NativeVex commited on
Commit
b494114
1 Parent(s): a386cc2

best effort

Browse files
midterm/Pipfile CHANGED
@@ -11,6 +11,7 @@ scikit-learn = "*"
11
  numpy = "*"
12
  ipython = "*"
13
  seaborn = "*"
 
14
 
15
  [dev-packages]
16
 
 
11
  numpy = "*"
12
  ipython = "*"
13
  seaborn = "*"
14
+ imblearn = "*"
15
 
16
  [dev-packages]
17
 
midterm/Pipfile.lock CHANGED
@@ -1,7 +1,7 @@
1
  {
2
  "_meta": {
3
  "hash": {
4
- "sha256": "41f1b07bccc213aac71af44fbe1a7097b15931ef14d03a0b74d5df557b5e4422"
5
  },
6
  "pipfile-spec": 6,
7
  "requires": {
@@ -479,6 +479,21 @@
479
  "markers": "python_version >= '3.5'",
480
  "version": "==3.4"
481
  },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
482
  "ipykernel": {
483
  "hashes": [
484
  "sha256:24ebd9715e317c185e37156ab3a87382410185230dde7aeffce389d6c7d4428a",
 
1
  {
2
  "_meta": {
3
  "hash": {
4
+ "sha256": "404469f7b529f055b3656628796cd8f830288094ebc5eaaa200cb031d4f27f7b"
5
  },
6
  "pipfile-spec": 6,
7
  "requires": {
 
479
  "markers": "python_version >= '3.5'",
480
  "version": "==3.4"
481
  },
482
+ "imbalanced-learn": {
483
+ "hashes": [
484
+ "sha256:7b630516696c09b2b5a5398df8db19d73db0c13fa065d2f1407a6d2037794bc6",
485
+ "sha256:bc7609619ec3c38c442292928239ad3d10b5deb0af8a29c83822b7b57b319f8b"
486
+ ],
487
+ "version": "==0.10.1"
488
+ },
489
+ "imblearn": {
490
+ "hashes": [
491
+ "sha256:d42c2d709d22c00d2b9a91e638d57240a8b79b4014122d92181fcd2549a2f79a",
492
+ "sha256:d8fbb662919c1b16f438ad91a8256220e53bcf6815c9ad5502c518b798de34f2"
493
+ ],
494
+ "index": "pypi",
495
+ "version": "==0.0"
496
+ },
497
  "ipykernel": {
498
  "hashes": [
499
  "sha256:24ebd9715e317c185e37156ab3a87382410185230dde7aeffce389d6c7d4428a",
midterm/midterm/take_at_home_(1).ipynb CHANGED
@@ -1062,23 +1062,26 @@
1062
  },
1063
  {
1064
  "cell_type": "code",
1065
- "execution_count": 13,
1066
  "metadata": {
1067
  "tags": []
1068
  },
1069
  "outputs": [],
1070
  "source": [
1071
- "#from sklearn.pipeline import Pipeline #make_pipeline\n",
1072
- "#from sklearn import preprocessing\n",
 
 
1073
  "#from sklearn.linear_model import LogisticRegression, SGDClassifier\n",
1074
  "\n",
1075
  "#i = 20000\n",
1076
  "\n",
1077
- "#pipe = Pipeline([\n",
1078
- "# ('preprocessor', preprocessing.StandardScaler()),\n",
 
1079
  "# ('classifier', LogisticRegression(max_iter=i, class_weight='balanced')),\n",
1080
- "# ('classifier', MyLogisticRegression())\n",
1081
- "#])\n",
1082
  "\n",
1083
  "\n",
1084
  "\n",
@@ -1088,7 +1091,7 @@
1088
  },
1089
  {
1090
  "cell_type": "code",
1091
- "execution_count": 14,
1092
  "metadata": {
1093
  "scrolled": true,
1094
  "tags": []
@@ -1097,10 +1100,10 @@
1097
  {
1098
  "data": {
1099
  "text/plain": [
1100
- "{'batch_size': 61, 'lr': 1}"
1101
  ]
1102
  },
1103
- "execution_count": 14,
1104
  "metadata": {},
1105
  "output_type": "execute_result"
1106
  }
@@ -1116,12 +1119,12 @@
1116
  " #'classifier__solver': ['sag'],\n",
1117
  " #'classifier__class_weight': [None, 'balanced', *[{0: 1, 1:10**x} for x in range(-5, 5)]],\n",
1118
  " #'classifier__eta0': [10**x for x in range(-5, 5)],\n",
1119
- " 'batch_size': np.linspace(1, X_train.shape[0], 70, dtype=int)[:10],\n",
1120
- " 'lr': [10**x for x in range(-5, 5)],\n",
1121
  "}\n",
1122
  "\n",
1123
- "#grid_search = GridSearchCV(pipe, param_grid, cv=5, scoring='f1', n_jobs=-1)\n",
1124
- "grid_search = GridSearchCV(MyLogisticRegression(), param_grid, cv=5, scoring='f1', n_jobs=-1)\n",
1125
  "\n",
1126
  "grid_search.fit(X_train, y_train)\n",
1127
  "grid_search.best_params_"
@@ -1129,7 +1132,7 @@
1129
  },
1130
  {
1131
  "cell_type": "code",
1132
- "execution_count": 15,
1133
  "metadata": {
1134
  "tags": []
1135
  },
@@ -1137,10 +1140,10 @@
1137
  {
1138
  "data": {
1139
  "text/plain": [
1140
- "0.8993288590604027"
1141
  ]
1142
  },
1143
- "execution_count": 15,
1144
  "metadata": {},
1145
  "output_type": "execute_result"
1146
  }
@@ -1156,7 +1159,7 @@
1156
  },
1157
  {
1158
  "cell_type": "code",
1159
- "execution_count": 16,
1160
  "metadata": {
1161
  "tags": []
1162
  },
@@ -1164,11 +1167,11 @@
1164
  {
1165
  "data": {
1166
  "text/plain": [
1167
- "array([[ 31, 15],\n",
1168
- " [120, 134]])"
1169
  ]
1170
  },
1171
- "execution_count": 16,
1172
  "metadata": {},
1173
  "output_type": "execute_result"
1174
  }
@@ -1179,32 +1182,21 @@
1179
  },
1180
  {
1181
  "cell_type": "code",
1182
- "execution_count": 17,
1183
  "metadata": {
1184
  "tags": []
1185
  },
1186
- "outputs": [
1187
- {
1188
- "data": {
1189
- "text/plain": [
1190
- "array([0.87437186, 0.85714286, 0.84848485, 0.81818182, 0.88 ])"
1191
- ]
1192
- },
1193
- "execution_count": 17,
1194
- "metadata": {},
1195
- "output_type": "execute_result"
1196
- }
1197
- ],
1198
  "source": [
1199
- "from sklearn.model_selection import cross_val_score\n",
1200
- "\n",
1201
- "scores = cross_val_score(grid_search, X, y, cv=5)\n",
1202
- "scores"
1203
  ]
1204
  },
1205
  {
1206
  "cell_type": "code",
1207
- "execution_count": 18,
1208
  "metadata": {},
1209
  "outputs": [
1210
  {
@@ -1236,27 +1228,27 @@
1236
  " <tr>\n",
1237
  " <th>0</th>\n",
1238
  " <td>0</td>\n",
1239
- " <td>0.809935</td>\n",
1240
  " </tr>\n",
1241
  " <tr>\n",
1242
  " <th>1</th>\n",
1243
  " <td>1</td>\n",
1244
- " <td>0.856796</td>\n",
1245
  " </tr>\n",
1246
  " <tr>\n",
1247
  " <th>2</th>\n",
1248
  " <td>2</td>\n",
1249
- " <td>0.850208</td>\n",
1250
  " </tr>\n",
1251
  " <tr>\n",
1252
  " <th>3</th>\n",
1253
  " <td>3</td>\n",
1254
- " <td>0.846912</td>\n",
1255
  " </tr>\n",
1256
  " <tr>\n",
1257
  " <th>4</th>\n",
1258
  " <td>4</td>\n",
1259
- " <td>0.850349</td>\n",
1260
  " </tr>\n",
1261
  " </tbody>\n",
1262
  "</table>\n",
@@ -1264,33 +1256,33 @@
1264
  ],
1265
  "text/plain": [
1266
  " epoch f1 score\n",
1267
- "0 0 0.809935\n",
1268
- "1 1 0.856796\n",
1269
- "2 2 0.850208\n",
1270
- "3 3 0.846912\n",
1271
- "4 4 0.850349"
1272
  ]
1273
  },
1274
- "execution_count": 18,
1275
  "metadata": {},
1276
  "output_type": "execute_result"
1277
  }
1278
  ],
1279
  "source": [
1280
- "best_f1 = pd.DataFrame(grid_search.best_estimator_.f1_score_history)\n",
1281
  "best_f1.head()"
1282
  ]
1283
  },
1284
  {
1285
  "cell_type": "code",
1286
- "execution_count": 19,
1287
  "metadata": {
1288
  "tags": []
1289
  },
1290
  "outputs": [
1291
  {
1292
  "data": {
1293
- "image/png": "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",
1294
  "text/plain": [
1295
  "<Figure size 640x480 with 1 Axes>"
1296
  ]
@@ -1302,6 +1294,13 @@
1302
  "source": [
1303
  "plot_f1(best_f1)"
1304
  ]
 
 
 
 
 
 
 
1305
  }
1306
  ],
1307
  "metadata": {
 
1062
  },
1063
  {
1064
  "cell_type": "code",
1065
+ "execution_count": 41,
1066
  "metadata": {
1067
  "tags": []
1068
  },
1069
  "outputs": [],
1070
  "source": [
1071
+ "from sklearn.pipeline import Pipeline #make_pipeline\n",
1072
+ "from sklearn import preprocessing\n",
1073
+ "from imblearn.over_sampling import RandomOverSampler\n",
1074
+ "#from imblearn.pipeline import Pipeline\n",
1075
  "#from sklearn.linear_model import LogisticRegression, SGDClassifier\n",
1076
  "\n",
1077
  "#i = 20000\n",
1078
  "\n",
1079
+ "pipe = Pipeline([\n",
1080
+ "# ('oversampler', RandomOverSampler()),\n",
1081
+ " ('preprocessor', preprocessing.StandardScaler()),\n",
1082
  "# ('classifier', LogisticRegression(max_iter=i, class_weight='balanced')),\n",
1083
+ " ('classifier', MyLogisticRegression())\n",
1084
+ "])\n",
1085
  "\n",
1086
  "\n",
1087
  "\n",
 
1091
  },
1092
  {
1093
  "cell_type": "code",
1094
+ "execution_count": 42,
1095
  "metadata": {
1096
  "scrolled": true,
1097
  "tags": []
 
1100
  {
1101
  "data": {
1102
  "text/plain": [
1103
+ "{'classifier__batch_size': 82, 'classifier__lr': 0.1}"
1104
  ]
1105
  },
1106
+ "execution_count": 42,
1107
  "metadata": {},
1108
  "output_type": "execute_result"
1109
  }
 
1119
  " #'classifier__solver': ['sag'],\n",
1120
  " #'classifier__class_weight': [None, 'balanced', *[{0: 1, 1:10**x} for x in range(-5, 5)]],\n",
1121
  " #'classifier__eta0': [10**x for x in range(-5, 5)],\n",
1122
+ " 'classifier__batch_size': np.linspace(1, X_train.shape[0], 70, dtype=int)[:10],\n",
1123
+ " 'classifier__lr': [10**x for x in range(-5, 5)],\n",
1124
  "}\n",
1125
  "\n",
1126
+ "grid_search = GridSearchCV(pipe, param_grid, cv=5, scoring='f1', n_jobs=-1)\n",
1127
+ "#grid_search = GridSearchCV(MyLogisticRegression(), param_grid, cv=5, scoring='f1', n_jobs=-1)\n",
1128
  "\n",
1129
  "grid_search.fit(X_train, y_train)\n",
1130
  "grid_search.best_params_"
 
1132
  },
1133
  {
1134
  "cell_type": "code",
1135
+ "execution_count": 43,
1136
  "metadata": {
1137
  "tags": []
1138
  },
 
1140
  {
1141
  "data": {
1142
  "text/plain": [
1143
+ "0.9166666666666666"
1144
  ]
1145
  },
1146
+ "execution_count": 43,
1147
  "metadata": {},
1148
  "output_type": "execute_result"
1149
  }
 
1159
  },
1160
  {
1161
  "cell_type": "code",
1162
+ "execution_count": 44,
1163
  "metadata": {
1164
  "tags": []
1165
  },
 
1167
  {
1168
  "data": {
1169
  "text/plain": [
1170
+ "array([[ 33, 12],\n",
1171
+ " [123, 132]])"
1172
  ]
1173
  },
1174
+ "execution_count": 44,
1175
  "metadata": {},
1176
  "output_type": "execute_result"
1177
  }
 
1182
  },
1183
  {
1184
  "cell_type": "code",
1185
+ "execution_count": 45,
1186
  "metadata": {
1187
  "tags": []
1188
  },
1189
+ "outputs": [],
 
 
 
 
 
 
 
 
 
 
 
1190
  "source": [
1191
+ "#from sklearn.model_selection import cross_val_score\n",
1192
+ "#\n",
1193
+ "#scores = cross_val_score(grid_search, X, y, cv=5)\n",
1194
+ "#scores"
1195
  ]
1196
  },
1197
  {
1198
  "cell_type": "code",
1199
+ "execution_count": 46,
1200
  "metadata": {},
1201
  "outputs": [
1202
  {
 
1228
  " <tr>\n",
1229
  " <th>0</th>\n",
1230
  " <td>0</td>\n",
1231
+ " <td>0.813799</td>\n",
1232
  " </tr>\n",
1233
  " <tr>\n",
1234
  " <th>1</th>\n",
1235
  " <td>1</td>\n",
1236
+ " <td>0.823768</td>\n",
1237
  " </tr>\n",
1238
  " <tr>\n",
1239
  " <th>2</th>\n",
1240
  " <td>2</td>\n",
1241
+ " <td>0.833749</td>\n",
1242
  " </tr>\n",
1243
  " <tr>\n",
1244
  " <th>3</th>\n",
1245
  " <td>3</td>\n",
1246
+ " <td>0.837076</td>\n",
1247
  " </tr>\n",
1248
  " <tr>\n",
1249
  " <th>4</th>\n",
1250
  " <td>4</td>\n",
1251
+ " <td>0.837068</td>\n",
1252
  " </tr>\n",
1253
  " </tbody>\n",
1254
  "</table>\n",
 
1256
  ],
1257
  "text/plain": [
1258
  " epoch f1 score\n",
1259
+ "0 0 0.813799\n",
1260
+ "1 1 0.823768\n",
1261
+ "2 2 0.833749\n",
1262
+ "3 3 0.837076\n",
1263
+ "4 4 0.837068"
1264
  ]
1265
  },
1266
+ "execution_count": 46,
1267
  "metadata": {},
1268
  "output_type": "execute_result"
1269
  }
1270
  ],
1271
  "source": [
1272
+ "best_f1 = pd.DataFrame(grid_search.best_estimator_.named_steps['classifier'].f1_score_history)\n",
1273
  "best_f1.head()"
1274
  ]
1275
  },
1276
  {
1277
  "cell_type": "code",
1278
+ "execution_count": 47,
1279
  "metadata": {
1280
  "tags": []
1281
  },
1282
  "outputs": [
1283
  {
1284
  "data": {
1285
+ "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkAAAAGwCAYAAABB4NqyAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAABO1ElEQVR4nO3de3xT9cE/8E+aNmnSNik0peVS5FZuYgFBGRTLlE6m/rq58ZoIqIhTHx04hMcLIsiUYXE+8jAVxfkSnZsIKvo4xcsQBQS5FhAvpbRQpKO0JS1NmqZt2uT8/kjPadKmadKmOSnn8369+oIm35zzzWnSfPq9qgRBEEBERESkIFFyV4CIiIgo3BiAiIiISHEYgIiIiEhxGICIiIhIcRiAiIiISHEYgIiIiEhxGICIiIhIcaLlrkAkcrlcKC0tRUJCAlQqldzVISIiogAIgoCamhr069cPUVH+23gYgHwoLS1FWlqa3NUgIiKiTigpKcGAAQP8lmEA8iEhIQGA+wIaDAaZa0NERESBsFqtSEtLkz7H/WEA8kHs9jIYDAxAREREPUwgw1c4CJqIiIgUhwGIiIiIFIcBiIiIiBSHAYiIiIgUhwGIiIiIFIcBiIiIiBSHAYiIiIgUhwGIiIiIFIcBiIiIiBSHAYiIiIgUh1thkBeL3QGzzQFrfSMMuhiY4jQw6jVdfoyvMgCk24y6GMRpo2Grb2q3TKD1ISIi6ggDEElKq+vw6Nbj+LrQLN2WlW7CmpkZ6Jeo6/RjWpfRa9TYeOdVWP9lEb4uMkOvUeP52ePx+t5i7C2q9Fkm0PoQEREFQiUIgiB3JSKN1WqF0WiExWJRzGaoFrsDC98+6hVkRFnpJrwwe7zPVp2OHgOgTZk7Jl+GH0otyPup2uf37d3WUX2IiEjZgvn8ZgsQAXB3M/kKMgCwu9AMs83RJnAE8hgAbcq8ue8nv9+3d1tH9SEiIgoUAxABAKz1jX7vr/FxfyCP8dW8GB2lQpOr5Z4YtQqNTu+Svm7rqD5ERESB4iwwAgAYYmP83p/g4/5AHuOrzNO/GeP1/YbbJrQp4+u2jupDREQUKAYgAgCY4jXISjf5vC8r3QRTfNvupkAe46vMT1V2ZA5Lkr4/WlLt9X17t3VUHyIiokAxABEAwKjXYM3MDPSO8w4WWekmPDMzw+d4G/Exg016r9uvGZYkPUYsMyIlXrr/9b1n8MB16bimORht3FOM+ZmDMdUj8GzcU+xVJpD6EBERBYpjgEjSL1GHwUl6VNW6By9/vugapBpj/YaNfok6jOlnRLHZLt22Mudy9PWYpt4vUYcpQ00oKLfhl5en4OEZI2GK1+DF2eNhtjlQ07zGz3O3jIOtvgk19Y1IiI2RynyRX4H/fvdb9DXGcvYXERGFBAMQebE3uqT/9wpw0cHT5lqv70st9RiWkuB1W3Glu8zU9GQM7dPSGtT6+Ck+Zi1eN7IPAOC8pR5qNRstiYio6/hpQl48Z1ddsDV0WN7pElBYYQMAXN7PnV4KymralBNvG5ma0Oa+jvSK0yDFoG332ERERMFiACIvNfVN0v8rm9fx8edMZS0cTS7ExkThF6NTAAD5ZVavMhZ7I85b6gEAwzsRgABgRGr74YqIiChYDEAkEQQBtoaWAGQOoAXoZHMgGZ6SgFF9fYeUE82BqH+irsOp8+0Z1RycTrQKV0RERJ3BAEQSu8MJp8cChYG0AJ1oDjsjUhKk7q3CChuanC1jiQrKO9/9JRohBSC2ABERUdcxAJHEs/sLCKwFSGztGZGagLReeug1ajiaXDhT2TIw+oRHmc4SH1tQVgNuX0dERF3FAESS1ttLmANoATpZ3hJuoqJUGJ7StqVGGgDdt/Mbyw7rEw91lAqWukaUWes7fRwiIiKAAYg8WINsAapvdEotPWILzUiPlhrAPa6oKzPARNpoNYaY4gCwG4yIiLqOAYgkrVuAKmv9B6CiChtcAtBLH4PkePc0dTEI5Z93h5T/XKyDraEJMWoVBjcHmM4a0SpcERERdRYDEEnEMUB6jRoAYK7x3wXmObZHpVIBAEaK09XL3bO1xLAyNDkeMV1cxFBsQTpxnjPBiIioaxiASCIGILGlprK2we+A44LmKeli6HH/3x1SSqrcLT/itPVRXRj/03Js9zHYBUZERF3FAEQSsQtsUHMAanQKsNY1tVu+oNy9AvRwj20vesVp0CehZdXmUMwAE4nHOHXBhkaPafZERETBYgAiidgCZIrTIEHr3ibO7GcckNgC1DrcjPRYELEghAFoQC8d4rXRaHQKKG61/xgREVEwGIBIIrYAJcTGwNTcimOu8R2Aqu0OlFvd9w1Pife6T+wG++5ctbRR6qjUrneBqVQqj0HWHAdERESdxwBEErEFKCE2Gklx7l3aK2t9D4QWW3b6J+qQ0Gp7ixHNXWKffV8Gp0uAURcjbWbaVZwJRkREocAARBKrFIBiYGqe1t7eWkD+trcQQ8pFe6P0vThLrKtarzNERETUGQxAJGnpAotGUry7Bai91aD9je0RV20WjQrB+B8RZ4IREVEoRMtdAYocnl1g/lqALHYHvi2pBgAk6mNgsTtg1Guk+2Nj1List14a/9M7XtOmTGeJ3Wvnquuwp/AC+ibqYIrTtDm2xe6A2eaAtb4RRl0M4rTRsNU3wVrfCIMuBqa4loDnr4y/4/o6TjDHlqtMVx8Xip8jhYav12M4fz6BnP9SeM905/uKdZTvdwoDEElqGjwGQTe3AFW2CkCl1XV45L1v8X2pexDy05+cwJ5CM9bMzEC/RJ1UxuqxqvT/bi9E3pmLXmU6q9bRBE10FBxNLtz22kEAQFa6qc35H916HF8XmqHXqPH87PF4fW8x9hZVAnAv9Ljxzquw/ssifF3ku0xHx/V1nECPLWeZrjzO1zUh+bR+PQLh/fkEcv5L4T3Tne8rpdfR12smnFQCt9Zuw2q1wmg0wmKxwGDo+uylnmLcU/9Gtb0R2xdnoajChvvfOoIJl/XC1vunAHD/Jbfw7aNev/BEWekmvDB7PAB0WKazab8z579uZB+crapFUUXLtPnWt/kqE8jz6uyx5SzTlce1viZsCZJPIO+F7vz5dPZ3QaS9H+R+Xym9jqJQvmaD+fxmAPJBiQFIEAQMe/xTOF0C9j82HWer7LjllX0YlKTHzoevBQCcqrBh+tpd7R5jx5JpANBhmaF94tu9359QnL8zuuu4PVlXfo7UdYG8F7rz5yPXe5EuXaF6zQbz+c1B0AQAqGt0wulyZ2H3GKC2g6A9u7V8qalvDKhMZ3Xm/MkJbaffj+lv8Pt9IMcFgEFJ+k4dW84yXXmcp678HKnruvN9Fqrz+yrja0PkSH/PdOf7Sul19CTH7xQGIALQMgBaHaWCXqNGUvMgaFtDE+obnQAAQ6v1flpLiI0JqExndeb8N4xJbVNucfZwv98HclwAmJ85uFPHlrNMVx7nqSs/R+q67nyfher8vsr8V9aQNrdF+numO99XSq+jJzl+pzAAEYCW9B2vjYZKpYIhNhqa5t3bxZlgpngNstJNPh+flW6CKV4TUJnO6sz57Q4nMocleZU9WlLtdVvr7wM5LgCUWes7dWw5y3TlcaKu/hyp67rzfRaq8/sqU3LRHlHvB7nfV0qvo0iu3ykMQATAcxFE98RAlUolrQVU2dwNZtRr8Oebx6D1koZZ6SY8MzMDRr17OuOamRltfvF5lumsQI5t1Gvw2I2jpPu2HS/F/MzBmOrxxtu4pxgPXJeOa5qPs3FPcZsyADB5SO82z+vyvi1rGv39mzNexwn02HKW6crjWl9rko/4emy9xtZVg3qF5ecjnn9Iqy6tqUOT2rxnPOv4+t6e957pzveV0usIyPs7hYOgfVDiIOhdJy9g3saDGNXXgE8XXQMAyHlhD747Z8Fr8yZi+qgUAO4FEGes2404jRr//P3VMOjcf+m1t/ZHTX2jNK0+VC9wi92BxVu+xZcFFbh76mA8cN0wr2PvO1WJ2a/uR19jLF6eeyUMHmtPeNYHgFRHzzK3vrofF2oasHHeRFzX/LxFz31egBe+KsK0dBOeyLm8zXE6OnYklAn2cU99/CO2HjmH347vj5U5oxl+IsiaT/KxYfdp6ft1s8bi5vEDwnb+RZuP4sNjpdL3m++ZhJ8N9Q7NT2/7EX/7uhjZo/rgsRtG9cj3THe8r1jH0H82AMF9fnMdIALgvQq0qHULEACcaN4BflRfA8Zf1rvd44l/AXYHo16DiYN74cuCClTUNLQ5j1jHMf2NGDewl3R7io/3QuvHphiAsQOM+CK/AiUX69qUL650T+GcPMzkNWPB13P1dexIKhPo4342JAlbj5zDeUs9w0+EOVNpBwBEqQCXAJRUtX3Ndqfi5sVOxfOfq65vt46Th/b890wo31eso/zYBUYAWgZBGzwCkLga9AWPxRBP+NkCI5z87Qkm3uZrn7JAiM/N13Yb/rYAuVSJ248UlNeADcaR5WTznnyZw9ytLuIefeHgcgltzn/Sx/lP+tk3kEhODEAEwLMFqGUkvq8WoK6Gi1AZ0fyhfOqCDY4ml9d9XQ1p4rELmluSRA1NTml7D7mffzilp8QjSgVU1Tq8wjDJq77RiTPNLZI5Y/sBCO8mwWer7KhvdEETHYXrR7u7ilv/0WB3NOGnKncLkJL+aKCegQGIAHjvAyZK9rEf2Inz7lAwsq+8bZj9jLFIiI1Gk0vAqQs26XbPv0o7G1LEx50st8HlamnxKKqwwekSYIiNRqohtgu171liY9QYlOQe7BrOD1jyr7DcBpcA9NLHYGpzC8xpcy0ampxhOb8YdtL7xGNUX/GPBu/XR2G5DYIAJMVppBZlokjBAEQAfAcgqQWo1h2ALHWNKLW4+/iHp8j715xKpfLZDVZy0Q67wwlNdJT0oR2swaY4aNRRsDU04Vx1y5gKqfWrrwEqVeu5cJe2kc2z306cZwCKFGJ314jUBPRt/oPA6RJw+kLbrQa6w0mP8w9vfi+WWethsbcsaOdZR6JIwwBEAFpWdvXsApN2hK9xd4GJv/D6GWNh1Mm/EJ6vsTqef5VGqzv38o5RR0mDNT2PHSndf3IY0Txq0de4KJKH2EU7MtXQ7h8E3Xv+lveDITYG/Zs3s/Qch6TEMXPUczAAEYB2WoDi3AFIbAGKlO4vkThW54THWJ1Q/cJt+TBpOXa+gn+Zi8+5oNzaQUkKl4Jyd9ev2Bor/huugdDieVrOH9/m/FIrkcwtxkS+MAARAN+DoE0J7i6wqloHnC4hYmaAiXz9xRuqVhpfrUstf3FHxvMPp1F9W8ZFNTldHZSmcBBfj+JrNZwtQA1NTmkKvDhL0NfkgUj7nUHkiQGIAPhuAeqt10DVvL7HRbsj4rqAxF+q5y0t4w7yPboFumJkqwB0sdaBcqu7JUzu8U9ySOulh16jhqPJJa3rQvKptnu+Ht0tLy0BpPsDkDghwKiLQYrB3VLcOoBV1TpwoUa57xmKfAxABMD3OkDR6ij0al7A6kJNg0cAiowuMM9xByfKrO5pwSGapi4+x+LmWTViEBrQS6fIjUCjolRITwnvGBNqn/gz6J/Y8noUu5nOVdd1+87anl1b4oSA4R6vD0EQpDqm9dYhTss1dynyMAARAN9dYIB7+ioAfPcfC2oamhCjVmFIcudmV3WHlrEpNSiqaJkWnJzQtSm3KQYtjLoYOF0Ciipsiu7+Eo1MaTsuiuRR4GOpB6M+RlqewdeChKHkq2traJ84qKNUsNY3ocxa39JFFynL/hK1wgBEEATBZxcY0DITbE+RGQAwNDkeMZ2cXdUdPLuqTni0UHV1mnrrWTUtHzjK/WUuToXPZwuQ7MTX+vBWgXy4j7Fr3aHAx/m10WoMbt4c9YTHe2ZEanzbAxBFgMj5JCPZ1De60NS84F+bFqDmtYC+OeUOQJHWAiINVj5vlWaphWrApWe4yj/PwZwjwjjIlvw72c54PGkRz27+GbV3/hEe52+ZkancPxoosjEAkdT9FaUC4jRqr/uktYCat8OItF9mYovMyXKbRwtQaEKK+Fzzz1u5nxFarvXZKjtqG5pkro1yCYLQ7gKD4jig7mwB8rcg6kiP859snqav5PcMRTYGIJIWQYzXRrfpOjLFe+/iK3aDRIohyXGIUatga2jCwTNVAELXSiMe50BxlXt1aXUUBpkiZ/xTuPWO06BP89iqcG66Sd7OW+pRU9+E6CgVhpi8u5ekFphu3Li2sPln39fHgqhil9jukxdgax4zOFjB7xmKbAxABKs0/qft7KbW+/dE2l9zMeooDE12fwg4mlxQqUI35Vb8MBE3Wx3WJ7LGP8mB3WDyE6/9kOQ4aKK9X4/D+rg3rr1ob5SmoIeav7V9xN8PlbXuFuNIGzNI5ImvTGp3ADQAJHkEoEjdBNRzVlpfY2zIFuqL10ZL0+wBoH8vHSx2R0iO3VMNbt5fbU/hBZy6YFP89ZDDCT9ja2Jj1FIrZXd1g/lbbT2tlx66mJZudCWPmaPIx8UZSBoDZPDRAhTr8RfmIFMcrHWNMOo1bcrJpbS6Dj+WWj2+r8cDbx/FmpkZ6OcRXjp77LrGlp21t/9YHrJj90Sl1XXYX1wJANj2XRm2fVeGrHRTm+thsTtgtjlgrW+EQRcDU/NSCuJtRl0M4rTRsNU3tVsmmMcF8nr0rFMg5w/VuTr7/P2VyfvJ3dWbHK+Bxe5oU6chpjicvlCLL09UoH8vXcjP/21JNQCgl77t+aOi3Mtk/ND8njTFa33WkSgSMABRuy1ApdV1eP7LQun74/+xRFQAsNgdeHTr8TYrE+8uNGPp1uN4Yfb4Tv/iFY9dVevdwhGKY/dE4vUQB7aKWl+P0uo6PLr1OL4udM8a1GvU2HjnVVj/ZRG+LjJDr1Hj+dnj8freYuwtqvRZJtDHAfAZwFrzrFMg5w/VuTr7/AO9Rhv3nkFRhc2rTqXVddLP6I1vzuCdwyUhPf+LOwpx/JwFALDm0xP4psjc5vxmW0vX22t7ilFYXhMxvzOIPMneBbZ+/XoMGjQIsbGxmDRpEg4ePOi3/Lp16zBixAjodDqkpaVh8eLFqK+vl+7/05/+BJVK5fU1cuTI7n4aPVrLIogtAUj8wDt05qJXWfEDLxK6Psw2h/RB09ruQrM0cy3Sjt0TBXI9xNeMZ7ns0SlY9fEP0od29ugU/O/2Aq9g0bpMoI8Tz+3v9di6ToGcP1Tn6uzzD/Qata6TeP6zVfYuHdtfmT2n2r8m4vnFLToCvW5EcpG1BWjLli1YsmQJNmzYgEmTJmHdunWYMWMGCgoK0KdPnzblN23ahKVLl2Ljxo2YMmUKTp48iTvvvBMqlQpr166Vyl1++eX44osvpO+jo9nQ5U+Nj0HQgXzgyd0CYu1guf+ubAfQncfuiQK9Hq1fM/86Vur3+0Bv81VG5O/12Pp1HM5zBXrszl6j1nUCQnP9Q3n+1mXk/p1B5EnWFqC1a9finnvuwfz58zF69Ghs2LABer0eGzdu9Fn+m2++QWZmJubMmYNBgwbh+uuvx+zZs9u0GkVHRyM1NVX6MplM4Xg6PZavLrCeEAB8jVny1JU9u7rz2D1RINfD12umT6stSQb0atsNEshtvsp4au/12LpO+lbrXHXnuQI9dmfK+KpTJJ6/dRmiSCJbAHI4HMjLy0N2dnZLZaKikJ2djX379vl8zJQpU5CXlycFntOnT+OTTz7BjTfe6FWusLAQ/fr1w5AhQzB37lycPXvWb10aGhpgtVq9vpRE/MVl8FjToycEAFO8BlnpvsNtVrqpzRpGkXLsniiQ6+HrNbP8plFe3z/5q8vblAnkNl9lPLX3emxdp+mj2rYsd9e5Aj12Z8r4qlMknr91GaJIIlsAMpvNcDqdSElJ8bo9JSUFZWVlPh8zZ84cPPXUU5g6dSpiYmIwdOhQ/PznP8eyZcukMpMmTcIbb7yBzz77DC+//DKKi4txzTXXoKam/Smhubm5MBqN0ldaWlponmQP4asFqCcEAKNegzUzM9rUMyvdhGdmZnSpub07j90TBXI9fL1mTlbYkDksSfr+aEm11/eB3uarjGcd2ns9tq5To1MI27kCPXZnyviqUySev3UZokgi+yDoYOzcuRNPP/00XnrpJRw5cgTvv/8+tm3bhlWrVkllbrjhBvzud79DRkYGZsyYgU8++QTV1dV455132j3uY489BovFIn2VlJSE4+lEDF87wfeUANAvUYcXZo/HjiXT8H9/mIIdS6bhhdnj0TcEM06689g9kXg93px/NQBAo47CulnjpOshvmaGeazLtHFPMR64Lh3XNL+ONu4pxvzMwZjq8WHaukygjwM6fj0a9Rqs/s0ViGpe4HxnQUWH52/vXFcP6tXhudbMzPDq9uvM8w/0GrV+/r7es3Kfv3UZokiiErprvfQOOBwO6PV6vPfee7j55pul2+fNm4fq6mp8+OGHbR5zzTXX4Gc/+xmeffZZ6bZ//vOfuPfee2Gz2RAV5TvPXXXVVcjOzkZubm5AdbNarTAajbBYLDAYImvvq+5w0/Nf44dSK16ffxWuHeHdRSCuaVJT34iE2BiY4gNbC4UuXS6XgNErP0N9owtf/vc0DEn23o7h0fe+xZbD/8FvxvfDwmvTpb/8xdeRwWONGc/XlWcZX7eJjztZVoPbNx5EjFqF/Y9dh6R4/4tzFptrce3/7IQ2Ogqb7p6EXnGaDs/vWce73jiEn6rseOHWccgZ17/D6/PLdbtxoqwGj90wEtmjUjr1/IO5Rq3fj77es3Kfn78zKFyC+fyWbXqURqPBhAkTsGPHDikAuVwu7NixAwsXLvT5GLvd3ibkqNXuQY3t5TibzYZTp07h9ttvD13lLzFiF5jBx0rQ4l92RKKoKBWGpyTg+H8sKCiraROATptrAQDThvfB0D4t97V+HaX4+N3k67XW+rbkeC30GjXsDicu2hs7DEAnzrvH9I1ITcCEQb2DOn+KARibloifquwoqa7zex4AaHK6pOf/yzGpuCyppTWsM88/0GvU+v5ArmO4z08UaWTtAluyZAleffVV/P3vf0d+fj7uv/9+1NbWYv78+QCAO+64A4899phUPicnBy+//DI2b96M4uJibN++HStWrEBOTo4UhB566CHs2rULZ86cwTfffIPf/OY3UKvVmD17tizPsSfw1QVG5E97u44LguB3r6hQEAOYr/P7ItWnk3vEBbP/2ZlKOxxNLuhi1Ejrpe/U+YgoPGRdIGfWrFm4cOECnnjiCZSVlWHcuHH47LPPpIHRZ8+e9WrxWb58OVQqFZYvX45z584hOTkZOTk5WL16tVTmP//5D2bPno3KykokJydj6tSp2L9/P5KTk8P+/HoCQRD87gVG5Et7oaDUY6fyoa1ahkJpZGoCjpVUo6CsBv8vw39Zf3tXBUIMToEEoJPNO6UPT4lHlDjwiIgikuyfeAsXLmy3y2vnzp1e30dHR2PlypVYuXJlu8fbvHlzKKt3yatvdKHJ5e4+ZAsQBWpk80ac4ge+qKDM3d00NDm+zU7loT2/O5Tknw+kBchdp1F9OzeeTwxOpy7Y0Oh0+d3dvLtbv4godHrULDAKPbH7K0oFxPlYJI7Il+Gp7tadM5W1qPfYMDZcAUDcCb2g3P+aXXZHE35q3hqis3Xqn6hDnEaNRqeAM83je9pzskxsAWIAIop0DEAKZ23u/orXRkOlYpM9BSY5XovecRq4BKDQY4PUE+fDE4DEFqCSqjrYGpraLVdYboMgoHmdGm275fyJilJheGpgY44KmlvExBYyIopcDEAKxwHQ1BkqlaplbIxHN5g4TmZU3+4NQL3iNEgxaL3O6YvY/dXVQCIGrtZdfp7qG504U+luIWIXGFHkYwBSOA6Aps5qGQjtDhmOJhdOXbA139f9LSBSN5jfABSaFqlAZp2JrU294zRc9ZioB2AAUriWNYDYAkTBGdGqW+jUBRuaXAISYqPRz+h/bZ5QGNkqgPnS1RlgokCmwoutTSNSEtidTNQDMAApXEsXGFuAKDgjWnULFXistxOOACB2weW3E0o81yQa2dUA1Hyus1V22B2+xxyJ14HdX0Q9AwOQwrELjDpL7BYqtzag2u5oCRvdPP5HJJ6noKzG50rwF2wNqKp1IEoFpPfpWp2S4rXSIOqTHoO+PXEKPFHPwgCkcBwETZ0Vr43GgF7ujVBPlNW0dAGFaQbUsD7xUEepYKlrRLm1oc39YovUoKQ46EKwxMOI5qn/J9tpcWpZBJEBiKgnYABSOCtbgKgLPGdHFYSouylQ2mg1Bpvce23l+xgHFOop+SOaN8byNRC62u6QQtjwlO5bAZuIQocBSOFausDYAkTBE1s7DhRX4bylHkB4u4BG+hmc3DL+JzQtUlILkI+p8OL5+yfq+F4i6iEYgBSOg6CpK8SwsyO/HIA7AIRzRqG/ACSuEh2yFqDU9luAWhZAZPcXUU/BAKRwHARNXSG2rtQ3ugCEfwCwGEryz3t3gTU5XdJg5VCFErFry2xrQKXNe8xRqKbbE1H4MAApmMXugLn5F3ltQxMsdofMNaKeZrApDtEeu56nGLRhfR2J4aaowoZDxZU4fcGGcms9vi40w9HkQmx0FIy60IR7vSYaA3vrAXivfg0wABH1RAxAClVaXYeFbx9FYYX7r+RlH3yPB94+itLqOplrRj2J2dbgtev72wdLwvo6ilIB6igVmlwCbn/tIE6ba7HknWOY/8YhAEB9kwuLNh8LWX2GJrsHXe/Ir8CpCzZY7A5U1zZIg7D1GjX/kCDqIRiAFMhid+DRrcfxdaHZ6/bdhWYs3Xqcv8ApIOLryO5wet0erteRxe7A0ve/g9PlXgMoa0QyXt9bjL1Fld1Sn9LqOhQ1/8Hw2p5i5LywB/llNbj3zTzUNrivwT1v5vEPCaIeggFIgcw2R5vwI9pdaIbZxgBEHZP7ddT6/MUXatuEn1DVRwx7JRdbgs3vJg7A/3x+Agd/utjmXPxDgijyceSrAlmbZ361p6aD+4kA+V9Hrc8vdue2pyv18RX2/v7NT+2WFwOXUc9NUYkiFVuAFKijacpcx4QCIffrqPX5f3l5it/yXamPr7AXo/a/3xn/kCCKbAxACmSK1yAr3eTzvqx0E0zx/KuVOib366j1+YelJCBzWFK31MdX2Ntw2wS/j+EfEkSRjQFIgYx6DdbMzMDlrTatzEo34ZmZGWy2p4CIr6PWIShcr6PW59+4pxjzMwdjaqsQFIr6+Ap7R0uquy1wEVH3Uwm+tlFWOKvVCqPRCIvFAoMhPBs7yuH1Pafx5Mf5mHhZLzwzMwOmeA3DDwXNvZ6UAzX1jUiIjQn768jz/AZdDOK00bDVN4W8PqXVdVi69Th2N48F0mvU2HjnVVj/VZHX+CAxcPVN1HX5nEQUnGA+vzkIWsHszav3DjbFYWgfbuBInWPUyxucfZ0/pRv+bumXqMMLs8e3CXsv+riNf0gQRT4GIAW7UONeBdqUoJW5JkQ9Q3thj4GHqOfhGCAFq6x1r1OSFMdf3kREpCwMQApmbm4BSmYLEBERKQwDkIJV1roDUFIcAxARESkLA5CCiVsDmBLYBUZERMrCAKRQTU4XLtrFMUBsASIiImVhAFKoKrsDggCoVEBvDoImIiKFYQBSqMrm7q/eeg3UUf73NCIiIrrUMAAplNnWvAZQPLu/iIhIeRiAFEpsAUrifkVERKRADEAKxRYgIiJSMgYghTKzBYiIiBSMAUih2AJERERKxgCkUJVSAGILEBERKQ8DkEJJq0CzBYiIiBSIAUihxBagJAYgIiJSIAYgBRIEwaMFiF1gRESkPAxAClTT0ASH0wWAXWBERKRMDEAKZK5xd3/Fa6MRG6OWuTZEREThxwCkQJW1XAOIiIiUjQFIgcQWIHZ/ERGRUjEAKZBZbAGKYwsQEREpEwOQAkktQAlsASIiImViAFKgytrmAMQWICIiUigGIAUy1zSvAcQWICIiUigGIAUSW4CS4hiAiIhImRiAFIirQBMRkdIxACmQmfuAERGRwjEAKUx9oxM19U0AgGQGICIiUigGIIWpal4DKEatgkEXLXNtiIiI5MEApDBS91ecFiqVSubaEBERyYMBSGEqbdwHjIiIiAFIYS7YuA8YERERA5DCsAWIiIiIAUhxxDFAnAFGRERKxgCkMJXSGkBsASIiIuViAFKYllWg2QJERETKxQCkMFwFmoiIiAFIcbgPGBEREQOQorhcAqpqOQ2eiIiIAUhBLtodcAnu//eOYwsQEREpFwOQglQ27wOWqI9BjJo/eiIiUi7uhnmJstgdMNscsNY3wqiLQZw2GsdLqgEAhtgYWOwOGPVsBSIiImViALoElVbX4dGtx/F1oRl6jRrPzx6P1/cWY29RJQDgbJUdD7x9FGtmZqBfok7m2hIREYUf+0EuMRa7Qwo/AHDX1MFe4Ue0u9CMpVuPw2J3yFFNIiIiWckegNavX49BgwYhNjYWkyZNwsGDB/2WX7duHUaMGAGdToe0tDQsXrwY9fX1PsuuWbMGKpUKDz74YDfUPDKZbQ4p/ADA7pMX2oQf6b5CszQtnoiISElkDUBbtmzBkiVLsHLlShw5cgRjx47FjBkzUFFR4bP8pk2bsHTpUqxcuRL5+fl47bXXsGXLFixbtqxN2UOHDuGVV15BRkZGdz+NiGKtb/T6/vh/LH7L17QqT0REpASyBqC1a9finnvuwfz58zF69Ghs2LABer0eGzdu9Fn+m2++QWZmJubMmYNBgwbh+uuvx+zZs9u0GtlsNsydOxevvvoqevXqFY6nEjEMsTFe38dr/Q/zSmhVnoiISAlkC0AOhwN5eXnIzs5uqUxUFLKzs7Fv3z6fj5kyZQry8vKkwHP69Gl88sknuPHGG73KLViwADfddJPXsf1paGiA1Wr1+uqpTPEaZKWbpO+1MVHIHJbks2xWuokrQhMRkSLJNgvMbDbD6XQiJSXF6/aUlBScOHHC52PmzJkDs9mMqVOnQhAENDU14b777vPqAtu8eTOOHDmCQ4cOBVyX3NxcPPnkk517IhHGqNdgzcwM/P6NQ8gvq8HFWgfmZw6GCsAej7FAWekmPDMzg1PhiYhIkXrUNPidO3fi6aefxksvvYRJkyahqKgIixYtwqpVq7BixQqUlJRg0aJF2L59O2JjYwM+7mOPPYYlS5ZI31utVqSlpXXHUwiLfok6/G7iADz1cT6uHNgLQ0xxeO6WcbDVN6GmvhEJsTEwxWsYfoiISLFkC0AmkwlqtRrl5eVet5eXlyM1NdXnY1asWIHbb78dd999NwDgiiuuQG1tLe699148/vjjyMvLQ0VFBa688krpMU6nE7t378aLL76IhoYGqNXqNsfVarXQai+tvbGamve8GNBLhyHJ8QCAFIOcNSIiIoocso0B0mg0mDBhAnbs2CHd5nK5sGPHDkyePNnnY+x2O6KivKssBhpBEDB9+nR89913OHbsmPQ1ceJEzJ07F8eOHfMZfi5VNfVNADjImYiIyBdZu8CWLFmCefPmYeLEibj66quxbt061NbWYv78+QCAO+64A/3790dubi4AICcnB2vXrsX48eOlLrAVK1YgJycHarUaCQkJGDNmjNc54uLikJSU1Ob2S50YgAy6HtXLSUREFBayfjrOmjULFy5cwBNPPIGysjKMGzcOn332mTQw+uzZs14tPsuXL4dKpcLy5ctx7tw5JCcnIycnB6tXr5brKUQscT0gtgARERG1pRIEQZC7EpHGarXCaDTCYrHAYOiZA2fuefMwtv9YjtW/GYO5ky6TuzpERETdLpjPb9m3wqDuUcMWICIionYxAF2irHXiIGiOASIiImqNAegSVdPgbgEyMAARERG1wQB0ieI0eCIiovZ1KgB9/fXXuO222zB58mScO3cOAPCPf/wDe/bsCWnlqHMEQfAIQGwBIiIiai3oALR161bMmDEDOp0OR48eRUNDAwDAYrHg6aefDnkFKXh1jU44m1eCbr07PBEREXUiAP35z3/Ghg0b8OqrryImpuXDNTMzE0eOHAlp5ahzxNYfdZQKeo1yVr8mIiIKVNABqKCgAFlZWW1uNxqNqK6uDkWdqIvEKfDx2mioVCqZa0NERBR5gg5AqampKCoqanP7nj17MGTIkJBUirrGyvE/REREfgUdgO655x4sWrQIBw4cgEqlQmlpKd566y089NBDuP/++7ujjhQkzgAjIiLyL+gmgqVLl8LlcmH69Omw2+3IysqCVqvFQw89hAceeKA76khBalkFmi1AREREvgT1Cel0OrF3714sWLAADz/8MIqKimCz2TB69GjEx8d3Vx0pSOIq0FwEkYiIyLegPiHVajWuv/565OfnIzExEaNHj+6uelEzi90Bs80Ba30jDLoYmOI0ANDmNqNeIz2G+4ARERH5F3QTwZgxY3D69GkMHjy4O+pDHkqr6/Do1uP4utAMANBr1Nh451VY/2URvi4yS+Wy0k1YMzMD/RJ1AFrGALEFiIiIyLdOrQP00EMP4eOPP8b58+dhtVq9vig0LHaHV/gBgLEDErF063Gv8AMAuwvNWLr1OCx2BwC2ABEREXUk6CaCG2+8EQDwq1/9ymuNGUEQoFKp4HQ6Q1c7BTPbHF7hBwD2na5st/zuQjPMNgeMeg23wSAiIupA0J+QX331VXfUg1qxNrfieJp4WS8c/uliu48RW36snAZPRETkV9ABaNq0ad1RD2rF1x5e/zVtCA6/mdfuY8TAw2nwRERE/nXqE7K6uhqvvfYa8vPzAQCXX3457rrrLhiNxpBWTslM8RpkpZuw26Mb7Nv/WJA5LAl7i9p2hWWlm2CKd88EYxcYERGRf0EPgj58+DCGDh2K//3f/0VVVRWqqqqwdu1aDB06lJuhhpBRr8GamRlI66WTbtu4pxgPXJeOa9JNXmWz0k14ZmaGNBW+poGDoImIiPwJuolg8eLF+NWvfoVXX30V0dHuhzc1NeHuu+/Ggw8+iN27d4e8kkrVL1GHocnxKLlYh7unDsbsqwfCFK/Bi7PH49fr9+JMpR1/mZmBGZeneK0DxIUQiYiI/Av6E/Lw4cNe4QcAoqOj8cgjj2DixIkhrRwBp8w2AMD0USkY2qdlte2+Rh3OVNqhjYnyCj+CIMDW0ByAdGwBIiIi8iXoLjCDwYCzZ8+2ub2kpAQJCQkhqRS52RqaUFJVBwAYmep9bU0JWgDAhZoGr9vtDiecLgEAxwARERG1J+gANGvWLPz+97/Hli1bUFJSgpKSEmzevBl33303Zs+e3R11VKyT5TUAgD4JWvSK03jdl9T8fWWtw+t2cQC0OkoFXYw6DLUkIiLqeYJuIvif//kfqFQq3HHHHWhqcn/YxsTE4P7778eaNWtCXkElKyhzB6ARqW1b1pKbW4DMrVqAPKfAey5USURERC2CDkAajQZ//etfkZubi1OnTgEAhg4dCr1eH/LKKZ0YgFp3fwHttwBZOQWeiIioQ0F/SlosFjidTvTu3RtXXHGFdHtVVRWio6NhMBhCWkElO1Hm3lttRGrba2qKb24BsrXTAqTlAGgiIqL2BD0G6NZbb8XmzZvb3P7OO+/g1ltvDUmlyD2by28LUPOih5U232OA2AJERETUvqAD0IEDB3Dttde2uf3nP/85Dhw4EJJKkXt210V7I6JUwDCP6e8isQXogq0BgiBIt9dwHzAiIqIOBR2AGhoapMHPnhobG1FXVxeSShFworn1Z5ApDrE+ZnOJAcjR5JLW/QFaNlE16NgCRERE1J6gA9DVV1+Nv/3tb21u37BhAyZMmBCSSpH/AdAAoNOoEadxByOzRzeYOAbI12aqRERE5BZ0M8Gf//xnZGdn49tvv8X06dMBADt27MChQ4fw73//O+QVVCqxBWhESvuDypPitaitssNsa8BgUxwAjgEiIiIKRNAtQJmZmdi3bx/S0tLwzjvv4KOPPsKwYcNw/PhxXHPNNd1RR0UqKBdngLW/urZJGgjdMhOMAYiIiKhjnfqUHDduHN56661Q14WaOV0CCsvde4C11wUGeA6EbtsFxkHQRERE7Qu6BejIkSP47rvvpO8//PBD3HzzzVi2bBkcDoefR1KgzlTWoqHJBV2MGgN7t7/AZFJzAPJsAeJCiERERB0LOgD913/9F06ePAkAOH36NGbNmgW9Xo93330XjzzySMgrqETiAOjhKfGIimp/O4vk5i4ws88uMLYAERERtSfoAHTy5EmMGzcOAPDuu+9i2rRp2LRpE9544w1s3bo11PVTpBN+9gDz1NIC5KsLjC1ARERE7Qk6AAmCAJfLBQD44osvcOONNwIA0tLSYDabQ1s7BbLYHTjy00UA7jE+Fnv73Yq+tsMQW4AMDEBERETtCjoATZw4EX/+85/xj3/8A7t27cJNN90EACguLkZKSkrIK6gkpdV1WPj2UewpcgfJl3aewgNvH0Vpte8FJltvhyEIAtcBIiIiCkDQAWjdunU4cuQIFi5ciMcffxzDhg0DALz33nuYMmVKyCuoFBa7A49uPY6vC71b0XYXmrF063GfLUGe22EAQK3DCVfzrhgcA0RERNS+oPtJMjIyvGaBiZ599lmo1W23bKDAmG2ONuFHtLvQDLPNAaNe43W7uA5QTX0TGpqcUutPdJQKsTFBZ1siIiLFCNlAkdjY2FAdSpHEPbzaU+PjfqMuBtFRKjS5BFTaHNKeYAmx0VCp2p89RkREpHRsJogQHY3Z8dWlpVKpvMYBcRFEIiKiwDAARQhTvAZZ6Saf92Wlm6TurraPa5kJxkUQiYiIAsMAFCGMeg3WzMzAgF46r9uz0k14ZmZGm/E/oiSPAMR9wIiIiALDT8oI0i9Rh0mDe+M/F89h9lVpuPuaITDFa9oNP0DLQGizzQGDzv3jZBcYERGRfyFrASopKcFdd90VqsMpVkOTe5HJ9JQEDO0T7zf8AC1dYJUeLUBcA4iIiMi/kAWgqqoq/P3vfw/V4RQr2G4sk8d+YNY6boNBREQUiIA/Kf/1r3/5vf/06dNdrgwh6JlcSXHNLUC1Dukx3AaDiIjIv4A/KW+++WaoVCoIgtBuGa4903XB7uVlSmheDbqmAUlx7tYgjgEiIiLyL+AusL59++L999+Hy+Xy+XXkyJHurKditHSBBdoC1DIImrPAiIiIAhNwAJowYQLy8vLavb+j1iEKTEsXWGAhJrm5BaiqtgGWOi6ESEREFIiAmwoefvhh1NbWtnv/sGHD8NVXX4WkUkrldAmodTgBBB6Aeje3ALkEoOSiPajHEhERKVVAn5THjx9HZmYmoqLabzCKi4vDtGnTQlYxJbI1d2EBgbfixKijkKiPQbW9EeXWhubHMgARERH5E1AX2Pjx42E2u3cqHzJkCCorK7u1UkolboiqjY6CJjrwFQrEcUAig45dYERERP4E9CmbmJiI4uJiAMCZM2fgcrm6tVJKFewAaJG4GKKILUBERET+BfRJOXPmTEybNg19+/aFSqXCxIkToVarfZblekCdJw6ADnYdn9YBiCtBExER+RfQJ+3f/vY3/Pa3v0VRURH++Mc/4p577kFCQkJ3101xOjuN3XOn+Bi1Ctogus+IiIiUKOBP2l/+8pcAgLy8PCxatIgBqBvUNHRuGnuSRwtQQmwMF6QkIiLqQNCDRV5//fXuqAehKy1AngGI43+IiIg6wr6SCNLZAJTk0QXGAERERNQxBqAIYg1yI1SRVwuQlgOgiYiIOsIAFEFCMQjaoGMLEBERUUcYgCJIZ9cB0qi9Bz1b7I6Q1YmIiOhSxAAUQYLdCBUASqvr8PB7x6XvP/+hHA+8fRSl1XUhrx8REdGlggEogogtQIEuhGixO/Do1uPYU+S9NcnuQjOWbj3OliAiIqJ2MABFkJogB0GbbQ58XWj2ed/uQjPMNgYgIiIiX2QPQOvXr8egQYMQGxuLSZMm4eDBg37Lr1u3DiNGjIBOp0NaWhoWL16M+vp66f6XX34ZGRkZMBgMMBgMmDx5Mj799NPufhohEewgaHHWWPvH838/ERGRUskagLZs2YIlS5Zg5cqVOHLkCMaOHYsZM2agoqLCZ/lNmzZh6dKlWLlyJfLz8/Haa69hy5YtWLZsmVRmwIABWLNmDfLy8nD48GFcd911+PWvf40ffvghXE+r04IdBN3Rnl/BDqYmIiJSClkD0Nq1a3HPPfdg/vz5GD16NDZs2AC9Xo+NGzf6LP/NN98gMzMTc+bMwaBBg3D99ddj9uzZXq1GOTk5uPHGG5Geno7hw4dj9erViI+Px/79+8P1tDrF6RJgawiuBcgUr0FWusnnfVnpJq/p8URERNRCtgDkcDiQl5eH7OzslspERSE7Oxv79u3z+ZgpU6YgLy9PCjynT5/GJ598ghtvvNFneafTic2bN6O2thaTJ09uty4NDQ2wWq1eX+Emhh8g8ABk1GuwZmZGmxCUlW7CMzMzYNQzABEREfki26p5ZrMZTqcTKSkpXrenpKTgxIkTPh8zZ84cmM1mTJ06FYIgoKmpCffdd59XFxgAfPfdd5g8eTLq6+sRHx+PDz74AKNHj263Lrm5uXjyySe7/qS6QByvo4mOgjZaHfDj+iXq8MLs8TDbHKipb0RCbAxM8RqGHyIiIj9kHwQdjJ07d+Lpp5/GSy+9hCNHjuD999/Htm3bsGrVKq9yI0aMwLFjx3DgwAHcf//9mDdvHn788cd2j/vYY4/BYrFIXyUlJd39VNoIdgq8J6Neg6F94jFuYC8M7RPP8ENERNQB2VqATCYT1Go1ysvLvW4vLy9Hamqqz8esWLECt99+O+6++24AwBVXXIHa2lrce++9ePzxxxEV5c5zGo0Gw4YNAwBMmDABhw4dwl//+le88sorPo+r1Wqh1Wp93hcunV0FmoiIiIInWwuQRqPBhAkTsGPHDuk2l8uFHTt2tDtex263SyFHpFa7u4sEQWj3XC6XCw0NDSGodffpzCrQRERE1DmyftouWbIE8+bNw8SJE3H11Vdj3bp1qK2txfz58wEAd9xxB/r374/c3FwA7hlea9euxfjx4zFp0iQUFRVhxYoVyMnJkYLQY489hhtuuAEDBw5ETU0NNm3ahJ07d+Lzzz+X7XkGorMboRIREVHwZP20nTVrFi5cuIAnnngCZWVlGDduHD777DNpYPTZs2e9WnyWL18OlUqF5cuX49y5c0hOTkZOTg5Wr14tlamoqMAdd9yB8+fPw2g0IiMjA59//jl+8YtfhP35BUNqAdKyC4yIiKi7qQR/fUcKZbVaYTQaYbFYYDAYwnLO9V8V4dnPC/C7CQPw7O/GhuWcREREl5JgPr971CywSxkHQRMREYUPA1CE4CBoIiKi8GEAihAcBE1ERBQ+DEARQmwB6miDUyIiIuo6BqAIwRYgIiKi8GEAihAcBE1ERBQ+DEARgoOgiYiIwocBKEKwC4yIiCh8GIAigMslwOZgFxgREVG4MABFAJujCeJ63GwBIiIi6n4MQBFA7P7SqKMQG6OWuTZERESXPgagCMAB0EREROHFABQBOACaiIgovBiAIkBLCxAHQBMREYUDA1AEYAsQERFReDEARQArAxAREVFYMQBFAHaBERERhRcDUARgFxgREVF4MQBFALYAERERhRcDUAQQW4AMbAEiIiIKCwagCMAuMCIiovBiAIoA7AIjIiIKLwagCMAWICIiovBiAIoALQGILUBEREThwAAUAazcDJWIiCisGIBk5nIJsDWwC4yIiCicGIBkVutogiC4/29gFxgREVFYMADJTBz/E6NWQRvNHwcREVE48BNXZp4DoFUqlcy1ISIiUgYGIJnVcAA0ERFR2DEAyYxrABEREYUfA5DMpCnwWg6AJiIiChcGIJmxBYiIiCj8GIBkxlWgiYiIwo8BSGYcBE1ERBR+DEAyE1uADAxAREREYcMAJLOWFiB2gREREYULA5CMLHYHyqz1AIC6RicsdofMNSIiIlIGBiCZlFbXYeHbR7H/dBUAYO32k3jg7aMora6TuWZERESXPgYgGVjsDjy69Ti+LjR73b670IylW4+zJYiIiKibMQDJwGxztAk/ot2FZphtDEBERETdiQFIBuLqz+2p6eB+IiIi6hoGIBkYOpjxxRlhRERE3YsBSAameA2y0k0+78tKN8EUrwlzjYiIiJSFAUgGRr0Ga2ZmYNKgXl63Z6Wb8MzMDBj1DEBERETdiQFIJv0SdXjklyMBuLfB2LFkGl6YPR59E3Uy14yIiOjSx/0X5KRy/9NLr8HQPvHy1oWIiEhB2AIkI6u0EzxzKBERUTgxAMmohgGIiIhIFgxAMuJGqERERPJgAJIRW4CIiIjkwQAkI7EFqKOFEYmIiCi0GIBkxBYgIiIieTAAyYgBiIiISB4MQDLiIGgiIiJ5MADJiOsAERERyYMBSEYtXWBsASIiIgonBiAZtXSBsQWIiIgonBiAZCS2ABkYgIiIiMKKAUgmgiDA1sAuMCIiIjkwAMnE7nDC6RIAsAuMiIgo3BiAZCJ2f6mjVNDFqGWuDRERkbIwAMnEcwC0SqWSuTZERETKwgAkE64BREREJB8GIJlILUBaDoAmIiIKNwYgmXAfMCIiIvkwAMmEq0ATERHJR/YAtH79egwaNAixsbGYNGkSDh486Lf8unXrMGLECOh0OqSlpWHx4sWor6+X7s/NzcVVV12FhIQE9OnTBzfffDMKCgq6+2kETewC4yKIRERE4SdrANqyZQuWLFmClStX4siRIxg7dixmzJiBiooKn+U3bdqEpUuXYuXKlcjPz8drr72GLVu2YNmyZVKZXbt2YcGCBdi/fz+2b9+OxsZGXH/99aitrQ3X0woIu8CIiIjkI+un79q1a3HPPfdg/vz5AIANGzZg27Zt2LhxI5YuXdqm/DfffIPMzEzMmTMHADBo0CDMnj0bBw4ckMp89tlnXo9544030KdPH+Tl5SErK6sbn01wWqbBswuMiIgo3GRrAXI4HMjLy0N2dnZLZaKikJ2djX379vl8zJQpU5CXlyd1k50+fRqffPIJbrzxxnbPY7FYAAC9e/dut0xDQwOsVqvXV3djCxAREZF8ZPv0NZvNcDqdSElJ8bo9JSUFJ06c8PmYOXPmwGw2Y+rUqRAEAU1NTbjvvvu8usA8uVwuPPjgg8jMzMSYMWParUtubi6efPLJzj+ZTrByEDQREZFsZB8EHYydO3fi6aefxksvvYQjR47g/fffx7Zt27Bq1Sqf5RcsWIDvv/8emzdv9nvcxx57DBaLRfoqKSnpjup78VwJmoiIiMJLtk9fk8kEtVqN8vJyr9vLy8uRmprq8zErVqzA7bffjrvvvhsAcMUVV6C2thb33nsvHn/8cURFteS5hQsX4uOPP8bu3bsxYMAAv3XRarXQarVdfEbBYRcYERGRfGRrAdJoNJgwYQJ27Ngh3eZyubBjxw5MnjzZ52PsdrtXyAEAtdq9kaggCNK/CxcuxAcffIAvv/wSgwcP7qZn0DU1DRwETUREJBdZmx+WLFmCefPmYeLEibj66quxbt061NbWSrPC7rjjDvTv3x+5ubkAgJycHKxduxbjx4/HpEmTUFRUhBUrViAnJ0cKQgsWLMCmTZvw4YcfIiEhAWVlZQAAo9EInU4nzxP1QWwB4jpARERE4Sfrp++sWbNw4cIFPPHEEygrK8O4cePw2WefSQOjz54969Xis3z5cqhUKixfvhznzp1DcnIycnJysHr1aqnMyy+/DAD4+c9/7nWu119/HXfeeWe3P6dACILAlaCJiIhkpBLEviOSWK1WGI1GWCwWGAyGkB/f7mjC6Cc+BwD88OQMxGnZCkRERNRVwXx+96hZYJcKsfVHHaWCXqOWuTZERETKwwAkA3EKfLw2GiqVSubaEBERKQ8DkAzERRANOnZ9ERERyYEBSAbSAGgtB0ATERHJgQFIBlwFmoiISF4MQDLgFHgiIiJ5MQDJQGwB4iKIRERE8mAAkgH3ASMiIpIXA5AM2AVGREQkLwYgGVg5CJqIiEhWDEAyYAsQERGRvBiAZMBp8ERERPJiAJIBB0ETERHJiwFIBuwCIyIikhcDkAysXAeIiIhIVgxAYSYIAluAiIiIZMYAFGZ1jU44XQIAjgEiIiKSCwNQmImtP+ooFfQatcy1ISIiUiYGoDATp8DHa6OhUqlkrg0REZEyMQCFmZVT4ImIiGTHABRmHABNREQkPwagMOMq0ERERPJjAAozsQWIawARERHJhwEozFpagNgFRkREJBcGoDDjPmBERETyYwAKMwYgIiIi+TEAhZmVXWBERESyYwAKM7YAERERyY8BKMw4CJqIiEh+DEBhxhYgIiIi+TEAhRnXASIiIpIfA1CYsQuMiIhIfgxAYSQIArvAiIiIIgADUBjVN7rQ5BIAsAWIiIhITgxAYSR2f0WpgDiNWubaEBERKRcDUBhZm7u/4rXRUKlUMteGiIhIuRiAwogDoImIiCIDA1AYlVnqAQAxahVOXbDBYnfIXCMiIiJlYgAKk9LqOvx1RyEA4EylHdOf24UH3j6K0uo6mWtGRESkPAxAYWCxO/Do1uM4UVbjdfvuQjOWbj3OliAiIqIwYwAKA7PNga8LzT7v211ohtnGAERERBRODEBhYG0e/Nyemg7uJyIiotBiAAoDQwezvjgrjIiIKLwYgMLAFK9BVrrJ531Z6SaY4jVhrhEREZGyMQCFgVGvwZqZGW1CUFa6Cc/MzIBRzwBEREQUTtyRM0z6JerwwuzxMNscqKlvREJsDEzxGoYfIiIiGTAAhZFRz8BDREQUCdgFRkRERIrDAERERESKwwBEREREisMARERERIrDAERERESKwwBEREREisMARERERIrDAERERESKwwBEREREisMARERERIrDrTB8EAQBAGC1WmWuCREREQVK/NwWP8f9YQDyoaamBgCQlpYmc02IiIgoWDU1NTAajX7LqIRAYpLCuFwulJaWIiEhASqVKqTHtlqtSEtLQ0lJCQwGQ0iPTd54rcOH1zp8eK3Dh9c6fEJ1rQVBQE1NDfr164eoKP+jfNgC5ENUVBQGDBjQrecwGAx8Q4UJr3X48FqHD691+PBah08ornVHLT8iDoImIiIixWEAIiIiIsVhAAozrVaLlStXQqvVyl2VSx6vdfjwWocPr3X48FqHjxzXmoOgiYiISHHYAkRERESKwwBEREREisMARERERIrDAERERESKwwAURuvXr8egQYMQGxuLSZMm4eDBg3JXqcfLzc3FVVddhYSEBPTp0wc333wzCgoKvMrU19djwYIFSEpKQnx8PGbOnIny8nKZanzpWLNmDVQqFR588EHpNl7r0Dl37hxuu+02JCUlQafT4YorrsDhw4el+wVBwBNPPIG+fftCp9MhOzsbhYWFMta4Z3I6nVixYgUGDx4MnU6HoUOHYtWqVV57SfFad87u3buRk5ODfv36QaVS4f/+7/+87g/kulZVVWHu3LkwGAxITEzE73//e9hstpDUjwEoTLZs2YIlS5Zg5cqVOHLkCMaOHYsZM2agoqJC7qr1aLt27cKCBQuwf/9+bN++HY2Njbj++utRW1srlVm8eDE++ugjvPvuu9i1axdKS0vx29/+VsZa93yHDh3CK6+8goyMDK/bea1D4+LFi8jMzERMTAw+/fRT/Pjjj3juuefQq1cvqcxf/vIXPP/889iwYQMOHDiAuLg4zJgxA/X19TLWvOd55pln8PLLL+PFF19Efn4+nnnmGfzlL3/BCy+8IJXhte6c2tpajB07FuvXr/d5fyDXde7cufjhhx+wfft2fPzxx9i9ezfuvffe0FRQoLC4+uqrhQULFkjfO51OoV+/fkJubq6Mtbr0VFRUCACEXbt2CYIgCNXV1UJMTIzw7rvvSmXy8/MFAMK+ffvkqmaPVlNTI6Snpwvbt28Xpk2bJixatEgQBF7rUHr00UeFqVOntnu/y+USUlNThWeffVa6rbq6WtBqtcLbb78djipeMm666Sbhrrvu8rrtt7/9rTB37lxBEHitQwWA8MEHH0jfB3Jdf/zxRwGAcOjQIanMp59+KqhUKuHcuXNdrhNbgMLA4XAgLy8P2dnZ0m1RUVHIzs7Gvn37ZKzZpcdisQAAevfuDQDIy8tDY2Oj17UfOXIkBg4cyGvfSQsWLMBNN93kdU0BXutQ+te//oWJEyfid7/7Hfr06YPx48fj1Vdfle4vLi5GWVmZ17U2Go2YNGkSr3WQpkyZgh07duDkyZMAgG+//RZ79uzBDTfcAIDXursEcl337duHxMRETJw4USqTnZ2NqKgoHDhwoMt14GaoYWA2m+F0OpGSkuJ1e0pKCk6cOCFTrS49LpcLDz74IDIzMzFmzBgAQFlZGTQaDRITE73KpqSkoKysTIZa9mybN2/GkSNHcOjQoTb38VqHzunTp/Hyyy9jyZIlWLZsGQ4dOoQ//vGP0Gg0mDdvnnQ9ff1O4bUOztKlS2G1WjFy5Eio1Wo4nU6sXr0ac+fOBQBe624SyHUtKytDnz59vO6Pjo5G7969Q3LtGYDokrFgwQJ8//332LNnj9xVuSSVlJRg0aJF2L59O2JjY+WuziXN5XJh4sSJePrppwEA48ePx/fff48NGzZg3rx5Mtfu0vLOO+/grbfewqZNm3D55Zfj2LFjePDBB9GvXz9e60scu8DCwGQyQa1Wt5kNU15ejtTUVJlqdWlZuHAhPv74Y3z11VcYMGCAdHtqaiocDgeqq6u9yvPaBy8vLw8VFRW48sorER0djejoaOzatQvPP/88oqOjkZKSwmsdIn379sXo0aO9bhs1ahTOnj0LANL15O+Urnv44YexdOlS3Hrrrbjiiitw++23Y/HixcjNzQXAa91dArmuqampbSYKNTU1oaqqKiTXngEoDDQaDSZMmIAdO3ZIt7lcLuzYsQOTJ0+WsWY9nyAIWLhwIT744AN8+eWXGDx4sNf9EyZMQExMjNe1LygowNmzZ3ntgzR9+nR89913OHbsmPQ1ceJEzJ07V/o/r3VoZGZmtlnO4eTJk7jssssAAIMHD0ZqaqrXtbZarThw4ACvdZDsdjuiorw/CtVqNVwuFwBe6+4SyHWdPHkyqqurkZeXJ5X58ssv4XK5MGnSpK5XosvDqCkgmzdvFrRarfDGG28IP/74o3DvvfcKiYmJQllZmdxV69Huv/9+wWg0Cjt37hTOnz8vfdntdqnMfffdJwwcOFD48ssvhcOHDwuTJ08WJk+eLGOtLx2es8AEgdc6VA4ePChER0cLq1evFgoLC4W33npL0Ov1wj//+U+pzJo1a4TExEThww8/FI4fPy78+te/FgYPHizU1dXJWPOeZ968eUL//v2Fjz/+WCguLhbef/99wWQyCY888ohUhte6c2pqaoSjR48KR48eFQAIa9euFY4ePSr89NNPgiAEdl1/+ctfCuPHjxcOHDgg7NmzR0hPTxdmz54dkvoxAIXRCy+8IAwcOFDQaDTC1VdfLezfv1/uKvV4AHx+vf7661KZuro64Q9/+IPQq1cvQa/XC7/5zW+E8+fPy1fpS0jrAMRrHTofffSRMGbMGEGr1QojR44U/va3v3nd73K5hBUrVggpKSmCVqsVpk+fLhQUFMhU257LarUKixYtEgYOHCjExsYKQ4YMER5//HGhoaFBKsNr3TlfffWVz9/P8+bNEwQhsOtaWVkpzJ49W4iPjxcMBoMwf/58oaamJiT1UwmCx3KXRERERArAMUBERESkOAxAREREpDgMQERERKQ4DEBERESkOAxAREREpDgMQERERKQ4DEBERESkOAxAREREpDgMQEREAdi5cydUKlWbzV6JqGdiACIiIiLFYQAiIiIixWEAIqIeweVyITc3F4MHD4ZOp8PYsWPx3nvvAWjpntq2bRsyMjIQGxuLn/3sZ/j++++9jrF161Zcfvnl0Gq1GDRoEJ577jmv+xsaGvDoo48iLS0NWq0Ww4YNw2uvveZVJi8vDxMnToRer8eUKVNQUFDQvU+ciLoFAxAR9Qi5ubl48803sWHDBvzwww9YvHgxbrvtNuzatUsq8/DDD+O5557DoUOHkJycjJycHDQ2NgJwB5dbbrkFt956K7777jv86U9/wooVK/DGG29Ij7/jjjvw9ttv4/nnn0d+fj5eeeUVxMfHe9Xj8ccfx3PPPYfDhw8jOjoad911V1iePxGFFneDJ6KI19DQgN69e+OLL77A5MmTpdvvvvtu2O123Hvvvbj22muxefNmzJo1CwBQVVWFAQMG4I033sAtt9yCuXPn4sKFC/j3v/8tPf6RRx7Btm3b8MMPP+DkyZMYMWIEtm/fjuzs7DZ12LlzJ6699lp88cUXmD59OgDgk08+wU033YS6ujrExsZ281UgolBiCxARRbyioiLY7Xb84he/QHx8vPT15ptv4tSpU1I5z3DUu3dvjBgxAvn5+QCA/Px8ZGZmeh03MzMThYWFcDqdOHbsGNRqNaZNm+a3LhkZGdL/+/btCwCoqKjo8nMkovCKlrsCREQdsdlsAIBt27ahf//+XvdptVqvENRZOp0uoHIxMTHS/1UqFQD3+CQi6lnYAkREEW/06NHQarU4e/Yshg0b5vWVlpYmldu/f7/0/4sXL+LkyZMYNWoUAGDUqFHYu3ev13H37t2L4cOHQ61W44orroDL5fIaU0REly62ABFRxEtISMBDDz2ExYsXw+VyYerUqbBYLNi7dy8MBgMuu+wyAMBTTz2FpKQkpKSk4PHHH4fJZMLNN98MAPjv//5vXHXVVVi1ahVmzZqFffv24cUXX8RLL70EABg0aBDmzZuHu+66C88//zzGjh2Ln376CRUVFbjlllvkeupE1E0YgIioR1i1ahWSk5ORm5uL06dPIzExEVdeeSWWLVsmdUGtWbMGixYtQmFhIcaNG4ePPvoIGo0GAHDllVfinXfewRNPPIFVq1ahb9++eOqpp3DnnXdK53j55ZexbNky/OEPf0BlZSUGDhyIZcuWyfF0iaibcRYYEfV44gytixcvIjExUe7qEFEPwDFAREREpDgMQERERKQ47AIjIiIixWELEBERESkOAxAREREpDgMQERERKQ4DEBERESkOAxAREREpDgMQERERKQ4DEBERESkOAxAREREpzv8HD9A/g5mPbV4AAAAASUVORK5CYII=",
1286
  "text/plain": [
1287
  "<Figure size 640x480 with 1 Axes>"
1288
  ]
 
1294
  "source": [
1295
  "plot_f1(best_f1)"
1296
  ]
1297
+ },
1298
+ {
1299
+ "cell_type": "code",
1300
+ "execution_count": null,
1301
+ "metadata": {},
1302
+ "outputs": [],
1303
+ "source": []
1304
  }
1305
  ],
1306
  "metadata": {