blstm_adam_snli.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "F2hv12elHUkk"
   },
   "source": [
    "# Training BiLSTM on SNLI using ADAM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "x4cJFGwS4Foe",
    "outputId": "875322fc-97c8-4f04-8752-d7a143425cf8"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mounted at /content/drive\n",
      "/content/drive/.shortcut-targets-by-id/1rxvBmpHEy23c0WZ6V3NFhmsmCCJ4quW5/DL_project\n",
      "/content/drive/.shortcut-targets-by-id/1rxvBmpHEy23c0WZ6V3NFhmsmCCJ4quW5/DL_project\n"
     ]
    }
   ],
   "source": [
    "from google.colab import drive\n",
    "drive.mount('/content/drive', force_remount=True)\n",
    "%cd /content/drive/MyDrive/DL_project/\n",
    "!pwd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "nLURgE-65F61"
   },
   "outputs": [],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import torchvision\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "from torch.utils.data.dataloader import DataLoader\n",
    "from torchvision.transforms import *\n",
    "import os\n",
    "import time\n",
    "import gc\n",
    "import shutil\n",
    "import h5py\n",
    "\n",
    "from dfw.dfw import DFW\n",
    "from dfw.dfw.losses import set_smoothing_enabled\n",
    "from dfw.dfw.losses import MultiClassHingeLoss\n",
    "from dfw.experiments.InferSent.data import get_nli, get_batch, build_vocab\n",
    "from dfw.experiments.InferSent.models import NLINet\n",
    "\n",
    "from snli_train_utils import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "g9ywtAAt-8qH",
    "outputId": "7c1f8052-aca3-48cc-c422-3deeb67a4bb8"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "** TRAIN DATA : Found 549367 pairs of train sentences.\n",
      "** DEV DATA : Found 9842 pairs of dev sentences.\n",
      "** TEST DATA : Found 9824 pairs of test sentences.\n",
      "Found 38957(/43479) words with glove vectors\n",
      "Vocab size : 38957\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py:12: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray\n",
      "  if sys.path[0] == '':\n"
     ]
    }
   ],
   "source": [
    "GLOVE_PATH = \"dfw/experiments/InferSent/dataset/GloVe/glove.840B.300d.txt\"\n",
    "nlipath = 'dfw/experiments/InferSent/dataset/SNLI'\n",
    "train, valid, test = get_nli(nlipath)\n",
    "word_vec = build_vocab(train['s1'] + train['s2'] +\n",
    "                       valid['s1'] + valid['s2'] +\n",
    "                       test['s1'] + test['s2'], GLOVE_PATH)\n",
    "\n",
    "for split in ['s1', 's2']:\n",
    "    for data_type in ['train', 'valid', 'test']:\n",
    "        eval(data_type)[split] = np.array([['<s>'] +\n",
    "            [word for word in sent.split() if word in word_vec] +\n",
    "            ['</s>'] for sent in eval(data_type)[split]])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "EPhF30-jBOfu"
   },
   "outputs": [],
   "source": [
    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "word_emb_dim = 300\n",
    "config_nli_model = {\n",
    "    'n_words'        :  len(word_vec)         ,\n",
    "    'word_emb_dim'   :  word_emb_dim   ,\n",
    "    'enc_lstm_dim'   :  2048   ,\n",
    "    'n_enc_layers'   :  1   ,\n",
    "    'dpout_model'    :  0.    ,\n",
    "    'dpout_fc'       :  0.       ,\n",
    "    'fc_dim'         :  512         ,\n",
    "    'bsize'          :  64     ,\n",
    "    'n_classes'      :  3      ,\n",
    "    'pool_type'      :  'max'      ,\n",
    "    'nonlinear_fc'   :  0   ,\n",
    "    'encoder_type'   :  'BLSTMEncoder'   ,\n",
    "    'use_cuda'       :  True                  ,\n",
    "\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "cB6cyM6FNnbO",
    "outputId": "78c3c52e-c2fd-4ec3-dfc5-9675397e10c4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cuda Tesla T4\n",
      "Epoch:0   Loss:0.136   Val loss:1.37   Accuracy:46.078   Best val loss:1.37   Time taken:335.065\n",
      "Epoch:1   Loss:0.071   Val loss:1.079   Accuracy:50.438   Best val loss:1.079   Time taken:336.949\n",
      "Epoch:2   Loss:0.061   Val loss:1.075   Accuracy:55.531   Best val loss:1.075   Time taken:343.199\n",
      "Epoch:3   Loss:0.059   Val loss:1.05   Accuracy:60.031   Best val loss:1.05   Time taken:334.443\n",
      "Epoch:4   Loss:0.05   Val loss:1.02   Accuracy:59.5   Best val loss:1.02   Time taken:334.954\n",
      "Epoch:5   Loss:0.052   Val loss:0.955   Accuracy:59.859   Best val loss:0.955   Time taken:336.21\n",
      "Epoch:6   Loss:0.059   Val loss:1.111   Accuracy:56.156   Best val loss:0.955   Time taken:337.421\n",
      "Epoch:7   Loss:0.055   Val loss:0.911   Accuracy:59.844   Best val loss:0.911   Time taken:342.586\n",
      "Epoch:8   Loss:0.052   Val loss:1.002   Accuracy:56.781   Best val loss:0.911   Time taken:342.11\n",
      "Epoch:9   Loss:0.057   Val loss:1.186   Accuracy:50.016   Best val loss:0.911   Time taken:340.661\n"
     ]
    }
   ],
   "source": [
    "trainer = Net_Trainer(device, 'nli_adam_snli', config_nli_model, word_vec, train, valid, test, lr=1e-4, optm='adam', loss='svm', batch_size=64, epochs=100, \n",
    "                          lr_decay=0.99, lr_shrink=5.0, weight_decay=0, mom=0, smooth=False)\n",
    "best_epoch, losses, errors, accs, testacc = trainer.train()\n",
    "plot_train(losses, errors, accs)\n",
    "print('Best epoch:',best_epoch)\n",
    "print('Test accuracy:',testacc,'%')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "nuge5xfw4-iU",
    "outputId": "c915581c-e25b-4c51-bfe8-0f3991aa077b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cuda Tesla T4\n",
      "Epoch:10   Loss:0.058   Val loss:0.84   Accuracy:66.422   Best val loss:0.84   Time taken:330.26\n",
      "Epoch:11   Loss:0.05   Val loss:0.811   Accuracy:65.891   Best val loss:0.811   Time taken:327.447\n",
      "Epoch:12   Loss:0.047   Val loss:0.947   Accuracy:62.062   Best val loss:0.811   Time taken:329.669\n",
      "Epoch:13   Loss:0.051   Val loss:0.864   Accuracy:65.062   Best val loss:0.811   Time taken:328.246\n",
      "Epoch:14   Loss:0.059   Val loss:0.913   Accuracy:63.938   Best val loss:0.811   Time taken:332.778\n",
      "Epoch:15   Loss:0.054   Val loss:0.912   Accuracy:60.188   Best val loss:0.811   Time taken:330.77\n",
      "Epoch:16   Loss:0.049   Val loss:0.985   Accuracy:62.5   Best val loss:0.811   Time taken:331.698\n",
      "Epoch:17   Loss:0.056   Val loss:0.758   Accuracy:68.031   Best val loss:0.758   Time taken:336.511\n",
      "Epoch:18   Loss:0.048   Val loss:1.017   Accuracy:59.531   Best val loss:0.758   Time taken:332.229\n",
      "Epoch:19   Loss:0.057   Val loss:0.801   Accuracy:67.797   Best val loss:0.758   Time taken:328.381\n",
      "Epoch:20   Loss:0.062   Val loss:0.777   Accuracy:65.469   Best val loss:0.758   Time taken:327.469\n",
      "Epoch:21   Loss:0.055   Val loss:0.945   Accuracy:56.859   Best val loss:0.758   Time taken:331.931\n",
      "Epoch:22   Loss:0.046   Val loss:0.979   Accuracy:60.922   Best val loss:0.758   Time taken:328.972\n",
      "Epoch:23   Loss:0.048   Val loss:0.819   Accuracy:66.172   Best val loss:0.758   Time taken:330.005\n",
      "Epoch:24   Loss:0.048   Val loss:0.734   Accuracy:69.891   Best val loss:0.734   Time taken:333.971\n",
      "Epoch:25   Loss:0.049   Val loss:0.931   Accuracy:62.781   Best val loss:0.734   Time taken:334.663\n",
      "Epoch:26   Loss:0.047   Val loss:0.899   Accuracy:66.516   Best val loss:0.734   Time taken:333.906\n",
      "Epoch:27   Loss:0.045   Val loss:0.961   Accuracy:63.453   Best val loss:0.734   Time taken:338.26\n",
      "Epoch:28   Loss:0.056   Val loss:0.918   Accuracy:61.203   Best val loss:0.734   Time taken:331.976\n",
      "Epoch:29   Loss:0.048   Val loss:0.7   Accuracy:73.156   Best val loss:0.7   Time taken:332.728\n",
      "Epoch:30   Loss:0.053   Val loss:0.676   Accuracy:69.734   Best val loss:0.676   Time taken:337.358\n",
      "Epoch:31   Loss:0.048   Val loss:0.666   Accuracy:69.938   Best val loss:0.666   Time taken:335.286\n",
      "Epoch:32   Loss:0.045   Val loss:0.783   Accuracy:69.109   Best val loss:0.666   Time taken:341.091\n",
      "Epoch:33   Loss:0.047   Val loss:0.693   Accuracy:70.516   Best val loss:0.666   Time taken:335.911\n",
      "Epoch:34   Loss:0.037   Val loss:0.873   Accuracy:68.078   Best val loss:0.666   Time taken:336.1\n",
      "Epoch:35   Loss:0.051   Val loss:0.739   Accuracy:66.906   Best val loss:0.666   Time taken:332.763\n",
      "Epoch:36   Loss:0.045   Val loss:0.823   Accuracy:64.922   Best val loss:0.666   Time taken:329.799\n",
      "Epoch:37   Loss:0.053   Val loss:0.696   Accuracy:65.766   Best val loss:0.666   Time taken:329.298\n",
      "Epoch:38   Loss:0.046   Val loss:0.663   Accuracy:74.125   Best val loss:0.663   Time taken:331.846\n",
      "Epoch:39   Loss:0.047   Val loss:0.922   Accuracy:65.953   Best val loss:0.663   Time taken:330.976\n",
      "Epoch:40   Loss:0.049   Val loss:0.7   Accuracy:75.562   Best val loss:0.663   Time taken:333.121\n",
      "Epoch:41   Loss:0.048   Val loss:0.736   Accuracy:67.938   Best val loss:0.663   Time taken:332.315\n",
      "Epoch:42   Loss:0.045   Val loss:0.8   Accuracy:69.812   Best val loss:0.663   Time taken:333.85\n",
      "Epoch:43   Loss:0.042   Val loss:0.708   Accuracy:76.109   Best val loss:0.663   Time taken:334.966\n",
      "Epoch:44   Loss:0.043   Val loss:0.714   Accuracy:74.875   Best val loss:0.663   Time taken:326.296\n",
      "Epoch:45   Loss:0.046   Val loss:0.704   Accuracy:70.594   Best val loss:0.663   Time taken:334.426\n",
      "Epoch:46   Loss:0.037   Val loss:0.661   Accuracy:73.531   Best val loss:0.661   Time taken:328.772\n",
      "Epoch:47   Loss:0.046   Val loss:0.931   Accuracy:62.172   Best val loss:0.661   Time taken:331.82\n",
      "Epoch:48   Loss:0.05   Val loss:0.734   Accuracy:69.656   Best val loss:0.661   Time taken:335.868\n",
      "Epoch:49   Loss:0.044   Val loss:0.722   Accuracy:72.25   Best val loss:0.661   Time taken:332.02\n",
      "Epoch:50   Loss:0.041   Val loss:0.712   Accuracy:72.422   Best val loss:0.661   Time taken:337.378\n",
      "Epoch:51   Loss:0.048   Val loss:0.883   Accuracy:60.5   Best val loss:0.661   Time taken:340.118\n",
      "Epoch:52   Loss:0.048   Val loss:0.76   Accuracy:72.781   Best val loss:0.661   Time taken:330.553\n",
      "Epoch:53   Loss:0.047   Val loss:0.802   Accuracy:67.484   Best val loss:0.661   Time taken:328.786\n",
      "Epoch:54   Loss:0.045   Val loss:0.651   Accuracy:70.734   Best val loss:0.651   Time taken:329.846\n",
      "Epoch:55   Loss:0.045   Val loss:0.714   Accuracy:74.375   Best val loss:0.651   Time taken:329.839\n",
      "Epoch:56   Loss:0.04   Val loss:0.665   Accuracy:73.734   Best val loss:0.651   Time taken:333.189\n",
      "Epoch:57   Loss:0.047   Val loss:0.914   Accuracy:69.703   Best val loss:0.651   Time taken:331.098\n",
      "Epoch:58   Loss:0.048   Val loss:0.753   Accuracy:69.016   Best val loss:0.651   Time taken:329.366\n",
      "Epoch:59   Loss:0.047   Val loss:0.773   Accuracy:68.203   Best val loss:0.651   Time taken:331.479\n",
      "Epoch:60   Loss:0.049   Val loss:0.818   Accuracy:70.875   Best val loss:0.651   Time taken:334.575\n",
      "Epoch:61   Loss:0.046   Val loss:0.687   Accuracy:73.391   Best val loss:0.651   Time taken:335.109\n",
      "Epoch:62   Loss:0.041   Val loss:0.96   Accuracy:59.656   Best val loss:0.651   Time taken:335.945\n",
      "Epoch:63   Loss:0.053   Val loss:0.652   Accuracy:70.75   Best val loss:0.651   Time taken:335.987\n",
      "Epoch:64   Loss:0.045   Val loss:0.627   Accuracy:74.453   Best val loss:0.627   Time taken:333.297\n",
      "Epoch:65   Loss:0.04   Val loss:0.704   Accuracy:72.969   Best val loss:0.627   Time taken:332.059\n",
      "Epoch:66   Loss:0.049   Val loss:0.687   Accuracy:71.875   Best val loss:0.627   Time taken:340.625\n",
      "Epoch:67   Loss:0.043   Val loss:0.681   Accuracy:72.031   Best val loss:0.627   Time taken:340.922\n",
      "Epoch:68   Loss:0.04   Val loss:0.661   Accuracy:72.766   Best val loss:0.627   Time taken:335.273\n",
      "Epoch:69   Loss:0.043   Val loss:0.815   Accuracy:67.609   Best val loss:0.627   Time taken:335.512\n",
      "Epoch:70   Loss:0.049   Val loss:0.804   Accuracy:68.125   Best val loss:0.627   Time taken:330.031\n",
      "Epoch:71   Loss:0.042   Val loss:0.629   Accuracy:73.922   Best val loss:0.627   Time taken:333.847\n",
      "Epoch:72   Loss:0.04   Val loss:0.606   Accuracy:72.469   Best val loss:0.606   Time taken:331.831\n",
      "Epoch:73   Loss:0.039   Val loss:0.548   Accuracy:76.375   Best val loss:0.548   Time taken:338.87\n",
      "Epoch:74   Loss:0.044   Val loss:0.711   Accuracy:69.109   Best val loss:0.548   Time taken:332.565\n",
      "Epoch:75   Loss:0.039   Val loss:0.564   Accuracy:77.047   Best val loss:0.548   Time taken:335.364\n",
      "Epoch:76   Loss:0.042   Val loss:0.634   Accuracy:73.984   Best val loss:0.548   Time taken:334.932\n",
      "Epoch:77   Loss:0.043   Val loss:0.706   Accuracy:75.453   Best val loss:0.548   Time taken:339.337\n",
      "Epoch:78   Loss:0.041   Val loss:0.588   Accuracy:77.578   Best val loss:0.548   Time taken:333.403\n",
      "Epoch:79   Loss:0.044   Val loss:0.617   Accuracy:73.922   Best val loss:0.548   Time taken:330.889\n",
      "Epoch:80   Loss:0.044   Val loss:0.541   Accuracy:74.812   Best val loss:0.541   Time taken:337.39\n",
      "Epoch:81   Loss:0.044   Val loss:0.702   Accuracy:72.266   Best val loss:0.541   Time taken:330.962\n",
      "Epoch:82   Loss:0.047   Val loss:0.817   Accuracy:65.562   Best val loss:0.541   Time taken:344.434\n",
      "Epoch:83   Loss:0.045   Val loss:0.721   Accuracy:70.5   Best val loss:0.541   Time taken:341.891\n",
      "Epoch:84   Loss:0.037   Val loss:0.905   Accuracy:61.844   Best val loss:0.541   Time taken:337.254\n",
      "Epoch:85   Loss:0.038   Val loss:0.701   Accuracy:68.0   Best val loss:0.541   Time taken:334.875\n",
      "Epoch:86   Loss:0.041   Val loss:0.662   Accuracy:74.078   Best val loss:0.541   Time taken:334.436\n",
      "Epoch:87   Loss:0.033   Val loss:0.718   Accuracy:73.625   Best val loss:0.541   Time taken:340.9\n",
      "Epoch:88   Loss:0.044   Val loss:0.658   Accuracy:73.812   Best val loss:0.541   Time taken:335.661\n",
      "Epoch:89   Loss:0.045   Val loss:0.647   Accuracy:72.594   Best val loss:0.541   Time taken:340.815\n",
      "Epoch:90   Loss:0.042   Val loss:0.655   Accuracy:74.953   Best val loss:0.541   Time taken:343.877\n",
      "Epoch:91   Loss:0.041   Val loss:0.607   Accuracy:75.297   Best val loss:0.541   Time taken:338.311\n",
      "Epoch:92   Loss:0.041   Val loss:0.764   Accuracy:71.922   Best val loss:0.541   Time taken:336.818\n",
      "Epoch:93   Loss:0.039   Val loss:0.629   Accuracy:77.266   Best val loss:0.541   Time taken:333.139\n",
      "Epoch:94   Loss:0.036   Val loss:0.605   Accuracy:76.375   Best val loss:0.541   Time taken:334.156\n",
      "Epoch:95   Loss:0.042   Val loss:0.626   Accuracy:77.25   Best val loss:0.541   Time taken:335.271\n",
      "Epoch:96   Loss:0.048   Val loss:0.827   Accuracy:64.031   Best val loss:0.541   Time taken:338.939\n",
      "Epoch:97   Loss:0.05   Val loss:0.614   Accuracy:71.328   Best val loss:0.541   Time taken:338.937\n",
      "Epoch:98   Loss:0.043   Val loss:0.749   Accuracy:71.828   Best val loss:0.541   Time taken:341.526\n",
      "Epoch:99   Loss:0.041   Val loss:0.647   Accuracy:74.609   Best val loss:0.541   Time taken:336.218\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX4AAAEWCAYAAABhffzLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOy9d3xb1f3//zyy5L3tDGc6TkIG2QkJqwQIlPVpGAU+ZRQKLW2hUCjQlg8dH/rp+LZ0l09/9NPBalkN0AUNu6UtZWSQHUISO8Px3rZs2ZZ0fn9cnesr6UqWbcmS4/N8PHhgS1dXx471uq/7Ou/zPkJKiUaj0WjGD45kD0Cj0Wg0o4sWfo1GoxlnaOHXaDSacYYWfo1GoxlnaOHXaDSacYYWfo1GoxlnaOHXjFmEEBuFENcnexwazVhDC79mVBFCdFn+8wsheizfXzOUc0kpL5BSPpqoscaKEOITQggphPjPZI9Fo4kFoRdwaZKFEOIQ8Ckp5as2zzmllN7RH9XQEUL8DVgCvC2lvGgU33fM/I40qYV2/JqUQAhxphCiWgjxZSFEHfCwEKJICPG8EKJRCNEa+Hqa5TV/F0J8KvD1J4QQ/xJC/CBwbJUQ4oII7/VlIcQzIY/9VAjxM8u5KoUQnYHzRLwTEULMBNYCnwbOE0JMtjyXJoS4VwhxMHCuLUKI6YHnThRCvCKEaBFC1Ash7g08/ogQ4luhvxfL94cC498BuIUQTiHEPZb32COEuDRkjDcJIfZanl8hhPiiEOLZkON+JoT4aaSfVXP8oIVfk0pMBoqBmRhC6gAeDnw/A+gB/jfK69cA+4BS4H7gN0IIYXPcU8CFQog8MAQauBJ4QgiRA/wMuEBKmQecCmyL8p7XAZullM8CewHrReJO4CrgQiAfuBHoDrzvq8CLwBRgDvBalPcI5SrgIqAw4PgPAh8CCoBvAL8TQpQFfrYrgPsC48wH1gPNwO+A84UQhYHjnMDHgMeGMA7NGEULvyaV8AP/LaXslVL2SCmbpZTPSim7pZSdwLcx3HUkDkspfyWl9AGPAmXApNCDpJSHga2AcsZnA91Syrct41gkhMiSUtZKKXdHec/rgCcCXz8R+F7xKeCrUsp90mC7lLIZ+A+gTkr5QymlR0rZKaV8J8p7hPIzKeVRKWVP4OfZIKWskVL6pZRPA/uB1ZYx3C+l3BQYwwEp5WEpZS3wD+CKwHHnA01Syi1DGIdmjKKFX5NKNEopPeobIUS2EOL/hBCHhRAdGEJVGHDodtSpL6SU3YEvcyMc+wSGcwa4OvA9Uko38J/AZ4FaIcQLQoj5dicQQpwGzMK4g1DnXCyEWBb4fjqGGw8l0uOxcjRkHNcJIbYJIdqEEG3AIoy7nsHe61Hg2sDX1wK/HcGYNGMILfyaVCK00uAuYB6wRkqZD5wReNwuvhkqG4AzA3MGlzLg2pFSviSlPBfjjuF94FcRznF9YCzbAvMS71geB0OgZ9u87ihQEeGcbiDb8v1km2PM31NgjuFXwK1AiZSyENjFwO8o0hgA/ggsEUIswrgLeTzCcZrjDC38mlQmDyPXbxNCFAP/Ha8TSykbgb9jzCFUSSn3AgghJgkhLg5k/b1AF0b0E4QQIhNjXuDTwDLLf7cBVwcy818D3xRCzBUGS4QQJcDzQJkQ4g4hRIYQIk8IsSZw6m0Y8w/FgYniOwb5UXIwLgSNgXHdgOH4Fb8G7hZCrAyMYU7gYkHg7uoZjIveu1LKI7H+/jRjGy38mlTmJ0AW0AS8jTEZGk+eAM7B4vYxPhN3AjVAC8acws02r70E46L0mJSyTv0HPAQ4MTLzHwG/B14GOoDfAFmB+YpzgY9gxFP7gbMC5/0tsB04FHjd09F+ACnlHuCHwFtAPbAYeNPy/AaMuZEngE4Ml19sOcWjgdfomGccoev4NZpxjBBiBkacNVlK2ZHs8WhGB+34NZpxihBC3d08pUV/fOFM9gA0Gs3oE5jDqAcOY8RSmnGEjno0Go1mnKGjHo1GoxlnjImop7S0VJaXlyd7GBqNRjOm2LJlS5OUckLo42NC+MvLy9m8eXOyh6HRaDRjCiHEYbvHddSj0Wg04wwt/BqNRjPO0MKv0Wg044wxkfHb0d/fT3V1NR6PZ/CDNQkjMzOTadOm4XK5kj0UjUYTI2NW+Kurq8nLy6O8vBz7vTY0iUZKSXNzM9XV1cyaNSvZw9FoNDEyZqMej8dDSUmJFv0kIoSgpKRE33VpNGOMMSv8gBb9FED/G2g0Y48xLfwajeb45d9H/82WGr0TZCLQwj9MzjrrLF566aWgx37yk59w8812rdsNzjzzTL0QTaOJkev+cB1feOkLyR7GcYkW/mFy1VVX8dRTTwU99tRTT3HVVVdFeEXy8Xq9yR6CRhMTdV11HGw9yMHWkWxNnBw212zmL/v+EvZ4XVcdLx14yeYVo48W/mFy+eWX88ILL9DX1wfAoUOHqKmp4UMf+hA333wzq1at4sQTT+S//3vw3QL/53/+h5NOOolFixbx6U9/GtUx9cCBA5xzzjksXbqUFStWcPCg8SH43ve+x+LFi1m6dCn33HMPEHw30dTUhOpt9Mgjj7B+/XrOPvts1q1bR1dXF+vWrWPFihUsXryYP/3pT+Y4HnvsMZYsWcLSpUv5+Mc/TmdnJ7NmzaK/vx+Ajo6OoO81muHS7mnH5/dFfP7NI8YmYjWdNfT094zWsOLCN//xTS7fcDlVrVXmY1JKrnnuGj7y5EdIhY7IY7ac08odL97BtrptcT3nssnL+Mn5P4n4fHFxMatXr2bjxo1cfPHFPPXUU1x55ZUIIfj2t79NcXExPp+PdevWsWPHDpYsWRLxXLfeeitf//rXAfj4xz/O888/z0c+8hGuueYa7rnnHi699FI8Hg9+v5+NGzfypz/9iXfeeYfs7GxaWloG/Vm2bt3Kjh07KC4uxuv18oc//IH8/Hyampo4+eSTWb9+PXv27OFb3/oW//73vyktLaWlpYW8vDzOPPNMXnjhBS655BKeeuopLrvsMl2zrxkRfulnzgNz+OqHvsrtJ99ue8ybR83dI6lqq2LhhIWjNbwRU9tZS5+vj3tfv5cnP/okAM9/8DyvV70OQJ+vjwxnRjKHqB3/SLDGPdaY5/e//z0rVqxg+fLl7N69mz179kQ9z9/+9jfWrFnD4sWLef3119m9ezednZ0cO3aMSy+9FDAWSmVnZ/Pqq69yww03kJ2dDRgXoME499xzzeOklNx7770sWbKEc845h2PHjlFfX8/rr7/OFVdcQWlpadB5P/WpT/Hwww8D8PDDD3PDDTcM9dek0QTR0dtBU3cTb1W/FfGYN4++SX5GPgCVrZWjNbS4UO+ux+Vw8dSup9h0bBP9vn6++MoXzee7+7uTODqD48LxR3PmieTiiy/mC1/4Alu3bqW7u5uVK1dSVVXFD37wAzZt2kRRURGf+MQnota5ezwebrnlFjZv3sz06dO57777hlUX73Q68fv95jmt5OTkmF8//vjjNDY2smXLFlwuF+Xl5VHf77TTTuPQoUP8/e9/x+fzsWjRoiGPTaOx0uZpA2Bv017b57v7u9lau5XrllzHQ9seGlPCL6WkrquOT634FM/ufZYvvvJFLl94Ofua93HR3It4Yf8L9Hh7KKIoqePUjn8E5ObmctZZZ3HjjTeabr+jo4OcnBwKCgqor69n48aNUc+hRLe0tJSuri6eeeYZAPLy8pg2bRp//OMfAejt7aW7u5tzzz2Xhx9+mO5uwzWoqKe8vJwtW4zSN3UOO9rb25k4cSIul4u//e1vHD5sdG09++yz2bBhA83NzUHnBbjuuuu4+uqrtdvXxAUl/Pua9tnm/JuObcLr93LpgkvJceWMKeFv722nz9fHnOI53Lf2Pt44/AZ3v3w3Z5WfxZUnXgmkhuPXwj9CrrrqKrZv324K/9KlS1m+fDnz58/n6quv5rTTTov6+sLCQm666SYWLVrEeeedx0knnWQ+99vf/paf/exnLFmyhFNPPZW6ujrOP/981q9fz6pVq1i2bBk/+MEPALj77rt58MEHWb58OU1NTRHf75prrmHz5s0sXryYxx57jPnz5wNw4okn8pWvfIW1a9eydOlS7rzzzqDXtLa2pnTFkib1aO1pZfGDi3mv9r2gx9s97QD0+no53B7eLl7l+6dOP5XZxbPHVGVPXVcdAJNzJ/OpFZ9iXsk8+nx9/PDDPyTbZcSzKTFZLaVM+f9WrlwpQ9mzZ0/YY5rEsGHDBnnttddGfH6s/Fv0efvk7obdyR7GuOHvVX+X3If8v83/F/T4H/f+UXIfkvuQz+97Pux1Fz5+oVzwvwuklFJe8tQlcuHPF47KeOOB+plfPfiqlFLKPQ175HN7npNSSvn8vucl9yHfPvr2qI0H2CxtNFU7fk1UbrvtNu655x6+9rWvJXsoI2bDng0s+v8Wxb0CTGPPobZDgOH8raioB8Jzfr/08++j/+a06cadckVhBZWtlSlRAhkLVscPsGDCAi5dYBRomI7fm3zHr4VfE5UHHniAAwcOcMIJJyR7KCPmSPsRJJJfbflVsocyLqhqM+rYWz32wp/lzGJvY7Dw723cS5unjdNmGMI/u3g2Hq+H2q7aURjxyKl31wMwKXdS2HNZrixAZ/wjZqy4gOOZsfRv0NRtzH38bufvcPe5kzya4x/l+Ft6gteaKOFfOWVlmONX+b7p+IsqgLFT0lnXVYfT4aQ4K7zMOpUy/jEr/JmZmTQ3N48p4TnekIF+/JmZmckeCk/ufJLbN9ovBlI0dTfhEA46ejv4/e7fj9LIxi9m1BPi+Nt728lLz2PRhEXsbdob9Bl+8+ibTMiewJziOcDYE/76rnom5kzEIcKlNcuZOo4/YXX8Qoh5wNOWhyqArwOFwE1AY+Dxe6WUfx3q+adNm0Z1dTWNjY2DH6xJGGoHrmTzwv4XeGH/C/z0gp9GPKapu4mlk5bS4+3hV1t/xQ3LdXlqIomW8RdkFrBgwgLaPG3Uu+uZnDsZKSV/P/R3Tp9xutnuu7ywHIFISeF397lxCIcZ4QDmz2JHKmX8CRN+KeU+YBmAECINOAb8AbgB+LGU8gcjOb/L5dK7PmlMOvs6afe045d+W7cFhvCXZpdywZwLuPPlO9nVsItFE/WCtETg9Xup7qgG7KOewsxCFpQuAIxcf3LuZDbXbOZI+xG+ceY3zGPT09KZXjA95Uo6n9j5BLe8cAuXzL+ERy55xHy8rquOSTnh+T6Mz4x/HXBQShletKvRxIHO3k4kks7ezojHKOH/+NKPk56Wrid5E0h1RzU+6cMhHLaTu4WZhSyYEBD+QM6/Yc8GXA4XF8+7OOj4iqKKlHH8rT2tXPXsVVzz3DV09nWyqWZT0PMxOf5xlPF/DHjS8v2tQogdQoiHhBC2a5eFEJ8WQmwWQmzWcY5mMDp6O4DwPNmKEv7S7FI+uuCjPLbjsZT4EB6PqJhnQekC26inMLOQqXlTyUvPY2+jkfNv2LOBcyrOoSgrWBJUSaei39eftMn58x8/n2f2PMM3z/omt6+5nYMtB/FLo1WKlJL6rvqIjj8jLQOBGB+OXwiRDqwHNgQeehCYjRED1QI/tHudlPKXUspVUspVEyZMSPQwNWOczj7D6VtrxK30+/pp722nNNtoQvexRR+jzdPG1tqtozbG8YQS/uVly2nvDW7B3N7bTmFmIUII5pfOZ2/TXjbXbOZQ2yGzrYGV2cWzqeuqMwXzmueu4bSHoq+ITwTN3c28e+xd7lt7H18946vML51Pr6+Xo+1HAcN09Pv7bUs5wdimNMuVlRIZ/2g4/guArVLKegApZb2U0iel9AO/AlaPwhg0xzkq4gl1l4rmHqMHkRL+2UWzATjacXQURjf+qGqtQiBYMtFoR269ILd52ijIKACMBU57m/ZGjHkguLLnn4f/yYY9G9hev91s/TBabK4x9rs4ZfopAJxQYqxt2d+yHwhfvGVHljMrZse/u2E3U344hdcqXxv2mCMxGsJ/FZaYRwhRZnnuUmDXKIxBc5wzmONv7g4W/ukF0wFMt5YsmrubzT7to8EdL97B/W/en/D3OdR+iKn5U033qyI4KaUZ9YARBdV01vC7Hb/j3NnnhsU8MCD8B1sOcvcrd5Mm0gBGfQW2yvNXlq0EYG7xXAD2NxvCX98VWLwVIeoBI+eP1fHvbNhJbVet+TcbTxIq/EKIHOBc4DnLw/cLIXYKIXYAZwF6U03NiPBLP119XUDkjF8t3lIfovyMfPIz8jnSfmR0BhmBn77zU9Y9to5n9zw7Ku/3+M7H+f6/v4/Xn9htOA+1HaK8sNxcyKTuxLr6uvBLf5DwA9R21XLFwitsz6Xuzr7/7+/z7rF3+dbZ3wIY9ZhuU80m5pXMoyDTuFuZkjeFbFc2HzR/AMTo+F2xO/6d9TtJE2nML50/wpGHk1Dhl1K6pZQlUsp2y2Mfl1IullIukVKul1KOjbXYmpTgR2/9iJ++HVyrb53oi+T4Q4UfYEbBjKRHPUo0bvjTDaZzTBRqA5Sm7ibeOPRGQt/rUNshZhXOoijTcPDqgqz+fUzhD1T2RIp5AIqzisnPyOfNo2+yeOJivnjqFynLLeO9uvdsj48Vr9/Lj976kSnY0ZBS8u6xd1k9dSCZFkIwp3iOGfVEa9egyHZlx1xQsKNhB/NK5yVkt64xu3JXMz55cteTPLnryaDHVMwDkTN+O+Gfnj896cJf2VrJkklLcKW5uHzD5QmtMrLuAfvMnsh7NgwVd5+bH731I/p8xv7T/b5+qjuqKS8sN6MbVcvf3mt4QCX8FUUVpKelR4x5wBBYFffcf+79pDnSWF62fMTCv7V2K3e9fBenP3R60O/GjmOdx6jrquOkKScFPT63eG5Qxu9yuMyLnR1Dyfh31u9k8cTFMR07VLTwa8YU7Z52UzwU1tr9wRx/SVaJ+dj0/OlJj3oOth7k1Gmn8vhlj7Ozfie3/vXWhL2Xapo2t3guz73/XNTNzofCxgMbuevlu3hur5HoVndU45d+Q/iV4+8JdvxqctfpcPL4ZY/zg3Ojr+e8eN7FXLvkWs6bfR4AyycvZ2/j3hFdKBvcDQAcbj/MaQ+dxq6GXdR01vDdf32X1b9azcb9A5sobTpm5PsnTQ0X/srWSrx+L/XueiblTjJXHdsRa8bf0dvB4fbDWvg1GjAcY2g1R5Djj5Lx56XnBd02Ty+YTlN3U5h49Pv64zjiyLR52mjpaaGiqILz55zPnafcyUPbHqK2MzHpp6qFv/vUu2lwN/DPI/+My3mVgD692+jQoko5rY4/UtQDcPnCy83IJxL3nXkfv730t6aorihbgU/62Nmwc9jjVmbgD//5B4QQrPn1Gqb/eDr/9dp/sbNhJ9/8xzfNYzfVbMLpcLJs8rKgc5xQcgJev5fDbYep66qLmu+DkfHHcrHa1WDUvCyepIVfo6Hd0x7m6mNy/D1NYdURMwpmAJitBdTX+d/N55+H4yOK0VBCPLvYmLw8e9bZwIAzHyo1nTVRL1qVrZUUZBRwzeJryHJmsWH3hojHDgUl/Bv3b6SjtyNI+DOdmWQ5s8Icv1X4h8PyycsBwnb3GgqNbmNh6NqZa/nXDf9i7cy1fPm0L7Pv1n18++xv81b1W+ysNy4sm2o2sXjiYjKdwQ0J55YEKnta9kddvKXIdmXHFPWo910yacmQf65Y0MKvGTP0envp9fXS4+0JEjjl+AsyCqJGPSXZJUGPTc8PlHRacv5Nxzbh8Xo40HIg3sMPQwm/yq9nFswE4HDb0DubtHvamfvAXB7b/ljU96soqiAnPYcL514Yt7hHCX+vr5e/7PsLVW1VOISDaflG876irCIz44+X8JcXllOYWTiinL+xu5GMtAxy03OZVTSLv17zV76z7jucUHIC1y+9noy0DP5vy/8hpWRzzeawfB8GSjo/aP4gNsfvjG0B186GneSl55l/E/FGC79mROxu2M3Nz98ct7w4GtZs3/q1cvwzCmZEjXpCHb+q5bfm/LsbdwPg7k98S4CDLUbjMVP4CwPCb7MP7WC8V/ce3f3d1HTWRDxGCT/AFQuvoK6rzux/PxIauxuZXzqfqXlTeXr30xxqO8TUvKmkp6UDUJRZZP67qJhOlUQOFyEEyyePbIJX/U3YZfIl2SVcvvByfrvjt+yo30Gbpy0s3weYmDORvPQ89jXto8HdED/H37CTRRMXRZ0vGAla+DUj4sUDL/KLLb8YFYdszfatzl45/hkFM6I6/lDhV47UuohrT+MegFHpBVPZWklpdin5GfkA5KbnUpJVMizHr2raI7lJv/RT1VZlCv9FJ1xEpjMzLnGPErwrT7ySlw6+xPb67ZQXlpvPF2cVB2X8Wc4s86IwEpZPXs6O+h3DXpPQ2N3IhJzI7WA+s/IzdPR2cPcrdwPYOn4hBHNL5vLOsXfwSV9sjn+QjF9KmdCKHtDCrxkhyhmrkrZEEuT4PeGOf3r+9MgtG7qbKc0KFv5MZyYTcyYGRT3xdPyVrZW83/R+xOcPth40FycpZhbO5FD7oSG/lyn8EUSlprOGPl+fKfy56bmsn7eex3c+HvUi5+5zmxfDSDS4G5iYM5ErT7ySPl8fO+p3BAl/UVZRUMY/0phHsbxsOR6vJ+rvOBqN7kYmZEcW/tNnnM6C0gW8WvkqWc4sTpx4ou1xc4vnmnce0Wr4YcDxR9tAqqazhlZPa8ImdkELv2aEqNvWRC8+gmCxD4p6+joRCKbmT6XH20Ovtzfodb3eXjr7Om2Xvltr+b1+ryki8XD8d7x4B5/44yciPm+NXhQzC2aOyPFHihFUnbr1/e5YcwetnlYe3vaw7WuklFy+4XLW/HqN2YHSDiWga6auMSfMg4Q/05Lx98ZR+Ec4wWt3F2hFCMFnVn4GMKqInA777UvmFs81fz+xVPVIpLnmwQ5VqaQdvyZlUQI5KlFPb4Sop7eT3PRcsz1AaNwT2qDNyvSCgVr+gy0HzQ/kUBx/V1+XbdzQ0tMScYFYv6+fI+1Hwh1/wUwOtx8e0pai7j63ecGKFPWETiSD0Wzs5Gkn85O3f2I7R/Pwtod58cCLdPV1hW2movD6vTT3NDMxZyJCCK5caHTXnFU4sEmSNeOPp+OfVzqPLGfWsHP+xu7ojh/guqXXkePK4fQZp0c8RjVrg+h9emCgJ3+0nF9V9GjHr0lZTMc/GlGPJ0LU09dJXkaeKSihwm+3alcxI3+GmfFbI43BhL+rr4vHdzzORU9cRNH3ivjyK18OO6a7v5sGd4OtWz7SfgSf9IU7/sKZdPd3mxerWNhevx2JcaGIJvwCYTpyxV2n3MXB1oP8ed+fgx4/2n6UL7z0BfLS8wAitjVQv1uVlV+/7Hry0vNYNWWVeUxxVjFdfV1Ga2xPe9yE3+lwsmTSkmH17Onz9dHR2xE14wcjptp1yy6+dsbXIh6jSjohBscf2Hc3WmXPjoYdTMmbYrthe7zQwq8ZEUnL+C1fd/R2kJeeF9YXRhFN+KcXTDe3bVT5fnlhedSoR0rJKb85hWv/cC27GnaR7cq2zeW7+7vx+r22bjm0hl8xnJJOJXyTciZFzPgr2yqZXjA9bFL10vmXMqtwFj98a2BbDCklN/3lJrx+Lw9c8ABAxEVlqhZ+Ys5EABZNXET7Pe1BblUt4mrztJn77caLZZOXDWsRV7S/iVDKC8vJSc+J+Lwq6cxIyzAn6iMRq+NPVP2+Qgu/ZkSoP+Aj7UfCsvV4E62qZ7iO31rLv7txN+WF5UzMmRjV8R/rPMauhl3ct/Y+qm6vYl7JPNsPsnrMzi2rPWTtHD8MraTzvdr3mJA9gTnFc6I6/tD3AkhzpHH7mtt58+ibvF39Nu8ee5ebX7iZlw6+xPfO+R6nTj814s8AAzX8SviBsBJEdUFu6Wkxop6M+Dh+gLLcMlp6WoZc2aMuWINFPbFQkl1CUWYRk3MnD1p+qfbdjXSB7vf1s7dpb0LzfdDCrxkhSiBVuWAiae9tJzc9l7z0vLCqnrz0vIH2AD2xO34VfRxpP8Luht2cOOFEclw5UR2/6gN/TsU5OISDnHT749XvRvVpt1LZWklGWgZT8qYEPT4sx1+3lRVlK6K2A6hsraSiMFz4AW5cfiMFGQV86OEPsebXa/jNe7/h+qXXc8tJt1CWZ2yfMZjwRxNQa9uGeGb8gLkoL9IcRCSG4vhjYV7pPPN3FY3BHP/+lv30+fq08GtSm+7+bvP2NtGVPe2edgoyCijMLAyr6onF8dtlpmoRV1VrFfua9xnCn54T1fG/V/uesbtU4HY80qKcwRz/rKJZOETwR7A4q5jc9NyYHX+vt5ddDbsM4Y+wKrS7v5u6rjpbxw+Ql5HH98/9PpfMv4THLnmMhrsbeOSSR3AIB7npueS4cqjtihD1dAdHPXao3/uxjmP0+/vjK/xZwxN+Ne7BMv5Y+fmFPzdjsWhEy/h7+nv4/MbP4xAOTp52clzGFQn7+iSNJkbcfW6WTlrKP4/8M+GVPe297RRkFuAQjrCqnrz06MJfmFmIK80Vds6y3DLSRBp/P/x3+nx9LJywkCMdR6I7/vptzCmeQ16GMfGZ4wq/UPilH4/XA9gLf6ToRQhhVvbEwq6GXXj9XlaUreBg60Fbx69650QSfoCbVt7ETStvsn1ucu7kqI4/TaRFbKkMA1GPuiNMhONXO6zFSjyjHjDKPWMhkuP3eD1c+vSlvF71Oo9c8kjQhHEi0I5fMyK6+7uZXjCdwszChE/wtvcajr8goyDc8afnkenMJNOZaTu5G+mWPs2RxpS8Kbxa+SoAJ0480VbIrbxX+15Ql0Y7x28VYLVBh0JKycGW8MVbipmFsdfyq4ld5fjt7jzsSjmHQlleWVThL80uDbtzsaIuCmotQTwnd9XdxFCqoMD4mxCIhFbO2GGX8ff5+rj895fz0sGX+PX6X3Pd0usSPg4t/JoR4e53k+PKCdqQIlG0ewzHX5hZGJ7xB9x3YWahbR2/tQ9/KNZWDwtKF0TN+Ns8bVS1VQUJv93xVgEOFc3mnmY6+zojCvHMgpmmSx+MrbVbKcgoYFbhrIi93kcq/JNzJ0eNeqLFPJBgx581TMff3UhxVjFpjrS4jSUW7Bz/c30prU8AACAASURBVHuf44X9L/DABQ9w4/IbR2UcWvg1I6K7v5scVw5ziueMTtSTUUBB5kAXzn5fP72+XrPe3LpYSDHYCk2V86uyvWgZ/476HcDAqlGwd/zW14cKv2rOFtHxF8yk1dMa1G46Eu/VvcfysuUIISL2galsrSTHlTPsiczJOdGjnsGE35XmIjc9N6FRz3Amd+OV7w8Fu4xfNda7ZvE1ozYOLfyaYSOlxN3nJtuVzdziuQkv6VSTu9aoRzVoi+b4BxX+QEnniROMXiw5rhy8fq/tsnrVHiDI8afn0OPtCVqoFc3xD+bAYy3p9Pq9bK/fzorJRr6c5TImd0NX/ar5hOF2eizLK6PN02bOWVhpcDfEJKBFmUXmXUw8hT8vPQ+nwznkqKexuzFuFT1Dwc7xt/a0IhBxjcAGQwu/Ztj0+/vxSZ8h/CVzE17SqSZ3lbhLKU1XrCqLrA3BFIMJvyrpNIU/sFjHLu7ZVr+NiTkTg1Zoqg+z1W2rD/aUvClhGb+q4Z9VNAs7Yi3p3Ne0D4/Xw/Iy4+4jy5mFX/rp9wdvxhJpIjlW1M9q5/ob3Y1MzI7u+MH4d1EXjngKvxBGTj+cyd14TewOBbuMv9XTSmFmYdR5knijhV8zbJQw5qTnmKsXYynpbHA3DKkXDRgTYB6vx3T8Xr+XHm/PgONPt3f83f3ddPd3x+T4F05YaPw8roDw28Q92+q2sXzy8iD3bHe8Ev6Kogoa3Y1BC4wOtBxgSt4U84IRSqyOX60nUHcfkRYH1bvrKcsdvMY8EpGEv9fbS3tve0yO3zqJqvbbjRclWSW0eIKjHill1L+xwcxAoshIy0Aggh2/pzVqVVQi0MKvGTbqjzfblc2c4jnA4K0b9jfvZ9qPpvHMnmeG9F7WDTzULXG7p910/CrqCc34lROM9iE/bcZpXDj3Qj48+8NAZMff5+tjd8PusH1X7W7frcIvkWb5IBi7Nc0rmRdxPJNzJ5Oelj6o499Rv4P0tHTzXJFqxN19bnLTc6OeKxqRhD+WGn6FmuBNT0sP275wpJRkl4Q5/uc/eJ6S+0vo6usKO94v/UbGnwTHL4QwIzlFa0+r+fsZLRIm/EKIeUKIbZb/OoQQdwghioUQrwgh9gf+P7o/sSZuKIeb48oxl60PNsH7+M7H6ff388bhN4b0XirTVwu4wKiwieT4lduLZYVmaXYpL1z9grnyMpLj39O4h35/f5jw210o1NeqS6U17tnXvC+oo2MoDuFgRsGMQR3/9vrtLJyw0FyfoBy/9QLk8/vo8faMSPjV3UJov57QPj3RUMJWmFkY912lirOKwzL+LbVbaPW02sZTbZ42fNKXlMldCC8GOK4cv5Ryn5RymZRyGbAS6Ab+ANwDvCalnAu8FvheMwaxOn4wuhRGc/xSSp7Y+QRgfDCHQpDjD0QF7b3hjr8wsxC/9JsXhOEszY/k+FW0Yq3ogcEdPwy45ebuZlp6WqI6fiCmRVzb67ezdNLSsHHYzTVEazI2GBNyJiAQYSIaS7sGhRK2eOb7ipKskrCqHnWR6ujtCDs+3u0ahkroCuuWnpbjx/GHsA44KKU8DFwMPBp4/FHgklEagyYCje7GiDtXRcOa8QPMKZ7D/ub9SCl5u/pt/t8//1+QeG6t3cr+lv1Mzp3MtrptQ2qsZXX8QVFPiONXHyCV8w9L+CM4/vdq3wuKtaIdrwRXlWwq0dzXvA8gquOHwTdkaXA3UNdVFyT8dlGP9a5suDgdTibkTBhR1KMy/kQJf2jUU9NllEha13so4r1qd6iEOf6e1lFfSDZawv8x4MnA15OklOqesQ6w3blACPFpIcRmIcTmxsZGu0M0ceKy31/GzS/cPOTXhTn+QEnnyl+u5JTfnMK9r9/Lt/7xLfP4J3c9icvh4isf+goer2fQLf2sWB1/UNRj4/jVczAyxx+aD2+r38aSSUvCFv1Ec/yqckc1atvXZAj/vNJBHH/hTGq7aiOWx26v2w7A0skW4beZ3FU/w0iiHjDintBFXHadOSOhLsjxntgF46LS4+0J+rmjOf549+kZKtZmelJKI+o53hy/ECIdWA+E7eosjSDWdupdSvlLKeUqKeWqCROS8w80HpBSsqN+h7kL1VAwY4SAm1w+eTkSidfv5cGLHuSqRVfxw7d+yAfNH+CXfp7e/TTnzTmPcyvOBWBLTexxT5Djt0Y9oY4/pEPn1rqtFGcVD8lRmQ7ecrcipTQresKOt4mG1O9mQvYEctNzTbf8QfMHuByuoK0J7VBzA28efdP2+e31AeEfzPGH3JUNF7t+PQ3uBlwO16A96CHBUY/NIi61KMra2kOR7KjHusLa3e/G6/cePxm/hQuArVJKNbtVL4QoAwj8v2EUxqCJQKunlY7ejiGvfISBGEE53vXz1lN1exXbP7udz676LD8670dkOjO548U7+NeRf1HdUc3Vi65mbslc8tLz2FyzOeb3ilbV43K4yHBmAMGOX0rJa5WvcVb5WUOqkTaF3BLddPZ10tHbEbSloMLO8bv73TgdTlxpLkM03QNRz+zi2RH3b1VcMv8SZhXO4pN//qRtXLG9fjtT86aaogf2jj8eUQ/YC3+ju9HccnEwrJO78cZs2xCY4PX5feZkuq3jD0Q9ycz41d+KMijHneMHrmIg5gH4M3B94OvrgT+Nwhg0EVCrSIcj/KETh0IIygvLTSGYnDuZb5z5DTYe2Mjn/vo5sl3ZrJ+3HodwsKJsxZAmeJVzy8/IJ8eVQ5pIo83TZuy+FYh5gKBduA60HOBox1HWzVo3pJ/LzvGbH1AbZxYp41cXhEk5k4Ic/2ATu2BEV49f9jhH249y68Zbw57fXrc9KOYBe8cfz6inrqsuqDa+oTu2VbuQ2IzfbNQWyPkbuxvNVdR2F82m7iayXdkR11EkmmxXtnlxVp+748rxCyFygHOB5ywPfxc4VwixHzgn8L0mSaiOia2e1iEvqlLCGO0DdOvqW1k4YSG7Gnaxft568yKxsmwl2+q20e/rj/haK+2ednJcOTgdToQwlrerqEfFPBDs+F+reg2AdRVDE37181iFXM0Z2Dkz9TOFZvzqgjA5dzL1XfX4/D4OtBwYdGJXccr0U/j62q/zux2/M6uhwFg4tbdpb1DMA/blnPGMevr9/UEGIZY+PYrRiHqU41cxD0TO+JM1sQvGv5Pp+D3HoeOXUrqllCVSynbLY81SynVSyrlSynOklEO3mpq4oRy/1++1XewSjdCM3w5XmosHLniANJHGDctuMB9fNWUVvb7emCd4VbsGherXozZhUai8ubWnldeqXmNa/jRzVXGspDnSyHRmBjt+T2THr5x2aMavLiAqJjnSfoReX29Mjl9x74fu5bTpp3HzCzeb8zB7m/bi9XvDhN+unDOeUQ8EL+JSUU+sry/IKBjSzx4roZuxWNcb2GX8jd2NSZvYheCMX91JHq9VPZoURQk/DD3ucfe7cQhH2AbeoZw962xavtxirowFWDllJUDMOb/qzKlQC7XUJiyKNEcaBRkFtPS08Leqv7Fu1rphLRgK7ckfLYtVF4rQjN8q/K2eVnNT8FgdPxillL+99Lf0+/q586U7AfuKHoge9YzU8dttwdjgbojZOeem51J7Vy2XLbhsROOwIzTqUY4/05kZsY4/Wfk+hGT8UQxFItHCP86pbBsQ/tB2xoOh4oxYhDW08mNO8Rzy0vNizvlVL35FQWaBWcdvdfxgXBTeOPwGzT3NQ873FaGtmdXvJlJUke3KjprxA/zj8D+AwUs5Q5lVNIuvnvFVnt37LC8deInt9dvJcmaF3cnYTu4G7kJGmvErx69KOrv7u3H3u2N2/Gp88V61q86b5cwyox41xjnFcyJO7iYz6rFm/Mfz5K4mhalqrTI/1EN2/H3uYU+QOYSDlVNWxi78IY7fjHpCHD8Y7km566Hm+4rQzVXMjD+CM8tx5YRl/FbHD4bwF2YWDkt07jrlLk4oOYHbNt7Gu8feZdHERWHrCZwOJ06HM+4LuCA86hlKu4bRoCS7JCjqKckqYUL2hMhRTzIz/oDjl1LS0tOCQzjCzEui0cI/jvH6vRxuP8yqKauAoQt/t7d7RBHCyrKVbK/bHtMEb6jjN6OeCI4fYH7pfKbkTRnW2MIcf6BneqSadTvHr343SjS31m7lhJIThuV6M5wZPHDBA+xv2c+bR98My/cVoZuxuPvcZKRljHinqbz0PLJd2abwD6Vdw2hQklUyMLnbVcOUvCnkZ+SHOf5YurUmmmxXNhJJn68vKS2ZQQv/uKa6oxqv38vKMiNvH2rbhpE4fhiY4N3duHvQY20df6COPz89WIzVbfNwYx4Id/yDfUBz0iM7/km5RtTjk74RTW5+ePaHuXzh5UB4vq8I7fzY1dc14pgHjFJd6xaMQ1m1OxpYe/LXdtZSlldmxoFW1OKtZE7uWquvWj2j364BtPCPKTp7O/nuv76Lz++Ly/nUxK4S/iE7fkvJ4nBQ7/t29duDHqt231IUZBbQ0dsR1fGfPevsYY/NLuOPVoqY7coO684ZmvHD0CZ27fjxeT/m/Dnnc+HcC22fD91w3d3vHvHErkJVJ0kp2bDHWIg/NX9qXM49UqxRT01nwPGnhzv+ZC/eAkv1lbcnKS2ZQQv/mOL5D57nv177L7NLpBW7bfEGQ9XwL5q4iPS09GFV9YzE8c8pnsMJJSfw6PZHox7X7+unx9sTFvVIJH7pD8v4S7NLcQgHZ5afOeyx2WX80SovbDN+p/G7yXBmmB/ukZYzTsufxsZrNkbcUSt0w3V3v3vE+b6iLLeM2s5a7nr5Lh7d/ij3nn4v0/KnxeXcI0VFPX7pNzeeUebAuj7FdPxJzvhhwPGPdkUPaOEfU6h8NXTCqqazhoLvFvBa5WtDOl9layVOh5PpBdNtNykfDGuOPRyEENyy6hbern6brbVbIx5n7dOjsH4d6vhvW30bf/7Yn0d0C52bnhuW8UdzZtGqemAg7hlqRc9QsTYAg/hFPWA4/r1Ne/nx2z/m86s/z7fO/tbgLxolVGtmtdtZWW4Z+Rn5+KQv6IKcClGPdb2FdvyaQVH9R0Jzy0Nth+jz9fFK5StDOl9lWyUzCmbgdDgpzioe1aoexfXLrifblc3P3/15xGOsfXoU1q9DHf/0gulcdMJFIxqXXcYf1fFbMn6/9NPj7Qm6KKoJ3tCWzvEmtNe7uy9+UY/akOXGZTfy4/N/nJDSzOFSnFWM1+81215PyZsS1MxPoTpzJrWO35LxJ6MXP2jhH1OoCbVQx6+E8d1j7w7pfNZNuIuzioddxz8SCjMLuXbxtTyx64mIFx47x2/N2xNRCmdX1VOYESXjdw5k/Cp2s14UywvLmVM8J+H9YUIdfzyjnuuWXscDFzzALz/yy1GvQhkM1bZhV8MuwFhwpiqwrDl/U3cTaSItIa0jYsXa1G+wCDFRpNa/niYqkRy/EsbNNZuHNPFb1VpFRaEh/EVZRaOe8Ss+t/pzeLweHn7vYdvnbR1/RmTHHw9yXDn0+frMzWIGzfgtjt+uh9H959zPi9e8GPdxhhLq+OMZ9UwvmM6tq28dcWloIlBtG5TwT8mbEtTFVdHU3URJdklSL1wq429wN+CTPl3Vo4mO2swjtFJB/WF39nWat7qD0dnbSWN3o7lRyHCinng4foAlk5Zw+ozTeXDzg2ZXRSu2GX9m5Iw/Hlh77Pf099Dr640p45dShm1QA0amPLt4dtzHGUqY4++Ln+NPZZR4KuGfnDvZ1vE3djcmNeaBgajnWOcxYPRX7YIW/jFFpKhHrSqF2OOeqjajoseMejKLh1TH7/P78Hg9cYsuPnfS5zjYepCXDrwU9pyd4w+KehLk+MG4q4mln0qOKwe/9NPn67MV/tEiLOOPYzlnKmONeooyi8h0Ztpm/Mnu0wMDfxfHOgLCr6MeTSSklFGjnjSRRn5GPpuObYrpfKqGXwl/UVYRnX2dMbdJVuISL1G5bMFlFGUW8cyeZ8KeG2pVTzywOn51QRysjh8MoY2la2miCN3Pdbw4fhX1tHpazdXakTL+ZK82VlGPdvyaQWnvbafP12d+HfRcoJ3BqimreLcmRsffGuL4A7fK1gleKWXEHv2x9OIfCulp6Zwx8wzeOPxG2HPqQmdtl5DhzCAjzdh1K9GOP1ovfvN4y4UidGey0cTasqHP10e/vz9uGX8qY3XNqpOoXcbf6E5+1KP+LlQXUe34jyNerXx1SHvKDobK98Em4w+0M1g9ZTXb67bHtJirsrWS/Ix8U8xM4bfEPV965Uuc9ehZtq9PhKs9s/xMDrYepLqjOujx9t52sl3ZuNJcQY8rB54IYQty/DFEPdZKjaRGPYGWDVLKuG3CMhZwOpzmXaAqO1WGQH1e/NJPc09z0oU/NOPXk7vHEZ/f+HnWPrJ2yCWWkVD5vtPhDHf8gU1KTpp6Ev3+frNfezQq24xSTlWLrS4A1gnet6rf4s2jb9rGP4lwtWtnrgXgjUPBrj+0XYOiILOAbFd2QqpMgjL+GFrnWo9Pdsbvl376/f1x68w5VlA5v4p60hxp5Kbnmp+X1p5W/NKf9KgnIy0DgRhw/DrqOX5Q/covePwC9jbuHfH5VL4/u2h2WMbf5mmjMLOQ1VNXA7FN8Fa2VgZtHK5ch1X4K1sr8fq9HGw9GPb60P1248GSSUsoyCgIi3tCd99SFGYWJiTmAXvHH0vGb3X8yXDa1p788dpvd6ygcn7l+IGgDp1q1W6yHb8QgixXFh6vhzSRlpR/Hy38CcLj9fDh2R/G5XBx7m/P5XDb4RGdT0U9c0vm2mf8GQVMzZtKWW5ZTDl/TWcN0/Onm9+HZvw9/T1mJ0a7C1e8M34wHNqHZn7IXvjtHH9GQcL6mNtl/NGE33qhSLbjB2PyfTxFPTDwN2xtxa32bYDUaNegUH8bRVlFSVkBrYU/QXi8HhaULuDlj79MV18Xn3/x8yM6X4O7AYGgorDCPuPPLEAIweqpqwd1/N393XT0dphtBGAgv1aO/1DbIfO5vU3hwp+oypW1M9fyQfMHwfumeuwd/zkV53De7PPi+v6K0Kqe3PTcsDkGK1bHn4iLYqxYHf94jXrU5C4EO/5UaNegUBfoZMQ8oIU/YfT6esl0ZrJk0hLOn3M+uxsG7zkfjXp3PaXZpRRnFdPV1xW0Qteaga+eupoPmj8Iqu0PO1fg7kE1DoMBN6uEX9X5g73wJ6pyRXXUVK6/wd3AnsY9TMsL7wL5pdO+xP9e+L9xfX9FaB3/YB/QVMn4rRcgHfUQ1JM/VaIeCHb8yUALfwKQUuLxesxyw1mFszjSfmREffTr3fVMyp1kOl9rpUJHb0eQ8EP0TczVfIHV8auqCDWRqer8l0xaYhv1JCrHXjZ5GXnpeeYE71de+wo93h6+eNoX4/o+gxGa8Q/W2yU0408Tabgcke8QEsV4jnrKcstwOVwRHX8qCb+6M0tGRQ8kWPiFEIVCiGeEEO8LIfYKIU4RQtwnhDgmhNgW+M9+R4kxjKq3z3RmAsZm2f3+fnMWfzg0uBuYmDMxbDViV18XEmkK0+KJiwF4v+n9iOdS7Z2twg+Bfj2egONvrSLLmcXamWt5v+n9sFYKiXK1ToeT02eczhuH32BLzRZ+895v+PzqzzO/dH5c3yeWcaSnpZtVPYM5s9CMPyc9tk3o4814jnpuOekW3vjEG0F/k9aMv9HdSLYrOyl3YqGYjv84jXp+CrwopZwPLAWUdfyxlHJZ4L+/JngMo06vrxcYEP7ywnIgOD4ZKvVd9UzKmRS2GlFFOupOQF0AQit/rEQS/uKsgbYNqtxz4YSFuPvdYbX1pptMgKisnbmWvU17+eSfP0lpdilfW/u1uL9HLKjWzG2etkE/oKGOP1niYuf4x0vUU5BZwCnTTwl6LMjx9yS/XYPiuM34hRAFwBnAbwCklH1SysjB83GEWkCV4RyIemBgtaziR2/9iPN+d17E1bFW6t2G8IeuRjT72ATuBNSK1tAJYCt1XXUIRFg9s7VRW2VrJbOKZrGgdAEQXtmjHL9ymPFkbblRz7+9fjvfWfedpLXQVa2ZY9klKT0tHafDibvfHbeupcPBrpxzvEQ9dhRkFJhzYqnQrkGR8hm/EOIjQgyrh+ksoBF4WAjxnhDi10II9Rd4qxBihxDiISGE7U8uhPi0EGKzEGJzY2PjMN4+eSjhV45/RsEMBCLM8f/lg7/w8sGXea0q+s5ZaqLOLuox+9hYql7yM/LDSj6t1HXVUZpdGlalUpRptGaWUpotm1XEEjrB6+53k+nMTEh725VlK8lNz2VF2QpuWHZD3M8fKzmuHDPqidaLX6H65KSM4+93IxDmY+MRdYfc2deZEu0aFOoCncqO/z+B/UKI+4UQQwlancAK4EEp5XLADdwDPAjMBpYBtcAP7V4spfyllHKVlHLVhAmpcZWOlV6vEfWoyd0MZwZT86eGCb9y0T9956dRz6dW7VondyM5fghUMgwi/KExDwxsxtLc00xnXycVRRVMzJlIUWZR2JxBvFoy2+FKc/HytS/zh//8Q1J7v+ek59DmacPd747JmaloKKnCb834A7tvpdJOWaON9fPS1N2UEjX8MOD4U3ZyV0p5LbAcOAg8IoR4K+DGB1s5Uw1USynfCXz/DLBCSlkvpfRJKf3Ar4DVIxh/ShLq+MHI+a218c3dzdS765mcO5nnP3ie/c37I57PLL+0yfiVwFvjEGuuaUc04W/paTEremYVzUIIwYIJC2wdfyLF7ZTppzCjYEbCzh8LOa6cgda5MTizbFc23d7uhF4UYxkDDNwljpeJ3UhYPy+N3Y2UZqWI41cZf6pGPQBSyg4M4X4KKAMuBbYKIW6L8po64KgQQu0uvQ7YI4Qosxx2KbBrOANPZUInd8HI+a0ZvxLS75z9HVwOFw+8+0DE86nyS7uoJ3RyFwKVDINM7toJf1FmEV6/l531O4GBzp0LShfYZvzHe3ack55jTmrH5PjTU8Dxh0Q9x/u/0WCoz0uDu4Guvq6UiXpSvqpHCLFeCPEH4O+AC1gtpbwAo0rnrkFefhvwuBBiB0a08x3gfiHEzsBjZwFfGMH4U5LQyV0whL+6o9os9dzTuAeAs2adxccWfYyHtz0cUaytUU+mMxOXwzVo1BPJ8au+/pNyJoU9p247t9QaXUVVNdKC0gU0djfS3N1sHhuPjdZTnRxXju0dVSRUxp8qk7vx3G93rKIcv7qLTZWoZyw4/o9ilF8ullJ+X0rZACCl7AY+Ge2FUsptgZx+iZTyEillq5Ty44FzLZFSrpdS1kY7x1jELuqZVTQLieRI+xHAyPezXdnMKJjB7Wtup6uvi4fee8j2fCrqmZgzESFEUIbf3tuOy+EKeq9ok7sdvR14vJ6IUQ8Yi78m5kw0ywAXTAhU9ljinmTGGaOF1S3H4szUZHAyHb/T4cTpcNLj7YnrfrtjFXUnrBoNasdvEIvw3weYzV+EEFlCiHIAKWX0cpRxiprcDc34YaCkc0/THhaULsAhHKycspLTZ5zOzzf93PZ89e568jPyzfNZM3zVx8Y6gVeQEdnxR6rhhwH3sb1+uxnzALYlncl0taNFrmtANGNxZqlQ1QMDm7Goyd3xjHL8qSb8yyYvY9HERUzMmZiU949F+DcA1mWbvsBjmgiYUU9acNQDA83P9jTuYeGEhebz/zH3PzjYepDO3s6w8zW4G4KiGetqRLvOlerCYLc+IJrwK8ff5+sLatk8s3AmWc6scMd/nIvKkB2/JeNP5t2Q2oxFRz0DEejBFkP4U6WO/4K5F7Dz5p1BcfBoEovwO6WUfeqbwNfpiRvS2Mcu6pmWPw2nw0lVWxUdvR1Ud1QHCb9y2Hare1WfHoW18VR7b3tY/lyQUYBf+s0FPFZiEX7reAAcwsG80nlBwj9eMn5FLBl/jiuHrr6u1HD8gZW74z3qyXZlkybSUs7xJ5tYhL9RCLFefSOEuBhoStyQxj52VT1pjjRmFMygqq3KrIlXEQoYcwAwMAllpb6rPuiW0Or42zxtYS2L7TaZVkSNeiyu1ur41VjVhDSMr4w/Iy0jphXK2a5smnuaza+ThYqcdDmnsemJugMWiKTVzacasQj/Z4F7hRBHhBBHgS8Dn0nssMY2dlU9YOT8Va1VpoDaOv7WCI7fEvXkZ+QHVfWERj3mohWbCd66rjpcDpdtZp3tyiY9LT1oPIplk5dxpP0IjW5jFfV4yPiVaMZaeZHjyjGrtpLq+F1ZA1U9x3kcFwvKCBVnFSd1QWAqEcsCroNSypOBhcACKeWpUsoDiR/a2MUu6gHDRR9qO8Sexj1kpGWYLh8Mt52fkR/m+Pt9/bT0tIRl/NYFXKGOX10IbB2/u45JuZNsWy0IIUzXHyr8J087GYB3jr2DlHJcOf5YKy+sYp/sqEdtCDPeox4YMEI65hnAGctBQoiLgBOBTFU9IqX8nwSOa0xjV9UDhvDXu+vZXLOZeaXzcDoGfv1CCGORV0jGr3YNCop6AnX6Ukpbx68cjt26gEiLtxTFWcU09zQzLT9445OVZStJE2m8U/0O51Scg1/6x43jj7VJnNVdJ9NpZ7myaOpuQiKP+4tzLKjPQ6rU8KcCsSzg+gVGv57bAAFcAcxM8LjGNMrxh27EoRz+v478KyjmUVQUVYQ5frvdsgoyCvBJHx29HXT2dYZP7mZGdvyqvXMkirOKmVkwM+yWOCc9h8WTFvP2sbeTupn4aGI6/hijnlRy/GrTkeP93ygWlDHSjn+AWDL+U6WU1wGtUspvAKcAJyR2WGMbj9dDpjMzrDmWquXv9/ezsDRc+JXjt5Zhmqt2QzJ+wGwnENHxR8j4ozn+KxZewY3Lb7R97uSpJ/NO9Ttmyel4cfyxRj1Wd53sjF/NxeioZ+DzkCp9elKBWKIeT+D/3UKIKUAzRr8eTQR6fb1BNfwKa6WMWg1rpaKoAo/XQ11Xnbl9nKrC81Lx/wAAGWdJREFUCY16AHMVcKwZv8/vo8HdEFX4bz/59ojPnTztZH6x5RdmS4fjPUYYyxm/qiw73v+NYkF9HnTUM0Asjv8vQohC4PvAVuAQ8EQiBzXWUY4/lMm5k83H7aIeFQVZc/6dDTvJSMtgZuFAuqb+kI92HA36XqFcXmjG39zTjE/6ogp/NNQE7+tVrwPjx/EPJ+NPdjmnQkc9Fsevox6TqMIf2IDlNSllm5TyWYxsf76U8uujMroxSiThF0JQXliO0+FkTvGcsOdVJY0153/32LssL1tullnC4I4/zZFGXnpeWNQTrYY/FuaWzKUos8gU/uNdVEaS8Sd15a5l4xXt+Ac+H6myajcViCr8gZ75P7d83yuljNzvVwMEop4IS7Hnlcxj4YSFQUKuCO3n4/V72VK7hdVTgrcsUA5GCb+dI7XryT9S4XcIB2umrTFX8B7vjn9q3lS+dsbX+OiCj8Z0fCpl/Aqd8WvHb0csGf9rQoiPAs/JWDaH1UR0/AA/v/Dn9Hh7bJ/LdGZSlltGZZvh+Pc07qG7v5vVU4OFX0U7puMPiXrAfheukQo/wJqpa3jxwIvA8e8mhRD8z1mxVy2nUsavON7vymJBV/WEE4vwfwa4E/AKITwYJZ1SSpmf0JGNYXq9vRGFf2r+1KivrSiqMB3/O9XG5mVhwj9I1AP2HTrjIfwq54fj3/EPlVTJ+K2O/3i/OMfCORXncNOKm1g8aXGyh5IyDCr8UsrBtljUhODxemyremJhVtEs3jj0BmDk+0WZRWHzAbnpuQhExHJOMG5vW3pagh6r66ojx5Uzott/60VIu8lglNiniTTbKG+0sDp+HfVAWV4Zv/zIL5M9jJQilgVcZ9j9NxqDG6tEi3oGo6Kwwtyp692ad1k9dXXYegCHcJCXkUe/v5+MtAzb+QS7XbhCu3wOh+KsYuaVGLtpascfjHLX2a7spG5wHuT49cVZY0MsUc8XLV9nYmyOvgU4OyEjOg7o9fVS6hxenqh26trbuJddDbu4eN7FtsepKMcu5gHITw/fhWuwxVuxcvK0k9nXvE/HCCFkOjMRiKRfENX7Ox3OpN55aFKXWKKej1i/F0JMB36SsBEdB3i8nmFvsKBKOp/Z8wx+6Q/L9xUFmQUc7Tgascbc2rNfUddVx/zS+cMal5XPrvosJVkluNJcgx88jhDCEP1kC7+KenTMo4lETE3aQqgGwpedakxGEvWo1b1P734aCJ/YVahc3y7fByPj7/H20O/rNwW6trOWtTPXDmtcVk6ednLQJK9mgJQQ/kDUo+/INJEYVPiFEA8AqozTASzDWMGriUCvt5fMtOEJ/5S8KbgcLva37Ke8sDzinpwq4okU9VjbNpRkl+Duc9PqaWV6/vRhjUsTGznpOUnP1ZXjT/Y4NKlLLI5/s+VrL/CklPLNBI3nuGAkUU+aI43ywnL2t+yP6PZhYFFKJMdv7dBZkl3C4fbDAEGtHzTxJ5Ucv456NJGIRfifATxSSh+AECJNCJEtpewe7IWBHj+/BhZh3DXcCOwDngbKMfr+XCmlbB3W6FOUkUQ9YEzw7m/ZH7Zi10osUQ8MdOg83BYQ/gIt/Imkoqgi6QuFTMevox5NBGJp0vYaYN1wNAt4Ncbz/xR4UUo5H1gK7AXuwej/Mzdw7ntiH+7YIFJ3zlipKDQmeKM5fiX4ESd3Qzp0asc/Omy4YgO/uOgXSR2DmfHrqEcTgViEP1NK2aW+CXw96L2sEKIAOAP4TeB1fVLKNuBi4NHAYY8Clwx10PGkp7+Hjfs3Eq9uFF6/F6/fOyLHf9LUkyjJKmFF2YqIxwyW8YfuwnWk/QhOh5OyXN1RO5FkOjOHHfPFCxU1aceviUQswu8WQpgKJIRYCdg3mwlmFtAIPCyEeE8I8WshRA4wSUpZGzimDhjZiqIRsmHPBi584kK+9MqXgsTfL/1srd1qe0Go7azF5/fZni/StotD4YZlN1B9Z3VUxxZrxm9GPe2HmZY/TW82PQ7Q5ZyawYhF+O8ANggh/imE+BdGPn9rDK9zAiuAB6WUywE3IbFOoOmbrdUWQnxaCLFZCLG5sbExhrcbHofaDgHwg7d+wP1v3g8Y9e7n/e48Vv5yJa9UvhJ0fLunnTkPzOGh9x6yPZ/aAGMkrk8IMeiFw8z4B3H8ZtTTdljn++MEXc6pGYxBhV9KuQmYD9wMfBZYIKXcEsO5q4FqKeU7ge+fwbgQ1AshygAC/2+I8L6/lFKuklKumjAhcX20qzuqKc0u5apFV3HPa/fwhRe/wNJfLOWfh/8JGB0yrRxoOUB3fzebajbZnk/ttzsSxx8LZtQTyfEHHldRz+H2wzrfHyc4HU6ynFkxbyCjGX/E0qvnc0COlHKXlHIXkCuEuGWw10kp64CjQoh5gYfWAXuAPwPXBx67HvjTsEYeJ451HmNGwQweueQRzp9zPj955ydMyJ7Alk9vIS89L2zzc/W96kkfSjyinliYUTAj6P+hZDozcTqcdPR20O/rp6azRjv+ccTzVz/PbWtuS/YwNClKLOWcN0kprZuxtAohbgL+vxheexvwuBAiHagEbsC42PxeCPFJ4DBw5dCHHT+qO6opLywnPS2dZ698lj+9/ycumX8JWa4sZhXNiiz8jfbCrxz/SKp6YmHZ5GVU3V5lbt4SihCCggyjJ391RzV+6dfCP444e5ZupaWJTCzCnyaEEGoTFiFEGhBT5ycp5TZglc1T62IfYmKp7qjm9OmnA0Y1xFWLrzKfqyiqYF/TvqDjlfA39zTT6G4M28B5tKIeIKLoK1SHTtW3X0c9Go0GYpvcfRF4WgixTgixDngS2JjYYY0O3f3dtPS0MC1/mu3zFYUVVLVVBVX2VLZV4hDGr80u7lGTu6Mh/IORn2F06DRr+LXj12g0xCb8XwZex5jY/Sywk+AFXWOWYx3HACILf1EFHq/H3LkKDMe/ZuoawD7uMaOeJNdygzHB2+5pN1ftTi/QfXo0Gk1sVT1+4B2M9gqrMfrw2wfcYwy1g1Wk7RBVi2QV73j9Xg63HWbtzLVku7JtHf9oRj2DoTZcP9x+mEk5k1JiTBqNJvlEFH4hxAlCiP8WQrwPPAAcAZBSniWl/N/RGmAiUcIfzfHDgPAfbT+KT/qYUzyH+aXz7aOeUarqiQW14bou5dRoNFaiOf73Mdz9f0gpT5dSPgDYL1cdo5iOP8/e8c8snIlAmMKv/l9RVMGC0gXRo54EV/XEQn56wPHrxVsajcZCNOG/DKgF/iaE+FVgYjd5G4kmgGOdxyjKLIrYGiHTmcnU/KlUttkL/9GOo3T1dQW9JpWiHrUL15H2I1r4NRqNSUThl1L+UUr5MYxVu3/DaN0wUQjxoBDiw6M1wERS3VEdMeZRVBRVBDl+p8PJtPxpLJhgbEL2ftP7QcenUlVPQUYBPumj19erox6NRmMSy+SuW0r5RGDv3WnAexiVPmOeWIR/VuHAIq7KtkrKC8tJc6SxoNQQ/tC4J5WqelS/HtClnBqNZoBYyjlNpJStgR46KbMAayTE6vhrOmvweD1UtlaaE75ziufgdDjDJnhTLepRaMev0WgUQxL+44k+Xx/17vqIE7sKJfSH2g4Zwh/YJMWV5mJO8Zww4VdVPSkxuWtx/JF6+mg0mvHHuBX+ms4aIHIpp0IJ/9barbT0tJjfA7aVPR6vB6fDmRJ971WHzvyMfN2pUaPRmIxb4R+shl+hhP61yteCvgdD+A+2HqTf128+NtL9duOJcvw639doNFbGrfAP1q5BMSlnElnOLF6tMrYZtgr//NL5eP1eDrQcMB8b6X678URl/Drf12g0Vsat8Mfq+IUQVBRVmB0ugxx/oKTTmvNrx6/RaFKdcS38uem5QROgkVBiX5xVHFQpM790PhBcy9/r600Z4S/IKGDJpCWcMfOMZA9Fo9GkELH04z8uqe40SjmFGHwxshJ+q9sHYzPrgowC6rvqzcc8Xk9K1PADpDnS2P7Z7ckehkajSTHGteMfrJRTEUn4wbgLaO5pNr9PpahHo9Fo7BjXwj9Yvq+YVTgLwKzht1KSXUJLT4v5fa83daIejUajsWNcCr/P76O2szZm4T+h5AQA5pbMDXvOzvGnSlWPRqPR2DEuhb/eXY9P+mIW/nml83jh6he4evHVYc+VZJXQ3K2jHo1GM3YYl5O7sZZyWrlw7oW2jxdnFQdHPSlU1aPRaDR2jEvHPxzhj0RJVgltnjZ8fmOPmlSq6tFoNBo7xrXwx1rVE43irGIkkjZPG6CjHo1Gk/okVPiFEIeEEDuFENuEEJsDj90nhDgWeGybEMI+Q0kgtZ21uBwuSrJLRnwudQ41wdvrTZ2WDRqNRmPHaGT8Z0kpm0Ie+7GU8gej8N621HbVMil3Eg4x8utecVYxgJnza8ev0WhSnXEZ9dR11VGWWxaXc5VkBRx/oLJHC79Go0l1Ei38EnhZCLFFCPFpy+O3CiF2CCEeEkIU2b1QCPFpIcRmIcTmxsbGuA6qtquWybmT43IuFfW09LQgpUyp7pwajUZjR6KF/3Qp5QrgAuBzQogzgAeB2cAyoBb4od0LA1s8rpJSrpowYUJcBxVPx6+inuaeZvp8ff9/e/cbY9dR3nH8+9tdx944dvw3i5M1dRBuUKCNEy2R2yJEA4qgiQiIKhRQG6FIKagqaVUgQbyq1L5oVTUlECEZKBgIfw2GiBcWwYS2UtuQTesaQwKkwXTXWdvrJLZ3k3jjP09fnDn2zfoudsidvXfP/D7S6p47d+85Mxrr8exzZuYAvfHYRTOzuWQN/BGxL70eBLYD10bEgYg4GRGngE8B1+asw2wnTp1g8plJ1i3rTOBfsWQFQjz57JPMnKweu+jAb2a9LFvgl7RU0rL6GLge2COpNeK+HdiTqw7tHJg+QBAdS/X0qY+Vgyt56rmnTj9o3fP4zayX5ZzVMwRsT9seDwBfiogdkr4gaRNV/n8v8KcZ63CW/dP7ATqW6oG0bcNzT54O/B7xm1kvyxb4I+Jx4Ko25X+c65rnY2J6AqBjqR44s23DzAmnesys9xU3nXNiqgr8nUr1QDWzp3XE71k9ZtbLigv8dapnaOlQx85Zj/id6jGzhaC4wD8xPcGqwVUdvQFbb83sWT1mthAUF/g7OYe/tnpwNVPPTzE1MwV4Vo+Z9bbiAv/E9ERHb+zCmUVc9Y1jj/jNrJeVF/inOrddQ63etmHf0X2Ab+6aWW8rKvBHRJZUTz3if2LqCcAjfjPrbUUF/sPHDjNzciZLjh/giWkHfjPrfUUF/joH3+lUz+wRv2/umlkvKyrwn96uocM3d2fn+D3iN7NeVlTgz7FqF2DZBcsY6BvgwDMHAAd+M+ttRQX+HBu0AUhi1eAqTsUpABb1Lero+c3MOqmowD8xPcHgwCDLFy/v+LnrPP+SgSWkHUnNzHpScYH/ZRe9LEtgrmf2OM1jZr2uqMC/f3p/x2/s1uobvF68ZWa9rqjAn2PVbq011WNm1svKCvzTEx2/sVtzqsfMFopiAv+xE8c4fOxwtsBfj/i9eMvMel0xgb+eypkr1eMRv5ktFMUF/lw3d53jN7OFopjAX6/azZbj96weM1sgygn8mTZoq3nEb2YLxUDOk0vaC0wBJ4ETETEiaRXwVWADsBe4OSKezlkPqEb8fepj7dK1Wc5f5/h9c9fMet18jPh/PyI2RcRIen8nsDMiNgI70/vsxqfGWXfROgb68vxfV6d6POI3s17XjVTPTcDWdLwVeNt8XHTsyBjrL16f7fyDA4Ms7l/Mkn4HfjPrbbkDfwDflfSwpNtS2VBETKTj/cBQuy9Kuk3SqKTRycnJl1yRsaNjrF+eL/BL4sbfvJHNw5uzXcPMrBOy5viB10XEPkmXAPdLerT1w4gISdHuixGxBdgCMDIy0vZ3zldEMH50nBs23vBSTnNO227elvX8ZmadkHXEHxH70utBYDtwLXBA0jqA9HowZx0Anj72NM8efzbriN/MbKHIFvglLZW0rD4Grgf2APcBt6RfuwX4dq461MaOjAFkzfGbmS0UOVM9Q8D2tPf9APCliNgh6SHga5JuBX4J3JyxDkCV3wc84jczI2Pgj4jHgavalD8JvDHXddsZPzoOwPDy4fm8rJlZTypi5e7YkTEG+gayrdo1M1tIygj8R8e4dNml9Pf1d7sqZmZdV0zgd37fzKxSRuDPvGrXzGwhaXzgrxdvDS/zjV0zMygg8B969hAzJ2c84jczSxof+D2H38zshZof+L1q18zsBRof+L14y8zshRof+MeOjrGobxGXLL2k21UxM+sJRQT+4eXD9KnxTTUzOy+Nj4aew29m9kKND/zjR8c9o8fMrEWjA/+pOFUt3vKNXTOz0xod+A8+c5Djp457xG9m1qLRgd9z+M3MztbowF/P4feI38zsjEYH/nq7Buf4zczOaHbgPzLGkoElrLlwTberYmbWMxod+K9YcwXvfs27SQ98NzMzQBHR7Tqc08jISIyOjna7GmZmC4qkhyNiZHZ5o0f8ZmZ2Ngd+M7PCZA/8kvol/bek76T3n5P0C0m70s+m3HUwM7MzBubhGrcDjwDLW8o+FBHb5uHaZmY2S9YRv6Rh4Abg0zmvY2Zm5y93quefgA8Dp2aV/62k3ZLukrS43Rcl3SZpVNLo5ORk5mqamZUjW+CXdCNwMCIenvXRR4BXAa8FVgF3tPt+RGyJiJGIGFm7dm2uapqZFSfniP/3gLdK2gt8BbhO0hcjYiIqM8BngWsz1sHMzGaZlwVckt4AfDAibpS0LiImVC2nvQs4FhF3nuP7k8Avf83LrwEO/ZrfXchKbHeJbYYy211im+HFt/s3IuKslMl8zOqZ7V5JawEBu4D3nesL7Sp+viSNtlu51nQltrvENkOZ7S6xzdC5ds9L4I+IHwA/SMfXzcc1zcysPa/cNTMrTAmBf0u3K9AlJba7xDZDme0usc3QoXYviN05zcysc0oY8ZuZWQsHfjOzwjQ68Et6s6SfSnpM0q9cK7BQSVov6QFJP5H0Y0m3p/JVku6X9PP0urLbde20Nju/Xi7pwdTfX5V0Qbfr2GmSVkjaJulRSY9I+p2m97Wkv0z/tvdI+rKkJU3sa0n/LOmgpD0tZW37VpW7U/t3S7rmxVyrsYFfUj9wD/AW4ErgXZKu7G6tsjgB/FVEXAlsBv4stfNOYGdEbAR2pvdNU+/8Wvs74K6IeCXwNHBrV2qV18eAHRHxKuAqqvY3tq8lXQZ8ABiJiNcA/cAf0cy+/hzw5lllc/XtW4CN6ec24JMv5kKNDfxUW0E8FhGPR8TzVNtG3NTlOnVc2gLjv9LxFFUguIyqrVvTr20F3tadGuYxe+fXtBL8OqDe7ruJbb4YeD3wGYCIeD4iDtPwvqZabzQoaQC4EJiggX0dEf8KPDWreK6+vQn4fNr+5j+BFZLWne+1mhz4LwPGWt6Pp7LGkrQBuBp4EBiKiIn00X5gqEvVymX2zq+rgcMRcSK9b2J/Xw5MAp9NKa5PS1pKg/s6IvYB/wD8H1XAPwI8TPP7ujZX376k+NbkwF8USRcB3wD+IiKOtn4W1Zzdxszb/RU7vzbdAHAN8MmIuBp4hllpnQb29Uqq0e3lwKXAUs5OhxShk33b5MC/D1jf8n44lTWOpEVUQf/eiPhmKj5Q/+mXXg92q34ZnLXzK1Xue0VKB0Az+3scGI+IB9P7bVT/ETS5r98E/CIiJiPiOPBNqv5vel/X5urblxTfmhz4HwI2prv/F1DdELqvy3XquJTb/gzwSET8Y8tH9wG3pONbgG/Pd91yiYiPRMRwRGyg6tfvR8R7gAeAP0y/1qg2A0TEfmBM0hWp6I3AT2hwX1OleDZLujD9W6/b3Oi+bjFX394H/Ema3bMZONKSEjq3iGjsD/AHwM+A/wU+2u36ZGrj66j+/NtNtdvprtTu1VSzAH4OfA9Y1e26Zmr/G4DvpONXAD8EHgO+Dizudv0ytHcTMJr6+1vAyqb3NfDXwKPAHuALwOIm9jXwZar7GMep/rq7da6+pdrd+J4U235ENevpvK/lLRvMzArT5FSPmZm14cBvZlYYB34zs8I48JuZFcaB38ysMA78VjRJJyXtavnp2AZnkja07rRo1ivm5WHrZj3suYjY1O1KmM0nj/jN2pC0V9LfS/qRpB9KemUq3yDp+2kP9J2SXp7KhyRtl/Q/6ed306n6JX0q7Sf/XUmD6fc/kJ6hsFvSV7rUTCuUA7+VbnBWquedLZ8diYjfAj5BtRsowMeBrRHx28C9wN2p/G7gXyLiKqr9c36cyjcC90TEq4HDwDtS+Z3A1ek878vVOLN2vHLXiiZpOiIualO+F7guIh5Pm+Dtj4jVkg4B6yLieCqfiIg1kiaB4YiYaTnHBuD+qB6igaQ7gEUR8TeSdgDTVNsufCsipjM31ew0j/jN5hZzHL8YMy3HJzlzX+0Gqr1WrgEeatlp0iw7B36zub2z5fU/0vG/U+0ICvAe4N/S8U7g/XD6WcAXz3VSSX3A+oh4ALgDuBg4668Os1w8yrDSDUra1fJ+R0TUUzpXStpNNWp/Vyr7c6onYH2I6mlY703ltwNbJN1KNbJ/P9VOi+30A19M/zkIuDuqRyiazQvn+M3aSDn+kYg41O26mHWaUz1mZoXxiN/MrDAe8ZuZFcaB38ysMA78ZmaFceA3MyuMA7+ZWWH+H8ErmRj0THQxAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best epoch: 80\n",
      "Test accuracy: 67.5 %\n"
     ]
    }
   ],
   "source": [
    "# Conitnue training by loading checkpoint\n",
    "# Best epoch = epoch with least val loss\n",
    "trainer = Net_Trainer(device, 'nli_adam_snli', config_nli_model, word_vec, train, valid, test, load_ckpt='v1_ckpt_nli_adam_snli', lr=1e-4, optm='adam', loss='svm', \n",
    "                      batch_size=64, epochs=100, lr_decay=0.99, lr_shrink=5.0, weight_decay=0, mom=0, smooth=False)\n",
    "best_epoch, losses, errors, accs, testacc = trainer.train()\n",
    "plot_train(losses, errors, accs)\n",
    "print('Best epoch:',best_epoch)\n",
    "print('Test accuracy:',testacc,'%')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "WXnom5Zqq51A",
    "outputId": "caf69c24-9030-497f-a183-d750efefefde"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test accuracy of best model: 66.765625\n"
     ]
    }
   ],
   "source": [
    "best_acc = test_model(device, config_nli_model, 'v1_ckpt_nli_adam_snli_best', test, 64, word_vec)\n",
    "print('Test accuracy of best model:',best_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 573
    },
    "id": "GJaWdZ9x1I1K",
    "outputId": "16038468-e0a9-49a0-8cab-291e91a58d80"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "checkpoint = torch.load('checkpoints/v1_ckpt_nli_adam_snli_best', map_location=torch.device('cpu'))\n",
    "losses = checkpoint['loss_hist'][:40]\n",
    "errors = checkpoint['err_hist'][:40]\n",
    "accs = checkpoint['acc_hist'][:40]\n",
    "plot_train(losses, errors, accs)"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "blstm_adam_snli.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}