZvWlmCy3whrrTguuvMLOPzOwTM3sn+N4JzOxHZvasmc0Bng0u/8HM/mlmG8zsayHHPhD834nB7S8GewDPBaeWxswuC65baGaPmlmd74Nw90PAYgKzfGJmd5nZAjNbYmYvmVl7MxsHXAn8ItjrGHCidorUlwJC2qJpwFfdfQxwH/B4cP2HwNnuPgqYDnw75DNDgYvcfUpw+TQCU4+PBX5oZolhzjMK+Ebws/2Bc80sBXgSuDR4/syTFWtmnYFc/jVN+8vufqa7jwBWEZieYS6B+Xrud/eR7r6+jnaK1Ium+5Y2xcxSgXHAC8F/0MO/Xj6UDfzVzHoCScCnIR+dEfyX/BGvu/th4LCZ7QK6c/xrUD9298LgeRcDfQm8PWyDux859l+Au09Q7nlmtoRAOPzK3XcE1w8zs58QeC9GKjCrge0UqRcFhLQ1cUBx8Np+bY8ReJXpjOALcX4Usu1grX0Ph3xfTfjfpfrsU5fZ7n65mfUD5pvZ8+6+GHgauNrdlwRfOjQxzGfraqdIvegSk7Qp7l4KfGpm10PglZNmNiK4OZ1/zbt/W4RKWAP0D3nN5I0n+0Cwt/EQ8J3gqo7A9uBlrdD3cu8PbjtZO0XqRQEhsa69mRWGfH2LwB/VO4OXb1YAVwX3/RGBSzILgd2RKCZ4meoe4M3gefYDJfX46BPAhGCw/D/gI2AOsDpkn+nA/cFB9gGcuJ0i9aLpvkWamZmluvuB4F1NU4F17v5ItOsSqU09CJHmd1dw0HoFgctaT0a3HJHw1IMQEZGw1IMQEZGwFBAiIhKWAkJERMJSQIiISFgKCBERCUsBISIiYf1/GZJmyQjOpe4AAAAASUVORK5CYII=\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": [
    "from sklearn.model_selection import StratifiedKFold\n",
    "from sklearn.metrics import accuracy_score, balanced_accuracy_score, f1_score, roc_auc_score, matthews_corrcoef\n",
    "\n",
    "FOLD_EVAL = True\n",
    "\n",
    "shape_reformatter = BlockInputShapeReformatter()\n",
    "# BLOCKNAMES = \"Dense\"\n",
    "PATIENCE = 10\n",
    "SEED = 255\n",
    "\n",
    "model_fold_eval = {blockname : {'acc' : list(), 'f1' : list(), 'auroc' : list(), 'mcc' : list()} for blockname in models}\n",
    "\n",
    "if FOLD_EVAL:\n",
    "    \n",
    "    for blockname in tqdm(models):\n",
    "        \n",
    "        PARAMS = model_eval[blockname].iloc[0,:].to_dict()\n",
    "        \n",
    "        _ = PARAMS.pop(\"final_metric\")\n",
    "        LR_OPTION = PARAMS.pop(\"lr_max_arg\")\n",
    "        BATCHSIZE = PARAMS.pop(\"batchsize\")\n",
    "        MAIN_ACT = eval(PARAMS.pop(\"act\"))\n",
    "\n",
    "        main_model_params = {key : value for key,value in PARAMS.items() if 'rna' not in key and 'prot' not in key}\n",
    "        rna_block_params = {key.replace(\"rna_\",\"\") : value for key,value in PARAMS.items() if 'rna' in key}\n",
    "        prot_block_params = {key.replace(\"prot_\",\"\") : value for key,value in PARAMS.items() if 'prot' in key}\n",
    "\n",
    "        rna_block_name = determine_block_name_from_params(rna_block_params)\n",
    "        prot_block_name = determine_block_name_from_params(prot_block_params)\n",
    "        \n",
    "        skf = StratifiedKFold(n_splits = 5, random_state = 255, shuffle = True)\n",
    "        skf.get_n_splits(data[0], data[-1])\n",
    "\n",
    "        for train_index, test_index in skf.split(data[0], data[-1]):\n",
    "\n",
    "            #####################################################\n",
    "            # Generate train, test, val data\n",
    "            #####################################################\n",
    "\n",
    "            train_data, test_data = zip(*[(X[train_index], X[test_index]) for X in data])\n",
    "\n",
    "            test_idx, val_idx = train_test_split(range(len(test_data[0])), random_state = 255, test_size = 0.5, \n",
    "                                                 stratify = test_data[-1])\n",
    "\n",
    "            test_data, val_data = zip(*[(X[test_idx], X[val_idx]) for X in test_data])\n",
    "\n",
    "#             print(f\"\"\"Train data: {len(train_data[0])} | + samples: {sum(train_data[-1])} | - samples: {len(train_data[-1]) - sum(train_data[-1])}    \n",
    "#         Test data: {len(test_data[0])} | + samples: {sum(test_data[-1])} | - samples: {len(test_data[-1]) - sum(test_data[-1])}\n",
    "#         Val data: {len(val_data[0])} | + samples: {sum(val_data[-1])} | - samples: {len(val_data[-1]) - sum(val_data[-1])}\n",
    "#         ---\"\"\")\n",
    "\n",
    "            #####################################################\n",
    "            # Preprocess train, test, val data\n",
    "            #####################################################\n",
    "\n",
    "            refine_train_data = [shape_reformatter.reformat_input(blockname, train_data[i].float()) for i in range(2)] + [train_data[-1].float()]\n",
    "            swap_seq_input_order(refine_train_data)\n",
    "            refine_train_dataset = EncodedRPIDataset(refine_train_data)\n",
    "\n",
    "            refine_val_data = [shape_reformatter.reformat_input(blockname, val_data[i].float()) for i in range(2)] + [val_data[-1].float()]\n",
    "            swap_seq_input_order(refine_val_data)\n",
    "            refine_val_dataset = EncodedRPIDataset(refine_val_data)\n",
    "\n",
    "            refine_test_data = [shape_reformatter.reformat_input(blockname, test_data[i].float()) for i in range(2)] + [test_data[-1].float()]\n",
    "            proteins, rnas, y_true = refine_test_data\n",
    "            y_true = y_true.reshape(-1,1)\n",
    "\n",
    "            #####################################################\n",
    "            # Generate model\n",
    "            #####################################################\n",
    "\n",
    "            block_initializer = BlockInializer(SEED)\n",
    "\n",
    "            # Temporarily initializes blocks to calculate input size\n",
    "            rna_block = block_initializer.initialize_block(rna_block_name, 1024, rna_block_params)\n",
    "            prot_block = block_initializer.initialize_block(prot_block_name, 1024, prot_block_params)\n",
    "\n",
    "            dense_input_calculator = MainModelDenseInputSizeCalculator(refine_train_data[0], refine_train_data[1])\n",
    "            dense_input_size = dense_input_calculator.calcualte_input_shape(rna_blocks  = [rna_block.cpu()], \n",
    "                                                                            prot_blocks = [prot_block.cpu()])\n",
    "\n",
    "            rna_block = block_initializer.initialize_block(rna_block_name, 1024, rna_block_params)\n",
    "            prot_block = block_initializer.initialize_block(prot_block_name, 1024, prot_block_params)\n",
    "\n",
    "            main_model_params = eval(block_initializer._reformat_list_str(str(main_model_params)))\n",
    "\n",
    "            set_seed(255)\n",
    "            main_model = MainModel(rna_blocks  = [rna_block],\n",
    "                                   prot_blocks = [prot_block],\n",
    "                                   dense_input_size = dense_input_size,\n",
    "                                   act = MAIN_ACT,\n",
    "                                   **main_model_params).cuda()\n",
    "\n",
    "            #####################################################\n",
    "            # Train model\n",
    "            #####################################################\n",
    "\n",
    "            set_seed(255)\n",
    "            loaders = DataLoaders.from_dsets(refine_train_dataset, refine_val_dataset,\n",
    "                                             bs = BATCHSIZE,\n",
    "                                             device = 0)\n",
    "\n",
    "            learner = Learner(loaders, main_model,\n",
    "                              splitter = main_model_splitter,\n",
    "                              cbs = [EarlyStoppingCallback(patience = PATIENCE)],\n",
    "                              loss_func = nn.BCELoss())\n",
    "\n",
    "            lr_min, lr_steep = learner.lr_find()\n",
    "\n",
    "            max_lr_arg = initiate_lr_option(LR_OPTION, lr_min, lr_steep)\n",
    "            learner.fit_one_cycle(1000, lr_max = max_lr_arg)\n",
    "\n",
    "            #####################################################\n",
    "            # Evaluate performance on HIV-1 data\n",
    "            #####################################################\n",
    "\n",
    "            learner.model.eval()\n",
    "            y_pred = learner.model(rnas.cuda(), proteins.cuda()).cpu().detach().round()\n",
    "            \n",
    "            \n",
    "            model_fold_eval[blockname]['acc'].append(round(accuracy_score(y_true, y_pred),3))\n",
    "            model_fold_eval[blockname]['f1'].append(round(f1_score(y_true, y_pred),3))\n",
    "            model_fold_eval[blockname]['auroc'].append(round(roc_auc_score(y_true, y_pred),3))\n",
    "            model_fold_eval[blockname]['mcc'].append(round(matthews_corrcoef(y_true, y_pred),3))\n",
    "            \n",
    "#             print(f\"ACCURACY: {round(accuracy_score(y_true, y_pred),3)}\")\n",
    "#             print(f\"F1 SCORE: {round(f1_score(y_true, y_pred),3)}\")\n",
    "#             print(f\"ROC AUC: {round(roc_auc_score(y_true, y_pred),3)}\")\n",
    "#             print(f\"MCC: {round(matthews_corrcoef(y_true, y_pred),3)}\")\n",
    "#             print(\"---\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dense\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>acc</th>\n",
       "      <th>f1</th>\n",
       "      <th>auroc</th>\n",
       "      <th>mcc</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.984</td>\n",
       "      <td>0.985</td>\n",
       "      <td>0.984</td>\n",
       "      <td>0.969</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.003</td>\n",
       "      <td>0.003</td>\n",
       "      <td>0.003</td>\n",
       "      <td>0.007</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.980</td>\n",
       "      <td>0.980</td>\n",
       "      <td>0.980</td>\n",
       "      <td>0.960</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.982</td>\n",
       "      <td>0.983</td>\n",
       "      <td>0.982</td>\n",
       "      <td>0.965</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.985</td>\n",
       "      <td>0.985</td>\n",
       "      <td>0.985</td>\n",
       "      <td>0.970</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.987</td>\n",
       "      <td>0.987</td>\n",
       "      <td>0.987</td>\n",
       "      <td>0.974</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>0.988</td>\n",
       "      <td>0.988</td>\n",
       "      <td>0.988</td>\n",
       "      <td>0.976</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         acc     f1  auroc    mcc\n",
       "count  5.000  5.000  5.000  5.000\n",
       "mean   0.984  0.985  0.984  0.969\n",
       "std    0.003  0.003  0.003  0.007\n",
       "min    0.980  0.980  0.980  0.960\n",
       "25%    0.982  0.983  0.982  0.965\n",
       "50%    0.985  0.985  0.985  0.970\n",
       "75%    0.987  0.987  0.987  0.974\n",
       "max    0.988  0.988  0.988  0.976"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "StackedRes\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>acc</th>\n",
       "      <th>f1</th>\n",
       "      <th>auroc</th>\n",
       "      <th>mcc</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.982</td>\n",
       "      <td>0.982</td>\n",
       "      <td>0.982</td>\n",
       "      <td>0.964</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.004</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.007</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.978</td>\n",
       "      <td>0.978</td>\n",
       "      <td>0.978</td>\n",
       "      <td>0.956</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.980</td>\n",
       "      <td>0.981</td>\n",
       "      <td>0.980</td>\n",
       "      <td>0.961</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.981</td>\n",
       "      <td>0.981</td>\n",
       "      <td>0.981</td>\n",
       "      <td>0.962</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.986</td>\n",
       "      <td>0.986</td>\n",
       "      <td>0.986</td>\n",
       "      <td>0.971</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>0.986</td>\n",
       "      <td>0.986</td>\n",
       "      <td>0.986</td>\n",
       "      <td>0.971</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         acc     f1  auroc    mcc\n",
       "count  5.000  5.000  5.000  5.000\n",
       "mean   0.982  0.982  0.982  0.964\n",
       "std    0.004  0.004  0.004  0.007\n",
       "min    0.978  0.978  0.978  0.956\n",
       "25%    0.980  0.981  0.980  0.961\n",
       "50%    0.981  0.981  0.981  0.962\n",
       "75%    0.986  0.986  0.986  0.971\n",
       "max    0.986  0.986  0.986  0.971"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DenseSAE\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>acc</th>\n",
       "      <th>f1</th>\n",
       "      <th>auroc</th>\n",
       "      <th>mcc</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.976</td>\n",
       "      <td>0.976</td>\n",
       "      <td>0.976</td>\n",
       "      <td>0.952</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.003</td>\n",
       "      <td>0.003</td>\n",
       "      <td>0.003</td>\n",
       "      <td>0.006</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.973</td>\n",
       "      <td>0.973</td>\n",
       "      <td>0.973</td>\n",
       "      <td>0.946</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.973</td>\n",
       "      <td>0.974</td>\n",
       "      <td>0.973</td>\n",
       "      <td>0.947</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.974</td>\n",
       "      <td>0.975</td>\n",
       "      <td>0.974</td>\n",
       "      <td>0.950</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.979</td>\n",
       "      <td>0.979</td>\n",
       "      <td>0.979</td>\n",
       "      <td>0.958</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>0.979</td>\n",
       "      <td>0.979</td>\n",
       "      <td>0.979</td>\n",
       "      <td>0.958</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         acc     f1  auroc    mcc\n",
       "count  5.000  5.000  5.000  5.000\n",
       "mean   0.976  0.976  0.976  0.952\n",
       "std    0.003  0.003  0.003  0.006\n",
       "min    0.973  0.973  0.973  0.946\n",
       "25%    0.973  0.974  0.973  0.947\n",
       "50%    0.974  0.975  0.974  0.950\n",
       "75%    0.979  0.979  0.979  0.958\n",
       "max    0.979  0.979  0.979  0.958"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Attention\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>acc</th>\n",
       "      <th>f1</th>\n",
       "      <th>auroc</th>\n",
       "      <th>mcc</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.968</td>\n",
       "      <td>0.969</td>\n",
       "      <td>0.968</td>\n",
       "      <td>0.937</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.004</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.008</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.965</td>\n",
       "      <td>0.966</td>\n",
       "      <td>0.965</td>\n",
       "      <td>0.931</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.966</td>\n",
       "      <td>0.966</td>\n",
       "      <td>0.966</td>\n",
       "      <td>0.932</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.966</td>\n",
       "      <td>0.967</td>\n",
       "      <td>0.966</td>\n",
       "      <td>0.934</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.969</td>\n",
       "      <td>0.969</td>\n",
       "      <td>0.969</td>\n",
       "      <td>0.938</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>0.975</td>\n",
       "      <td>0.975</td>\n",
       "      <td>0.975</td>\n",
       "      <td>0.950</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         acc     f1  auroc    mcc\n",
       "count  5.000  5.000  5.000  5.000\n",
       "mean   0.968  0.969  0.968  0.937\n",
       "std    0.004  0.004  0.004  0.008\n",
       "min    0.965  0.966  0.965  0.931\n",
       "25%    0.966  0.966  0.966  0.932\n",
       "50%    0.966  0.967  0.966  0.934\n",
       "75%    0.969  0.969  0.969  0.938\n",
       "max    0.975  0.975  0.975  0.950"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Reccurent\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>acc</th>\n",
       "      <th>f1</th>\n",
       "      <th>auroc</th>\n",
       "      <th>mcc</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.976</td>\n",
       "      <td>0.977</td>\n",
       "      <td>0.976</td>\n",
       "      <td>0.953</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.004</td>\n",
       "      <td>0.003</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.007</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.973</td>\n",
       "      <td>0.974</td>\n",
       "      <td>0.973</td>\n",
       "      <td>0.947</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.974</td>\n",
       "      <td>0.974</td>\n",
       "      <td>0.974</td>\n",
       "      <td>0.948</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.974</td>\n",
       "      <td>0.975</td>\n",
       "      <td>0.974</td>\n",
       "      <td>0.949</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.978</td>\n",
       "      <td>0.978</td>\n",
       "      <td>0.978</td>\n",
       "      <td>0.957</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>0.982</td>\n",
       "      <td>0.982</td>\n",
       "      <td>0.982</td>\n",
       "      <td>0.964</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         acc     f1  auroc    mcc\n",
       "count  5.000  5.000  5.000  5.000\n",
       "mean   0.976  0.977  0.976  0.953\n",
       "std    0.004  0.003  0.004  0.007\n",
       "min    0.973  0.974  0.973  0.947\n",
       "25%    0.974  0.974  0.974  0.948\n",
       "50%    0.974  0.975  0.974  0.949\n",
       "75%    0.978  0.978  0.978  0.957\n",
       "max    0.982  0.982  0.982  0.964"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for model in model_fold_eval.keys():\n",
    "    print(model)\n",
    "    display(pd.DataFrame(model_fold_eval[model]).describe().round(3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1 Score</th>\n",
       "      <th>AUROC</th>\n",
       "      <th>MCC</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "      <td>5.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.982</td>\n",
       "      <td>0.982</td>\n",
       "      <td>0.982</td>\n",
       "      <td>0.964</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.004</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.004</td>\n",
       "      <td>0.007</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.978</td>\n",
       "      <td>0.978</td>\n",
       "      <td>0.978</td>\n",
       "      <td>0.956</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.980</td>\n",
       "      <td>0.981</td>\n",
       "      <td>0.980</td>\n",
       "      <td>0.961</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.981</td>\n",
       "      <td>0.981</td>\n",
       "      <td>0.981</td>\n",
       "      <td>0.962</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.986</td>\n",
       "      <td>0.986</td>\n",
       "      <td>0.986</td>\n",
       "      <td>0.971</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>0.986</td>\n",
       "      <td>0.986</td>\n",
       "      <td>0.986</td>\n",
       "      <td>0.971</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Accuracy  F1 Score  AUROC    MCC\n",
       "count     5.000     5.000  5.000  5.000\n",
       "mean      0.982     0.982  0.982  0.964\n",
       "std       0.004     0.004  0.004  0.007\n",
       "min       0.978     0.978  0.978  0.956\n",
       "25%       0.980     0.981  0.980  0.961\n",
       "50%       0.981     0.981  0.981  0.962\n",
       "75%       0.986     0.986  0.986  0.971\n",
       "max       0.986     0.986  0.986  0.971"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Update with any changes\n",
    "import pandas as pd\n",
    "\n",
    "fold_results = pd.DataFrame({\"Accuracy\" : [0.986, 0.986, 0.98, 0.978, 0.981],\n",
    "                             \"F1 Score\" : [0.986, 0.986, 0.981, 0.978, 0.981],\n",
    "                             \"AUROC\" : [0.986, 0.986, 0.980, 0.978, 0.981],\n",
    "                             \"MCC\" : [0.971, 0.971, 0.961, 0.956, 0.962]})\n",
    "\n",
    "fold_results.describe().round(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "#####################################################\n",
    "# Load training data\n",
    "######################################################\n",
    "\n",
    "train_idx, val_idx = train_test_split(range(len(data[0])), random_state = SEED, test_size = 0.1, \n",
    "                                       stratify = data[-1])\n",
    "\n",
    "train_data, val_data = zip(*[(X[train_idx], X[val_idx]) for X in data])\n",
    "\n",
    "train_data = [shape_reformatter.reformat_input(block_name, train_data[i]) for i in range(2)] + [train_data[-1]]\n",
    "val_data = [shape_reformatter.reformat_input(block_name, val_data[i]) for i in range(2)] + [val_data[-1]]\n",
    "\n",
    "swap_seq_input_order(train_data)\n",
    "swap_seq_input_order(val_data)\n",
    "\n",
    "train_dataset = EncodedRPIDataset(train_data)\n",
    "val_dataset = EncodedRPIDataset(val_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "StackedResBlock(1024, **{'hidden_sizes': [70, 40, 10], 'act_cls': nn.Sigmoid, 'bn_1st': False})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [55, 10], 'act_cls': nn.ReLU, 'bn_1st': True})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [70, 40, 10], 'act_cls': nn.Sigmoid, 'bn_1st': False})\n",
      "StackedResBlock(1024, **{'hidden_sizes': [55, 10], 'act_cls': nn.ReLU, 'bn_1st': True})\n"
     ]
    }
   ],
   "source": [
    "block_initializer = BlockInializer(SEED)\n",
    "\n",
    "# Temporarily initializes blocks to calculate input size\n",
    "rna_block = block_initializer.initialize_block(rna_block_name, 1024, rna_block_params)\n",
    "prot_block = block_initializer.initialize_block(prot_block_name, 1024, prot_block_params)\n",
    "\n",
    "dense_input_calculator = MainModelDenseInputSizeCalculator(train_data[0], train_data[1])\n",
    "\n",
    "dense_input_size = dense_input_calculator.calcualte_input_shape(rna_blocks  = [rna_block.cpu()], \n",
    "                                                                prot_blocks = [prot_block.cpu()])\n",
    "\n",
    "set_seed(255)\n",
    "rna_block = block_initializer.initialize_block(rna_block_name, 1024, rna_block_params)\n",
    "prot_block = block_initializer.initialize_block(prot_block_name, 1024, prot_block_params)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'dense_hidden_layers': [3000, 2000, 1000, 10],\n",
       " 'p': 0.2,\n",
       " 'bn': False,\n",
       " 'lin_first': False}"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "main_model_params = eval(block_initializer._reformat_list_str(str(main_model_params)))\n",
    "main_model_params"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.272765</td>\n",
       "      <td>0.213641</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.190086</td>\n",
       "      <td>0.149383</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.170463</td>\n",
       "      <td>0.137746</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.155680</td>\n",
       "      <td>0.113873</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.138959</td>\n",
       "      <td>0.108333</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.134556</td>\n",
       "      <td>0.098267</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.132672</td>\n",
       "      <td>0.086774</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.108989</td>\n",
       "      <td>0.099645</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.082421</td>\n",
       "      <td>0.100140</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.118945</td>\n",
       "      <td>0.094318</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.093352</td>\n",
       "      <td>0.081650</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.087566</td>\n",
       "      <td>0.074937</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.097374</td>\n",
       "      <td>0.072265</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.083360</td>\n",
       "      <td>0.067570</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.077698</td>\n",
       "      <td>0.076471</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.069805</td>\n",
       "      <td>0.071761</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.073305</td>\n",
       "      <td>0.062144</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.072788</td>\n",
       "      <td>0.062365</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.078970</td>\n",
       "      <td>0.063378</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.062467</td>\n",
       "      <td>0.064209</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.063822</td>\n",
       "      <td>0.066883</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.054934</td>\n",
       "      <td>0.061285</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.056949</td>\n",
       "      <td>0.058648</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.060117</td>\n",
       "      <td>0.052769</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.068028</td>\n",
       "      <td>0.057182</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.067520</td>\n",
       "      <td>0.059561</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.051490</td>\n",
       "      <td>0.055375</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.056461</td>\n",
       "      <td>0.053287</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.058263</td>\n",
       "      <td>0.047864</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.049690</td>\n",
       "      <td>0.052224</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.052308</td>\n",
       "      <td>0.052196</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.055172</td>\n",
       "      <td>0.053076</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.063783</td>\n",
       "      <td>0.050873</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.048371</td>\n",
       "      <td>0.053978</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.050568</td>\n",
       "      <td>0.050541</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.035784</td>\n",
       "      <td>0.058573</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.041942</td>\n",
       "      <td>0.055830</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.054647</td>\n",
       "      <td>0.052314</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.046907</td>\n",
       "      <td>0.054074</td>\n",
       "      <td>00:09</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No improvement since epoch 28: early stopping\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": [
    "#####################################################\n",
    "# Generate and train model\n",
    "#####################################################\n",
    "\n",
    "PATIENCE = 10\n",
    "\n",
    "set_seed(255)\n",
    "main_model = MainModel(rna_blocks  = [rna_block],\n",
    "                       prot_blocks = [prot_block],\n",
    "                       dense_input_size = dense_input_size,\n",
    "                       act = MAIN_ACT,\n",
    "                       **main_model_params).cuda()\n",
    "\n",
    "set_seed(255)\n",
    "loaders = DataLoaders.from_dsets(train_dataset, val_dataset,\n",
    "                                 bs = BATCHSIZE,\n",
    "                                 device = 0)\n",
    "\n",
    "learner = Learner(loaders, main_model,\n",
    "                  splitter = main_model_splitter,\n",
    "                  cbs = [EarlyStoppingCallback(patience = PATIENCE)],\n",
    "                  loss_func = nn.BCELoss())\n",
    "\n",
    "lr_min, lr_steep = learner.lr_find()\n",
    "\n",
    "max_lr_arg = initiate_lr_option(LR_OPTION, lr_min, lr_steep)\n",
    "learner.fit_one_cycle(1000, lr_max = max_lr_arg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0035"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def calculate_avg_train_val_diff_across_final_epochs(learner, epochs) -> float:\n",
    "    return np.diff(np.array(learner.recorder.values[-epochs:]), axis = 1).mean().round(4)\n",
    "\n",
    "calculate_avg_train_val_diff_across_final_epochs(learner, 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(1,1, figsize = (3,3))\n",
    "\n",
    "learner.recorder.plot_loss()\n",
    "sbn.despine()\n",
    "plt.xlabel(\"Batches\")\n",
    "plt.ylabel(\"Loss\");\n",
    "plt.title(\"Non-HIV-1 training\")\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.savefig(\"Non_hiv_RPI_training_curve.png\", dpi = 300)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Max validation loss: 0.2136\n",
      "Min validation loss: 0.0479\n",
      "\n"
     ]
    }
   ],
   "source": [
    "_, val_loss  = zip(*learner.recorder.values)\n",
    "\n",
    "max_val_loss = max(val_loss)\n",
    "min_val_loss = min(val_loss)\n",
    "\n",
    "print(f\"\"\"\n",
    "Max validation loss: {round(max_val_loss,4)}\n",
    "Min validation loss: {round(min_val_loss,4)}\n",
    "\"\"\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# even though loaded models != learner.model, it results in identical predictions\n",
    "# learner.model.eval()\n",
    "# torch.save(learner.model, \"../Models/NonRefinedMainModel.torch\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Predict HIV-1 data w/ unrefined"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"../Data/TrainingData/FullEmbeddedHIVTrainingData.pickle\",'rb') as handle:\n",
    "    data = pickle.load(handle)\n",
    "\n",
    "shape_reformatter = BlockInputShapeReformatter()\n",
    "BLOCKNAME = \"StackedRes\"\n",
    "\n",
    "refined_data = [shape_reformatter.reformat_input(BLOCKNAME, data[i].float()) for i in range(2)] + [data[-1].float()]\n",
    "proteins, rnas, y_true = refined_data\n",
    "y_true = y_true.reshape(-1,1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<function sklearn.metrics._ranking.average_precision_score(y_true, y_score, *, average='macro', pos_label=1, sample_weight=None)>"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import average_precision_score\n",
    "\n",
    "average_precision_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "hiv_train_idx, hiv_val_idx = train_test_split(range(len(data[0])), random_state = 255, test_size = 0.2, \n",
    "                                       stratify = data[-1])\n",
    "\n",
    "hiv_train_data, hiv_val_data = zip(*[(X[hiv_train_idx], X[hiv_val_idx]) for X in data])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "#####################################################\n",
    "# Load training data\n",
    "#####################################################\n",
    "\n",
    "refine_train_data = [shape_reformatter.reformat_input(BLOCKNAME, hiv_train_data[i].float()) for i in range(2)] + [hiv_train_data[-1].float()]\n",
    "swap_seq_input_order(refine_train_data)\n",
    "\n",
    "refine_val_data = [shape_reformatter.reformat_input(BLOCKNAME, hiv_val_data[i].float()) for i in range(2)] + [hiv_val_data[-1].float()]\n",
    "swap_seq_input_order(refine_val_data)\n",
    "\n",
    "refine_train_dataset = EncodedRPIDataset(refine_train_data)\n",
    "refine_val_dataset = EncodedRPIDataset(refine_val_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "def freeze_layers(model_block):\n",
    "    for layer in model_block:\n",
    "        for param in layer.parameters():\n",
    "            param.requires_grad = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1.947425</td>\n",
       "      <td>0.727569</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.965631</td>\n",
       "      <td>0.209985</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.554396</td>\n",
       "      <td>0.000288</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.382601</td>\n",
       "      <td>0.000334</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.343534</td>\n",
       "      <td>0.031163</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.265505</td>\n",
       "      <td>0.000406</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.191364</td>\n",
       "      <td>0.000256</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.180988</td>\n",
       "      <td>0.000035</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.167628</td>\n",
       "      <td>0.000002</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.123989</td>\n",
       "      <td>0.000002</td>\n",
       "      <td>00:00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#####################################################\n",
    "# Generate and train model\n",
    "#####################################################\n",
    "\n",
    "REFINE_EPOCHS = 10\n",
    "BATCHSIZE = 8\n",
    "LR = 1e-3\n",
    "\n",
    "set_seed(255)\n",
    "loaders = DataLoaders.from_dsets(refine_train_dataset, refine_val_dataset,\n",
    "                                 bs = BATCHSIZE,\n",
    "                                 device = 0)\n",
    "\n",
    "model = torch.load(\"../Models/NonRefinedMainModel.torch\")\n",
    "model.train()\n",
    "freeze_layers(model.blocks.rna)\n",
    "freeze_layers(model.blocks.prot)\n",
    "\n",
    "# for param in model.dense.parameters():\n",
    "#     param.requires_grad = False\n",
    "    \n",
    "# for param in model.final.parameters():\n",
    "#     param.requires_grad = False\n",
    "\n",
    "#freeze_layers(model.dense)\n",
    "# freeze_layers(model.final)\n",
    "\n",
    "refine_learner = Learner(loaders, model,\n",
    "                         splitter = main_model_splitter,\n",
    "                         cbs = [EarlyStoppingCallback(patience = 10)],\n",
    "                         loss_func = nn.BCELoss())\n",
    "\n",
    "# lr_min, lr_steep = refine_learner.lr_find()\n",
    "\n",
    "#max_lr_arg = initiate_lr_option(LR_OPTION, lr_min, lr_steep)\n",
    "# slice(lr_steep, lr_min)\n",
    "refine_learner.fit_one_cycle(REFINE_EPOCHS, lr_max = LR)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(1,1, figsize = (3,3))\n",
    "\n",
    "refine_learner.recorder.plot_loss()\n",
    "sbn.despine()\n",
    "plt.xlabel(\"Batches\")\n",
    "plt.ylabel(\"Loss\")\n",
    "plt.title(\"HIV-1 refinement\")\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.savefig(\"HIV_RPI_training_curve.png\", dpi = 300)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-0.4153"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def calculate_avg_train_val_diff_across_final_epochs(learner, epochs) -> float:\n",
    "    return np.diff(np.array(learner.recorder.values[-epochs:]), axis = 1).mean().round(4)\n",
    "\n",
    "calculate_avg_train_val_diff_across_final_epochs(refine_learner, 100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Max validation loss: 0.7276\n",
      "Min validation loss: 0.0\n",
      "\n"
     ]
    }
   ],
   "source": [
    "_, val_loss  = zip(*refine_learner.recorder.values)\n",
    "\n",
    "max_val_loss = max(val_loss)\n",
    "min_val_loss = min(val_loss)\n",
    "\n",
    "print(f\"\"\"\n",
    "Max validation loss: {round(max_val_loss,4)}\n",
    "Min validation loss: {round(min_val_loss,4)}\n",
    "\"\"\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ModuleList(\n",
       "  (0): StackedResBlock(\n",
       "    (input_layer): ResBlock(\n",
       "      (convpath): Sequential(\n",
       "        (0): ConvLayer(\n",
       "          (0): Conv1d(1024, 70, kernel_size=(3,), stride=(1,), padding=(1,), bias=False)\n",
       "          (1): Sigmoid()\n",
       "          (2): BatchNorm1d(70, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        )\n",
       "        (1): ConvLayer(\n",
       "          (0): Conv1d(70, 70, kernel_size=(3,), stride=(1,), padding=(1,), bias=False)\n",
       "          (1): BatchNorm1d(70, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        )\n",
       "      )\n",
       "      (idpath): Sequential(\n",
       "        (0): ConvLayer(\n",
       "          (0): Conv1d(1024, 70, kernel_size=(1,), stride=(1,), bias=False)\n",
       "          (1): BatchNorm1d(70, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        )\n",
       "      )\n",
       "      (act): Sigmoid()\n",
       "    )\n",
       "    (hidden_layers): ModuleList(\n",
       "      (0): ResBlock(\n",
       "        (convpath): Sequential(\n",
       "          (0): ConvLayer(\n",
       "            (0): Conv1d(70, 40, kernel_size=(3,), stride=(1,), padding=(1,), bias=False)\n",
       "            (1): Sigmoid()\n",
       "            (2): BatchNorm1d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          )\n",
       "          (1): ConvLayer(\n",
       "            (0): Conv1d(40, 40, kernel_size=(3,), stride=(1,), padding=(1,), bias=False)\n",
       "            (1): BatchNorm1d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          )\n",
       "        )\n",
       "        (idpath): Sequential(\n",
       "          (0): ConvLayer(\n",
       "            (0): Conv1d(70, 40, kernel_size=(1,), stride=(1,), bias=False)\n",
       "            (1): BatchNorm1d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          )\n",
       "        )\n",
       "        (act): Sigmoid()\n",
       "      )\n",
       "      (1): ResBlock(\n",
       "        (convpath): Sequential(\n",
       "          (0): ConvLayer(\n",
       "            (0): Conv1d(40, 10, kernel_size=(3,), stride=(1,), padding=(1,), bias=False)\n",
       "            (1): Sigmoid()\n",
       "            (2): BatchNorm1d(10, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          )\n",
       "          (1): ConvLayer(\n",
       "            (0): Conv1d(10, 10, kernel_size=(3,), stride=(1,), padding=(1,), bias=False)\n",
       "            (1): BatchNorm1d(10, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          )\n",
       "        )\n",
       "        (idpath): Sequential(\n",
       "          (0): ConvLayer(\n",
       "            (0): Conv1d(40, 10, kernel_size=(1,), stride=(1,), bias=False)\n",
       "            (1): BatchNorm1d(10, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          )\n",
       "        )\n",
       "        (act): Sigmoid()\n",
       "      )\n",
       "    )\n",
       "  )\n",
       ")"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learner.model.blocks.rna"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "shape_reformatter = BlockInputShapeReformatter()\n",
    "BLOCKNAME = \"StackedRes\"\n",
    "\n",
    "refined_data = [shape_reformatter.reformat_input(BLOCKNAME, data[i].float()) for i in range(2)] + [data[-1].float()]\n",
    "proteins, rnas, y_true = refined_data\n",
    "y_true = y_true.reshape(-1,1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ACCURACY: 1.0\n",
      "BALANCED ACCURACY: 1.0\n",
      "F1 SCORE: 1.0\n",
      "ROC AUC: 1.0\n",
      "MCC: 1.0\n"
     ]
    }
   ],
   "source": [
    "refine_learner.model.eval()\n",
    "y_pred = refine_learner.model(rnas.cuda(), proteins.cuda()).cpu().detach().round()\n",
    "\n",
    "print(f\"ACCURACY: {round(accuracy_score(y_true, y_pred),3)}\")\n",
    "print(f\"BALANCED ACCURACY: {round(balanced_accuracy_score(y_true, y_pred),3)}\")\n",
    "print(f\"F1 SCORE: {round(f1_score(y_true, y_pred),3)}\")\n",
    "print(f\"ROC AUC: {round(roc_auc_score(y_true, y_pred),3)}\")\n",
    "print(f\"MCC: {round(matthews_corrcoef(y_true, y_pred),3)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "# even though loaded models != learner.model, it results in identical predictions\n",
    "refine_learner.model.eval();\n",
    "torch.save(refine_learner.model, \"../Models/TatRefinedMainModel.torch\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}