├── .gitignore ├── LICENSE ├── README.md └── models ├── __init__.py ├── diffusion_models ├── __init__.py ├── ddim_mnist.ipynb ├── ddpm_cifar10.ipynb ├── ddpm_mnist.ipynb └── improved_ddpm_mnist.ipynb ├── gated_pixelcnn.ipynb ├── pixelcnn.ipynb ├── stable_diffusion ├── README.md ├── assets │ ├── blend_sd.png │ ├── img2img │ │ └── sketch-mountains-input.jpg │ ├── inpainting_examples │ │ ├── overture-creations-5sI6fQgYIuo.png │ │ └── overture-creations-5sI6fQgYIuo_mask.png │ └── sd.png ├── random_walks_with_stable_diffusion-no-images.ipynb ├── stable_diffusion-v1-5.ipynb └── stable_diffusion.ipynb ├── vae.ipynb ├── vq_vae.ipynb └── vq_vae_ema.ipynb /.gitignore: -------------------------------------------------------------------------------- 1 | # Byte-compiled / optimized / DLL files 2 | __pycache__/ 3 | *.py[cod] 4 | *$py.class 5 | 6 | # C extensions 7 | *.so 8 | 9 | # Distribution / packaging 10 | .Python 11 | build/ 12 | develop-eggs/ 13 | dist/ 14 | downloads/ 15 | eggs/ 16 | .eggs/ 17 | lib/ 18 | lib64/ 19 | parts/ 20 | sdist/ 21 | var/ 22 | wheels/ 23 | pip-wheel-metadata/ 24 | share/python-wheels/ 25 | *.egg-info/ 26 | .installed.cfg 27 | *.egg 28 | MANIFEST 29 | 30 | # PyInstaller 31 | # Usually these files are written by a python script from a template 32 | # before PyInstaller builds the exe, so as to inject date/other infos into it. 33 | *.manifest 34 | *.spec 35 | 36 | # Installer logs 37 | pip-log.txt 38 | pip-delete-this-directory.txt 39 | 40 | # Unit test / coverage reports 41 | htmlcov/ 42 | .tox/ 43 | .nox/ 44 | .coverage 45 | .coverage.* 46 | .cache 47 | nosetests.xml 48 | coverage.xml 49 | *.cover 50 | *.py,cover 51 | .hypothesis/ 52 | .pytest_cache/ 53 | 54 | # Translations 55 | *.mo 56 | *.pot 57 | 58 | # Django stuff: 59 | *.log 60 | local_settings.py 61 | db.sqlite3 62 | db.sqlite3-journal 63 | 64 | # Flask stuff: 65 | instance/ 66 | .webassets-cache 67 | 68 | # Scrapy stuff: 69 | .scrapy 70 | 71 | # Sphinx documentation 72 | docs/_build/ 73 | 74 | # PyBuilder 75 | target/ 76 | 77 | # Jupyter Notebook 78 | .ipynb_checkpoints 79 | 80 | # IPython 81 | profile_default/ 82 | ipython_config.py 83 | 84 | # pyenv 85 | .python-version 86 | 87 | # pipenv 88 | # According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control. 89 | # However, in case of collaboration, if having platform-specific dependencies or dependencies 90 | # having no cross-platform support, pipenv may install dependencies that don't work, or not 91 | # install all needed dependencies. 92 | #Pipfile.lock 93 | 94 | # PEP 582; used by e.g. github.com/David-OConnor/pyflow 95 | __pypackages__/ 96 | 97 | # Celery stuff 98 | celerybeat-schedule 99 | celerybeat.pid 100 | 101 | # SageMath parsed files 102 | *.sage.py 103 | 104 | # Environments 105 | .env 106 | .venv 107 | env/ 108 | venv/ 109 | ENV/ 110 | env.bak/ 111 | venv.bak/ 112 | 113 | # Spyder project settings 114 | .spyderproject 115 | .spyproject 116 | 117 | # Rope project settings 118 | .ropeproject 119 | 120 | # mkdocs documentation 121 | /site 122 | 123 | # mypy 124 | .mypy_cache/ 125 | .dmypy.json 126 | dmypy.json 127 | 128 | # Pyre type checker 129 | .pyre/ 130 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | Apache License 2 | Version 2.0, January 2004 3 | http://www.apache.org/licenses/ 4 | 5 | TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 6 | 7 | 1. Definitions. 8 | 9 | "License" shall mean the terms and conditions for use, reproduction, 10 | and distribution as defined by Sections 1 through 9 of this document. 11 | 12 | "Licensor" shall mean the copyright owner or entity authorized by 13 | the copyright owner that is granting the License. 14 | 15 | "Legal Entity" shall mean the union of the acting entity and all 16 | other entities that control, are controlled by, or are under common 17 | control with that entity. For the purposes of this definition, 18 | "control" means (i) the power, direct or indirect, to cause the 19 | direction or management of such entity, whether by contract or 20 | otherwise, or (ii) ownership of fifty percent (50%) or more of the 21 | outstanding shares, or (iii) beneficial ownership of such entity. 22 | 23 | "You" (or "Your") shall mean an individual or Legal Entity 24 | exercising permissions granted by this License. 25 | 26 | "Source" form shall mean the preferred form for making modifications, 27 | including but not limited to software source code, documentation 28 | source, and configuration files. 29 | 30 | "Object" form shall mean any form resulting from mechanical 31 | transformation or translation of a Source form, including but 32 | not limited to compiled object code, generated documentation, 33 | and conversions to other media types. 34 | 35 | "Work" shall mean the work of authorship, whether in Source or 36 | Object form, made available under the License, as indicated by a 37 | copyright notice that is included in or attached to the work 38 | (an example is provided in the Appendix below). 39 | 40 | "Derivative Works" shall mean any work, whether in Source or Object 41 | form, that is based on (or derived from) the Work and for which the 42 | editorial revisions, annotations, elaborations, or other modifications 43 | represent, as a whole, an original work of authorship. For the purposes 44 | of this License, Derivative Works shall not include works that remain 45 | separable from, or merely link (or bind by name) to the interfaces of, 46 | the Work and Derivative Works thereof. 47 | 48 | "Contribution" shall mean any work of authorship, including 49 | the original version of the Work and any modifications or additions 50 | to that Work or Derivative Works thereof, that is intentionally 51 | submitted to Licensor for inclusion in the Work by the copyright owner 52 | or by an individual or Legal Entity authorized to submit on behalf of 53 | the copyright owner. For the purposes of this definition, "submitted" 54 | means any form of electronic, verbal, or written communication sent 55 | to the Licensor or its representatives, including but not limited to 56 | communication on electronic mailing lists, source code control systems, 57 | and issue tracking systems that are managed by, or on behalf of, the 58 | Licensor for the purpose of discussing and improving the Work, but 59 | excluding communication that is conspicuously marked or otherwise 60 | designated in writing by the copyright owner as "Not a Contribution." 61 | 62 | "Contributor" shall mean Licensor and any individual or Legal Entity 63 | on behalf of whom a Contribution has been received by Licensor and 64 | subsequently incorporated within the Work. 65 | 66 | 2. Grant of Copyright License. Subject to the terms and conditions of 67 | this License, each Contributor hereby grants to You a perpetual, 68 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 69 | copyright license to reproduce, prepare Derivative Works of, 70 | publicly display, publicly perform, sublicense, and distribute the 71 | Work and such Derivative Works in Source or Object form. 72 | 73 | 3. Grant of Patent License. Subject to the terms and conditions of 74 | this License, each Contributor hereby grants to You a perpetual, 75 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 76 | (except as stated in this section) patent license to make, have made, 77 | use, offer to sell, sell, import, and otherwise transfer the Work, 78 | where such license applies only to those patent claims licensable 79 | by such Contributor that are necessarily infringed by their 80 | Contribution(s) alone or by combination of their Contribution(s) 81 | with the Work to which such Contribution(s) was submitted. If You 82 | institute patent litigation against any entity (including a 83 | cross-claim or counterclaim in a lawsuit) alleging that the Work 84 | or a Contribution incorporated within the Work constitutes direct 85 | or contributory patent infringement, then any patent licenses 86 | granted to You under this License for that Work shall terminate 87 | as of the date such litigation is filed. 88 | 89 | 4. Redistribution. You may reproduce and distribute copies of the 90 | Work or Derivative Works thereof in any medium, with or without 91 | modifications, and in Source or Object form, provided that You 92 | meet the following conditions: 93 | 94 | (a) You must give any other recipients of the Work or 95 | Derivative Works a copy of this License; and 96 | 97 | (b) You must cause any modified files to carry prominent notices 98 | stating that You changed the files; and 99 | 100 | (c) You must retain, in the Source form of any Derivative Works 101 | that You distribute, all copyright, patent, trademark, and 102 | attribution notices from the Source form of the Work, 103 | excluding those notices that do not pertain to any part of 104 | the Derivative Works; and 105 | 106 | (d) If the Work includes a "NOTICE" text file as part of its 107 | distribution, then any Derivative Works that You distribute must 108 | include a readable copy of the attribution notices contained 109 | within such NOTICE file, excluding those notices that do not 110 | pertain to any part of the Derivative Works, in at least one 111 | of the following places: within a NOTICE text file distributed 112 | as part of the Derivative Works; within the Source form or 113 | documentation, if provided along with the Derivative Works; or, 114 | within a display generated by the Derivative Works, if and 115 | wherever such third-party notices normally appear. The contents 116 | of the NOTICE file are for informational purposes only and 117 | do not modify the License. You may add Your own attribution 118 | notices within Derivative Works that You distribute, alongside 119 | or as an addendum to the NOTICE text from the Work, provided 120 | that such additional attribution notices cannot be construed 121 | as modifying the License. 122 | 123 | You may add Your own copyright statement to Your modifications and 124 | may provide additional or different license terms and conditions 125 | for use, reproduction, or distribution of Your modifications, or 126 | for any such Derivative Works as a whole, provided Your use, 127 | reproduction, and distribution of the Work otherwise complies with 128 | the conditions stated in this License. 129 | 130 | 5. Submission of Contributions. Unless You explicitly state otherwise, 131 | any Contribution intentionally submitted for inclusion in the Work 132 | by You to the Licensor shall be under the terms and conditions of 133 | this License, without any additional terms or conditions. 134 | Notwithstanding the above, nothing herein shall supersede or modify 135 | the terms of any separate license agreement you may have executed 136 | with Licensor regarding such Contributions. 137 | 138 | 6. Trademarks. This License does not grant permission to use the trade 139 | names, trademarks, service marks, or product names of the Licensor, 140 | except as required for reasonable and customary use in describing the 141 | origin of the Work and reproducing the content of the NOTICE file. 142 | 143 | 7. Disclaimer of Warranty. Unless required by applicable law or 144 | agreed to in writing, Licensor provides the Work (and each 145 | Contributor provides its Contributions) on an "AS IS" BASIS, 146 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or 147 | implied, including, without limitation, any warranties or conditions 148 | of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A 149 | PARTICULAR PURPOSE. You are solely responsible for determining the 150 | appropriateness of using or redistributing the Work and assume any 151 | risks associated with Your exercise of permissions under this License. 152 | 153 | 8. Limitation of Liability. In no event and under no legal theory, 154 | whether in tort (including negligence), contract, or otherwise, 155 | unless required by applicable law (such as deliberate and grossly 156 | negligent acts) or agreed to in writing, shall any Contributor be 157 | liable to You for damages, including any direct, indirect, special, 158 | incidental, or consequential damages of any character arising as a 159 | result of this License or out of the use or inability to use the 160 | Work (including but not limited to damages for loss of goodwill, 161 | work stoppage, computer failure or malfunction, or any and all 162 | other commercial damages or losses), even if such Contributor 163 | has been advised of the possibility of such damages. 164 | 165 | 9. Accepting Warranty or Additional Liability. While redistributing 166 | the Work or Derivative Works thereof, You may choose to offer, 167 | and charge a fee for, acceptance of support, warranty, indemnity, 168 | or other liability obligations and/or rights consistent with this 169 | License. However, in accepting such obligations, You may act only 170 | on Your own behalf and on Your sole responsibility, not on behalf 171 | of any other Contributor, and only if You agree to indemnify, 172 | defend, and hold each Contributor harmless for any liability 173 | incurred by, or claims asserted against, such Contributor by reason 174 | of your accepting any such warranty or additional liability. 175 | 176 | END OF TERMS AND CONDITIONS 177 | 178 | APPENDIX: How to apply the Apache License to your work. 179 | 180 | To apply the Apache License to your work, attach the following 181 | boilerplate notice, with the fields enclosed by brackets "[]" 182 | replaced with your own identifying information. (Don't include 183 | the brackets!) The text should be enclosed in the appropriate 184 | comment syntax for the file format. We also recommend that a 185 | file or class name and description of purpose be included on the 186 | same "printed page" as the copyright notice for easier 187 | identification within third-party archives. 188 | 189 | Copyright [yyyy] [name of copyright owner] 190 | 191 | Licensed under the Apache License, Version 2.0 (the "License"); 192 | you may not use this file except in compliance with the License. 193 | You may obtain a copy of the License at 194 | 195 | http://www.apache.org/licenses/LICENSE-2.0 196 | 197 | Unless required by applicable law or agreed to in writing, software 198 | distributed under the License is distributed on an "AS IS" BASIS, 199 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 200 | See the License for the specific language governing permissions and 201 | limitations under the License. 202 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # nngen 2 | 3 | ## Deep Generative Models 4 | 5 | - [VAE](https://arxiv.org/abs/1312.6114) [[code](models/vae.ipynb)] 6 | - [PixelCNN](https://arxiv.org/abs/1601.06759) [[code](models/pixelcnn.ipynb)] 7 | - [GatedPixelCNN](https://arxiv.org/abs/1606.05328) [[code](models/gated_pixelcnn.ipynb)] 8 | - [VQ-VAE](https://arxiv.org/abs/1711.00937) [[code](models/vq_vae.ipynb)] [[EMA version code](models/vq_vae_ema.ipynb)] 9 | 10 | ### Diffusion Models 11 | - [DDPM](https://arxiv.org/abs/2006.11239) [[ddpm_mnist code](models/diffusion_models/ddpm_mnist.ipynb)] 12 | - [DDIM](https://arxiv.org/abs/2010.02502) [[ddim_mnist code](models/diffusion_models/ddim_mnist.ipynb)] 13 | - [DDPM+](https://arxiv.org/abs/2102.09672) [[improved_ddpm mnist code](models/diffusion_models/improved_ddpm_mnist.ipynb)] 14 | 15 | ### Text-to-Image Models 16 | - [Stable Diffusion](https://arxiv.org/abs/2112.10752) [[code](models/stable_diffusion)] 17 | -------------------------------------------------------------------------------- /models/__init__.py: -------------------------------------------------------------------------------- 1 | 2 | -------------------------------------------------------------------------------- /models/diffusion_models/__init__.py: -------------------------------------------------------------------------------- 1 | 2 | -------------------------------------------------------------------------------- /models/pixelcnn.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "markdown", 5 | "id": "2db6c902-9d8f-4ed3-8f42-7ff59612141e", 6 | "metadata": {}, 7 | "source": [ 8 | "## PixelCNN\n", 9 | "\n", 10 | "PixelCNN is an autoregressive likelihood model for the task of image modeling.\n", 11 | "\n", 12 | "### Reference\n", 13 | "- https://keras.io/examples/generative/pixelcnn/\n", 14 | "- Pixel Recurrent Neural Networks: https://arxiv.org/abs/1601.06759" 15 | ] 16 | }, 17 | { 18 | "cell_type": "code", 19 | "execution_count": 1, 20 | "id": "ff8aa709-5726-4581-9d2a-f0503a591008", 21 | "metadata": {}, 22 | "outputs": [], 23 | "source": [ 24 | "import os\n", 25 | "import numpy as np\n", 26 | "import torch\n", 27 | "import torch.nn as nn\n", 28 | "import torch.nn.functional as F\n", 29 | "from torchvision import datasets, transforms\n", 30 | "from sklearn.manifold import TSNE\n", 31 | "import pandas as pd\n", 32 | "import matplotlib.pyplot as plt\n", 33 | "\n", 34 | "%matplotlib inline" 35 | ] 36 | }, 37 | { 38 | "cell_type": "code", 39 | "execution_count": 2, 40 | "id": "5bb28827-5f13-4a50-91bf-146b18ce9a6d", 41 | "metadata": {}, 42 | "outputs": [], 43 | "source": [ 44 | "class MaskedConv2d(nn.Conv2d):\n", 45 | " \"\"\"\n", 46 | " Implements a conv2d with mask applied on its weights.\n", 47 | " \n", 48 | " Args:\n", 49 | " mask_type (str): the mask type, 'A' or 'B'.\n", 50 | " in_channels (int) – Number of channels in the input image.\n", 51 | " out_channels (int) – Number of channels produced by the convolution.\n", 52 | " kernel_size (int or tuple) – Size of the convolving kernel\n", 53 | " \"\"\"\n", 54 | " \n", 55 | " def __init__(self, mask_type, in_channels, out_channels, kernel_size, **kwargs):\n", 56 | " super().__init__(in_channels, out_channels, kernel_size, **kwargs)\n", 57 | " self.mask_type = mask_type\n", 58 | " \n", 59 | " if isinstance(kernel_size, int):\n", 60 | " kernel_size = (kernel_size, kernel_size)\n", 61 | " mask = torch.zeros(kernel_size)\n", 62 | " mask[:kernel_size[0]//2, :] = 1.0\n", 63 | " mask[kernel_size[0]//2, :kernel_size[1]//2] = 1.0\n", 64 | " if self.mask_type == \"B\":\n", 65 | " mask[kernel_size[0]//2, kernel_size[1]//2] = 1.0\n", 66 | " self.register_buffer('mask', mask[None, None])\n", 67 | " \n", 68 | " def forward(self, x):\n", 69 | " self.weight.data *= self.mask # mask weights\n", 70 | " return super().forward(x)\n", 71 | " \n", 72 | "\n", 73 | "class ResidualBlock(nn.Module):\n", 74 | " \"\"\"\n", 75 | " Residual Block: conv1x1 -> conv3x3 -> conv1x1\n", 76 | " \"\"\"\n", 77 | " \n", 78 | " def __init__(self, in_channels):\n", 79 | " super().__init__()\n", 80 | " \n", 81 | " self.conv1 = nn.Sequential(\n", 82 | " nn.Conv2d(in_channels, in_channels // 2, 1),\n", 83 | " nn.ReLU(inplace=True)\n", 84 | " )\n", 85 | " # masked conv2d\n", 86 | " self.conv2 = nn.Sequential(\n", 87 | " MaskedConv2d(\"B\", in_channels // 2, in_channels // 2, 3, padding=1),\n", 88 | " nn.ReLU(inplace=True)\n", 89 | " )\n", 90 | " self.conv3 = nn.Sequential(\n", 91 | " nn.Conv2d(in_channels // 2, in_channels, 1),\n", 92 | " nn.ReLU(inplace=True)\n", 93 | " )\n", 94 | " \n", 95 | " def forward(self, x):\n", 96 | " inputs = x\n", 97 | " x = self.conv1(x)\n", 98 | " x = self.conv2(x)\n", 99 | " x = self.conv3(x)\n", 100 | " return inputs + x" 101 | ] 102 | }, 103 | { 104 | "cell_type": "code", 105 | "execution_count": 3, 106 | "id": "56c8d21a-232c-4558-9ac9-da658f576572", 107 | "metadata": {}, 108 | "outputs": [], 109 | "source": [ 110 | "class PixelCNN(nn.Module):\n", 111 | " \"\"\"\n", 112 | " PixelCNN model\n", 113 | " \"\"\"\n", 114 | " \n", 115 | " def __init__(self, in_channels=1, channels=128, out_channels=1, n_residual_blocks=5):\n", 116 | " super().__init__()\n", 117 | " \n", 118 | " # we use maskedconv \"A\" for the first layer\n", 119 | " self.stem = nn.Sequential(\n", 120 | " MaskedConv2d(\"A\", in_channels, channels, 7, padding=3),\n", 121 | " nn.ReLU(inplace=True)\n", 122 | " )\n", 123 | " self.res_blocks = nn.Sequential(\n", 124 | " *[ResidualBlock(channels) for _ in range(n_residual_blocks)]\n", 125 | " )\n", 126 | " self.head = nn.Sequential(\n", 127 | " MaskedConv2d(\"B\", channels, channels, 3, padding=1),\n", 128 | " nn.ReLU(inplace=True),\n", 129 | " MaskedConv2d(\"B\", channels, channels, 3, padding=1),\n", 130 | " nn.ReLU(inplace=True),\n", 131 | " nn.Conv2d(channels, out_channels, 1)\n", 132 | " )\n", 133 | " \n", 134 | " def forward(self, x):\n", 135 | " x = self.stem(x)\n", 136 | " x = self.res_blocks(x)\n", 137 | " x = self.head(x)\n", 138 | " return x" 139 | ] 140 | }, 141 | { 142 | "cell_type": "code", 143 | "execution_count": 4, 144 | "id": "f3d8d7da-625d-4506-8db7-5f582bf42985", 145 | "metadata": {}, 146 | "outputs": [], 147 | "source": [ 148 | "image_size = 28\n", 149 | "in_channels = 1\n", 150 | "out_channels = 1\n", 151 | "channels = 128 # hidden channels\n", 152 | "n_residual_blocks = 5\n", 153 | "\n", 154 | "batch_size = 128\n", 155 | "epochs = 50\n", 156 | "\n", 157 | "transform=transforms.Compose([\n", 158 | " transforms.ToTensor()\n", 159 | "])\n", 160 | "\n", 161 | "dataset1 = datasets.MNIST('/data', train=True, download=True,\n", 162 | " transform=transform)\n", 163 | "dataset2 = datasets.MNIST('/data', train=False,\n", 164 | " transform=transform)\n", 165 | "train_loader = torch.utils.data.DataLoader(dataset1, batch_size=batch_size, shuffle=True)\n", 166 | "test_loader = torch.utils.data.DataLoader(dataset2, batch_size=batch_size)\n", 167 | "\n", 168 | "model = PixelCNN(in_channels, channels, out_channels, n_residual_blocks).cuda()\n", 169 | "optimizer = torch.optim.Adam(model.parameters(), lr=5e-4)" 170 | ] 171 | }, 172 | { 173 | "cell_type": "code", 174 | "execution_count": 5, 175 | "id": "7158b9d0-9dd1-4060-8d7e-dd6f0f185d4e", 176 | "metadata": {}, 177 | "outputs": [ 178 | { 179 | "name": "stdout", 180 | "output_type": "stream", 181 | "text": [ 182 | "Start training epoch 0\n", 183 | "\t [468/469]: loss 0.09943267703056335\n", 184 | "Start training epoch 1\n", 185 | "\t [468/469]: loss 0.08993204683065414\n", 186 | "Start training epoch 2\n", 187 | "\t [468/469]: loss 0.08632215857505798\n", 188 | "Start training epoch 3\n", 189 | "\t [468/469]: loss 0.0857618898153305\n", 190 | "Start training epoch 4\n", 191 | "\t [468/469]: loss 0.08323406428098679\n", 192 | "Start training epoch 5\n", 193 | "\t [468/469]: loss 0.08515626192092896\n", 194 | "Start training epoch 6\n", 195 | "\t [468/469]: loss 0.09080298990011215\n", 196 | "Start training epoch 7\n", 197 | "\t [468/469]: loss 0.0851127877831459\n", 198 | "Start training epoch 8\n", 199 | "\t [468/469]: loss 0.08575288951396942\n", 200 | "Start training epoch 9\n", 201 | "\t [468/469]: loss 0.08795469254255295\n", 202 | "Start training epoch 10\n", 203 | "\t [468/469]: loss 0.08480170369148254\n", 204 | "Start training epoch 11\n", 205 | "\t [468/469]: loss 0.08301049470901489\n", 206 | "Start training epoch 12\n", 207 | "\t [468/469]: loss 0.08615297824144363\n", 208 | "Start training epoch 13\n", 209 | "\t [468/469]: loss 0.08480261266231537\n", 210 | "Start training epoch 14\n", 211 | "\t [468/469]: loss 0.08743501454591751\n", 212 | "Start training epoch 15\n", 213 | "\t [468/469]: loss 0.08629097044467926\n", 214 | "Start training epoch 16\n", 215 | "\t [468/469]: loss 0.08073026686906815\n", 216 | "Start training epoch 17\n", 217 | "\t [468/469]: loss 0.08512020856142044\n", 218 | "Start training epoch 18\n", 219 | "\t [468/469]: loss 0.08142197132110596\n", 220 | "Start training epoch 19\n", 221 | "\t [468/469]: loss 0.0807342529296875\n", 222 | "Start training epoch 20\n", 223 | "\t [468/469]: loss 0.08460844308137894\n", 224 | "Start training epoch 21\n", 225 | "\t [468/469]: loss 0.08269714564085007\n", 226 | "Start training epoch 22\n", 227 | "\t [468/469]: loss 0.08292073011398315\n", 228 | "Start training epoch 23\n", 229 | "\t [468/469]: loss 0.08164182305335999\n", 230 | "Start training epoch 24\n", 231 | "\t [468/469]: loss 0.08252546936273575\n", 232 | "Start training epoch 25\n", 233 | "\t [468/469]: loss 0.08031551539897919\n", 234 | "Start training epoch 26\n", 235 | "\t [468/469]: loss 0.07918865978717804\n", 236 | "Start training epoch 27\n", 237 | "\t [468/469]: loss 0.08197366446256638\n", 238 | "Start training epoch 28\n", 239 | "\t [468/469]: loss 0.08056250959634781\n", 240 | "Start training epoch 29\n", 241 | "\t [468/469]: loss 0.0826050266623497\n", 242 | "Start training epoch 30\n", 243 | "\t [468/469]: loss 0.08208416402339935\n", 244 | "Start training epoch 31\n", 245 | "\t [468/469]: loss 0.08351639658212662\n", 246 | "Start training epoch 32\n", 247 | "\t [468/469]: loss 0.07901784777641296\n", 248 | "Start training epoch 33\n", 249 | "\t [468/469]: loss 0.08264601230621338\n", 250 | "Start training epoch 34\n", 251 | "\t [468/469]: loss 0.0835525318980217\n", 252 | "Start training epoch 35\n", 253 | "\t [468/469]: loss 0.08255098015069962\n", 254 | "Start training epoch 36\n", 255 | "\t [468/469]: loss 0.0846540629863739\n", 256 | "Start training epoch 37\n", 257 | "\t [468/469]: loss 0.08017908781766891\n", 258 | "Start training epoch 38\n", 259 | "\t [468/469]: loss 0.0825449600815773\n", 260 | "Start training epoch 39\n", 261 | "\t [468/469]: loss 0.08127618581056595\n", 262 | "Start training epoch 40\n", 263 | "\t [468/469]: loss 0.07939404249191284\n", 264 | "Start training epoch 41\n", 265 | "\t [468/469]: loss 0.08136747777462006\n", 266 | "Start training epoch 42\n", 267 | "\t [468/469]: loss 0.08235418796539307\n", 268 | "Start training epoch 43\n", 269 | "\t [468/469]: loss 0.0813743844628334\n", 270 | "Start training epoch 44\n", 271 | "\t [468/469]: loss 0.08264637738466263\n", 272 | "Start training epoch 45\n", 273 | "\t [468/469]: loss 0.07891112565994263\n", 274 | "Start training epoch 46\n", 275 | "\t [468/469]: loss 0.08211299031972885\n", 276 | "Start training epoch 47\n", 277 | "\t [468/469]: loss 0.08249464631080627\n", 278 | "Start training epoch 48\n", 279 | "\t [468/469]: loss 0.07970467209815979\n", 280 | "Start training epoch 49\n", 281 | "\t [468/469]: loss 0.08078432828187943\n" 282 | ] 283 | } 284 | ], 285 | "source": [ 286 | "print_freq = 500\n", 287 | "for epoch in range(epochs):\n", 288 | " print(\"Start training epoch {}\".format(epoch,))\n", 289 | " for i, (images, labels) in enumerate(train_loader):\n", 290 | " images = (images > 0.33).float() # convert to 0, 1\n", 291 | " images = images.cuda()\n", 292 | " logits = model(images)\n", 293 | " loss = F.binary_cross_entropy_with_logits(logits, images)\n", 294 | " optimizer.zero_grad()\n", 295 | " loss.backward()\n", 296 | " optimizer.step()\n", 297 | " if (i + 1) % print_freq == 0 or (i + 1) == len(train_loader):\n", 298 | " print(\"\\t [{}/{}]: loss {}\".format(i, len(train_loader), loss.item())) " 299 | ] 300 | }, 301 | { 302 | "cell_type": "code", 303 | "execution_count": 6, 304 | "id": "6d148f2e-25b4-461e-bc9c-79769f2e57b0", 305 | "metadata": {}, 306 | "outputs": [ 307 | { 308 | "data": { 309 | "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkgAAAJICAYAAACaHhuvAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/MnkTPAAAACXBIWXMAAAsTAAALEwEAmpwYAAEAAElEQVR4nOydeVxc5fX/P3dWmIUBZoAZGJawBAJJSCBNAmoSE6PRNgGX1Ma4VOvWr8bEaqxxqYk/E7Uaa03UqlVb19pYs9mYmFXNRhZMCBASCIFhmWGHYQZmv78/8p37ZWAYBpgVnvfrdf/g3jv3Hs59lvM8zznnoWiaBoFAIBAIBALh/2D5WwACgUAgEAiEQIMYSAQCgUAgEAj9IAYSgUAgEAgEQj+IgUQgEAgEAoHQD2IgEQgEAoFAIPSDGEgEAoFAIBAI/eC4ukhRFMkBAICmacrZeaKfKxD9uIboxzVEP64h+nEN0Y9riH5cM5h+ADKDRCAQCAQCgTAAYiARCAQCgUAg9IMYSAQCgUAgEAj9cOmDRCAQCAQCwbOEhYUhPz8fUVFRzDmVSoWioiIYDAY/SjZ8FAoFZs+eDZFINOBaS0sLjh49Cq1W6wfJRg8xkAgEAoFA8CFyuRyrVq3CrFmzmHM7d+5EeXl50BlIGRkZePHFF6FUKgdcO378OKqqqoiB5AyZTAaZTOb0Wnt7O1paWkA2yyUQCMFCWFgY5HI5WKyB3gnd3d3QaDSwWq1+kIwQqFAUhZiYGISHhzPnUlJSBpwTiUROy1Wgw+VyERYW5vC/2AnW/8mO1wwkFouFwsJC3HPPPeBwHF9D0zS++OILvPfeezCbzd4SgUAgEDzKVVddhccffxxisXjAtUOHDuG1115De3u7HyQjBCohISG4++67cfPNNzPnQkNDkZKS4kepCO7gNQOJoijEx8cjLy8PbDbb4RpN0zh8+DAoatD0AwQCwQ+wWCxwuVyHcxaLZdzPinA4HLDZbMjlcvziF79wOlpuamqCSCSCXq9365lWqxUWi8XDkhICBYqiwOVyIRAIkJaWhtmzZ3vkuVwuFywWi5QfH0B8kAgEAkNOTg4KCgogFAoBAGazGbt27cIPP/zgZ8n8B4/Hw5IlS5Cfn49JkyYhNDTU6X1ZWVl45pln0NPTM+QzaZrGoUOHsGvXrnFvfI5VYmNj8Zvf/AZJSUkOvkajQSwW47bbbsPkyZNJ+fEBxEAiEAgMkyZNwiOPPMLMkBiNRmg0Gvz444/j1l+Qx+Phuuuuw4MPPuhy1js1NdXtZROapmG1WrFnzx7SwY1RoqKicOeddyI7O9tjqyVCoRAFBQVYvHgxKT8+wOMGkkAgQE5ODuRyOSZNmhQwy2jR0dHIyclxGP1VVlaivLwcNpvNj5IR/AmLxUJmZibS0tIGXKNpGhUVFbhw4cK4Mg4oimLqbaDU30DAHV24qy+KojBx4kQUFhZCo9GguLgYOp1utCIS/ASXy0V2djbi4+OZc8nJyQgPD3dZJmw2G8rKylBVVYXjx48PGsFm778UCgViY2P9Xi8pikJ6ejoyMjIwffp0CAQCv8rjLTxuIEmlUqxcuRJz5syBQCAIGA/2zMxMbNiwAXFxcQCudH7vvPMOLly4QAykcQyHw8HSpUvx0EMPDWh0rFYrNm7ciMrKSjJKI3ichQsXIj8/H0VFRVi1ahUxkIIYgUCA++67D7feeitzjsPhICwszOXvLBYLvv76a7z33nswGAzo7u52ep+9/4qPjx/ymb6AxWJhyZIlWLVqFUJDQwNCJm/gcQOJzWYjIiIC0dHRA65ZrVa0tbVBr9ejo6PDa6NyDocDqVTqMFsUHx+PmJgYRi6apiESiXxuiTuTrbe3F62trX7phMPDwyGRSBg9WK1WtLa2ore31+ey+AKhUAipVMoY7jweD3FxcYiOjnZqIMXFxSEpKQk6nc5v38ifUBSFyMhIJCYmQq/Xo62tbdgDCj6fD5lMBhaLhfb2dredmP0Nj8eDTCZDeHi406i10SIQCCAQCBAeHj4gkMUXjLe67w3sZVsmkyE2NtZpv+cMs9mMtrY2aLVaNDQ0oLm52Wl/aP9G9v7LnjYnEGa0zWYzenp6wGKxAkIeb+BTHyStVovNmzfj6NGjqK2t9ZoHvlQqxRNPPIGcnBzmXGRkJKRSqVfeNxycyXbq1Cm88cYbaG5u9qksLBYLBQUFuOOOO5gGur29HW+++SaOHj3qU1l8RX5+Ph555BEm6yuLxUJKSopTQ9mun+zsbJw8edIv38jfcLlc3H777cjLy8Phw4fx17/+FR0dHcN6RmpqKp588kmEh4fjrbfewsGDB70krWdJSEjA6tWrMXHiRKSmpvpbHI8yHuu+N7CX7QkTJiA9Pd3t36nVamzcuBHnzp3DpUuXnBoYfb9RVFRUQPRfdmw2G7Zv344zZ85g5syZ+MMf/uC2cRhMeMxAoigKLBYLbDZ70FkZo9GIkpIS7N+/31OvdSqDSCRCbm4u5s+f7/Q+m80Gq9Xqk6U1FovloI++stE0DZvNBpPJBD6fP+gz7A2YzWbzqKVOURSSk5OxYMEC5h0ajQaff/65x97hC4Yz+o6Li8PcuXOdhmn3x66f5ORkWCwWhISEjEJKz2Mv786wly1PvCMlJQXJycno6OgAj8cb9jPCw8OZbRX+/e9/j1omXyESiTBz5kxMmzbNK8+3fyN/LPGPlbrvb+xle+LEiU6vDzbjrNVqceLECRw/ftzpdXu6jdTUVIdvBPi2/xoMmqZRXV2N6upqcLlcr2T/7t93eqpNGw4eM5DS09OxePFiKJVKTJgwwVOP9bgMOp0OO3fuxPnz53Hs2DGvLpmw2Wxcd911yMvLYz50eHg4JkyYwIT5/vjjj6isrERXV5fTZyQkJKCwsBACgQDffvstSktLvSZvMGLXj7ujq6lTpwacoTMSKIrCvHnzMGfOnAFGEk3TOHr0KPbv3z/q8m21WrFv3z4cO3YM5eXlxE/Ggxw5cgQHDhxATU0N2tra/C0OwcNoNBps3boVGo1mwLWWlhbU19c7/V14eDgKCgqQmpo6oH77sv/yJzKZDIWFhQ5O7+fPn8fOnTt9ukTvMQMpIyMDq1atGjQNvy9wRwa9Xo8tW7Zgx44dHp+R6Q+Hw8HChQuxcuVKh6gge5KvH3/8ERs2bIDFYhnUMlYqlXj44YcRFRWF2tpaYiD1w66fwUZw/ek/KglWWCwW5s6dizVr1jhNxPrGG2/g0KFDo25ALRYL9u7di7/+9a9+m+0Yi9A0jaKiIrzyyiswGo1Er2MQjUaDv//97zh79uyAa65mQyQSCZYvX4758+cPaK982X/5E5lMhnvuuQd5eXnMua1bt+LgwYPBYyCxWCykpaUhMTGRCfXztXHkrgxtbW0oKytDY2Mj1Gq1Vy1voVCIrKwsREVFYcKECeBwOGhubkZ5eTlMJhOAKyPzyspKl8YRcMWgYrPZCAkJQXZ2Ntra2qBSqUYVWRUSEoKsrCxER0cjOTk5aAwGHo+HzMxMh7XurKwsiMXiYTu56nQ6lJeXo7OzkzkXExODSZMmjWgZyVf0/XYpKSngcDhOZ5CSk5OxcOFCtLS0oLS01K3khYNhn9IfbmOclJSE1NRUTJkyZcyGAfelfx13hT2FhNlsDnrjiM1mIz09HUqlEo2NjaioqBjzGZ5dle36+npcuHABFy5cQGdnp9vtdHR0NDIzMxEfHw+ZTObQpvmy/3KHvu4Hubm5gyZPHS4KhQIZGRlISUlBZGSkgw7kcjnmzZuHhoYGlJeX+2RLn1EZSBwOB7/+9a/xu9/9DkKh0C+hfu7KUFZWhj/+8Y+oq6vzumIVCgWefvppzJgxg/F1KS4uxrPPPovW1lYAVxrIrq4utxtHgUCA+++/H8uWLcNnn32Gl156acTRJjKZDE888QSuvvpqSCSSgEnFMBQSiQSPPvooFi5cyJzj8/kjcl7UaDR45ZVXcPr0aebcokWLsH79+kE3WA4E3Pl2FEXhhhtuwKxZs3Dy5EmsXr0a1dXVPpXTLsPq1ashEokglUoHDWEeK/Sv466gaRrd3d1jwpDg8/m46667cMcdd+Cbb77BCy+8ELS7t7uDs7Ldl0OHDuGll15CV1fXsJZOc3JysH79esjl8gHP9GX/5Q72MP8VK1ZAJBIhMjLSI8+dPXs2XnzxRUil0gE6yM3Nxeuvvw6VSoWnnnoKR44c8cg7XTEqA4miKERERCAhIcHlLITJZEJXVxeam5s97szlTAar1QqtVguj0cjc19jYiLq6OjQ0NHj0/c7gcDiIiYlBfHw8s8O3/f0tLS0jeiaLxWIKTWRk5KhmfTgcDqKjox3WdwMRkUgEoVDI/K/R0dFQKpVISEhw6/cWiwVdXV1ON0RWq9Woq6uDSqWCWCyGUCiE2WwO2ClrPp8PiUQCuVwOpVI55LcTi8UQi8Wora0dsFm0HQ6HA4lE4rD3Wv/EdhRFMTvYD0c3FEUhLi6O+VadnZ1obW31ijOnu7BYLISFhTn4oNnbpr4jcnteF5lMNqju+tLd3Q29Xj/qOh6ssFgspg3um0JjLGN3ouZyuQPaYp1Oh7q6Ordnbe3tj0KhQHx8PKKiohye1bds+aL/cpfw8HAkJCSMOkUFRVEQi8UQCARM2yaRSAbcFxoaCqVSCZqmERsbC7lczlwzGo3DmnBwF5+E+ZeXl2Pz5s2or693uh7radra2rB582aUlJQw55qamnxueZvNZma9uKGhYUyPqjwNi8XC4sWLsXTpUqbBtS8zuotKpcKmTZtw+fLlAde0Wi0uXboELpeL2267DQUFBYiNjQ3YhGfTp0/Hww8/jNjYWEyaNMkjz0xISMCKFSscAhri4+Mdlgw4HA5uu+025OTkDNtASk1NBYfDQUVFBTZt2oSamhqcO3fOI7KPhLCwMPz+97932Bfr/Pnz2LRpExobG5lzc+fOxb333ouYmJghjXFSx8cfNE1j7969qK+vR1ZWFlasWIHY2NgRPctV+2Oz2bBz505s2bIFarU6IGaOvAGfz8fy5ctxww03DGh/nCGTybBq1SosX76cOXfmzBls3rzZrdnb4eATA6m5uRl79+6FSqXy6HMHC3Pu6elhonh8jV0miqJgs9lQWlqKHTt2jGhmou+WD+MNiqKQkZGBJUuWDDlCGUy3nZ2dOHjwoEujnM/nY8qUKViyZElA61qhUOCGG25ATEyM23K6KnMURSE8PBzXXnutS6OTxWJh8uTJmDx58rBltsvQ1taG/fv34+LFiyN6hqcICQnBrFmzUFBQwJyTSqX4+OOPHe5LSkrCr371qyEbapqmYbFYRlXHgx17GxXIdccbVFVVoaqqCu3t7fjtb3/r8O3dLQcURYHD4Thtf+x79Z0/fx47duzwu8+Rt6AoClwuF9OmTXOol64IDQ1Ffn7+gHMfffSRx+UL2s1qIyMjccMNNyApKQkzZswIiAqamZmJBQsWICEhYcQjCjszZszANddcg9TUVI+t745F6uvrsWfPHqejq8bGxjGz3FFZWYl3330XcXFxWLRo0ZAzGydPnsRPP/3ENOJ9yc7OxrXXXovExESH6XxP40qGYKe9vR27d+9GbW0tTp48OS6No+joaNx4441QKpXDmtkdSzQ0NOCjjz5y8Fs8duyY02X9vrjqv7RaLXbv3o3Lly/j2LFjQe/EPxjBUH6C1kCKiorCQw89hLy8PLf8BHzB9OnT8dxzzyEiIgIcDsetaBZnUBSFa665Bi+++CJCQkIC5v8LRFQqFd58802nsxP2Ef5YoLy8HBcuXEBiYiLS0tJcGkg0TeOnn37CCy+8AIPBMEAHM2fOxJ/+9CeIRCKvla2hZAh2Wlpa8P777+PYsWNj7n9zl9jYWDzyyCPIzs4et21UbW0t/vKXvzgYOFardcgZH1f9V0dHBz7++GMcOHBgRNGjwUIwlJ/AlMoN7FNzfUOy29racOnSJahUKofwbV/BYrHA4/FgsVhQUVGB1tZWNDY2jqiAczgc8Hg8puBYLBZUV1ejubkZly5dGrOjCjs0TUOlUuHIkSMunT7PnTuHrq6uYRujoaGhSElJQVRUlMvdsVUqFerr61FWVubg9O9r7DmIdDodSkpKBjhWT5w40aEuWK1WmEwmpvPmcDhITk5GVFQUJk6ciNDQUIdnuIO9ftlTa/R1pOzs7ERVVZWDE7b9b38bEOHh4UhNTUVsbCwTGVNfXw+VSoVz584NiAbVaDQ4duwYpFIpUlNTmW1p+kLTNMxm84gHQWOF/m1wVFQUZs+ejebmZlRWVo75yEV7OXCXqKgoJjxeKpUO2n+1tLQEZNni8/lITU2FVCpFfHz8qFdu+pef4eLt9jloDSRnFBcXY+3atWhsbPTrnllqtRobNmzAqVOnPJYht6enB++//z527tyJzs5Ov3bWvsBms2Hbtm346aefXN7X29vrNFPtUCgUCjz77LPIzc0dNKzfarVi27ZtePfdd9Hd3R0Q2Y5bW1vx+uuvO/jHXHXVVVi/fj0UCsWgvxMIBHjwwQexePFihIeHu9zaZjDs9YvP52P9+vUOSdxKS0vx/PPPOzg7t7e3+904Aq7kynrppZeQkJCA6Oho0DSNPXv24M0330RXV9eA8nPw4EGUlpZi8uTJePnll5GRkeEnyYOP/Px8pKamorKyEs888wzOnDnjb5ECivz8fDz33HOQyWQOUVhA4PRfrpDJZHjyySeRn5/v94hFX7TPQWcg8Xg8CIVChIeHD5iW0+l0uHTpEpqamvwim8lkQkdHB5qbm1FdXY2qqiqPPdtqtaKxsdHvjq6+pK2tzeOF3l5+7Eky09LShpRhNEk5PY3FYhmwRUFSUtKgo1gOhwOhUAipVIrExES3M47bZ6v6GjhqtRqVlZUICQmBRqNx8CtqampCZWVlQIUh2xEKhUhOTmaWJWmaRnt7Oy5evOh0lN7V1YWuri6IRKJB0xKw2WyEhYU59Q+0Wq3Q6/UBYRwOB/s+lpGRkejt7R1RnrWwsDCEhYXBYrGMi+SgwyUsLAypqakOe0H29PTAYDAw9SsQ/Sa5XC6EQiEzAzZUO2IwGNDb24vu7u4RtZ329sdqtSI0NNQhNYfNZoNer4fBYEB9fb1X2+egM5BmzpyJu+++GwqFAsnJyf4Wx4GioiI89dRTMBgMPk/MR3CPQC4/3iA5ORkPPvggEhMTMWPGDLd/p9Fo8MEHH+DChQvMubq6OnR3d6OnpwebNm3Cli1bHO7v6OjwqOyBTExMDB5//HGnYcUtLS344IMPUFZW5gfJRk5YWBgefPBBLFq0CLt27cKWLVuCzsgLNiwWC7755hvG4T9QlySzs7Nx3333ITY21q1B1r59+/D111+PeDbM3v5cvnwZt99+O2688UbmWnt7Oz744AOcO3cO586d86q7yagNJF87kCUnJ2Pp0qVOd2P3tzNbTU0Nampq/CoDwTWuys9Ywu4bEBUVhcWLF7s9c2Snq6sL33//PY4ePer0+sGDB0ctYzATFhaG66+/3um1mpoa7Ny5M+gMpJCQEMyZMwc0TUOj0eA///nPkL8ZzAfF321xoNJfX1arFcXFxfjiiy8CWmfx8fG45ZZbEBMTM+S9NE3j/Pnz+PLLL4f0oxqs/Gi1Wnz//fc4ffo0srOzsWjRIuZenU6H/fv3+ySNz6gMJKvVimPHjiE0NBTJycmYM2eOx/ZkAa6sd86bN8/BR2T27NkO/hNWqxVFRUUoKSnBzz//PKo9pwhjC3fKT7BCURRyc3ORm5vL7EfX91p2djYzczSUMWg2m3Hs2DGUl5cz5zQazYh8u8YSbW1t+Prrr3H69Gnk5eUhMzPTrd+FhYVh8eLFSE1NxalTp1BcXBzQnV9/+paf+vp6HDx4EF1dXcz1zMxM5OfnY8KECQO2g7h48SKOHDmC2tracVV++Hw+rr76aqSlpaG0tBTHjx9nZt+4XC5TfsZK+zMaXJUfO+Hh4bj55puRm5uLqVOngqIoXL58GT/++KNPM4qPykCyWCzYsWMHdu3ahYKCAkyfPt2jBlJcXByeeOIJTJ06lTlnj+7qL8Pbb78Ns9k85p2XCe7jTvkJVlgsFhYtWoTVq1eDz+cP+J/mzZuHq666CiwWa8gG2Wg04uuvv8ZHH33EdOQ0TY/7utTQ0ICNGzciLCwML7/8stsGUkREBB566CGYTCa88sorOHPmTMD4sLmLvfwcO3YMpaWlDgZSfn4+NmzYALFYPKDcnT59Gs899xw6OjrGVfkJDQ3F8uXLsXTpUnz44YcoLi5mDCQej4fbbrsN9957L3g83phof0aDq/JjJyYmBo8++ihsNhtzz7lz57B27Vo0NTX5rGyNeonNZDLBZDLBYDCMapTEZrMRHx/v4PSYnp6OyMhIl85+FEUhOjoa6enp6OjogEqlIuvmgxAeHg6lUomkpCSHGQd/Yd+vKzo6Gp2dnSP6dkKhEImJiQ5OfHbcKT/OMJlMqK2tRWdnJ9RqdUCN/nk8HhISEph9kEQikdNIEvs+Ua7o7e1FTU0N2traUF9fD71e7y2xgxKbzQaDwQCKolBVVYXi4mJIpVIolUqX2d0pikJISAi4XC4SEhIwffp0xk/CZDKhrq7OweDwF2q1GsXFxYiIiEBiYqJDebGXHz6fP6B82R3/ndU5i8WCnp6eEW+kHWzY25/o6GgoFAqIRCIolUpMnz6d0YFAIEBcXJxDugiDwYDa2lomqCeQ2hhv46r82LHXob6IxWKkpaVBJBL5zF8rYJy0hUIhHnzwQQdnLIFAMGTGYC6Xi9/85jdYsGAB9u/fj/Xr14+5rL2ewp7I0r4por/hcrm4/fbbceedd+LAgQMj+nZpaWlYu3YtEhMTB1xzp/w4wx5KX1RUBI1GE1A5p2QyGVavXo1Zs2ZBLpePKsy2trYW69atQ1lZGdRqtQelHFsYjUZ8+umn+O677/CrX/0KTz31lFsDDDabjcLCQuTl5TEdYFtbG1566SUcOnTIy1K7xh4ifezYMeTn5+NPf/rTgLBzwtDY25+UlBTExcUBuDL7NnHiRGbWkM1mD9hZQa1W4+WXX8bPP/9M6p6b5OTk4I033oBKpcK6detw4sQJr7/TbwaSPUGU3fEqLCwMKSkpmDZt2rCeQ1EUYmNjERsbC5VKhfDwcBgMBhiNRodpbTabDT6fD4qiYDAYYLVaB8hgX1bw1wwURVHg8/lOk2fZLeq+oxCr1Qqj0ThkB26f1pXL5Zg8eTKio6MH3GN/li/Dk+0zSNnZ2airqxv027lCKBRi0qRJw3ZCdoXJZMKlS5d8srHycOHxeEhJSRlVan77UnRrayvKyspQWlrqQQmDAx6PB5FIhN7e3kFnv+3Lk2w2G52dnejs7ERTU9OwDGa5XO5geDQ1NSEmJgYikYiZffcXdj+zmJiYgExKGAw4a3+kUumgvjV2zGYz6uvrcenSpXGje3vfNlIfLJFIhKSkJLBYLJ+lkPCbgXTttdeisLCQyWXE5/ORk5MzqmdOmTIF69atQ319Pf75z3+ioqKCuZaeno67774bbDYbn332Gc6ePTtAhp6eHnzxxRc+sUydIRaLsXz5cuTk5GDKlCkOeZ4EAgHuuusuh+R8KpUK//jHPwbkxekLh8NBQUEBrrvuOkyYMGHQke+FCxfwySefoL6+3i8dpqtvR/AsR48exddff42GhoZxOXqlKArXXnstxGIxLly4gH/+859Oc6fFxcXht7/9rcNsa1pa2qj8LMPCwnDfffdh/vz5+P7777Ft27ag808ijJ7o6Gg89thjuPXWW/HNN99g3759/hbJqwgEAixbtgyzZs3CpEmTRuSHVVJSgs8//xwNDQ0+ywfoFwOJoihMnToVv/vd7zzqsJaUlISkpCTU1NRg3759Dp2sUqnEsmXLwOPx8NNPP6GkpGSADB0dHSgqKvKbgSQQCHDDDTc43dWYx+Ph2muvxbXXXsuc+/nnn/Htt9+6NJDYbDby8vLwwAMPuEwLX19fjy+//BIqlWp0/8QIcfXtCJ7l/Pnz+Pjjj8e1z9HUqVMxdepUHD58GNu2bXNqIEmlUtxyyy3DntV2RWhoKK6//nrQNI2uri7s3LmTGEjjkPDwcCxZsgRGoxGVlZXYv3//mPZD4vP5mD9/Pu64444RP6OmpgafffaZTxNBe8xAUqlU2LJlC2JjY5Gfn+8wrRwbG4tbbrmFyYpsD1H2VppykUiEhQsXOqz7TpkyBUKhEBwOB9deey3Cw8O9KoMviIyMxK9+9StMmTJl0Hs4HA4mTZo06j1zApWmpiZs27bN5TYbzpBIJMjLy3PYzV6lUqGoqAh1dXVBObNiMBhw/Phx1NXVISsrC9OmTWPKt8lkwokTJ1BdXY3jx48Pa/+oYKexsRHffPMNlEolZs+eDaVSyVyLiYlBYWEhamtrUVRU5LM8ZhRFISsrC8uXL4darcbRo0eh1Wp98u7+uNLPWIbFYiEnJweTJk1i6v5gmdO9BZvNxi9+8QvcddddAwwkmqZRUlKCc+fOBZXx1Lds2901hEIhJkyY4PYztFotjh496pBV/NSpUz53/veYgVRSUoLnnnsOiYmJeOuttxwMpIyMDKxbt85h7Z7P53ttB9/IyEg88sgjDr40ds95ALj//vthsVi8KoMvUCqVeOqpp4YcgXoy9UKgcfnyZaxfv37Yhu7EiROxadMmBwPJXoY1Gk1Q5tPSarV4//33sXv3bqxYsQKTJ09mZkd7enrwySefYMuWLTAajePG7wEAKioq8MILL0ChUOAvf/mLgwGQnJyMZ599Fu3t7Vi9erVPE70uWLAA+fn5OH78OKqqqvxmILnSz1iGy+WioKAAjzzyCL799luUl5f73EDicDgoLCx0CE6yY7FY8Nprr6GsrCzoZhntZdsORVHD8hvSaDR48803UVRUxJwzm80+b5c9Zh2YzWZ0dXVBq9UOcPLlcDgICwvzyHu0Wu2gkUUikQgKhQJsNtvpDtx97wOupCxXqVSMdd7V1RVwqd57e3uhVqtddmgsFgvR0dEIDw93qR9XqFQqn88q0DSNpqYmVFRUQCwWQ6FQMGHsGRkZaG1tdbqVQ1+sVuuwOpawsDDI5XJERUUNWN7tW4aDEZvNhp6eHnR2dg4YadE0DZ1Oh87OTv8I50csFgu0Wi14PB4uXbqEiooKREZGIioqitlTDQAmTJjgsDHthAkTXIYiA1f02tLSMqrIWX9v6mvXj0AgGNDOCAQCpKSkALjSaQVr3egLj8eDQqFAeHg44uPjERERAYVCgfT0dISFhUGtVg/ZEdvLT0JCwoB2xO7M72zWRyKRICYmxmFAFxoa6nQQa7VaER8fj4yMDGi1WqjV6qBJYcPn80fkjG3vvy5evIimpia/t1dBN31y9OhRvPHGG04NmXnz5mH16tVON5DsD03T2LVrF9577z2m0FksloDbKqSyshLr16936RsUGhqKxx57DIWFhS7144rOzk6f71ZvNpvxr3/9Cz/88APz7aRSKVavXo22tjZ89NFH+Oc//+nRMPurrroKjz/+OKKjo5mGnzA+6OrqwubNm/H5559j+fLleOihh5jcP/Y0I7fccgtzv1AoHDJNhMViwb/+9S98+eWXI5aru7s7YLNOp6WlYcOGDWhubsbGjRvx/fff+1ukUaNQKPDMM88gOzub+b45OTl48803oVKpsGHDBpw6dWrQ31MUhRtvvBEPPfQQIiIiBqRHOHjwIDZt2uR0OejGG2/EH/7wB5cDeDv2NBEzZsxAUVERXn75Zb9txO4r7P1Xc3MzLl265G9xPG8g0TQNk8kEo9EIDofjMqGaO88ym80OlnhjYyNOnjzp1LKMiYmBTqdjltKGenZdXR1OnDgR0MsNWq0WZ86ccem1LxQKUV9fD6PR6FI/gQZN06itrUVtbS0UCgXMZjP4fD4mTZoEq9WKgwcPIiQkhJlepmkaFotlVAaTTCbDL37xi6Ddi61v/erPcNIjjEfMZjMuXLgAiqKQl5cHg8HAlCWKopCcnIzU1FSXz7DZbLBYLEybZDQacenSJRQVFQWVn4i7iMViTJs2DR0dHUOGrgcLoaGhmDx5MmbNmsWci4yMRGRkJKRSKSQSyZDPiI2NxaxZswbs6mC1WtHQ0ICioiKns1ApKSnQ6/UOSTld9ZNKpRJKpRJardavW5TYbDZmaZ7L5Xrcp9Wuu0DrvzxuIHV0dODDDz/EgQMHsHDhQixcuHDEymxoaMBXX33lsO/K+fPnB3XUKisrw4YNG9xe6zx58mTQTFm6wmQyYfv27aiurnapn2CCxWLhuuuug1gsZjoerVaLLVu2BN0moJ7EXr/27t074FpPTw/Onz/vB6mCC5qm8dNPP+H5559nljoEAgFuueWWIVONFBcXY9u2bUznZ7Vacfjw4TFpHBHcp7e3F//5z39QXFyMkpKSQQfdp0+fxrp165ilWy6Xi5tuuglz5871pbjDxt63KpVK3H777UhLS/PYs00mE3bs2IGjR48GXP/lcQNJq9XiP//5D7hcLsLCwrBgwYIRzyK1tLQwOYvsuGqIqqqqhjUtN1YaNbPZjH379o2pUFH7KH/27NnMOY1Gg+Li4nFtINnrl6td1MdqxKInOXXqFE6fPs38HRERgczMzCENpPPnz+Odd95xGOGOlTpHGDkGgwG7d+/GF1984bI8VFRU4MKFC8zffD4fcrkcc+bMCeh6a+9bExMTkZeX53EDad++fXj//fcDri55zQfJarWivLwcW7duHbGBVF1djc7OzmEpLdAU7EuC+X9vbGzErl27EBsbi5ycHCa6rG+jERoairy8PLDZbFRWVqK8vNzpcltSUhKmTp3qtNzNmDHDIXLRZrOhrKwMVVVVOH78uM+jWEZKMH/rQKGvDo1GI06ePDlkxOepU6dgMpnGpP7taSJsNhvS0tKQmZkZ1GlQBqO7uxs//PADWlpakJWV5bCsKhQKcfXVV0MsFjPGjP1b8/l8TJ8+HQqFApMnTx6gG5vN5la56HuP1WpFSUkJtm7dyrR1AoHAoQ0ErriPLFq0CHV1dThz5oxf0pDQNA29Xo/Dhw+ju7sbGRkZSE9P95hhF4h1ymsGks1mw86dO3HgwIERP8MeXUEY+5w5cwZPPfUUEhMT8Ze//MWhcbATFhaG3//+97j33nvxzjvv4MKFC04NpLlz52Lt2rVOl1pDQkIcfNQsFgu+/vprvPfeezAYDAEXxUjwDT09Pfjoo4/wxRdfuLzPYDAEZQoId9BqtXj33Xfx8ccf43/+53+Qnp4+Jg0kjUaD1157DSKRCM8++yxSUlKYTl4qlWLVqlXo7e3Fxo0bUVlZyfj1SSQS/P73v8eiRYsgEAg8kiLGbDbj66+/xq5du5hz8fHxA9rASZMmYcOGDVCr1Vi9erXf8rS1tbXhzTffRGhoKJ544gmkpaWNys840PFqFJterx/X2XqHi9VqRXNzs0MknUajCWgnck9hNBphNBoREhKC2tpaZsf6vg6TLBYLEokEYWFhiIuLQ2JiolMfMqVSiZiYGJezAWazGW1tbdBqtWhoaBhzO2p3dXWhtrYWYrEYMpnM3+IEPDRNQ6vVjusBmc1mY9Jc6HQ6h/pgTyWSlJQEqVQa0MtBQ2G1WtHR0QG9Xo/6+nrU1NRAJBJBKpWCxWIhPDwcYrEYcXFxSEpKYgwkmUwGhULhsJel0WhES0sL2traRmw463Q66HQ65m9n6RZ4PB6kUikTyOIvbDYbOjs70d3djYaGBtTU1LhlILHZbERERDiN3mOxWIiMjERSUhJ0Oh3a29sDZoPwoAvzH8vYw5C/+uor5lxnZyeam5v9KJVvaW1txcaNG/Hpp5/izjvvxPLlyweMYimKwqJFi5CWlua0IsXFxQ3ZiKjVamzcuBHnzp3DpUuXxpRxRNM0du/ejcrKSkydOhVPPvmkW5GdBMJgCIVC3H///Vi8eLFb9SsYsKcZOXbsGK6++mqsXLkSERERAK502gUFBcjOzmbaBh6Ph0mTJjk8o7KyEhs3bsTly5cdfIvGOjabDdu3b8eZM2fcmmUUi8VYsWIF5s+fP+Aan8/H3XffjQULFmDv3r14++23HQxGf0IMpADCZDKhpKTE32L4FYPBgDNnzjB7yJnNZoepbBaLBYqikJCQMGSOGmfQNA2bzQatVosTJ07g+PHjnhQ/YFCpVFCpVDCZTOjp6WG22WGz2YwOCAR34XA4mDx5MiZPnuz0ut3/JpjSTNA0jaqqKlRVVUEsFqO3txdhYWFMG5OcnIzk5GSXz9BqtTh+/DgqKysBwCPLTWw2e8AMnb3OjjbNiaegaRrV1dWorq526/6IiAgsXbrU6TU2m42MjAxkZGSgsbExoHa3CBxJCIQ+2Gw2HDp0CFarlRmhiEQiLFmyBOnp6SN+bklJCXbt2oW6ujqXm/yONUJDQ3HzzTcjNTUVJ0+exJ49e8bVfmwE72EymbBnzx6cOnUKJSUlQRPo0JeKigq8/vrriI+PR0FBwZCGkZ34+Hg88sgjo8qk3p/w8PAB+5ZVV1dj+/btqK+v99lO9gRiIBECFJqm8eOPP+Lw4cPMOblcjvT09FEZSOfOncPGjRvR0dERECMxXxESEoLCwkIsWbIE7733Hg4cOEAMJIJHMJlM2LVrFz744AO3I7kCjYqKCly8eBFJSUnIzs4eloH0P//zPx79nymKGrBsVV1djU2bNkGlUo2rdsvfEAOJELD0n7Lv6enBzz//POT+WK4oKSmB0Wgcl40MRVFgs9ljMjKJ4FkuX76M77//HtHR0Zg8efKQyXeDbXnNGTabDXq9ftAEwnw+H5mZmQMibD1dnwwGA8rKyhy2fjp9+jT0en3QtVvh4eHIysqCXC6HQqFweo/VasWFCxdQX1+Pc+fOBVTyZmIgEYIGuxP7Rx99NOJnkMhKAsE1NE1jz549KCoqwowZM/D666+7PaMS7LS2tuKNN95wOgiLiorCyy+/jIULF3pdho0bN+LIkSPMud7eXo8u4/mK1NRUvPTSS0hNTR10j1Sj0YjPPvsMn3/+ecC1z8RAIgQNNpsNra2t/haDQBjzdHd3o7u7GzExMWhsbHSYQRIIBBCLxbBarejq6kJnZ+eYyQ1lT7XijN7eXjQ2Ng65sbBdP65SIdh15yyFi1qtRl1dncsNyoMFPp+P2NhYKJXKQe+x2Wxob28PyP+XGEgEAoFAcEp1dTXWrl3rkL9m0aJFuPfee9HQ0IBNmzahsrISpaWlfpTSN2i1WrzzzjvYunWry/vs+nGVCkGtVmPz5s2oqKgYcK2np8fpeYLvIQYSgUAgEJzS3t6O/fv3O5yTy+Uwm83o7OzEwYMHHfbKHMsYjUacOHFiyPvs+uHxeIPeo9Vq8cMPP4zZNCPAFZ/HoRKK0jTNHIEIMZAIBAKB4DZnz57FW2+9hZaWFrS0tPhbnIDDrh8ulzvoPc3NzX7bLsQXzJgxA9dcc41L36Pm5mZ89913qKurC1gjmxhIBAKBQHCbEydOoLi4GDRNB1TEUaBg148rxrLuKIrCNddcgxdffBEhISGDJn5sbGzE5s2bUVJSErC6IAYSgTCG6erqwunTp9HR0YHk5GRERUVBLpcjLy8PbW1tqKysDKioEULgY7PZxsX+kCNlvOqHy+UiJSUFMpkMqampgxpHGo0GNTU1OH/+PDo7OwNbV33XAPsfAGhygCb6IfoJVv2EhITQSUlJ9IwZM+gdO3bQNE3TnZ2ddGVlJb1161Y6IyNjXOsnGA6iH6KfYNCPTCaj33nnHfrChQt0S0sLPRhffvklnZ2dTScmJtJ8Pj9g9UPTNJlBIhDGMgaDATU1Nejq6kJ3dzcAQCKRQCKRoLu7e1RJNwkEAoHL5UIoFEImkyEpKQkTJ04ccA9N09Dr9TCZTGhoaEBlZWVQpIYgBhKBQCAQCIQRkZ6ejgceeABKpRJTpkxxeo9er8cnn3yCw4cPo7KyEkaj0cdSjgxiIBEIBAKBQBgRsbGxKCwsREJCwqD3mM1mHD58GF9++aUPJRs9xEAiEAgEAoHgNhRFITc3F7m5ucjKyoJYLHZ6n0ajwaFDh9DY2IjKykofSzl6iIFEIBAIBALBbVgsFhYtWoTVq1eDz+cPmhSzpqYGr7zySlAtq/WFGEgEAoFAIBCGhd05m81mD3qPxWJBb29vUDhkO4MK1BTfBAKBQCAQCP6C5W8BCAQCgUAgEAINYiARCAQCgUAg9IMYSAQCgUAgEAj9IAYSgUAgEAgEQj+IgUQgEAgEAoHQD5dh/hRFkRA3ADRNU87OE/1cgejHNUQ/riH6cQ3Rj2uIflxD9OOawfQDkBkkAoFAIBAIhAEQA4lAIBAIBAKhH8RAIhAIBAKBQOgHMZAIBAKBQCAQ+kEMJAKBQCAQCIR+kM1qCYQgRSaTQSaTMX9bLBZoNBrodDo/SkUIFvqXH1fYbDZoNBpotVovS0UgBA7EQCIQghAWi4XCwkLcc8894HCuVOPW1la8+uqrOHz4sJ+lIwQ6zsqPK7RaLTZu3Ijvv//eB9IRCIEBMZAIQQmbzQaHw4HNZoPFYgFNeyelh/09zrBarbBYLF5571BQFIX4+Hjk5eWBzWYDAJqamiCXy8Hn8/0q21iGxWKBy+W6fb/FYoHVavWiRMOHy+WCx+NhwoQJDuXHFR0dHZBKpT6QjkAIHIiBRAhKrrnmGtx0002oq6vDv/71L7S0tHj8HSwWC9dddx0WLlwIFsvRXY+maRw6dAi7du0KmA5QLBbjrrvuQl5eXsDJNlbIyclBQUEBhELhkPdarVbs3bsXe/fu9ZoBP1wiIiJw++23Iz09HbNnzx5QrgkEwv9BDCRC0EFRFGbMmIHHHnsMJ0+exJ49e7xiIFEUhby8PKxcuXLAKJumaVitVuzZsydgjBCBQIAlS5bAZrMFnGxjhUmTJuGRRx5BeHj4kPdarVZotVrs378/YL5DWFgYbr31VixYsAAUNWgCYQKBAB8ZSNHR0cjJyUFoaOiwftfZ2Yni4mJ0dXV5SbKxAZvNxpQpUzBhwgTmXHd3N4qLi9He3u7WM1x9o7a2NhQXF/vd+ZfP52P69OlQKBSYMmUK2Gw2ZDIZrr/+ekycOBFnz55FXV3dqN8jEAiQk5MDuVyOzMxMsFisoOpMWCwWJk6ciMLCQmg0moD4dsFM3/o1Y8YMcLlct8oDi8VCZmYmbr75ZqjVahQXF6O3t9cHEg8ORVHM0R+j0Yiff/4ZarUaaWlpTNknEMYrPjGQMjMzsWHDBsTFxQ3rd+fOncPKlSuJgTQEISEhWL58Oe6++27m3MWLF/H444+7bSC5+kYnTpzAqlWr/N7JSiQS/P73v8eiRYsgEAjA4XCQkpKCtWvXoq2tDU8//bRHDCSpVIqVK1dizpw5EAgEQdlJLFy4EPn5+SgqKgqIbxfM9K1fISEhbi2vAVcMpF/96leYP38+Dh06hD/84Q9oaGjwsrQjp6urC++++y727NmD//mf/0F6enpQln0CwVN41UAKDw+HRCJBfHw8YmJiEB0dPazfx8TEICEhAd3d3Whvbx9RI2+XwWAwoLW1NWCmuvsSEhICmUzm1BnYLndfh1uxWIyIiAim8RIIBIiNjUV0dDR0Oh3a29thNBphs9mGfLc73yg8PNwtR05PY5fNPtqVyWRQKBQOMnK5XEilUnA4HCQkJCApKQlarRYdHR3D9vsQCASQSqVITEyEXC53qgur1Yq2tjbo9foRvcNXCAQCCAQCyOVyJCQkwGQyob29HXq93t+iBR0URUEsFg+7/QIAkUgEkUgEhUKBxMREsFgstLa2+n0mqS/2Nqa5uRlWqxWhoaHDckQnEIZD//7LFTRNo6Ojw2/pJbxmILFYLBQUFOCOO+5AVFTUiCIgEhMTsXbtWjQ3N+Nvf/sb/vvf/45Yhp9//hlvvPEGmpubhy2Ht8nKysITTzzhtAEuLS3F66+/jvr6eubc3Llz8dBDDzHLYWw2G6mpqQCAY8eO4e2330ZTUxMuXbrk8r2e+Ebeoq9sduOMx+Nh0qRJTu8XCoW4//77sXjxYuzcuRPvvfceDAbDsN6Zk5ODlStXQi6XIysry+k9Wq0WmzdvxtGjR1FbWxvwkWJpaWlYv349mpqa8NZbb+HgwYP+FmlckpWVhVdffRVqtRpvvvkmjh496m+RGMrKyrBx40b09PRg0aJFuPfee5GYmOhW+D+BMFz691+uMJlM+PDDD/Gf//zHB5INxGs1gKIoJCcnY8GCBSOefRCLxZg5cyZ0Oh2+/fbbYf3WHo6bmpqKBQsWALgyUxOISKVSXHXVVUhISBhwjc/nQyAQOJxTKpWYP3++w3m7Y25DQwN++OEHdHZ2unxnf/30/UY2m81hZsSdmShP4kq2weBwOJg8eTImT56MysrKYS0N2H2M5HI55syZ43KmwGg0oqSkBPv373f7+f5EIpEgLy8PHR0d+Pe//+2191AUxei8f/kJROzy0jTtVvm216++/+dwiIyMxNVXX42mpiZ8+eWXYLPZbr/b27S1teHIkSMwmUy49957mfbSjtVqDciZd0JwwWKxwGKxEB8fP6D/Ggyj0YgDBw6Ay+W6XVc82f6MySFCeHg4CgoKkJqaijlz5oz5dXSTyYQ9e/bg1KlTKCkpGXLmxJV+dDoddu7ciYqKCuZcXV0d2travCa/u7J5Azabjeuuuw55eXnIzMx0q9ISBpKcnIwlS5aAw+Fgx44duHDhgr9Fckl+fj4WLFiAhoYGbNu2zWX5NplM+O9//wuNRoPs7GzcdNNNIx5sCYVC3H777cjOzsbRo0cDKsLNGSqVCtu2bUN9fT3OnTvnb3EIQUpISAh++ctfYurUqUyggztwOBwsXLgQYWFhbhk9Op3Oo+3PmDSQJBIJli9fjvnz5wddBNJIMJlM2LVrFz744AO3rGdX+tHr9diyZQt27NjBnPPlSNfX385eAVeuXMmMcAjDJzk5GY899hg4HA4uXLgQ0AYSRVGYPXs2nn76aZw6dQo//fTTkAbSrl27sHv3btx5552YP3/+iA0kkUiEpUuXwmaz4Y033sChQ4cC2kCqr6/Hu+++i4sXLwb8rCAhcAkNDUVhYSGWLVs2rHadzWZj4cKFA2Y1B0OtVnu0/fGpgWQymVBeXj6kH1B8fDwmTpzILK1wOBxkZmbi+uuvR2NjIyoqKpz6fkRHRyMzMxPx8fGQyWR+cSz2FhwOBxkZGYiNjUVmZuageXlc4Y5+7MaQrxvt0X47i8WC8+fPQ61Wo7y8fFjys1gssNlsl5W2ra0NZWVlaGxsRFNT07Bk8wY0TaOqqgp79+5FVFQUsrKyfL6EbF9GT05OxowZMyAUCpllqECEx+MhMzMTMTExyMjIcDtcH/i/+tXQ0IADBw4gOjoakydPdsiH1NzcjPLycphMJgBX9JOSkoLk5GSHZ9k7iEDRk1QqxZw5c2CxWJjl5ZqaGlRVVeHcuXPo7u4e8QCJxWIhLS0NiYmJA67RNI1Lly6hurp6VPITggN7OztcKIpy+3dCoRDTp0+HwWBATU0NKisrR2XY+9RA6urqwubNm7F3716X9y1fvhzPP/8848TF5/Nx1113obCwEN988w1eeOEFp17tOTk5WL9+PeRyeUA5HHsCoVCIBx54AIWFhRCLxeDxeMN+RiDrZ7Sy6fV6/P3vf8e2bdvQ3d3NdFKeoqysDH/84x9RV1fnduoEb2Kz2bBjxw4cOnQIc+fOxauvvorY2FifysBisbBkyRKsWLECIpEIkZGRXknY6SkkEgkeffRRXH/99RCLxSNyQj527BgqKiqQlpaG1157Dbm5ucy14uJiPPvss2htbQVwRT+rVq3Co48+GtCDNbsDuc1mg1QqBU3T2LNnD1577TXodLpRLa9zOBz8+te/xu9+97sBBqHVasVf//pXbN68OaBn0QjBg72O33vvvXjvvffw+uuvw2w2j/h5XjWQdDod1Go10xA1Nzejvr4eKpXK5e/6h09TFIWIiAhERERAKpUOugwSGhqKuLg4xMTEOMig1+vR3t4ecJVQJBJBKBQiMjJy0P+Jz+cjKioKLBYLSqXSwZHbYrGgq6sLnZ2d6OnpGfJ9zvQTKHhCNrPZDJPJ5JXvbDAY0NjYGFB5bLq6utDV1YXm5mafRtOx2WyEhYVBIBAwZTKQDQAejweJRAKFQgGlUon4+PgRP0uv10Ov1yMkJARqtRoajYa51tjYiLq6OsZIZLPZaGhogFqthlAohEQiCcgl3JCQEMTGxsJms0Gr1aKzsxMNDQ1QqVSj6lyA/2u7ExISnBpIEolkVM8nBDZcLhcSiQQymcwnM9wsFgsymQxSqRQRERGjnqX1moFkH+FWVFQwjYLBYMDZs2e99UqnMuzcuRNbtmyBWq0OiJG/HRaLhcWLF2Pp0qUuZ01SUlKwbt06mEwmTJkyxeGaSqXCpk2bUFlZidLSUl+IHbAIBALcf//9uOGGG7B79258/PHHMBqN/hZrTCKVSvHoo49i2rRpmDhxYkB2+n3JzMzEo48+CqVSiezsbI88U6PR4NVXX3Wotw0NDQ4z233bwJycHDz66KOQyWQeeb830Gq1eOedd1BUVISqqqqAT2FBCHxSUlLw2GOPISkpaUD/FQx4zUCiadqnzprOwm9pmkZFRQV27NgRcLNHFEUhIyMDS5YscTn6joyMdOqgRtM0Ojs7cfDgQbeMzpGGJ/sSmqYHzBy6C5fLRU5ODnJycqDRaMDlcmEymTziWGqXizipXkEgEDBRYMFAdHQ0Fi5cOCCNxmi+p06nw+HDh13e07cNNBqNuO+++0b8Pl9gMBhw4sQJhwANwtAMd5ZiPLUjUqkUCxYswMSJE4e8dyR68XoQj1ef7gPYbDauueYazJgxA1OmTHF7G4Bg5uzZszh06BBqamqG9PkIFv1UVlbi3XffhUgkAnBlhi0/Px+zZ88ediXIzs7GihUr0NDQgO+++27EfjE0TePIkSMoKipCRUUFuru7R/QcQuBx8uRJ/PTTT6iqqgqomWVCcBEXF4dFixYhMjLSrftLS0tx4MABMrvdB7PZjEOHDuHs2bPDMpJG00e4S9AbSBwOBzfddBMee+wxsNnscZH99cSJE1i3bh10Ot2Q0+DBop/y8nJcuHCBKegcDgdr1qzBzJkzh+3fMnPmTOTk5ODMmTM4c+bMiA0km82GAwcO4JVXXoHZbCZLDmMEmqbx008/4YUXXoDBYCDflTBiEhMTsWrVKrdmSADg888/x7Fjx4iB1Aej0YitW7fiww8/HNbvRtNHuP0OrzzVx3A4HPB4PKZzNRqNqKysRHt7O+rq6gJqSpPP5yM1NRVSqRTx8fFuW74WiwXV1dVobm7GxYsX0dvb67YDZX/9OKOzsxNVVVVobGz0WVLIvthsNodQYovFgpqaGhw5cgSRkZGYOHGi25F7LBYLPB4PERERyMnJgUAgwOXLl0cUnm+1Wr3m+O0pOjs7cerUKTQ1NSElJcXt0exwkUqlSElJQUJCgkN4u10Gf5YfZyQkJECpVCIrKwt8Ph9msxlVVVVoa2tDVVWVT4yj/jIEIt74dqGhoUhJSUFUVBRiY2MDJqWBp4iLi0NiYiKys7MhkUjcbpsUCgVmz57NzEZbrVaoVCo0NjZ6U1yfY9fP1KlTB2wpotFoUFNTw9Q9ewDMcCOPrVbrgPQTfev45cuXR52/b0wYSP1paWnBxo0bcfToUbS1tQVEOn87MpkMTz75JPLz811G5PWnp6cH77//Pnbu3InOzk6Pj0BKS0vx/PPPQ6VSBUSeH5vNhm3btuGnn37CVVddhfXr10OhUAzrGQkJCVi7di3a29vx6quv4quvvvKStP6ltLQUq1evhkKhwLp167zmG5STk4O1a9dCoVAMiDYMtPLDZrNRWFiIhx9+GGKxGFKpFJ2dndi0aRP279+P9vZ2rxtHzmQIRLzx7RQKBZ599lnk5uYGtGP6SKAoCosWLcKqVasgkUggl8vd/m1+fj5SU1OZPslkMmHjxo34xz/+4SVpfc9Q+jl06BBeeeUVZsNmm83msfQgXV1dHq3jXjWQQkNDERISAovFAr1e7zNDxWKxoK6uDhcvXvTJ+4YDi8Viwh4BoL29HVwuF0Kh0KmxZNddW1sbamtrPf4/9fb2wmAwoKmpCZWVlQEVxt7W1oa2tjbExcWhpaXF5QicxWJBIBA4jOT4fD4SExMhlUoRGxvrdGaFz+cPCD+12WzQ6/UwGAwBtev6YPT09KC6uho9PT3Q6XQeeSaLxYJIJHJYklUoFEhLS0NUVNSgMgyVwsOXSKVSh4SzFosFjY2NXm8XWCwWhEIhQkJCoFQqHWQIRDz57Xg8HoRCIaKjo5GcnIy0tDQPSOhf+rfPFEUhLi4O6enpzJYZNE1Dr9c7nQWhKAoCgQB8Ph9hYWEICwtjrhmNRiYcPZBWOkZL/1n/vvppaGhAZWWlW6lpnNG3fvWfnfJ0HfeagcRisXDTTTehoKAAZWVl+Pvf/x4wU+/+pL29HZs2bcKWLVuYc9nZ2bj//vudjjCrq6vx/vvvo7a2FqdOnfKoLDabDbt27cL27dtRX1+Pjo4Ojz7fU5w/fx4vvPCCy92fxWIx7rnnHuTn5w+4FhISgt/85jeYMWPGgGssFgvTpk1zWAJob2/HBx98gHPnzuHcuXMBNQPpK+RyOR544AEH34r4+HiIxWI/ShUcREZG4oEHHsCUKVMwZcqUgI8e9SQzZ87E3XffDYVCMSCDeLCSnp6OBx54wGEmbNKkSQ5Gr16vxyeffOI0sjEkJITZoma80lc/lZWVo1oB8WX98pqBRFEUpk6dijvuuAP79+/Hl19+6REDKditbL1ej4MHDzqca2trw7Jly5waSM3Nzdi5c+eoLOLBdGaz2VBSUoIvvvgioH1sNBoNtm3b5vKeiIgIzJ0716mBxOFwMHPmTMycOdOt9+l0Ouzfvx/79+8fibhjgrCwMFx//fVO9UlwjUgkwoIFC4ImDYInSU5OxtKlSwf4qDkjWNry2NhYFBYWDkgT0Rez2YzDhw/jyy+/HHBNJBJh1qxZ48pA6v9tXelnuPiyfvnEByk+Ph7Lly9HQ0MDfvzxR9TU1Azr92azGceOHUN5eTmOHz8Oo9GI8PBwzJs3D/Hx8Zg6dWpQOwGqVCp89tlnTpeALl26hM7OzhE/22q14tixYwgNDUVycjKz59IPP/wAlUqFkydPBk1D5Qqj0Yh9+/ZBq9UiKysLeXl5ARuxF8hMnDgRV111FRITEwf1reju7sYPP/yA+vp65lx5eXnAp0EQCARYtGgRFAoFzpw5gxMnToyb2cGkpCTMnTsXSqUScXFxoGkap0+fxunTp1FWVjaqb8flcpGXl4fMzEzMnj17SGd0e/kJ5PaHoijk5uYiNzcXWVlZA2ZOz549ixMnTjADy56eHlRWVjp9ltlsxg8//ACapjFx4kRcffXVI9oqKligaRpnz57F+++/z7TBrvTjLv3LsE/omwSv/wGAHunBZrPpdevW0RaLhbZYLHRPTw996dIl+pe//OWQv3344YdpvV5P2+nu7qZXrFhBC4VCmsfj0QDo9PR0ev/+/bRer6fNZjPdl8uXL9MLFiwYsez9D2/op7+uQkNDaYFAMOAICQmhKYoa1fN5PB4tEAjoZcuW0U1NTXRtbS39y1/+khYIBDSXyw14/bh78Pl8WigU0itXrnQoP8Ml2MqP/ZDL5fS2bduc/k/t7e30smXLhnzGsmXL6IaGBrqnp4e2Wq1On9W3/NgPPp8/4nLqDf30bX/s2Gw22mAw0N3d3fT69euZtsTTR1JSEr1v3z6XZcxms9F//vOf3ZLBE/pZsmQJXV1dzXxXi8VC/7//9//osLCwUX07ALRQKKTfeusturu7mzYajUPWr2Bof9hsNv3cc8/RXV1dtMFgoG02m8O3e/PNN+mIiAiHOsBmswd9nr0Nvv/+++nOzk7mWQaDgX788cdH3cb7Wj9DHVwud0Bf5ko/IynDzlCr1XRBQYFH9EPTtG9mkNhsNkJDQyGRSDBx4kSo1Wrmmk6nQ21t7ZBrkiaTCXq9nvmboiiEhIRAIBAw5zo7O1FXV4fa2tqAH832xWq1etUZ2GQywWQywWAwgKZpcLlcxMfHIyMjg7lHq9WitrZ21Hsv+ROj0Qij0Yj6+noUFxczZcO+j91Q0TTBWn5GSmhoKBITEx3qUGpqKsRisVN/r776aWlpGbGTpb+gKAp8Ph88Hg/x8fGYPn06Ojo63Gp/PIXFYoFKpUJHRwcaGhp8NnvCZrMhEAiY72q1WmE2m6HX60e8vG4vPzKZDHFxcUyS16Gw2WwwGAwBWX54PB6TxiIhIQEikYjxcen77S5fvgydTud2e2lvg41GY0DOmHkas9nskb7E7hAfHR2NiRMnQiKROG2buru7UVtbC7VaPaoVl/74dA1CIpFgxYoVuPvuu5lzZ86cwdq1a1FbWzvq5xcXF+Oll15CU1NTQEXTBBpSqRRPPvmkgxFw7NgxvPjiiw6bbwYrhw4dwoULF5jpXYFAgD/84Q+49dZbXf5uvJWfxMRErF27Funp6cy5yMhIB4OpL2NFP/Yw5OzsbBQXF3us/XEHexjywYMH0dLSEtRJKu3lJzMzE7Gxsf4WxyPIZDKsXr0as2bNglwud3AAHkvfLljgcrm4/fbbceeddyI8PNwhArAvpaWlePHFF1FXV4e6ujqPvd+rBpLJZIJOpwOfzwefzweHw8GECRMc7jEYDIiIiHBw4Ha2hs3n8x1GKEKhcEDobFdXF8rLywMiD0sgw+PxkJKS4nCuubl5zKyL29MD2BEKhWhoaBgyBF6j0aCsrAzNzc3eFtHn2Gdc+9YhqVSKzMzMITeRtI9+g00/9vanbzsRGhoKNpuNqKgoREVFoaenBxEREWhtbYXRaBxRp8fj8RzqjkAgGDSs32w24/Llyz7dtBu4MmOk1+vR09PDtK9cLhcikYiZeR1qZoPP5zNh7YD75ScYYLPZCAkJQXh4OCZOnOiwqbHVaoXRaERnZycqKytH9O3sugsJCQlqf1lfY59Bys7Odqo3e9vU3NyM0tJSB79IT+A1A8lms2H37t1oamrC5MmTcddddzl1Qp4wYQLWrFmDrq4u5tykSZMcGpyQkBAsXboUkydPZs6Fh4cjKSnJW+ITxhBGoxFbtmxBaWmpy/tqamrG7NKaQCDAXXfdhby8POacTCYbcuRvsViwfft27Nu3D5cvXw4a/fRtf+wNq0QiwZ133unQ+dnbH7VajS+++AInTpwY1ns4HA4KCgpw3XXXMe8Ri8Vubz3hK0pKSvDCCy9AqVTinnvuQXp6Om688UbI5XKUlpbi008/dbknnUAgwLJlyzBr1izmnDvlJ1iYMmUK7rzzTsTFxQ34dhcuXMAnn3yC+vr6IdsQZ/D5fCxduhRz5sxBWlqay3QlBPex1/Fdu3YxS58ex5WDEjzksHX99dfTtbW1QzrvjZZvvvmGjomJGRNObt44br75Zlqj0TjV3Z49e+iEhIRxrR9vHb7Sjysn7ZEylpxI5XI5vX37dqf/p7tO7P0PPp9Pv/HGGw5OvK7wpBPpSHTgzIHcnbofERFBf/755yMoQQMJxCAIb7WNAGiRSET/7W9/c/rssVS/vHG4ql8Wi4X+05/+NGrn78H0Q/vKSbuxsRHffPMNlEolZs+eDaVS6bFn22w2FBcX4/z58zh16lRQZD72FyqVClu2bEFsbCzy8/MdwrhjY2Nxyy23oL6+HsePH/f4VCUhOKiursbJkyeZjMBmsxnnz58fM46lg/0fXC4Xc+bMAYfDwcWLF3H69Gmny23Jycn4xS9+wcxwczgcTJo0achlE7VajWPHjqGhocGjPhLeJioqCvn5+VAoFAPcI4aLSqVCUVER6urqHAJ1xiNarRZHjx5FY2PjmKpfniIsLIwpd/3rl06nY+pSSUmJd3XnynqCh6xADodDh4WF0enp6fSuXbuGM9gYEqPRSD/77LN0REQELRQKvWKJB7MF3vfgcrm0RCKhp06dSh86dMhBj2azme7q6qIrKiroG2+8cVzqx1uHr/TjiRmkf/3rX3RSUhIdHh7OHHw+f8zrx2az0Xq9nu7o6KA3btxICwQCp8+4/fbb6cuXL9MdHR3MYTAYhtTrgQMH6MmTJ9MSiWTYoe2e1M9wZ5BmzZpFnzx5ku7s7KRNJtMIS9UVdu7cSU+cOJEOCwujORxOQJUfX88gXbhwgb7hhhvGVP3y5DFx4kR69+7dTutXbW0tXVBQQIeHh9MhISFe0w/tqxkki8UCrVYLHo+HS5cuoaKiApGRkYiKihq2w5pWq4VGo2ESvNnDugN1m4xAwmw2o6urC1qtdsDomMPhICwsDCaTCcnJycjIyEB7eztaWlrI6GaMYjQa0djY6BDiXlNTg7a2tqDxNRoO9j0aKyoqmHOhoaFQKBTg8XgQCAQQCASIi4tDRkaG0zD0pKQkSKVSt7dcaW1tRWtrKy5duoS2tjYHX0t/YDKZoFKpUFFRAZlMBplMBoFAgJSUFKfRi8nJyZDJZJBIJKN+d9/2JxiwfzuVSjWikHU+nw+FQgGpVIqIiAiHazabDd3d3R4NSR8LhIWFQS6XIy0tDTExMQ4Z2bu7u6HRaJhNlX2hO5+G+Xd1dWHz5s34/PPPsXz5cjz00EMOURHucPToUbzxxhtMA07TdFCHHAca9lQMy5cvxxdffIH33nsvqHMjEQantrYW69evd9jGprW1NSDz03iCvu2PnezsbDzzzDMO20jMmzcPEyZMcJpl225QuIPNZsO2bdvwj3/8Y0Bkpb9oa2vDn//8Z0ilUtx333347W9/iylTpuD11193utGqSCQa1m71YwVPfLvExEQ8++yzyMjIIAFFbnLVVVfh8ccfR3R09IBI6+LiYrz22mtoaGjA5cuXfSKPTw0ks9mMCxcugKIo5OXlwWAwXBGCw3GYSbJarYOG2zY2NuLkyZPE8h4FNE0zScvssFgscLlccLlcpKeng6ZpHD58mISkjkHs9aujowNnzpxBSUmJv0XyCfb2pz9arRYmkwkcDgcsFgsxMTGIiYkZ9buMRiOqq6tx/PjxgNnr0Gg0oqKiAmw2G9dffz1omoZEIkFOTo7X3mmxWGC1WmEymQJ2Ntpms8FoNMJkMoHL5YKmaajVavz888+wWq1MklE7rvooNpsNDoeDiIgITJs2DVOnTnV4j8VigdFoHDfb3AyHyMhITJ06lZk56ttHaTQanDx50qdpRvyyWRVN0/jpp5/w/PPPIyUlBb/5zW8QFRXFXC8qKsKOHTucjmjOnz9PHLFHSUdHBz788EPs3buXOZeRkYFf//rXbm0ySQhu7PWrvr5+3DvLqlQqbNy4EXFxcbjllls8Yih0dHTgq6++woULF3D8+PFx3RFaLBbs2rULhw4dQlVV1ZC5yPxFWVkZNmzYAKVSidtvvx2pqam47rrrIBaLBxh1NE3j0KFD2LVrl1PDd9asWViyZAni4uKgUCgcrhUXF2Pbtm2or6/3WXLSYOL06dNYt24dQkJCBlzzS/lx5aAELztiURRFX3311fSFCxccnLDeffddWiQS0RRFDTi8LZOzIxid3NzVv/244YYbaJVKxXwDX+8VNZYPX+nHXSdte/3yt14CqfxERkZ6LIy9pqaGvu666zzWXnlDP872qvMG9jB2FosV8OWHoqgBTuw2m23AYbVaXbaNDz/8MN3d3e1UH5988gkdEREx7urXcL+DL/v+wfRD+8pJ2xmpqanIyspCZmam0z18bDZbwE7HjhX66lej0eC7775jZvJomkZpaanPR78CgQA5OTkOM4quaGtrQ3FxccCOTAORCRMmYPHixdBoNCguLva743AgYDQacfLkSY8k8WtubkZzc3NAt180TeP8+fPYunUr5HI5cnNznf7vruqXRCJBbm6uUwfu7u5uFBcXQ6PRoKqqKihm0Wiahl6vx+HDh9Hd3Y2MjAykp6cPcDOgKAoTJ05EYWGhU//M6dOng8vloqurC6dPn3aoX6dOnXK6MkL4PwKp3lCuhPlfq83zL6UoPPzww3jmmWcgFAoRFhbmkJr/b3/7G5544omAcRaladqpI4639OMPeDwexGKxw3fo6elxy/DwpH7i4+PxxhtvYM6cOW7df+LECaxatQqXLl0a7qt8hq/Kj1wux9/+9jcUFBS4vK+3txc6nQ4lJSVYuXIlysrKPCnGsAmE+kVRFMRisdOp/eFitVqh1Wo9FtzgLf0IhUIIhULMmzcPGzdudJqf7sSJE1i5ciWqq6sHXJsyZQrefPNNhx0O7FRWVuLxxx/HiRMnoNPpvOoW4Un9sFgshIWFITQ0FE888QRWrVrldNuYnp4e6PV6p515aGgoxGIxSktL8dhjjznUL4PBgO7ubp8aAYFQvwKZwfQD+MkHCbgSHREdHc0kXLPZbGhvb4dOp0Nra2tAWZHjAZPJFBBRNjRNw2AwoLe3F+Hh4UOGF8vlciQkJDj4Ami1WnR0dJAyNAihoaEIDQ1FZGTksKNIxyo0TUOr1QZNCLon4HK5CA0NBZ/PB4vFgtVqRVtbm8PAVKVSQaPROHWMbWtrG9RR2Wq1oqOjAy0tLV6T3xvYbDZ0dnaiu7sbDQ0NqKmpYQwkNpsNmUyG0NBQJiWE0WgcsHFtT08P2traUFNTg6ampqDZu5AwEL8ZSP3p6enBBx98gP3796OhocHBe50wfmhtbcXGjRvx6aef4s4778Ty5csddtTuT1paGtavX+/QqO/cuRPvvfceEyVJIBAcYbFYKCgowB133IGoqChIpVJotVps3rwZR48eZe7r7Owclx28zWbD9u3bcebMGab9iYyMxKpVq5Cfn8/cV1lZiY0bNzrNjt7d3U1S0AQ5AWMg2bc0OHToEGiaDoo1a4LnMRgMOHPmDNhsNvLy8mA2m8HhDCymFEWBxWJBIpE4bMAKXIl24PF4MJvNARNe7Qvs9cYeluzKsASu6JDNZoPNZhOfv3GEPaVHamoqFixYwMyQdHR0oKSkBPv373f5e3vZ6p+eZSxB0zSqq6sdlhZjYmKwbNkyhzalvb0dR48edcglRhg7BIyBFBoaiptvvhmpqak4efIk9uzZQxIUjmNsNhsOHToEq9U6oKOnKAr5+fkOjXtfZsyYgTVr1qC2thbbtm2DRqPxldh+RafT4auvvsLZs2dd6seOXC7Hgw8+CJVKhT179uDUqVM+lJbgD8LDw1FQUIDU1FTMmTNnSCPaGbm5uVi0aBHi4+PHVRJJe/06c+YMc66uri4gXBMIXsJViBu8GMa3evVq2mg0OoRA2mw22mKx0G+//fageyH54wi2MMmxoh+Komg2mz3g4HA49FNPPTWg/PQtR1arlT516hSdnZ09ZvXj7GCxWEPqpy9Wq5Xu7OykH3jggXGhn2A8PKmfxMRE+vvvv6ctFsuAHdLVajVdUFAw5DMeeughWqvV0larddBydf78eTo/Pz/o9DPUwWKxHNoib6YvCEb9BOMxmH5of4b5O8M+5T+SUQ1h7EHTtNMlMoqiUF1dje+//x7R0dGYPHmyw/YPFEWBoiiEh4cjPz8fMpkMFy5cQH19vS/F9wv2pTLazeUyFotF6tw4wr481ndmsa2tDWVlZWhsbERTU5NbzxiszKjValRUVODSpUtjcrcD4voxvggoA4lAcAeaprFnzx4UFRVhxowZeP3115GcnDzgvoSEBPzpT39CZ2cnXnzxRXz55Zd+kJZACGzKysrwxz/+EXV1dWhvbx/Vs44dO4YXXnghYPaeIxBGg98MJJ1OB41GA5FIBIlE4jCiEQgEkMvl6OrqQldX16ChpITxS3d3N7q7uxETE4PGxkaIRCKEhYU55LHhcrmQy+XMtfGEvX7Z02i4Qq/XB0zOMYJ34PP5kEgkkMlkA8qEwWBAY2MjGhoaBv09RVFMfqCwsLBBnbP1ej3q6+vH5OwRYfzht73Y9u7di/r6emRlZWHFihWIjY1lrs+ZMwd//etfUVlZibfeegs1NTX+EJMQBFRXV2Pt2rWMw7G7CSbHMn3rlztLZxaLBaWlpT6QjOAvpk+fjocffhixsbGYNGnSsH8vFApx3333Yd68eUhKSnLL8CYQgh2/zSBVVVWhqqoK7e3tuPfeex2uJSUlITExET///DP++c9/+klCQjDQ3t6O/fv3IyIiAjfddJO/xQkY7PWLQAAAhUKBRYsWISYmxuF8P6fdQeFyuZgxYwaWLFky6D3D8X0jEIKBgPRBOnv2LA4dOoSampqgy8RKIBAIgQ5N0zhy5AiKiopQUVGB7u7uET/LbDbj0KFDOHv2LE6fPk2S/BLGDAFpIJ04cQLr1q2DTqcj/kcEAoHgYWw2Gw4cOIBXXnkFZrN5VO2s0WjE1q1b8eGHH8Jms5E2mzBmCEgDKSoqCjk5OWhra0NlZSX0er2/RSIQCIQxA0VRiIuLwy9+8Qu3ss2LxWJERUU5nNPpdKiqqkJraysaGxvJLvUEt5DL5UhKSnKZxLa3txeVlZWjmtn0BAFpIF177bWYPHkySktLsWbNGlRUVPhbJAKBQBgzsFgsFBYW4pprrnHrfjabjejoaIdzdXV1WLt2LcrKyogrBMFt5s2bh6effhqhoaGD3nPp0iU888wzDlnL/YHfDSSLxYLOzk6H/BshISFITU1Fb28voqOj0dTUhJ6eHrK2TSAQCCPAbDajs7MTISEhEAqF4HA4kEqlkEqlI36mwWBAbW0tCQYgDCAkJAShoaFO00HExcUhLS3NIblvfyiKglwuR0REBHp7e/228bjfDaTq6mq89NJLEIlEAK6MbBYsWIA77rgDSqUSTz/9NNRqNT777DMcPHjQz9ISCARC8HH27FmsWbMGCQkJuP/++zF58mR/i0QYwyxYsAC//vWvnW40npaWBj6f7/L3MTExePzxx6FWq/HVV1/hu+++85aoLvG7gdTS0oKdO3c6nBOLxVi6dCkiIiJw4403QqfToaioiBhIhBFDwo8J45m6ujrU1dUhKSkJixcv9piBROoVoT8URSEzMxO/+c1vRpwvKywsDNdffz2MRiPOnj2L3bt3+6Ws+d1AckZpaSk+/PBDxso0Go0oLy/3s1T+hc/n4+qrr0ZaWtqwfme1WlFUVISSkhIvSeZf5HI55s2bB4VCMUA3ra2tOHToEBoaGsZ9+SEQAECr1WLnzp2orKx06365XI5rr70WEomEOVdeXo6jR4+iurqabCdC8CpsNhv5+fkwGAyorq7Gjz/+iN7eXt8J4GonW/hpd10Oh0MLBAKHg8PhBNxuv76UITw8nP7oo49ovV4/rKOtrY1esWIFTVHUmNTP7Nmz6TNnztB6vZ62WCwOO4qfOXOGnj17Nik/AX4Q/fhOPxRF0SEhIQPa18GOBQsW0JWVlQ716oMPPqCjoqLokJCQgNjNnpSfwNIPRVH06tWraaPRSHsCo9FI6/V6+osvvqCjo6N9ph+apgNzBslisZBcGv2gKAp8Pt+lY5sz2Gy203XgsQKbzUZoaKhTvdhsNhgMBrLPGIHwv9A0PSyHV4PBMGAHe4vFAr1e7zfHWcL4gsfjgcfjISQkZNA9AL0FRZM1ZAKBQCAQCAQHht7JkkAgEAgEAmGcQQwkAoFAIBAIhH4QA4lAIBAIBAKhH8RAIhAIBAKBQOgHMZAIBAKBQCAQ+uEy/vt/c+eMe2iadhpbSPRzBaIf1xD9uIboxzVEP64h+nEN0Y9rBtMPQGaQCAQCgUAgEAZADCQCgUAgEAiEfhADiUAgEAgEAqEfxEAiEAgEAoFA6AcxkAgEAoFAIBD6QQwkAoFAIBAIhH4QA4lAIBAIBAKhHy7zIBECEzabDQ7H+aezWq2wWCw+lsh3cLlcsFiu7fqxrgMCgRB4cDgcsNlst+6laRpmsxk0TVIRBTLEQApCZs2ahSVLloDH4w24dvLkSWzbtg29vb1+kMy7iMVi3HbbbZgyZcqg99A0jWPHjmHHjh0wmUw+lI5AIIxXQkNDccsttyA3N9et+5uamvDVV1+hpqbGu4IRRgUxkIKQqVOn4pFHHoFQKBxw7dNPP8V33303Jg0koVCIgoICLFmyxOV9oaGh2LVrFzGQCASCTwgJCcGiRYuwfPlyt+6vqKjADz/8QAykAIcYSEEKi8UCRQ3MkO7sXLATHR2NnJwcKBQKxMbGDvk/Jicno6CgABqNBsXFxejq6vKRpIFDamoqsrKynC5HXr58GefOnYPVavWDZN5FIBAgJycHUVFRLu+rrKxEeXk5bDabjyQbHyQkJGDq1Kngcrkee2ZbWxuKi4uh0+k89kxvMFib7AyxWIxrr70WMTExKCsrQ1VVlZelI4wEYiARAp7MzExs2LAB8fHxCAsLG/L+OXPmICcnByUlJVi5cuW4M5AoisLChQvxzDPPOF2G/ec//4m1a9eip6fHD9J5F6lUipUrV2LOnDmD3kPTNN555x1cuHCBGEgeZtasWVi/fj0kEonHnnnixAmsWrUq4A2k4aBQKLB69WrodDqsX78ely5dIv5IAQgxkAgBD4/Hg1QqhUwmc+v+0NBQhIaGIjIyckQjWQ6HA6lUitDQ0GH/1o5Wq0VHR4dPGz02mw2pVAqBQID4+HjExMQ4/f/FYrHH3hkeHg6JRDJg5EzTNKMDXyAQCCCVSpGYmAi5XI7o6OhB76VpGiKRaEzOtg4Hd74dn8+HTCYDi8VCe3s79Ho9cx+Px4NMJnMwwpVKJWJiYtwayAxHTnednwMRmqbR3t6O7u5uiEQiREZGgs1mIyIiAkKhEEqlEomJidDr9WhrayNGewBBDCQCoR9SqRRPPPEEcnJyRvyMnTt34r333oPBYPCgZK4JCwvDo48+ivz8fCQmJg4a6egpWCwWCgoKcMcddwzowGiaxpYtW/DRRx/5JKIwJycHK1euhFwuR1ZWltffF+y4++1SU1Px5JNPIjw8HG+99RYOHjzI3JeQkIDVq1cjJSWFOadQKCAQCHz2fwQDBoMBn376Kb799lssXLgQjzzyCEQiEYArUbm333478vLycPjwYfz1r3/12aCCMDTEQCIELBRFgcVigc1m+2S0b3+fSCRCbm4u5s+fP+JnVVZWDpmOwFPY5RYIBJg2bRoWLFjg9XeyWCxwuVykpqZiwYIFDp2szWaD1WrF6dOnva4Du9+HXC7HnDlzXM4cjWfsZcQOh8Nx+u2AKwbS2bNnmZnb/Px8REVF4ZtvvnG4Nzw8HLNmzUJ2drZXZKZpGjabLahnVKxWK0wmE8rLy7F//34oFAoYDAaEhoYy7VpaWhrS0tKg1WqdLokHKvYyZf9OYxFiIBECEoqiMG/ePMyZMwepqake9WkYjPT0dCxevBhKpRITJkzw+vs8RXJyMpYsWQKlUomJEyd6/X3h4eEoKChAamoq5syZ49Dx6nQ67Ny5E+fPn8exY8e86gjOZrNx3XXXIS8vD5mZmWTmwgX2sm2fuWCxWAO+nR2KopCfn481a9YgLi6OWW6++eabkZqaytwnl8shl8u9JvORI0dw4MAB1NTUoK2tzWvv8RYajQZbt26FSqXC6dOnAQAlJSV47bXXEB8fj4KCAsTHx/tZypHRt32urq7Gjh07xqSvJzGQCAEJi8XC3LlzsWbNGnA4HJ/MxmRkZGDVqlWQy+U+m/3xBMnJyXjssceQkJDgE7klEgmWL1+O+fPnD4jc0ev12LJlC3bs2AGbzeZVHywOh4OFCxdi5cqVYLFYQfXNfI29bMfExDDnXEVdzZ49GzNnznSYeSosLHRIsdF/VsqT0DSNoqIivPLKKzAajUE5Q6HRaPD3v/8dZ86cYerBuXPnUFpaikmTJiE3NzdoDaS+7fOBAwdw6NAhYiD5CqVSifT0dJ875pnNZpw/fx4ajcan7yU4h6IosNlsr3Z8LBYLaWlpSExMxPTp0yEQCNx+X01NDaqqqpw23uXl5V6dPaEoCsnJyUhOTsaMGTMgFAqdyt3Z2YnS0lKHiLWRyhYdHY3MzEzEx8czjrvV1dWorq5mOoD29nao1WqfpRDw5RJsMGOvS+62qfb7hzo3GDqdDuXl5ejs7HT5jpSUFCQnJzu9bl+qDQbjyGw2o6SkxCGQpKqqCp2dnQ7y0zQNmqZhtVoHDB6ioqIwb948NDQ0oLy8HO3t7T6T311CQkKQlZWF6OhopKSkMIPXkdQ/sViMrKwspw79er0eZWVlLsuPLwhIA2nevHl47rnnRhVFNBK6urrwpz/9Cdu2bfPpewn+g8Ph4Ne//jV+97vfQSgUuh19Q9M09uzZg9deew1ms3nA9e7ubq8mqmSxWFiyZAlWrFjBRMY4o6qqCs8//zyqq6tHLVtOTg7Wr18PuVwOqVQKm82G7du3Y9OmTUwnYLVaA7JhJ/gWjUaDV155hVlacgaLxcKqVavw6KOPBnWUGgD09PTg73//O7788kvmnMlkGtbSYG5uLl5//XWoVCo89dRTOHLkiDdEHRUymQxPPPEErr76akgkklENXhMSEvDCCy8gMzNzwLWqqio89dRTLsuPL/C7gcTlciGRSBwibmJjY5GQkOAzA8lsNqOrqws9PT1kJBqk2EOT+2YQb2trc2q8AFf8VyQSCcRiMeLi4pCQkOD02+t0Ouj1+gGjPZqm0dDQAJVKNeg7vAGbzUZYWBgEAgGUSiUSEhJcdi4GgwGNjY1QqVQjfqdYLIZQKIRCoUB8fDyThNFqtaKrqwt1dXVBl3RSJBJBLpdDp9Ohq6sr6OQfLgaDAc3NzWCxWJBIJB51BqZpGt3d3Q6zlGq1GnV1dU7LXd8y3F/vNpuNqcfd3d1BkxvIZrOhra3NbYPIYrGgvb0dTU1NEIvFEAgECA0NhVKphMViQUhIiJclHhkcDgfR0dGIj4+HTqdDU1MT2tvbh1V/QkNDERYWBoVCwbRh/TEajVAoFE593Oztji92SvC7gZSSkoLHHnsMsbGxAK5MuyYlJfnUm7+qqgqbNm1CTU0Nzp0757P3EjyHXq/Hhx9+iB9//JE519nZibq6Oqf3KxQKPProo8jKysKkSZOcGkc2mw07d+7Eli1bBkzz0zSNqqoqn2+KK5VK8eijj2LatGmYOHGi1/1uuFwubrvtNhQUFCA2Ntaj+W38hT2RplKpRFlZGTZt2oTGxkZ/i+VVzpw5g9WrVyMhIQGPPvoopk6d6rFnG41GfP7559izZw9zTqvV4tKlS07vd1WGtVot3nnnHRQVFfmlfvkKjUaDV199FXK5HHfdddeQ2ycFGn3bRrVaPaxZ47lz5+Lee+9FTEyMU+MIuNI+P/XUU06f29zcjM2bN6OkpGTE8ruLXw0kiqIgk8mwYMECn0TfDDYaaWtrw/79+3Hx4kWvy+AJ7GGVNE2TGa//xWw2o7i4GDt27Bhy1ElRFMLCwjBnzhzk5eU5vcfuJ3D+/Hns2LEjIGYYKIqCQCBAfn7+oKH8nh5xs1gsTJkyBUuWLHEoa33LYDCSmpqK1NRURERE4OOPPx7185wlWwwk1Go11Go1EhMT8etf/3pQ+Vy1J4P9xmw248yZM9i+ffuQcgxVhg0GA06cOIEdO3YM+axgRqfT4ciRIwgJCUF+fn7QteU0TaOiomJEbWNSUhJ+9atfuYw6FYlEuOaaa5xeq6mpwVdffTWsd44UvxlIM2bMwDXXXIPU1NRB/Sc8SU1NDfbs2QOtVjvg2uXLl4PKb+Ls2bN46623oFQqceONNw6579R4ICQkBL/85S+hUChw9uxZHDp0yOnSV2ZmJhYsWICEhARm1rI/Wq0Wu3fvxuXLl3Hs2LGAcBJNTU3FwoULER8fj8TERKf36HQ67Nu3D5WVlcw5b5TtQNSPv2Cz2bjmmmswY8YMpoPr6urC7t27R7Ws6S20Wi22bNmCn3/+ecA1kUiEhQsXOoTy2zEajdi7dy/Onz/v9NrZs2eHfLc7ZXi8EBkZiRtuuAFJSUkOZYcQWPjFQKIoCtdccw1efPFFhISEeD3jL3BlGe21115zuuRis9mCair3xIkTKC4uxrRp0zBt2jRiIOHKuvbSpUtx66234v3338eRI0ecGkjTp0/Hc889h4iIiEHLXUdHBz7++GMcOHDAabSJP8jKysKaNWsgl8sHlVuv1+Ozzz7Dzp07mXPeKNuBqB9/weFwcNNNN+Gxxx5jOrna2lpUVlYGpIHU0dGBjz76yOnSrFwuh1KpdGog9fb24uuvv3ZwQu6LO2XMnTI8XoiKisJDDz2EvLy8ca+LQManX4bL5SIlJQUymQypqakOxpHFYkF1dTWam5u98u5z585Bp9P5xLHL29hsNphMJphMpnE9eu+PvSwplUrk5eU53eYjIyMDAoHA5R5tNE3DbDYHVFlhsVjg8XhO5e7s7ERVVRUaGxvR1NTkEblDQ0ORkpKCqKgoxMbGDlheCzT9uMKuH5qmkZKS4pEZ6776SUhIAJ/PZ66JxWJMnToVRqMRdXV1g/rB+YvBjBmdToeysjJEREQ4vdbY2Diqb+5uGQ7GpJDDhaIocLlcB1/btrY2XLp0CSqVyu/h7UNBURQSEhJw1VVXoa2tDZWVlU7LhlgsRlpamkPAVUpKyqh8J0NCQjB58mT09vZ6vX751ECSSCR47LHHsGDBAkRGRjpYzj09PXj//fcdRr+exL4RIGHsc+2112Ly5MlOjUd79MxYorS0FM8//zxUKhWampo88kyFQoFnn30Wubm5bm8SHKjY9WM2m7Fu3TqPbMXiSj8ymQxPPvkkuru78fbbb+Pdd98NioFMV1cXNm3a5NQny2q1em3wCninDAcbxcXFWLt2LRobG72qa0/AYrFQWFiIa665BocPH8Zzzz0HtVo94L60tDRs2LDBYWeC8PBwhwHFcPFl/fKpgcThcBAbG+vUIdtms6G7uxutra0wGAwOIaOE8Ulvby/a29sREhIyaCJEi8XChOELBALweDxIJJJRbU1iD0PuO9NgNpuh1+t93tEJBAKEhIRALBYP+P97e3thMBjQ1NSEyspKNDQ0eOy9PB4PCQkJSEtLY87ZddDZ2enT1AaDQdM09Ho92tvbXfpw2PVjtVqh0+k88m5n+rFjD6Xn8/kQCARB419itVq9Es3nrzIciPB4PAiFQoSHhw9YWtPpdLh06VLQGIhSqRRSqRQNDQ2IioqC0WgccE9MTAxSU1MdNjS2Y7Vaodfrnc5ostlsCIVCp8uPHA4HSqUSVqsVSqUSUqkUBoPBK+1zwCx+CgQC3HPPPZg7dy727duHL774wqnCCeMDm82Gb7/9FrW1tcjOzsb9998PqVQ64L7q6mq8//776O7uxj333IP8/PxRvzsiIgKPPvoobrvtNuZcSUkJ/v73v/t0FpLD4eDmm2/GokWLkJiYCLFYzFyz2WzYtWsXtm/fjvr6ep/sAH727Fl89NFHaGxsDIiIT4vFgq1bt6K8vNzlfRqNBh0dHT5LUdDe3o4PPvgA586dw7lz54Ji9shbBFoZ9jczZ87E3XffDYVCMWgG8WBj0qRJWLdundNJjaioqEE3kK6vr8d7772HmpqaAdeUSiUeeOABpwMQOywWC7/85S+RmJiIs2fPeqV99rmBNNhoisfjIT8/H/n5+dBqtT4L4yMEJjRNo7S0FKWlpWhra8OyZcucGkjNzc3YuXMnWlpaMHfuXI8YSCKRCPPnz3c4J5PJ8OWXX/rUQGKz2cjNzcXy5cudhpGXlJTgiy++8FkKgrq6OnzzzTcBM8K12WwoLi5GcXGxW/eHhYX5ZDZHp9Nh//792L9/v9ffFegEWhn2N8nJyVi6dCnCw8P9LYrHkMvlKCwsHPbv2tvb8d133+HMmTMDrmVkZKCwsNClgURRFKZMmYIpU6ZAKpV6pX32iYGkVCoxb948xMbGIikpyek9RqMRhw8fRmVlJX788cegiiojjH3i4+OxfPlyNDQ04Mcff3Q66glWuFwu8vLyHFL+e3undl/iTvvjLhMnTsRVV12FxMTEMaMfgnfpW79mz549qP/NhAkTcM899zhNReMKjUaDgwcP+mSzWK1Wi507dzqkErHXLzabjR9++MGhbZTL5bj22muH7fLQ2dmJbdu2OSSDnDBhAubMmePTLch8YiClp6fjueeeQ0JCwqAZsnt7e/H555/jq6++gsViCZoIGcL4YOLEiXj++eehVqvx2GOPjSkDicfj4bbbbsO9997L+IhQFDUqR8pAwp32x11yc3Px0ksvISIiYszoh+Bd+tYvHo83aBmcMmUK0tPTh50249ixYygtLfWJgdTR0YH33nvPwZfsuuuuw7Rp08DlcvHxxx/j22+/Za7l5eVh8uTJwzaQmpqasGnTJof3FBQUYPr06WPHQFIoFIiJiUFGRgYkEonDP9bU1ITGxkamMGi1WqjVaqfrmCwWC0qlclTRNFqtFrW1tQHhXOopent7mcRto9VPIKPValFaWorW1lbmXHR0NOLi4pgdoTs6OoYVvm0wGFBbWwu9Xj/gGo/HG+AvwWazERoaColEgokTJ0KtVg8ow4EMm81GfHy8Ux0JBALExcVBJBK5fEZ4eDimTp3KTGPbbDZoNBpoNBqvyOwp7N9uuA1raGgoEhMTHaIeU1NTIRaLfb6RdrAREhKCxMREREREICoqKmgc1b0BRVHg8XhD1i97OR0ufD7f61sO2aFpekD6lObmZpSUlIDL5aKpqcmhDzcYDAN88FpbW1FfX4/z588PGozl7D0Gg8Hnba3XDCQ2m43CwkLcf//9CA8Pd/AfoWkau3fvxubNm5mlNIvFMmgEA5/Px913341bb711xPIcO3YML774YsA35sNBpVJh3bp1kEql+MMf/jAq/QQyZWVlWL16tcPI65577sEjjzyClJQUbNiwARaLBXFxcW4/U61W4+WXX3aaAVihUGDt2rWYOXPmgGsSiQQrVqzAXXfdhU8++QRvv/12UBjdQqEQDz74IG688cYB19hs9qBZxfuSk5ODN954g6mzVqsV77//Pv7+97+PSUfkxMRErF27Funp6cy5yMjIMZcmwhvI5XKsWbMG06dPh0Kh8Lc4BC9ib58pinIrJ9GhQ4ewceNGtLe3B2Qy1b54dQZJLpcjOzvb6W7jdqvT2VKaPYGWfdQhEokwYcIETJs2bcSyNDc3+3QDXF/Q29uLCxcuQCgUoqGhATqdzi9Wtrfp7u52iFSiKIqZuRGJRMjIyBj2M41GIyorK506CLa3t6O5uRk6nW7AlDiHw8GECRNA0zT27dvn15Exl8uFSCSC0WiE0Wh0+O7961BYWBhSUlJGVYf6p0+wWq2Ij4+HWCx2KoO/sX+70NDQASNss9kMo9HoMMLl8/kOSQylUikyMzMxZcoUn8o9FggJCUFaWprTTXHtswP2ZLdjFRaLBT6fD5FI5DIxrR2r1Yre3t5hv8fZLI0v6d8+D0VraytKSkqCIpVPwIT59yU/Px+33XYbs8bP5XIxe/ZsP0sVuBiNRmzZsgWlpaWorKwcUSUj/B/t7e1455138O233+JXv/oVbrrpJp9NYbsLi8XCjTfeCLlcjtLSUnz66acOe65de+21KCwsZPKI8Pl85OTk+FQGf8JisbBo0SLcdNNNSEhIGBA1dPDgQWzbtg319fVoaWmBQCDAsmXLMGvWLOYemUzm1swaYXi0t7fjk08+QVlZGX7++ecxOfsIAHFxcfjtb3+L5ORkt/qvkpISfP755yNy0m5paRmpmAQXBKSBNGnSJPz2t78dcs2WcAWLxYLDhw/j8OHD/hZlTKDT6fDdd98xS0/OlqX8DUVRyM3NRW5uLr7//nts3bqVMU4oisLUqVPxu9/9zquzpq5k8Dd22e6///4BM9j28PIPP/yQmcGIiIjA/Pnzcccdd/hD3HFFd3c3/vvf/475NAhSqRS33HKL27O2NTU1+OyzzwImjQbBj5vVZmVlYfny5QPC+SmKwuzZs4ds2LVaLY4ePeq25Xzu3DmnDrkEQqBitVpx8uRJfPrpp0hMTMSsWbMQEhIy4L7Y2FjccsstjPO03Tjw9KyXSqVCUVER2Gw28vPzgzbMnaIoZGZmOrQ/QqHQYTuEvlgsFpw6dcohtFkikSAvL49sFE0YEc76r1OnTpHZ/36wWCzk5ORg0qRJmDFjhs+DI/w2g7RgwYJBk/rx+fwhDSSNRoM333wTRUVFbr3PviUFgRAsWCwWbNu2Dd999x0WL16MzMxMpwZSRkYG1q1b57BUwefzPb5LeElJCZ577jmEhITgrbfeCloDCRjY/lAUNajztcFgwFdffYV//OMfzLmJEydi06ZNxEAijAhn/ZfZbA4KvxxfwuVyUVBQgEceeQRcLhdCodCn7/eqgdTS0oLz589DIpFAoVA4NNh8Pn/IPCJGoxGNjY1OtxypqqpCU1NTwO96TPAcbDYbCoUCYWFhiI6OHhehw729vejt7YVOpxvUV4PD4Xh8Gw2TyQS1Wu0wor18+TLa2trA5/NRXV2NmJgY5ppKpQqIaD4ul8uUEZlMNmgZcdX+dHZ2oqmpiXE41+v1aGxsdGhrtFotSWbrApPJhNraWqdpJRoaGsbFYNVgMODy5ctOBzWk/3IPmqah0+mg0WiY9BrOgr68hdcMJKvVim3btuHUqVOYNWsW1qxZ49CgukNtbS3Wr1/vdN+n3t5eXLp0yVPiEoIAe4j9nDlzEBsb6/EZEsL/oVarsWHDBodMtu3t7dBqtWCxWPjzn//s4Pjc2dnp021YBkMqleLJJ5/EL37xCyiVyhEtMx48eBCbNm1ijEObzTamEoP6ArVajfXr1zvkErNjNxzGOiqVCmvXrnU6M0n6L/cwm83417/+hR9++AHz5s3D6tWrh5XvbrR4tYepr69HfX09RCIRuru7ERERAQ6HM2SjZbVaYbFY0NHRgTNnzjg00oTxC4/HQ3p6+ogiGi0Wi8N+T0ajMaiiZ2w2GxNKb4fD4XhsNNVfP52dnSgpKcHx48ed3l9RUeGR93oKFovFzKRNnjx5yDJib2Oc0dDQgKKiIrLcMQp6e3tRVlbmbzH8Sk9PD+m7RglN02hoaIBGo0FiYqLPZ219MgSvrKzEq6++iri4ONx2222YPHmyy/uLioqwY8cO1NfXQ61W+0JEwhjGZDJhx44dOHr0KHOuvb0dtbW1fpRqeJSVlWHDhg3MaJTNZmPhwoVYuHDhqJcax4J+MjMzsXTpUsTFxSElJWXI+3/66Sfs2rXLaYM7WH42AoHgWzgcDm666SbMnTsXaWlpPo9s94mBVFtbiw8//BByuRw5OTlDGkglJSV4++23odPpfCEeYYxjMpmwb98+vP/++8y5QEpo6A5VVVUOU/L22ZIFCxaMehZpLOgnLS0NDz74IGJiYoY0GGmaxqlTp/DWW285NYSC7X8nEMYqbDYbc+fOxapVq/ySi85nThw0TaO3txfHjh1zmMp3dt/PP/8cEA6fhMAgOjoaOTk5UCgUo0rc54mOz2g04ueff0ZjYyNKS0t9ukzXV36r1Yry8nJs3bp11AZSb28vLl++HPSGAUVRTo2j5uZmFBcXMz5FNE3j3LlzsFqtQf8/EwjBxoQJE1BQUMDstWa1WlFSUuLg5ycWi5GTkwO5XI60tDS/Jer1qZerVqvFu+++i48++sjlfb29vU4j1wjjk8zMTGzYsAHx8fEej9YaLl1dXXj33Xexe/du9PT0+C2SyWazYefOnThw4MCon2WPFBmrlJeXY82aNWhsbGTO+fPbEQjjmTlz5iAnJ8chSnTt2rUOBlJMTAyefPJJzJo1y68Jo31qINlsNnR1dfnylYQxAI/Hg1QqhUwm87coTBlubm72tyjQ6/XjIlx6pHR2dqKrqwt1dXVobm726zcLCQmBTCZjIi9pmoZWq0VHR4ffZCIQ/EFoaKhDwseenh7ExcUhKSmJOZeYmIjY2Ngh84zZ63hTU5NXBjwkTppAIIw5bDYbtm/fji+++ALNzc1+T0GQlZWFJ554AtHR0QCuGEhbtmzBRx99RGayCOMaPp+Pu+++GwsWLGDOCYXCIYMtfFHHiYFECFgoigKLxQKbzfZ5Ukh72HigbVJLcA+aplFdXY39+/e79Hn0BhRFgc1mO/iGRUVFIS8vD/Hx8QCuNO6nT58m5Ysw7mGz2cjIyEBGRsawfueLOk4MJEJAQlEU5s2bhzlz5iA1NRUSicRn7w4PD0dBQQFSU1MxZ84c0okRhkV4eDjuvPNOXHXVVcw5mqbx8ccfO/w9VMAKgUDwL8RAIgQkLBYLc+fOxZo1a3w+kyORSLB8+XLMnz8fLBZrXGxpQvAc9vLTN0Ju3759ePjhh1FXV8ecs9lsJIqOQAhgiIFECEhomsalS5dw4MABREVFISsry+meRq7o7u5GaWkpmpuboVKphrw/OjoamZmZiI+Ph0wmc1giaWtrQ1lZGRobG9HU1DTs/4cwvmCxWMwSQHV1NYqLi6HX68mMEYEQRBADiRCQ2Gw27NixA4cOHcLcuXPx6quvDjsHUm1tLV588UWUlpa6tSlkTk4O1q9fD7lcDqlU6nCtrKwMf/zjH1FXV4f29vZhyUEYn9idSDdt2gS9Xk/KDYEQZBADiRCwdHV1MWHaarUaXC4XYWFhg+7Cbqe3txdarRYajQZ1dXWor693eb9YLIZQKIRCoUB8fLxDaKlOp2N2c6+rq0NDQ4NH/jeC9xGJRJDL5ejp6UFXV5dHk3paLBa0t7dDo9G4vKe+vh51dXVk5ohACEKIgUQIeOyJ/mJjY/E///M/mDlzpsv7f/jhB3z88cdoampy8PlwBpfLxW233YaCggLExsY6JKK0J2PcsmUL1Go1mQEIIlgsFpYsWYKMjAwUFxdj8+bNaG1t9djzNRoNXn311QEzjX2x2Wy4ePFiUG2KTCAQ/g9iIBECnpaWFuzduxdyuRyFhYVDOrZevnwZ3377rVu7sbNYLEyZMgVLliwZ4IxN0zQqKiqwY8cOMgMQZFAUhfT0dKSnp4PP5+Pjjz9GW1ubx5yidTodDh8+7JFnjRWGE8xAnNPHDzRNj+p7OytXNE37JMiBGEiEoEGv12P79u2orKx0ed+xY8fIXn4EhqSkJDz44IOor6/H3r17UVVV5W+RxhRcLhfz5s1Ddna2W0ZSV1cXdu/e7VbgBCG4aWhowEcffTTiXRA4HA6uueYazJgxgzmn1Wqxe/duXL58GceOHfPqDC0xkAhBQ3d3Nz777LMhQ/6tViuZ8SEwpKWlYfXq1VCr1aivrycGkofh8XgoLCzE/fff79b9tbW1qKysJAbSOKC2thZ/+ctfRpwqhc/n4//9v/+H3Nxc5hkdHR34+OOPceDAAa9vOE0MJD+SkJAApVLJfHij0Yjq6mqf+roEggzDwVPbMoSGhiIlJQVRUVGIjY11qMBGoxGVlZVob29HXV0dWQ4IAtra2nDixAkoFAqkpaU5JBalKApcLhc8Ho8k/fQAcrkcSUlJTBqM0NBQxMXFgcfjufV7sViMqVOnwmQyMec6OztRWVnpcC5QCA8PR2pqKgAEdNsYiNA0ParZfIqiBswQ2Z/pi7JCDCQ/wWazUVhYiIcffphpaJqbm7F27Vrs379/3MjgLxQKBZ599lnk5uYOmP5taWnBxo0bcfToUbS1tREn2yCguLgYjz/+OBISErB+/Xrk5eX5W6Qxy7x58/D0008zG46yWKwhNxXti0wmw5NPPungI3j48GE899xzUKvVHpd3tGRlZeGll14Ch8MZF20j4f8gBpIHEAgEw05iyGazoVQqMXHiRMY4CQ8Ph1wuR2RkJHOf2WyGXq932kmz2WwIhUJmh/C+WK1W6PV6WCyWQeXzhAzBBo/Hg1AoRHR0NJKTk5GWlsZcs/+fra2tqK6uxsWLF/0oKWE46HQ66HQ6mEwmaDQah1E+l8uFUCgEi8WCSCRCZGQkent70dvb60eJA4e++nGHuLg4pKWlQSAQjOh9HA4HSqXS4VxDQwOioqLQ29sLvV4fED6EoaGhCAkJgVwuR2pqKng8HtM2kvLjP9hsNsLCwhAREYHe3l4YDAavvYtytXxAURRZWwBA07TTBVSKomg+n4877rgD11133bCeSVEUpkyZgqysLGZ5x2Aw4MSJEw55ey5cuID333/fab6VxMREPPTQQ0hMTBxwraGhAe+//z5qampw++23Y9GiRQMaQE/IALjWz1B68DVXX3017r77bigUCsyePdth9ujUqVP46KOP0NjYiKKiIpc5boZDMOnHH3hSP0KhEDNnzoRcLmfOZWdn4/7774dQKMSJEyegUqmwa9cubNmyxWNLtt7E2+Vn8uTJeOCBB9x2pE1LS0NOTo5DpvnRotFocPz4cTQ2NuLDDz9EcXGx27/1hn5YLBZuvvlmFBQUQKlUYtasWWCxWKT8+Bg+n4+XX34Zq1atYvoonU6HEydOoKGhAVu3bsX27dtHNXgfTD/2i4MeAGhygHalH5FIRP/tb3+jvcVPP/1ET5w40alc06ZNo8+cOeP0d+fPn6fz8/NpPp9Pv/HGG7TNZvOKDEPpJ9COu+++m+7o6HD6f/7nP/+hY2JifFp+yOF9/Vx//fV0bW0t851tNhv92muv0Twez+//eyDqx5+o1Wq6oKDA7/rhcDj0unXraIvFMkBGUn58d7jqv8xmM/2nP/2JZrPZXtEPTdNkic2TXLx4EUeOHIHRaBz1s6ZNm4aZM2dCoVBg2bJlTvf/io+PHzRRXXh4OG6++Wbk5uZi6tSpo9pw1S6DSqXCTz/9FHRRQFwuF3l5ecjMzMTs2bMdMnFbrVYUFRWhpKQEP//8s1u5kwjBDUVRyM7OZkL/Dx48iK6uLn+L5RPEYjHmzp3rsMSVmZkJsVg84meazWYcO3YM5eXlLu+bMGEC5syZw/guBSJ2/cTHx2PGjBlO282+5cfZDFJpaSmOHz8eFLNLgUp4eDjmzZuH+Pj4Af1Xd3c3fvjhB6hUKpw8edK7QTSurCcEgAUZCIcr/fSdQfriiy/o2NhYWiAQjOoQiUT0mjVraKPRSFutVrqnp4fW6/UDjt7eXtpqtTodidlsNrq3t5fW6/W02Wx2ewTnDLsMdXV19K9//eugG6EIhUL6rbfeoru7u2mj0ejwvxkMBvqPf/wjLRKJaD6f7/PyQw7/zJCYTCZar9fT+/bto1NTU/2uA1/pJyEhgd65c6dDO2IwGEY1w9zd3U2vWLGCFgqFLtu1ZcuW0U1NTS6f5e8ZpL76MZlMg8ppLz/9j+7ubvrNN9+kBQKB38uNN/TjqyM9PZ3ev3+/0/6rtraW/uUvf0kLBAKay+V6TT80mUHyLBaLBT09PaOehaAoihl9sFisEY24KIoatuP4YNhlEAqFHvU78BUURYHH40EkEjm9bjKZoNfrSTj/OILL5YLL5YLP54+r0H97XR6pg/Vg2OuQKwwGQ8DXMXf1Yy8/zhhqr0jC0Nj7L2ffwWazwWAw+GS236WTNoFAIBAIBMJ4ZPwMnQgEAoFAIBDchBhIBAKBQCAQCP0gBhKBQCAQCARCP4iBRCAQCAQCgdAPYiARCAQCgUAg9MNlmH8wpCL3BXQQp2r3BUQ/riH6cQ3Rj2uIflxD9OMaoh/XDKYfgMwgEQgEAoFAIAyAGEgEAoFAIBAI/SAGEoFAIBAIBEI/iIFEIBAIBAKB0A9iIBEIBAKBQCD0g2xW60dkMhlkMplb99psNmg0Gmi1Wi9LRSAQCAR7+6zX66FWq5kNxAnjB2Ig+QkWi4XCwkLcc8894HCG/gxarRYbN27E999/7wPpCAQCYfzSt30+ffo0Xn75ZTQ1NflbLIKP8ZuBxGazHQwDmqZhsVhgs9n8JZLP4HK54PF4mDBhAvLy8sBms4f8TUdHB6RSqQ+kIxAIw4HD4bhVh/tjtVrJrEQA0r997unpAZ/P97dYBD/gNwPpmmuuwU033cQYSXq9Hlu3bkVxcbG/RPIJERERuP3225Geno7Zs2eDxSJuYARCsMLhcHDTTTdh7ty5oKhB880NgKZpHDp0CLt27YLVavWihIThQNpnQl/8YiBRFIUZM2bgscceA4/HA3BlhqS8vHzMG0hhYWG49dZbsWDBgmE1qAQCIfBgs9mYO3cuVq1aNWwDyWq1Ys+ePcRACiBI+0zoi08NJD6fj+nTp0OhUGDKlClgs9lMIeTxeJg5cyYMBsOIn9/Y2IgzZ87AaDR6SmSP09PTgyNHjqC7u5s5FxUVhdzcXISGhg76Oy6Xy+jn8uXLOHfuHGlYvYRAIEBOTg6ioqKG9Tuz2YyzZ8+irq7OS5K5pm/9cofOzk4UFxejq6vLy5KNHRISEjB16lRwuVwAV9qttLQ0MtMwSqKjo5GTk+PQBlZWVqK8vNynbhcURTGHnZiYGCxatAh1dXU4c+YM1Gq1z+QZy0gkEuTk5CA8PJw5p1KpUFJSArPZ7D/B+uBTA0kikeD3v/89Fi1aBIFA4OCDJBQKcd999+GOO+4Y8fP/+9//4o9//CNaWlo8Ia5XaGtrw5tvvsnMnAHAvHnzsHHjRiiVykF/11c///znP7F27Vr09PT4QuRxh1QqxcqVKzFnzpxh/a6zsxNPP/203wykvvXLHc6dO4eVK1cSA2kYzJo1C+vXr4dEIgFwpUMViUR+lir4yczMxIYNGxAXFwfgygzbO++8gwsXLvjdL3XSpEnYsGED1Go1Vq9eTQwkDxEfH4/nnnsOkydPZs79+9//xnPPPRcwbZJPDKSQkBDIZDLI5XIoFApER0dDq9WitrYWfD4fMpkMHA4HYWFhCAsLY35nMpnQ2toKk8k06LMpikJERATCwsIQGxuLxMRECIVCAFcqWWdnZ8AoG7gSrt/Z2elwrrGxESqVinHYpCgKYWFhiIiIYO6xnwsLC4NYLPalyOMGgUAAqVSKxMREyOVyREdHD+v3fD4fCQkJSEpKglarRUdHB2jae/tBhoeHQyKRMKNdmUzG1C93iImJQUJCAvR6PXNOp9Ohvb3d751SIEFRFCIjIyEWi6FUKhETE+PQTnV0dKC5udmtb81msyGTyRASEuJNkYMOHo8HqVTKlF2apiESiXy+zGWxWKBWq1FbW8vUL7tsZrOZOGuPEBaLhcjISIfBRGJi4oD2yt6H29PZ0DSNjo4Ov6W38YmBlJWVhSeeeAJKpRKTJk0CTdPYvXs3Pv74Y6Snp+PJJ590OnuiUqnw2muv4dKlS4M+m8fj4Xe/+x1uvfVW5OTk4I033mAMKqvVis8++wyff/55QDf45eXleOqpp5hGk6IoLF26FPfdd59bKQAIniEnJwcrV66EXC5HVlbWsH8vFApx//33Y/Hixdi5cyfee++9US0Zu4LFYqGgoAB33HEHE0HF4/EwadIkt5+RmJiItWvXOiz37t27F2+//TZ0Op3HZQ5WQkJCcNddd+FXv/oVFAoFBAIBc81sNuOrr77Cf/7zH7cMpMjISKxatQp5eXneFJkwQlpbW7Fx40Z8+umnuPPOO7F8+XKyfOoBBAIBHnjgASxYsIA5JxaLER8f73Bffn4+3nrrLWaywGQy4cMPP8R//vMfn8prxye9r1QqxVVXXYWEhAQAV6zC2tpaHDhwAFqtFt3d3U79aTo7O1FUVISzZ88O+mw+n4/58+fDYrEgMjIS+fn5zDWLxYIjR46Ay+XCbDYHrJHU3t6OI0eOMH9TFIXs7GyYTCawWKwBFZSiKCa0mPghjRwWi+UwQpXL5ZgzZw6io6Nhs9mGrVuKojBp0iRkZGTgwoULXmtYWSwWuFwuUlNTsWDBgiFDzPv/HxRFgcViQSwWY+bMmQ7X1Go1QkJC0NvbOybLVv9v7g48Hg+ZmZkOjTtwZTbYaDTiwoUL2L9/v1sGUkxMDJYtWzZuUpoMhb0s9vVHtdc9f+jHYDDgzJkzYLPZuOqqq7w6Azwc7HrqC03TQVGG2Gw2QkJCMGXKlAF1qD9yuRxyuZz522g04sCBA+ByuX4pE36fnqirq8Pbb7/tNMePRqOBRqNx+XuLxYK9e/dCq9U6bfhYLBaeeeYZVFVVYceOHQG13DYYNE3jyJEj2LBhAyZMmIDCwkIH/cyYMQNr1qxBbW0ttm3bNqSOCAORyWQoLCx0GMFMmjQJAoEAnZ2d2L59O6qrq0f8/JMnT3rF0TA8PBwFBQVITU3FnDlzhjTCVCoVtm3bhra2NubcL37xC9xwww2Mo3Ffpk6ditWrV6Ourg7bt2/3mz+VNxCJRFi8eDEyMjKG9Tsej4fc3FyHc62trdi2bRuqq6tx/PhxtztSnU6Hr776Cj///DOOHTs2Jo1Qd6EoCvPmzcOcOXOQmpoKiUQCnU6HnTt34vz58+NeP31JTk7GkiVLGIdmmqZx9OhR7N+/P6B1lJCQgMLCQiiVSkyZMmXYv+dwOFi4cCHCwsJQXl6OnTt3OrgEeB2apgc9ANCeOK6//nq6traWtmOz2eg///nPNI/HowHQLBaLZrPZAw4Wi+XW8ymKcvp7Ho9Hr1u3jjYajfR3331Hx8fHj0h+b+vH1f909dVX0xcuXKD7YrPZaKvVSp86dYrOzs72mgyBrJ/RHhkZGfRPP/1EWywW5rBarTRN03RNTQ29cOFCp2XK3YOiKK/oJzExkf7+++9pi8VC22w2eiiOHDlCZ2RkOMj2+9//ntbr9U7vt5etc+fO0bNnzx5T5ScmJob+5ptvHL65O4fVah2g6/Pnz9NXX331gG/tzmFv79z9XTDWL3cONptNr127ljYajUzd02g09M033+x3/bDZbHrdunW0xWJhvrlaraYLCgr8oquFCxfSly9fZsqk2WymX331VaYPDdTyk5+fT5eXlzutQ+5is9loi8VCb9myhY6JifG4jIPph6Zp780gURSF5ORkJCcnMyHsJpMJ5eXlaGpqwsWLF5npstFOm9H/m1PETkhICLKyshAdHY3k5GRwOJwRTa37E/v/RDsZmdrDUIPtf/ImHA4HGRkZiI2Ndev+hIQEREZGgqIoVFZWora2lrnW1NSElpaWgBqZRUdHIzMzE/Hx8ZDJZE6X1ez1q7m5mTlXVlY2YAlbpVJh3759Th2FY2NjkZGRAYlEgtmzZ0MkEg3QTzDQt/2x15HIyEjI5fJhZ722Wq2oqKhAfX09c06lUqG9vX1EZSQYlkV8BUVRYLPZzEwo/b/LRoFU9+zw+XxMmzYNvb29qKmpQWVlpc+W4PouRQJX9JScnIyFCxeipaUFpaWlfo9qti9F93W6zsrKglgsdjrT3d3djdLSUgcfSHv709f31l5G5HI55s2bh4aGBpSXl6O9vd27/xC8uMTGYrGwZMkSrFixAiKRCJGRkWhvb8fmzZvx/fffo7u722tp9mUyGZ544glcffXVkEgkxMluHCAUCvHAAw+gsLDQrfu5XC6kUiksFgv+/e9/48MPP2QaO4vF4rAkFQjk5ORg/fr1kMvlg24509XVhc2bN2Pv3r3MOaPROOB/+eGHH3D27Fmn9eKWW27BunXroFAo8PTTT6O7uxuvvvqqg36Cgb7tj71TYbFYI9qux2g0MsEedsxmc8CVEYJ3kUgkePTRR3Hvvffivffew+uvv+63fD0UReGGG27ArFmzcPLkSaxevXpULgGewK6fhQsXMuf4fP6gda62thYvvvgiysvLmXP29qdvlKid3NxcvP7661CpVHjqqacc/Ha9hVd9kMLDw5GQkMA0UFarFa2trV7za+Dz+ZBIJJDL5VAqlYiPj4dOp0NTU9OIR3uEwIbL5UIikUAmk0GpVDKBAENhtVrR1dWF7u5uNDQ0QKVSBbQBEBoairi4OMTExAx6j71+qVQql8/S6XSDRqm1tbXBZrOBw+EgJiYG4eHhQZVWgs1mIywsDAKBgCkPrmaMzGYzurq6XA7W9Ho96uvrh9TrWEIoFDqE2dtsNmi1Wo9EZfb9Rvawb5PJhK6uLjQ3N3st8nO0sFgsyGQySKVSRERE+H32XiwWQywWo7a21q/RzjweDxKJBAqFYtA2mKZpaLVa9Pb2Muc0Gg3q6uoc6pW9/XFGaGgolEol/j97Zx7eZJX2/++TtUmaNm26JG26L5TWtlBQSmWpIoKogKOIy4zbuM1PcRlgVPRV8BV13hHHGXFmxIUZB0cYZpTFQRhAWaQsxQplKdB9TfclbZqkWc7vDybPNE2apm3aLD2f63quC56t5zk5y33ucy8mk2ncwmR43EjbnUydOhVPPvkkoqKiMHnyZFgsFuzevRvbt2+HWq0eF5UcZXxJSkrCM888g/j4+GEZAarVamzcuBEXLlxASUmJVwtHFNeRy+V4+umnMWXKFKSmpg6pPS4rK8P777+PhoaGQe8xmUw4f/68u4vq1cyfPx8//elP2Ym3u7sbH330EY4cOTLqdzv6jc6dO4eNGzeirq7OqdcyxftIT0/H008/DZVKhezsbIf3aLVafPLJJzbtp7Oz0+udQPxGQGIYBtHR0Vi4cCEiIyNBCIHJZEJJSQl27drlk9qjgSHv+zPA2G5CwjAMwsLCMG/ePKSmpg6rPjQaDQ4fPowTJ06McSnHF2dtxkr/Ohp4r6dXxaNFLBYjLy/Pzp14sHbR1taGgwcP4sqVK+NRPJ+AYRikpKTg9ttvt8mVuXfvXqfjkavvHvgbEULQ1NSEf//7314/YXobnh7/GYZBZGQkbr75ZruYRv3L1tfXh9OnT2Pnzp1O3+XK2DWe3+zzAhKXy8Xs2bMxffp0ZGZmQiKRQKPRYO/evaisrMTx48d90ihy+vTpmD17NpKTkxEaGmpz7ezZszh06BCqqqq8Oq3KWDKwfgwGA/bv34+SkhKXnm9ubva7lAESiQRLlixBSkrKoPcQQtj2ExISgltuucXGqDInJ8fvogVXVVVh3759DqPxVlZWUs2yCwQEBODWW2916ATR19eHQ4cODan5SU5Oxvz58xETE4O4uDg2nMnJkydx6dIlG2NdytAUFhbi6NGjKCsrG/c2zDAM8vLykJubi7S0NLtteGvZrIoJnU436NjM5/ORn5+P7OxsTJs2bdDxp7S0FPv370ddXd24OY34vIDE4/GwaNEiPPPMM+ByueDxeKiursbmzZvx7bffDuoJ5s0wDIPZs2fj9ddfR0BAgN3+8qlTp7Bu3Tr09PSMmaG7N+Oofjo7O/GPf/wDX3zxhUvvsGoY/QmpVIqf/vSnQy4INm3ahGPHjiEqKgpPPfWUjVqcw+H4XfT2srIy/OY3v3GonbBYLH7XDsYCkUiEZcuW4c4777S71tPTA61WO6SAlJGRgZdeegkKhQI8Hg8WiwXffvst3n77bRiNRvo7DANCCI4ePYrXXnsNer1+3OuOw+Fg3rx5ePHFF8Hn823GjP5l658mbLAyCgQCLF26FI8++qjT8efixYt4++230djYOG7f63MjYXh4OOu6D1yt3NjYWAiFQrS1taG8vBw1NTVoaWlxmsPNm5DJZEhOTmYlZ4ZhkJyc7FA4Aq4a4xoMBq/JeDxe8Pl8JCUlISwsjK0fnU6H0tJSNDc3o6GhwWd+8+HS1taGU6dOQalUIiUlhU2WOhBXhBuVSoWZM2ciNjYWISEhNomT/RGLxQKj0ei3bcNd9O9f8fHxdvZbg7UtkUiE1NRUXH/99U7fn56ejsDAQFgsFly4cAHt7e2oqqqCwWDwSS2/pzGbzejr6/OYYMnlciEQCBw6QVjL5qzPSSQSpKSkICwsDNHR0Q7HIUIIqqqqUF9fjwsXLqCnp2dc5z2fE5Dy8vLwyiuvsG6AVjsUACgqKsLatWvR0NBgEwvG28nIyMAbb7xho74ODQ31u5X8aAkODsYzzzyDefPmsfVTWlrKRkr3pd98uBQVFeH5559HbGws1q9fP6pcXjfccAOuueYa8Pl8m7D+lImNo/7lCgEBAXjggQewZMkSp/cFBgYiKCgIarUaGzZsQEFBgVOvJYp/ExMTg7Vr1yIjIwPh4eEO77HmOvzzn/+M7u7ucU9a69UzMJfLhUQisemoUVFRSElJgVQqZbeYCCFob2+HWq1GaWmpV9vlcDgcSCQSmzQPCoUCycnJDhP2OkIoFCI0NNSpq6PZbIZWq/ULtTWfz4dEImFXtqmpqey13t5eVFZWOk1o7A9YXfP7+vrQ2Ng4pM2BSCSCSCRyeC04OHhQDZQVi8XCbp14q9u1q/D5fMhkMqcJeC0WC3p7eye0lsmany8sLGzQtuMIq6GusxAU/TGZTKitraWG8X6MSCRCaGio0/4UGRmJpKQkJCcn213rP/7U1NTgypUrHjGV8WoBSaVS4YknnkBcXBx7LiEhAWKxGI2Njfjoo49w+fJl9lptba3XG/qFhobisccewzXXXMOeUyqVdobYzpg1axY2bNjgVPipr6/Hpk2bUFZWNqryegOTJk3CY489NuJ8Pv5Ee3s73n//fWzfvn3QezgcDm699VYsW7ZsxFpIa/+6dOkSzpw543N2fP2ZPHky1q1b5zTScE9PD/7yl7+goKBgHEvmXWg0Gnz00UfYu3fvqNsPZeLCMAzmzZuHsLAwp9rBkJAQREdHO7zmLePPuLf+4XxoSEgIFi1a5DC2QldXF/7973/73IAWGBiIefPmDZnV2BkpKSlOPZUA4NKlS9ixY4dfCEhRUVFYunSpy0Eg/RmtVovvvvvO6T0Mw0CpVDo0qHUVX+1fjlAoFENGWO/o6MCRI0f84ntHil6vx5EjR8AwDKKiokbVfigTm8mTJ2Py5Mkjft5bxp8xE5AIISgsLMSHH37IGvtpNBpUVVU5vF8oFGLWrFk2E39MTAzkcjmMRiOOHz9uE5K8sbHRr7PYnz17FqdOnWLdJB3VjzNkMhnuuOMOZGVlsecqKytx5MgRm2im3oZUKsXcuXNtthvT09MhlUqh0+lw+PBhmzZUXl6Ozs5Oh+8KCwtDfn4+goOD7dqPP2N15d+0adOoNEj+3L8GIhQKMW/ePEilUpw/fx4nTpzwi+3psaS1tRWHDh1Ca2vroPdwuVzMmDHDZhyi+D795/fY2Fjk5+ezUdFHw5UrV3Ds2DFUV1d7xfgzZgKSxWLBvn378O2337LnCCGD7kmKRCLcf//9WL58OXuOw+FAIBCgt7cX//jHP/Dpp5+yGihCCAwGw1gV36MQQnDo0CGsW7eO/caQkBC8++67LgtIkZGRePrpp21UnDt37sSPP/7o1QJSSEgInnjiCdx4443sOau3RHNzMzZv3oyvv/6avWaxWAZtB9HR0Vi5ciUSExPx4osvThgBCQAOHTo0qlxF/ty/HCEWi3Hfffdh2bJl+OSTT1BUVEQFpCGor6/Hhg0bUFxcPOg9AQEBWLduHTIzM30+CCnlv/Sf32+66SZMmTLFLQLSDz/8gFdeeQUdHR1eMf6M6Rab0Wh02SWPYRgIhUKIxWK0trairq6Ondx7e3tRV1cHrVY7lsUdFwwGA0pLSxESEjLoPYQQVFZW2rg0crlcXLlyBUVFRS79nYCAAMTGxto0WoVCgezsbKjValRXV3ulvZbRaERdXR0uX74MpVJp42XF4/EQHR2NtLQ0l941adIkhIaGQiqVIjk5GTk5Oew1vV6P6upqv2hTjhhO3/MnHPWvoKAgxMXF2ThGOEIoFEIoFEKlUmHq1Kloa2tDdXW1Vy8oPIF1fLZmVHdm22U2m+3aoVAoREpKCtra2uw0lTweD7GxsZDJZA7fVVNTg46ODrd9i69hMpnYOqivr/eobaB1jGlubkZxcbHLXsTh4eGIjo52mAbIZDKht7fXa/qcV1rgHTp0CBs2bGC9Z0wmk99EPW5ra8M777zjNAEoIQQtLS02K9je3l589NFH+Oc//+nS30lISMBrr71mY7+Vk5ODd999FzU1NVi3bh1OnTo18g8ZI6z1I5PJ8Pjjj7PBw4Cr3lcrVqzAAw884NK7xGIxYmNjIRAI8MADD+DWW29lB5TKykqsW7eO5n3yMxz1r5kzZ+LVV191OaRBfn4+UlNTceHCBaxdu9bGEYTy3/G5vb19RAl85XI5Vq1ahc7OTmzatAkff/wxuxi29vH8/Hy75zQaDd5++2188803o/0En6Wrqwvvv/8+vvvuO7s5wlNcuHABq1evdjme2p133olVq1aNW8LZ0eAxAYnP50MgELBqV4lEApPJhJ6eHtTX16O4uNjpysRX6evrG5FLunX15OqA1Nvbi9bWVvT09EAoFLJZ74ODgxEYGOi1Gdqt9cPlclFTU4Pu7m52Zc/j8ZCQkDDkO4xGI6ueNZlMMJlMUCgUNnZNYrEYYWFhkEgk6Ovrm5DaFn/EUf+Sy+Xo7OwccgvAGphVLpdDLpfDZDINy93d17Fq8a1j80CswWlHOz4LBAIkJSXBbDYjJiYGUqmUtbW0Bs2dMmUKe7/FYoFer0d7e/uYj1sMwyAgIAACgcCuDqzbzv3Hl/HGaDSisrLSqxZ23d3dLpkvWNtWa2urnXebtV71er1Xecx6TEC64YYbsHTpUtaI1JpY9pe//CVKSkomdDwSd9Dc3Izf//73iI6Oxk9+8hPcdNNNni7SsLBYLNi7dy+amppwzTXX4Gc/+5nLoRAKCgrwj3/8gx3EhEIhli1bhjlz5rD3RERE4JlnnsGdd96JL7/8EgcOHBiT76B4nkuXLuH11193KiBZbZCuu+66cSyZdyGVSnH//fcjJycHmZmZNgb+vb29+OKLL3Dy5Em3jc8cDge33HILFAoFOymKRCK7UB6VlZX485//jJqaGpdNDEZKaGgoHnjgAWRkZGDq1Kk220Dd3d3YsmULfvzxR5w7d84rtDe+glgsxr333osZM2Zg8uTJNsKnwWDA9u3bceTIEZSWlnrN9hoA26zwAw8AZCwOhmHIqlWriMFgIFba29vJvffeOyZ/b7THeNePOw+hUEjeffddYrFY2LqurKwk8+bN85n6ufnmm0l1dTVxlT/+8Y9EIpGwzwcGBpIPP/zQ4b16vZ48//zzhGEY2n48dHhD/YSEhJDPP//cpm0UFRWRKVOmTJj6USgUZMeOHQ77yUjHZ0fjz3A5evQoSU1NHZf6iY+PJwcOHHBYDrVaTZYsWeKRNmAdA0dSBm/tX1a6u7vJE0884ZF6dVY/hJDx1SAFBAQgNzcXsbGxmDZtGjgcDlpaWlBQUAC1Wo3KysrxLM6EJTAwEPPnz4dSqURxcTHOnTvnVWrNgTQ0NODLL7+EXC4f8l5CCE6cOGGzZWYNEyESiZCQkIDrrruOXcFwuVxce+21+NnPfobq6mqcPHnS5yNHUyjDQalUIjc3FyqVCjExMTbXRjs+m81mFBYW4q9//Svi4uIwY8YMp7Yner0eJ06csEksfOXKlXFPMeENMAyDrKwsZGZmIisrCxKJhJoCjDPjKiAFBQXhsccew6JFi1ibkoqKCrzxxhsoLS31S5sjbyQ0NBRPPfUU9Ho9fvOb3+DChQusDYA3cunSJbz22msOvR4cYTAYbLYADAYDtm3bhh07duCee+5BZmYmKyDxeDwsXboUt9xyC3bv3o2LFy9SAYkyoUhLS8Prr7+OmJgYiMVim2ujHZ9NJhN27NiBb775BrfffjvS09OdCkgajQabNm2yMcS2ejZNNDgcDhYtWoSVK1dCKBRCIpGgqanJ08WaUIyrgGTNQ9bfhdNoNEKj0aCrq2s8izKh4XA4CAwMhEgk8glPApPJNOoVpE6ng06nQ0NDAy5fvoywsDAolUo2Z5lIJEJgYKDLQhiF4i/w+XwEBQU5zM/njvHZ2vfUajUuX76Mtra2Qe9taWlBY2PjoMFfxwqpVIrIyEjExcVBIpGM6992Rnd3N9RqNWsP1tLS4lPaNJFIBKVSifDwcDbBvBWDwYCGhga0t7d7begGr3Tzp1DGioKCAjz11FNITk7GmjVrJnxuNwplvCgqKsJzzz3nNB6V0WhERUXFOJbqKjk5OVi1ahWio6ORmJg47n/fEWazGTt27MDp06fZc56qn5GSkpKCl19+GQkJCXYeyNXV1Vi/fj0uXbo0aIYNTzMuAhKHwwGPx4NQKASXywVw9cc3mUwwGo1OE9pRRgbDMHZ13h+rq+pE29Nub29Hd3c3jEaj3waJpFDchTWKvVAoHNZzhBAYjUYb28b29na0t7e7u4huITQ0FNdeey0iIyPtrlksFphMJhgMhnE3Rairq0NdXd24/k13IpVKMWXKFKSmptpd02q1KC4uxpkzZ8a/YC4yLgJSSkoK7r77bkRHR7MJ7I4ePYo9e/agpqbGqcqVMjLCw8OxfPlyJCcnY9asWTZh/js6OrBt2zZcvnwZJ06cmFACak5ODpYsWYKYmBjExcV5ujgUilcTFxeHlStXDnsLpKmpCdu2bfNazcBwKC0txd///nfU19ejpKTE08WhjCPjIiDFxcXh5z//OWJjY8EwDAghOH36NH7/+997Rb4VfyQ0NBR33303rr/+erscSBqNBv/85z9x8OBBr/ZeGwsmT56Mp556ymmqFwqFcpWoqCg89NBDwx4nLl26ZJdY2leprq7GJ598gpqamgk3Xk50xkxAYhgGkyZNQlpaGqZOnQqJRGIzUfeLxUAZIzgcjsMEkQNiYfg9XC4XmZmZSEhIwPTp021sICwWCy5cuICysjKcOHGCerBRxoSIiAjk5OQ4jMzd1taGoqIi9PT0eKBkV7U9e/fuZXPQKZVKu3uGm2hWKpXihhtusHlXQ0MDzpw545OL4ok0XrqD5ORkZGRkID093S5Aa2VlJc6dO4fS0tJxN8YfLmMmIHE4HCxevBjPPfccRCKRnQU7hTJeBAQE4P7778cDDzyAgIAAGy8Vk8mEf/zjH/jwww+h1+u9MoEvxfdJT0/Hm2++iejoaLtrp06dwnPPPecxAamkpARr1qyBUqnEb37zG4cC0nBRKpVYvXq1jY3jv/71L7zwwgtoaWkZ9fsp3gvDMJg/fz7WrFkDiURiN/cfOXIEa9euhUaj8XqPvDHdYpNIJIiIiLAxzG5ra4NWq0VHR8eoJHI+nw+5XO7UTZ0Qgs7OTq8PIRAQEICwsDCb0P6jJTo62q5uent70dbWhrq6ugmhKREIBAgLC4NMJkNUVBQiIiLYa0ajEW1tbdBoNKivr0dzczNdIU5AhEIhwsLC2Lx8/eHz+VAqlaxx8WgEGIFAALlcbtMGrchkMoeOFONFX18f2trawOFwUFtb65ZtMev39jfujoqKGtSNXqfTobW11avjsVGcw+VyIZfLIRaLERMTg8jISIceizqdDs3NzT4R22pc3fw1Gg02btyIgoICVFdXjyqXjVKpxMqVK5GRkTHoPWazGVu2bMHnn3/u1YbIGRkZWLlypcPBc6RIJBIkJSXZnCsqKsLvfvc7qNVql5IL+jqxsbFYvXo1UlNTkZycbHNNrVZjw4YNOHfuHMrLy6lwNEFJTk7GqlWrkJCQgEmTJtlci4uLw9q1a9Hc3Iw//elP+Ne//uWhUo4PXV1d2LhxI7Zt2zbqd8XGxmLVqlVIT09nz+Xk5ODdd991mMft9OnTePfdd9Hc3Dzqv03xDEFBQXj66aeRl5eHuLg4ty74PcWYfoHVPdKKVqvFmTNncPDgwRG/k2EYcDgcBAUF4dprr8XMmTMHvddsNuP48ePg8/leGU7A+i0RERGYNWuWXZh/d9PS0oKjR4/6bTTWgTZXMpkMM2bMQHZ2NnuOEAKLxQKNRoNTp07hxIkTnigqxUuQyWTIy8tz6IYslUpx3XXXoaenB19//fWI3m/t41wud9h2PONNX18fiouL3fKutLQ0tLW12WiEQkNDkZeXB8C+r1osFojFYnC5XFgsFrpg8UGEQiGysrIwb948h9ctFgt7+ApjJiBZLBYcOnQIZrOZjU7c3d2NK1eujOq9WVlZWLRokcO8QQPhcDjIz88Hl8vFlStXsGvXLq/ZbmMYBvn5+ZgzZw6Sk5MdRrGluE5AQABuvfVWZGVlsecUCgUUCoXNfcXFxdizZw9qa2t9Or4IxfuZyH28tbUVmzdvxoEDB+yuyWQyLFmyxCYgY1JSEp577jnU19dj165duHz58ngWlzLG6PV6/Otf/0JxcTEKCwt9Jv7emAlIhBAcOXIE33//vc250UqPmZmZWLlyJUJCQoZMC8EwDObMmYNZs2Zh//79OHTokNcISBwOB3PnzsVLL70EHo9HU1yMEpFIhKVLl+Lee+9lz1lX7/05d+4cNmzYgI6ODp9ayVB8j4ncx1tbW/GXv/zFodYsLi4O2dnZNgJSYmIiVqxYwaYCogKSf6HT6bBjxw588cUXPqUhHNMtNkKIW4zuuFwuJk2aBJVKhaysLAiFQnawsVgsKC0tRXV1NSIjIzF58mQ2ESlwdZLkcrkIDw/H3LlzUVtbi4sXL3rMkyIgIAAZGRmIiIhAUlLSmA2cPT09uHjxoo0bpa+62DoiPj4eycnJbN1JpVJER0fbGLt2d3fj/PnzNp5pxcXFMBgMVDia4FjbT2Zmpl2CVivW9tPc3Iyampoh3xkREYH09HR2/OFyuWwf7+jowIULFxw6R1y4cMEvo7oP1se0Wi0KCwttzC/kcjkyMjLA4/G8fiuS4jqdnZ04f/48mpqaUF9f73NG+D5hRSUUCvGzn/0M9913HyQSiZ2b9t///nd88sknWLBgAdavX4+wsDC7d2RkZODXv/411Go1XnrpJezfv388P4ElLCwMK1euxKxZsxAcHDxmq8rGxka8/fbb+OGHH9hzvb29Xu9W6QoMw2DBggVYvXo16yXB4XAQGhpqc191dTVef/11G4N0rVbrl5MRxXX6t5/AwEDI5XKH91nbz/nz512K15KTk2Mz/jAMw/bxCxcu4IUXXkBDQ4PdcwaDYUJlE2htbcW7775r42U7Z84c/PrXv55QWraJQFlZGf7nf/4HZWVlXptmxhleLSDxeDwEBwdDJpMhOjoasbGx7DWz2Yyuri50d3ejvr4eNTU1qK2thVqtZlcmDMNAIpEgMDAQAQEBiIqKAofDGXTFOB7weDxEREQMaT/V09MDrVYLPp+P4OBgp27AhBB0d3fbuE2q1WrU1ta6tPL1FazG+SKRiG0PzhJfAlfd+ft7zfD5fJt8S47qjuL/BAUFISYmxkbbbEWn00Gj0aCxsdElWzWpVAqJRAKlUomYmBiEhoZCo9HAYDCgt7cXvb29aGhoQG1tLerr68fqk3wGs9ls561mHbv5fP6ECEEyUdDr9WhoaPBZe0+vFpBiY2OxYsUKVhXeH7VajY0bN+LChQsoKSkBIQRnzpzB6tWr2ZUJh8PB8uXLsWzZMp9bmRw4cABbtmxBSkoKVqxYgaioqEHvNRgM+Pzzz7Fv3z72nEajQXl5+XgUddwICgrCL37xC+Tm5iI5OXlIN9KYmBi88sorTu3OjEYj/vrXv2LXrl3uLi7FRzl8+DA2b96MpqYm1NbWOr2Xz+fjrrvuwpIlSxAVFYWgoCC0tbVh48aNNh5hTU1NPrmCHi8uXryIl156CRwOB2fPnvV0cSgUAF4uIAUFBSE/Px9Tpkyxu9bT04OjR4+ioKCAPadWq6FWq9n/c7lcTJkyxc4gjGEYNifceDNYmg/rvrv1WmlpKXbv3o1rr712yFxIRqMRZ86cwc6dO8e07J6gvz2CSCRCbm4uFi9e7NKzwcHByM/Pd3qPwWBAQUEBOByOwzr2FWNCivuoqqrC119/7ZJWkcPhIDMzE4sXL2bbam9vLwoKCkYVzmSi0dLSwnq8eWOf8+Sc4Uv4W0oWrxaQxgKJRIIlS5YgOTkZhYWFOHr06Lga7Go0Gmzfvh0//vgjey4hIQELFiwAwzA4cOAASktLcfjwYZjNZtTX1+PTTz91aFdlxWAw+N2qi2EY5OXlITc318YQe2DAx9HC5XJxww03ONRGWSwWHDt2DCdPnvSrTk+heBvx8fFYsGABOBwO9u/fj7KyMk8XiSU+Ph6PP/446urqvK5s3oRWq8XOnTtRWlrKnqusrPRpzemEE5CkUil++tOfwmw247e//S2OHz/uMLLrWNHR0YFPP/3UZstv/vz5mDFjBgQCAbZs2YLdu3fDbDbDbDajuroav/3tb4f07BhNVHJvhMPhYN68eXjxxRdZ+yuGYdwenZXH42HRokVYsGCB3TWTyYQ333wThYWFPud9QaH4EsnJyazTRV1dnVcJISkpKVi9ejXUarXXlc2b6O7uxpYtW2zmtoHBon0NrxSQFAoF4uPjMXnyZLtMwO6Ax+OBy+V6LP/RwAbT0tKCwsJC8Hg8NDU12QhshBCfCarlDoRCIZKTkyGXyxEfH28T0sEZJpMJFRUVQ6YqiIyMRGJios1vP1hb4PF4SEhIwPXXX4+2tjaUlpaOqzA9XPh8PpKSkgb1yhoOOp0OpaWlNHnvCAgICMA111wDnU6H2traIe2YfBWJRIKUlBQbr+KmpiZUVla6vKCIjY2FSqVCZmYmAgMDQQjxOntRhmHA5/MhEAi8rmzehi8LQ47wSgEpPz8fL774ImQymV0kZH/k/PnzWL16NQD4bRoQVwkLC8OqVauQl5cHuVzu8oDU29uLTZs2Yffu3U7vW7p0KV5++WW7DNOO4HA4WLp0KWbPno3vv/8er7zyio2Nm7cRHByMZ555ZtBQ/8OhvLwca9aswZkzZ0ZfsAmGtQ13d3fjgw8+wB//+Ee/jLsVExODtWvXYvLkyey5HTt2YP369S6FE+FyuVi6dCmefPJJSKVSyOVytLa2jmWRKZRh4ZUCkkwmQ0pKikN3/L6+Pmi1WnR2dvqNtNrb24uKigpPF8Oj8Pl8SCQShIeHIzEx0SY3ltFoZGMXSSQS1hVYp9OxtkGdnZ2orq4eMpVNdXU12tranLYdLpcLiUQCHo8HuVwOuVyOqqqqIUMKeApr3YWFhSE+Pt5hXrHhwjAMFAqFTWwpa99zZo9lDa3R333e+vt5k5Cg0+nQ3t7u0M1/tHGyeDweVCoVzGYz5HK53wU+FAqFEIvFiIyMRFJSkk17i4uLQ2hoKBiGgVarddjPOBwOJBIJAgICoFKpkJqa6jFtvhWj0YjOzk4EBASwfZ9C8blWcOrUKXz22WdQq9UTXqjwJ7Kzs/HII48gKirKboK/dOkSPv74YxiNRjz66KPIycnBgQMH8I9//IPd8jIajTh9+vSQf+fkyZP41a9+5VTYiYiIwKOPPoprrrlmdB81TkyaNAmPPfYYu1XhDiIjI/H888/brOgLCwuxefNmp2ETgoKC8Mgjj2D69OnsucuXL2PTpk1obGx0S9lGCyEEBw4cQGtrq0PhpbS01G8izo8Fs2bNwv333w+FQoHo6Giba9dddx1+85vfoLKyEps2bXJorxMaGorHHnsMmZmZyMzM9Iptq7Nnz+Kll15CbGysT/V9ytjicwJSZWUltm/f7lJkW4rvEBMTg5/85Cc2QRytqNVq7NixA319fViwYAGmTp2KkpISfPHFF8O2CaqqqkJVVZXTe+Lj43H77bf7zCAZFRWFpUuX2gRSHS1BQUG4+eabbc6JRCJ88cUXTgUkkUiE/Px8m1AM33//PbZu3eo1AhJwVei+dOmSW9410TwcU1JSsHz5coca/oSEBCQkJODSpUvYsWOHQwEpMDAQ8+bNG3Qr2BP1abUV87W+TxlbfE5AGi06nQ6HDx9GZWUljh8/Tr2TfICYmBjcf//96OnpwaVLl9DQ0DCmv51Go8Hu3btRVlaG6dOnIycnhy1DfX09jhw5MqSQNVZIpVLMnTsXKpWKPZeeng6pVDrmfzshIQEPPvigU/uSoKAgxMfH25xTKpW49957UVNTg6NHj/qVF5DZbMbx48chEomQmJiIOXPmQCQSAbi63Xjttdfi8ccfR01NDQ4fPoyenh4Pl3hkCIVCzJo1CykpKZgzZ86QW1AymQx33HEHsrKy7K7J5XI7zVNdXR0OHTqEhoYGj/WtoRCLxVi4cCGUSiXOnDmDU6dOOdw2Tk9PR15e3qi36QghOH36NIqKiiacEO419A9cOPAAQDxxPPnkk0Sr1RJHfPbZZ0Qmk7n0Hi6XS9atW0dMJhP7fGNjI7n77ruJWCwmfD7fpfd4W/142+GO+rnjjjtIY2Ojw9/cZDKR3t5eUlFRQe68885h/XYjORiGIQEBASQoKIi89dZbxGQysWUoLy8nt95667jXj/WIjY0lu3fvJlqtlj30ej2xWCwO686dWOug/98eePT29tr0N0IIMZvNpLe3l9TW1pK7777b7/qXQCAgYrGY3HvvvaSpqcnm2/v6+ohWqyU7d+4kKpXKY/1rtIdMJiOffvop0Wq1xGAwDNlWLBYL0el0g7YRs9lsc/+BAwfIpEmTiEgkIlwu12P1Ex8fTw4cODDoN+n1etLd3U3Wr19PBAKBw3c8+uijpLm52Wk/ceXo6uoiL7300rDrwxvbjzcfg9UPIWTiaZAIIWyOJIpvwOVyIRKJIBKJYDKZxvy3I4RAr9fb5HGzlkEsFnvUgJPD4bDlGG+sdTBcrGWWSCQeN8YdC/r6+tDX1we9Xm+30ufz+eDz+QgICPAKW5uRwjAMa5zt6v39k9EOhdlshk6ng06nG2kRxxxrHQgEAqc2jDweDxKJZNR91Gw2e61jyESBGdihKRQKhUKhUCY6vrukoVAoFAqFQhkjqIBEoVAoFAqFMgAqIFEoFAqFQqEMgApIFAqFQqFQKAOgAhKFQqFQKBTKAJz6KzMMQ13cABBCHCZTovVzFVo/zqH14xxaP86h9eMcWj/OofXjnMHqB6AaJAqFQqFQKBQ7qIBEoVAoFAqFMgAqIFEoFAqFQqEMYMKlGqFQKJThwOPxMH36dKSkpNhds1gs+OGHH3Dp0iUPlIxCofQnICAAubm5iImJYc+1tLSgoKDAaZLtwaACEoVCoThBKBRi+fLleOihh+yuGQwGrFu3DpcvX6YZ1ykUDxMUFITHHnsMixYtYs+dOHECZWVlVEDyJ0JDQxEeHg6GsTew7+zsRFNTEx2QKZQxRCgUQqlUQi6XIyoqCjKZzO4eg8EwrKSsFMpEhMPhQKFQICgoyKX7+/r6oFarXU5eLJFIoFQqoVAooFAobPpqYGDgiBNFUwHJC2EYBrfccgueeOIJu2zOhBB89dVXeP/996HX6z1UQgrF/4mLi8PLL7+MtLQ0xMfHe7o4FIrPIpFI8Pjjj2PBggUu3V9VVYX169fj/PnzLt2fkZGBl156CTExMUhMTBxNUW2gApKXwefzweVyERsbixkzZkAgENjdc/78eYjFYlgsFhiNRqpJoow7fD5/yFWZ2WyGyWQapxK5Dy6XCx6Ph5CQEEyZMgVZWVl29xBCYDKZYDAYYDabPVDK8YfH44HL5bL/J4TQ8YfiEGsfsiKRSDBp0iTk5uay55y1H5FIhMDAQJf/nkwmQ05ODmJjY0dX8AFQAcmLkEqluOuuu5CZmYnp06fbNLD+zJgxA+vWrUNFRQW2bduGhoaGcS4pZSITEhKC5cuXIzU1ddB7CCE4dOgQ9uzZ43MCxIwZM7B48WJER0dDqVQ6vKelpQXbtm1DWVkZvv/+e78XEkQiEX7yk59g2rRp7Lna2lo6/lAcMnv2bCxatIidw4RCIaZMmWJzT319PbZt24b6+nq759vb21FdXT0eRXUOIWTQAwChB8h41Y9CoSA7duwgFouFDIXFYiE//PADyc7OnjD146uHv9VPXFwc2b9/P7FYLIMeZrOZ/N///R8RCAQ+Vz9PPvkk6e7udtr/SkpKyPXXX0/+E43Y79tPSEgI+etf/2rzG9PxxzeO8a4fhmHIqlWriF6vt2kvAykqKiJTpkwhDMPYHcP9mzfffDOprq522FePHj1KUlNTh10/hBCqQfIGIiIikJOTA6VSiaioKBvD7La2NhQVFcFgMCA7O5t1X2QYhj0oFHcRGxuLrKwsO9u3/kRERCAyMtJp22MYBqmpqVi6dCkaGxtRVFSEnp6esSiyW+Dz+Wz/mjp1qtPvt9JvovELhEIhpk6d6lBrFhgYiLi4OJvfnI4/ExeGYTBp0iSkpaU5bAOZmZngcrkOr1VWVuLcuXMoLS1FZ2fnsPuQWCxGTk4OwsPD2XNTpkyBWCwe/ocMARWQvID09HS8+eabiImJsbPyLy8vx6uvvor29na89dZbNvEdKBR3M2PGDKxfvx7BwcGD3sPlcl3yRpk/fz7y8vJw8uRJPPfcc14tIInFYjzyyCO48847IRKJIBQKPV2kcSc4OBi/+MUvsHDhQrtrHA4HUqnUA6WieCMcDgeLFy/Gc889Z2OXZkUsFg9qInLkyBGsXbsWGo1mRK73crkczz77LObMmcOeEwgELnvIDQe/F5BkMhmCg4PtJFlCCDQaDTo6OjxUsv+WLSYmBpGRkQgJCUFbWxt6e3vZe2pqatDY2IiOjg7U1taiqqqKvVZfX4++vj4PlJzir5hMJuh0OkgkEsjl8lEJCiaTCb29vdDr9V5rhyQUChEWFoawsDBERUUhIiLC6f09PT1ob29HfX2933iRBgQEICwsDAqFAkqlcsg6sMLn86FUKtHZ2cmes9aPxWIZo9KOP9b6GWzCHw6+Xj88Hg9yuRyBgYGIjo5GRESEQwHJisFgQEtLi42zRm1tLZqamlx24bciFoshl8sRFxcHhUIxZDu1zu+NjY0jnif9WkDicDhYsmQJ7rvvPrsfkRCC7du349NPP/WIp03/soWHh0Mul0Oj0WDjxo0oKChg7+vs7ERzczP6+vrw0UcfYffu3ey17u5u1NTUjHvZKf7L8ePH8cwzzyAxMRGrVq1Cenr6iN5DCMHevXuxefNmtLS0oLm52c0ldQ/JyclYtWoVEhISMGnSpCHvP378OD744AM0NTWhvLx8HEo49mRkZGDlypVQqVSYPHmyy8/FxcVh7dq16O7uZs/t378fH3zwgVdrC4eLtX5cFRyd4ev1I5fLsXLlSkybNg0JCQlOhSMAKC0txYYNG1BbW8ueq6+vh8FgGPbfzsnJwbPPPguFQoGMjAyn97pr/PFbAYnD4YDP5yM5ORnz5s1zKCD9+OOPIw4gNVoYhkFiYqJN2To6OlBcXIyDBw86fObChQu4cOHCeBbTa2EYZkS/HSHEZ1dv40FjYyMaGxvR1NSEtra2EWt+LBYLKisr8e2333qlltPafuRyOfLy8px65PWnqakJx44dYzXPA93efbFtyeVyXH/99cN2kZZKpbjuuutszqnVagQEBECn03mt1nAoOByOzY5DREQEZs2a5RbzhoaGBrdoojyFSCRCTk4ObrzxRofXLRaLjU1Re3s7CgoKcOXKlRH/TevvoVAoMGfOnCEFVYvFArPZ7Jbxx3d/KSfIZDIsWbIEycnJmDNnjseEIMrYkZeX51DwHYozZ87gm2++8ZvtkbGitbUVmzdvxoEDB0b0PCEEx48f99pJ0tp+4uPjIZfLXX4uKysLq1evdth+SkpKsHv3bmi1WncW1aew1k9tbS127txpoznwBaxzR/9gg4mJiU5t8ihX6enpwe7du23yEtbW1qKtrW3E7+Ryubjpppswc+ZMpKenD2mIbS1DSUmJW8YfvxSQgoODcf/99+PGG2+0Ww1QfB+GYZCbm4sXX3zRYSBNZ2zZsgXfffcdFZCGoLW1FX/5y19G1XcGria9hf7tRygUDmsBlZmZiWuuucbhd3311Vf47rvvJrSAZK2fixcv4vTp0z4nIPWfO6yMVFs90dBqtdi+fTt27drFnhutVpXH42H+/Pl49tlnweFwhvwd+pfBHeOPXwlIERERSE9PR0xMDMLCwhxqF/r6+nDx4kU0NTXhypUr464SDwgIQEZGBiIiIpCYmAiGYdDW1oYLFy6goaEBTU1NY/a3rfXjSKiwlsEbBQe5XI6MjAw25xXDMEhLS2Ojjg+H6Oho3HDDDWhubsaFCxdsDEytWH+j/poFb66fscIXt4ucIRAIkJ6ejsjISLb9DHfic+barlAokJ+fj/r6ely8eBHt7e3uKPaYYN3iT0xMxLRp0yASiZzebzKZUFJSArVabXeNy+Vi0qRJUKlUbP0EBwcjNzcXgYGBKC0t9Y6gf05wZe6gOKb//KVWq92iNZZIJMjIyEB4eDgSEhJc3pa0CmRu01w7C5IELwhyNZxj4cKF5IcffiD19fVEr9c7DBrV3NxMfv7zn5OYmBgik8nGPdCWSqUif/vb30hNTQ3p6uoihBBy+PBhkpubS6Kjo4lIJBrz+qmurrY7/vrXv5KoqKgRvXes28+cOXPI8ePH2bLW1NSQjo4Oh7/vUHR3d5Pa2lpy8OBBMm3aNKe/ka/Uj68f41U/4eHh5OOPPx5V+3FGb28vqa2tJceOHSPXX3+9V9cPl8slzz//PKmoqCDNzc3EZDI5/bbOzk7yzDPPkNjYWLsjLS2NfP755zb3G41G0tjYSEpLS8mjjz46pgE13VE/rswd7uCzzz5zed7xpvqxHvHx8eTAgQM23zQW81dycjL58ssvSU1NDdFoNC7Xr1qtJkuWLHFL/RBfDhTJ5XIRHBxsow1RKpWIiYmxCSBlpa+vD11dXVCr1airq/OY6pdhGPB4PPD5fPT29qK3txcNDQ2ora11GHJ9pAy3fmJiYqBUKmE0GqHRaEbkZeBuAgMDIZFIEBUVBZVKBZVK5dJzOp0OGo0GPB4PwcHBNquPwMBABAYGQqfTsdmfB6JQKKBSqWyMViMiInzauJJytU+EhYWNythWr9dDo9GAw+EgODjYJqCkSCSCSqWCyWRitZ3ejEwmQ2xsrFNtidFoRFdXF1pbW1FbW+vQazYwMNDGkw24ujUSGRkJmUzmE/GTRCIRoqOjERkZ6dL9PT090Gq1Trdw+Hy+3fjjjxByNadaX1/fiLTODMNAKpXa2BcpFArExMQM2VcH/g7Nzc1u1fL77C+nVCrx9NNPIy0tjT0XFRU1aLCoixcvYuPGjairq8PZs2fHq5h2tLa24r333sPnn3/OnmtqanK7On649ZOeno633noLDQ0N+MMf/oBTp065tTzDhcPh4Pbbb8eyZcugUCiGZUh7+PBhbN68GdHR0VixYgUSEhLs7lEqlfjVr37lsN7FYvGw3J0pE4cTJ05g06ZNCA0Ntetf/khZWRnef/99VFVV4dy5c54ujtdw4MABbNmyxWmImMTExEHHH39itHOHUCjE/fffjwULFrDnpFIpkpKSnD5nsViwe/dubN++nRXM9Hq9W+d3nxWQAgMDMXv2bOTl5Q15LyEETU1N+Pe//+1xo0GdTmcT52gsYBgGQUFBmDNnDmbOnOnwHqvEbbWnCA8Px/z589HY2IivvvpqTMvnCtZQ9rfddht4PJ5LxsLWb6qsrMTXX3+N1NRU/OxnPwMhxO55a/uhTAxcSc3jTBtgpaamBt988w0UCgXuueceh22r/99y5Z3jjdXoeKg+RQhBW1sbDh48OKibtr+nPHL0+xFCUFpait27dzt1IZ8yZQo7/jh7ny/Rf/vJ0dyxY8cOh0GZB9L/Hj6fjylTpmDJkiXDKofZbEZJSQl27do1Zt6yPisguQIhBMeOHcPJkydx6dIlOzWwP5Keno558+YhNjYWUVFRDu/RaDTYu3cvamtrkZeXh9zcXK8b4AghKCgowG9/+1skJiZiwYIFTlX1BoMB+/fvZ907jUYjmpub8dlnn+HIkSPIz89Hdnb2OH4BxVuIj4/HggULEB0djeTkZIf39G8/zrDmRWxvb8fWrVtRWFiI2bNnY/r06ew9QUFBWLZsGbKzs9nxx1smxqCgICxcuBAJCQmYOXPmoEbqhYWFOHr0KMrKygbVbkdEROCWW26BSqXyu77Vf+5wtG10+PDhISdl6/jTP1SGtf34KhqNBtu3b8fZs2ft5g6JRIIlS5YgJSWFvb+urg579+61yViRnZ2NG264gd2eFggEw2o/1vmrsrISx48fH1tnEmcGShhHQ7LhHmlpaeTYsWNODbZMJhNZt24dEYlEhMfjeYWR21gf999/P2lqaiJ9fX0OMygTQkhVVRVZuHAhEYvF5H//939tDDTdaeQ22m/hcrlEIBCQW2+9ldTU1Dj9rTs6OsiDDz5IBAIB4XK5BLiaVZrP55PQ0FDy0UcfOX1+KPbt20diY2P9vv144hjr+rnppptIWVmZ0z7Rv/04O6zjCMMwhMfjEalUSt577z279xqNRqLVasnLL7/MtkdvqJ+4uDjyzTffEIPBMKhhtsViIRs2bCCBgYFOx80pU6aQU6dOEYPBQMxms8N36fV68vzzz3u9kfYdd9xBGhsb2XL3nzsctQNXflPr+OOo/fhy/+LxeA7nDmu7NxgM7PHdd9+R5ORkm+cfe+wx0t7ebnPfYO3HEdb5y9XfYaT1Q3zFSJvD4SAhIcHGoDY2NnZQexqDwYDS0lK0t7ejqqoKBoPB71yW+9O/ftLS0iAWi1npnBCCqqoqGwPwxsZGtLS0OKwXgUCAyZMno7W1FfX19aiurvbY6tdsNsNsNqOlpQWFhYVOXYV7enrQ0NBgo/Im/zEe1Ol0uHLlCr7//ntERkYiMTHRZTfempoa1NXV4cKFCz698pvIWKPq9zeoHgghBH19fS5H3SWEwGQyQa/Xo7y8HMeOHUNYWBiSkpLA5/PZbWFvcxdnGAZ8Pt9hqA+j0YiysjK0tbWhrKwMer3eoY2NQqFAfHw8Jk+ejJCQEIfv0ul0KCsrQ2trKxoaGrxGgzYYbW1tOHXqFEJCQgBctW8Z7dxhHX/8DZPJxNbPsWPHWC2kUChEUlISQkND2Xt5PB44HA54PB4SExMRHh6O1NRUiESiYcewa2trQ3l5OWpqatDS0jIuEfp9QkAKCAjAAw88gHvuuYc9JxAIHHogAUBLSws2bNiAgoICtLW1+bVwBNjWT1BQkI03gNFoxLZt2/DnP/+ZHaT6+voGzU0THByMFStW4KGHHsKnn36K3/72tx7v5OfPn8fq1audeoOYzeZBv0mv1+Ozzz7Dzp07sXTpUrz88ssuZX42m83YsWMH/vjHP6K7u3tUEWEp/onJZMLWrVuxb98+3HzzzVi7du2wHAq8ia6uLrz//vs4ePAg2tvbBzVAzs/Px4svvgiZTDboGKxWq/Hmm2/i9OnTPtFvioqK8Pzzz9sItBNh7hgpFosFO3bswNGjR9lz4eHhWLduHebNm2d3v1gsxuOPP47bb78dMplsREmwi4qKsHbtWjQ0NIxbbkevFpC4XC4kEgmCg4MRFxc3ZL4ko9EIrVaL1tZWVFRUjCr/iy/gSv0QQtDc3IwrV67YreIcrW65XC6ioqJACEFYWJhX2Cb19vaioqJixM+T/xjpNzU1Qa1WDzroWduP9brZbEZdXR1KS0u9NmUGZXBEIhECAgIglUoHtbUxmUzQarXo6OgY0UKAEIKWlha0tLQgIyPDI4mvR0v/cbOqqsrhuMkwDMRiMYRCIaKjo5GSkuIw7UNfXx+0Wi2am5tRUVGBsrKy8fiEUdPT0+OzCWQ9RVtbm43w29nZCbVabWOz1t3dDbPZDC6XC6VSaTNHmc1maLVamz4jEAggkUgczjs9PT0oLy8f02DKA/FqAUmlUuGJJ55AUlIScnJyhrz/7Nmz+PTTT9HQ0OD3whEw/PqhOMfafvpH1z537hxdRfogHA4HixYtwpIlS6BSqWzU/v2pqKjApk2bUF1djdOnT49zKb2DS5cu4eOPP0Ztbe2grvxisRgPPPAAZs2ahZSUlEE1AKdOncJnn30GtVo9qkUNxffQaDT46KOPsGfPHvZca2srmpubHWr/6+rq8OGHH6Kqqoo9N2fOHDzwwAND5lwbL7xaQAoJCcGiRYtctnCvqanBl19+Oa4SpicJCQnBLbfcgilTpni6KH5BbW3thGo//gzDMMjKysJ9993n1A6oubkZu3fvnhALqsFQq9XYsWOHwyCQVgQCAWbNmoX77rvP6bsqKyuxfft2hyl8KP6NXq/HkSNHHF6z2nb1p729HXv27LGJW8Tj8XDPPfdQAckZ6enpyMvLQ0JCwpD7+WazGSdPnkRxcTF+/PFH9Pb2jlMpvZMrV67g2LFjrEGxyWRCcXHxoPE8CgsL8eGHHyI2Nhb5+fngcrk4fPgw60JJt5YmFqmpqbj++utHZCMwkDNnzuDUqVN+rYHjcrmYMWMGsrKyMHXqVK8Z2AdDo9Fg9+7dKC0tZc9dvHhxQoRAoYwtIpEIc+fORXx8vMNrVvf/ixcvoqCgABUVFV5vn+aVAlJeXh7efPNNSKXSIS3dTSYTdu3ahQ8++ABGo3HCexr98MMPeOWVV2xWcINZ+1ssFuzbtw/ffvstbrrpJkyZMgV8Ph+bN2/G119/DaPRSAWkCca0adPwxhtvQCaTjeo9FosFv//971FUVDQu3iaegsfjYfHixXjqqafA5/PdIliOJR0dHfjwww9tbLLMZrNf/0aU8SEoKAgPP/wwbrvtNofXrX2joKAAa9asQXd3t9e3O68RkLhcLmJiYhAaGorExERIpVKn+Yz0ej2qq6vR0dGBurq6CWVgFxYWBpVKhcmTJ9utWE0mE5vjzRWMRiMbVLG4uBh8Ph9NTU1+oYkTiUSIi4uzqaP4+HhwuVy2/Wi1WvZaeXm5xz32xhNH9ZOcnGyXF2kkEEKcutWPFQKBALGxsZDJZFAqlYM6GTQ1NbG2isNZVIWGhiImJobdthMIBIiOjkZgYCB7T19fH6qrq1mjVW9ycSeEuDVXFYUilUoRFxcHpVKJyMhIh2OH2WxGbW0t2tvbUVFRge7ubrYdcjgcqFQqhIWFseOzFUII6uvr0dzc7JHx2WsEJIlEgscffxyLFi2CXC4fUnOkVqvx1ltv4ccff4RarR6nUnoH+fn5WLlyJUJDQ20Sqo6GCxcuYPXq1WAYxuPpWNxFXFwc1q5di0mTJrHnQkNDIRaLUVNTg7feestm/7uzsxMajcYTRfUIzurHVwkLC8Pq1asxY8YMKBQKh95rhBDs3bsXGzduRGdnJxobG11+f15eHn71q1+xUd05HA6USqXNPa2trXjnnXdw8uRJNDY2+vUWI4WSkZGB1157zS7Bd3+0Wi02bdqEPXv2oK2tzUZzJBQK8cADD+DOO++0G3+sYWq2bNnikfHZ4wISh8NBQEAAgoKCkJSUNKRBtjU4W3t7Oy5duoTi4uJxKqnnEQqF4PP5iI6ORlZWlk1DMhgMMBqN0Ov1I1qxdnd34+LFi+4srscJCAjApEmTHBqxW4OJnjlzZtzL5WmsgQLlcjnS09ORmZnptncTQti26IntboFA4NI4YtWYOlPxczgcCIVCmxWtUqlEVlYWgoOD7e43m83Q6XTo7OzElStXPJoU211Yx2eJROL3WekpIyMoKAjp6ek2wtHA/q/RaFBeXu6wT3C5XKhUKofjtFWDdPbsWY9oYj3e4hMSEvDQQw8hNjbWJVf106dPY+vWrWhoaLBxD/R3xGIx7r33XsyYMQOTJ0+20bAZDAZs374dR44cQWlpKXQ6nQdLSvF28vLycNdddyE6OnrQfH0jpbu7G1u2bMGPP/6Ic+fO+WRcICvR0dF46KGHEBMTw55LSUmBSCRyeH9xcTE+//xz1NfX+41X3HDHZwoFAL777jvs2LGD7f8GgwFFRUUeLtXw8biApFAocPfddw8ZBNJKaWkp/vKXv0w4N1KhUIgbb7zRoZut0WjE999/j48++sgDJaP4GpMnT8ZDDz1kYzfjLnp7e/Hvf/8bO3fudPu7xxu5XI6f/OQnLofRqKqqwpYtW/wqTMRwx2cKhRCC4uJifPLJJ15vhD0UHhOQ0tPTMXXqVKSlpQ2Z9sFkMuH06dMoLS3F999/7/OV7i7a29tRUFAAtVrtNytWd5Camopp06YhOTnZLkBgRUUFCgsLUVVVNW7h6scCa5yfzMxM1hBZp9PhxIkTqKurs7tfIBDguuuuQ2JiInJzcwe18bPWj7M+JpFIkJeXZ5NmQq1W4/jx46ivr/d5GzZn7WcgFosFRUVFKCkpwenTp/1Oe9vU1IQdO3bY2FlZ64duuVEAoKGhAV9++SUbkocQgh9++GFI27vQ0FDk5eXZRdgGrm7JFRQUoKGhASUlJZ5zdHCWyRZjlF2YYRiyYsUK0tTURLq7u4fM5Nvd3U2ee+45IpPJiFgsHtPM0I6O8a4fR0dISAj5/PPPberl7NmzZNasWSQ4OJgIBIJxrRNvq5/+x8MPP0zq6+uJRqOxyza9detWEh8fT6RSqVsyQXuqfrhcLnnppZdIa2sr6ejoIB0dHeTSpUvklltucXi/TCYjmzZtIh0dHaS3t3fQvmatH5lMNuiRlZVFDh06ZPPct99+S6655hoSHBxM+Hy+x+onPj6eHDhwwOl4YrFYyP/93/8N2mectZ+BGAwG8vLLL5OQkBAikUjGZGzyZP/icrkkKCjI5vd//vnniVardVgfn332GZHJZOPSr7yhfnzhGOv64fF4dm1EJBIN+VxWVhY5cuQI6ezsJAaDwaYdXb58mSxYsIDIZDIiFAo9Uj+EEM9pkAICAiCTyZx6qxkMBjQ0NKC9vR0NDQ0TblsNuOqKrVQqER4ebqdpM5vN6OnpQVdXl4dK5z1wOBwoFAoEBQUhLi4OISEhrK0I+U8uts7OTlRVVaGtrc1nA+Px+XwolUoEBQVBpVIhJCSE9dQymUxITExEWloa2tvb0dLSwq68GIaBRCJxGN/I1fqRSqWIjIxEeHg4G9OktbUVra2tKC8vR1tbm1+0RaFQCJlM5tSbr6+vD2q1Gp2dnairq0NHR8c4lnD8MJvNdp5DzrRkwcHBSE1NRWtrK9Rqtd9p1Cj2mEymYXmXyWQyREZGIikpCeHh4TYODxqNBo2Njbhy5Qo7JnkSr9aRVldXY/369bh06dKEMsjuT0pKCl5++WUkJCQgISHB08XxWqxhIhYsWIDIyEibgH16vR6fffYZvvrqK7S2tvp0jCe5XI5Vq1bh2muvhUqlsnFjDw4OxooVK3D//ffjb3/7Gz788EOX4oa4Wj85OTlYtWoVoqOjkZiYyGb0/vOf/2yXuNLfsWarP3v2rNMUHRONvLw8bNy4EeXl5Vi/fj3Onz/v6SJRvIwbbrgBK1asQFhYmF1YgIKCArz77rts3CNPM+4CEp/PB5fLdWn/WqvVori42KddsXk8ntNcUP0hhMBkMsFisbB1JJfLMXXqVDZMO8UWhmHA5/MhkUgwadIk5ObmstcIITAajejt7UVpaSlOnDjhwZK6h4CAAKSnp9t8pxU+n49JkyaBEILvv/8eDMOw9SMUCgfNaG82m1FRUTFk/YSGhuLaa69FZGQk+1xtbS1OnDjhVxHXzWYz9Ho9eDwe+Hy+TbBJi8UCk8mEzs5OFBcX4+TJkx4sqWdwVj9hYWEICwtDUFAQ5HI5hEIhTCaTX7UPyuiIjIzEjBkzHGpoW1paUFhY6HHNkZVxFZCkUinuuusuZGZmYvr06X5v5Mfj8bBo0SLMnTt30Ii+/dFoNNi+fTsuXLiA2bNnY9GiRYiNjUVYWNg4lNY3iYuLwz333IOYmBg7b6P6+nps27YNVVVVfiEcjQRn9UNxzMmTJ7F27VrEx8fjnnvusQmFUFRUhB07dqCurg7V1dUeLKXncFY/VsLDw/GLX/wCixYtwp49e3D48GEPlJRCGR3jKqFIJBIsWbIEixcvdklg8HW4XC7mzp2L5557zqXvbWxsRFFRES5evIjp06fjmWee8frcTp7GGqsmNTXVro5bWlqwZcsWjwUZ8wac1Q/FMcXFxTh37hymTp2KefPm2QgAJSUl+MMf/uC3Nkeu4Kx+rISGhmL58uXQ6/VobGzEkSNHJmwfpPgu4yIgRUREICcnB0qlElFRUUMO1JWVlTh37hxKS0uHVLVxOBykp6ePaguqra0NRUVFbsvnJpVKkZOTA4VCgZSUlEHTHVy6dAmXL19GWFgYpk2bxm6JALD590CGUz/+SnJyMjIyMpCeng6pVOq0TU2UgdlgMODHH39EQ0MDzp8/z7rZOmtLA4mNjUVWVpbDPGq5ubk2+REZhsHkyZNxxx13ON1CcXf/GimD1Y8jCCGwWCwghMBsNuPcuXOorKzE6dOnaZgRXK2f9vZ2HDx4EDU1NcjMzHSYxZ0K5RTg6vZ/dnY2YmJiMHXqVBuzk4H9y5vyYY6LgJSeno4333wTMTExQ8Y8AoAjR45g7dq10Gg0Q1rH83g8LFu2DE888cSIO+OpU6fw3HPPuW0Aj4yMxKpVqzBjxoxBg/FZLBbs3r0b7733HmbPno0NGza4vOU4nPrxRxiGwfz587FmzRpIJBKX2tREoKurC3/84x+xd+9e9Pb2jiiK9YwZM7B+/XqHqTSEQiGbgwy4uji57bbbcOONNzp9p7v710gZaf3odDp8/vnn+Oyzz6DX633ayN+d1NXVYf369QgKCsLrr7/uUECiUICrmSAeeeQR3HnnnRCJRDY7I97cv8ZFQLLmfXJmS2OxWNDe3o6enh7U1taiqamJdRFlGAahoaE2g3P/d0dHRyMiImLEApJCoUBsbCz6+vrQ3t5uk+F9JPB4PISGhiI8PNzumslkQltbG3p6etgsxQ0NDaipqQGfz3epceh0OjQ3N3tVQxoPuFwu5HI5xGIxYmJiEBkZ6VDT0d3djY6ODtTX1/vNal8sFkMul0OlUg267WqxWNDV1WUTANNgMKC+vh4ikQghISEOBXYOh4PQ0FDEx8dDpVIhMjLSZaEzMDBwyIjcMpnMZUeFscRR/ThCKpUiJCQE0dHRbF13d3f7dGDRscBkMqG9vR16vR61tbWoqqpCYGAgQkNDWa25deyOi4uDVqtFW1ubXyXvFQqFCAsLsxmHNBoNOjo6Jozm2hnW+gkLC0NUVBQiIiLYa319fWhtbUVnZyc7F3obXmMl3dvbi48++ggHDx5EfX29TaK7gIAA/OxnP8Ntt91m9xyHw0FSUtKoVLkpKSlYv349mpqa8Pvf/x7ffffdiN81FG1tbdiwYQNOnz6NyspKmM1mXLx4Eb/61a/AMAwuXbo0Zn/b1wkKCsLTTz+NvLw8xMXFDapxO3z4MP70pz+hpaXFb1ywc3Jy8Oyzz0KhUCAjI8Pl50pLS7FmzRooFAqsWLHCobbHmk173rx5UCqVTuP/TATmzp2LJ554AhERETZ52CiOMRgM+Otf/4pvv/0W8+fPx1NPPcUKzXw+H8uXL8fMmTPx/fff43e/+51f2W8lJydj1apVNu1k9+7d+PDDD6HX6z1YMu/AWj8JCQmYNGmSzbWamhr85je/wZUrV1BaWuqhEjpnTAUkhmHA4XDA5XKHFGCMRiPOnTuHgwcP2l3jcrlIS0vDvHnzxqScwcHBmDlzJjo6OvD3v/99TP6GFZ1Oh6KiIhshrL29HceOHQNwVeDj8XgO7ZYsFgt7TESEQiGysrIGbQfWuqmpqcF3333nVxq28PBwzJ49m3Wx74/VXsYaIqI/XV1dKCwsREhICO6++26H77b2r7S0tGGXy/q3neHt7bW/dothGMTExODGG2+EWCxmg7F6+zd4ErPZjEuXLuHSpUuIioqy2bpkGAYpKSlISUmBRqNxGhjYl7DObXK5HHl5eTapMkpLSwcNqTHRkMlkdvVjpbu7G6dOnfLqMD5jJiAxDIP8/HzMmTMHycnJDm0aKLYEBgbi9ttvx+TJkzFz5kybgVuv1+Nf//oXiouLUVhY6FWGbN7ARK6fiooK7Ny5E3V1dXY5+WJjY7F06VKoVCpkZma69e8SQnDo0CEcOXLEqQBRW1vrtUEkrfVjzSMFANOnTwefz0djYyO++uor1NTU4IcffvBgKSneRl5eHubNm4f4+HibtkPxL8ZMQOJwOJg7dy5eeumlQTUiFFskEgmWLVuGxYsXg8Ph2GjddDodduzYgS+++IL1rqH8l4lcPxUVFXj//fdRU1NjJ6ioVCo8+eSTmDRpkts9iiwWC44cOYI333zTqRebK1omT2Gtn/4rXGvfa2xsxMcff4wzZ85MqPZEcQ7DMMjNzcWLL77oNAArxfcZMwGJEILy8nJ8++23CA8PR0ZGho2LsJXOzk6cP38eTU1NaGxstLkmlUqRkZGByMjIYdkC6PV6XLhwYdir1u7ubrsyjAStVotTp06ht7cXKSkpiIuLY6+JRCJMnz4dHA4H5eXlqKioYK/135J0hNlsnjARaRmGQWJiIhITE9mJPTQ01MbID7BtP/X19S7Xj0QiQUZGhsPcZFasbbj/b+SNWAUQR0IIwzDgcrluGcSbm5tx8eJF1vDdbDajtLTU4daeNyIUCjFlyhSb/GAZGRmQSqUO+5yzeqVMbKzjdP88iCUlJVCr1bh48eKEGacHIz4+HsnJycjMzPRpm8YxE5AsFgt27dqFQ4cOYe7cufj1r3/tMKBYWVkZ/ud//gdlZWVob2+3uRYbG4vXXnsN11xzjdOJbCCtra3YsGEDa9cznDIPLMNIUKvVePvttyGVSvHCCy/g5z//OTvJh4WF4Ze//CW0Wi1+97vfYePGjRO+MzmCw+Fg8eLFWLFiBTt5Wff8++Os/ThDqVTixRdfxLRp0wa9x2w209+oH0VFRXj55ZfR2toK4KoA0dXV5TMCRHBwMJ5++mk88sgj7DmhUEi3SCijRqvV4uOPP8aOHTvQ3d3tN96zI4FhGCxYsACrV69GYGCgT/evMTXS7urqQldXF2pra6FWqx2uYq0u7nV1dXbX+Hw+FAoFVCqV3TWz2Yyuri6HDVGtVqO2ttZjHkwmkwlNTU2sq3ljYyNEIhGCgoLA5XIREREBs9mM6OhoKBQKdvKNiIiw07IZjUZ0dXWhtbV1QnhFcLlcBAUFQSwWQ6VSITY21uHqvqenB1qt1mn7scIwDKRSqc1KRqFQICYmxi5ZYn/MZrNP2M4JhUKEh4c77AtyuXxEKX0c9a+GhgbU1taipaVlVOUdS8xmM9rb29HU1ASJRGITgoDD4QwrbQ+fz4dcLkdkZCQ0Gg3NTD/B4XA4CAoKgkgksgtOa7FYRuw1K5VKIZFI7M4TQthxztcICgpCTEyMQ6N8nU4HjUaDtrY2r7cVHRc3/4sXL+Kll15yqGpra2tDU1PTsN+pVquxceNGh27xvb29XuEubzKZsH37dhQVFWHmzJl48sknWU2YVUOSlpbGrsADAgKQnZ1t846ysjK8//77qKqqwrlz58b7E8YduVyOp59+GlOmTEFqauqgW0MHDhzAli1b0NTUNGT7EQqFuP/++7FgwQL2nFQqRVJSklvL7inS09Px1ltvOfTak8vldtuSruCof9XX13t9YNLW1la89957+OKLL7B8+XIsW7ZsxNuLMTExeOWVV9DU1ITNmzdj3759bi4txZcICgrCL37xC+Tm5iI5OdktuUT5fD7uuusuLFmyxO6axWLBtm3bsH37dp/R0rrC4cOHsXnzZjQ1NaG2ttbTxXHKuAhILS0t2L9//7CesdrjODIsJYRAo9Hg8OHDXp2E1GKx4MKFC7hw4QK4XC4eeughEELY1A+TJk2yiw3RH0II2tracPDgQTvvJH9FLBazHiKOIISAEILS0lLs3r17SFW2NZv9lClTHA5CvoT12wf2ifDwcMyfP9+tf8cX+pcjdDodCgoKwOVyMWXKlGEZVw+8NygoCHPnzkVPTw++/fZbdxeV4mMEBARgxowZWLx4sVvexzAMeDweMjMzHeYnNZvNOHPmjE+la7HO246w9q/Kykp8/fXXPhGGxWsCRfYnOzsbN9xwA+Li4uyiUV+8eJHN/6NWqz1UwuFTWlqKP/7xj4iOjsbChQudbu0AQGFhIY4ePTps2xp/pqenBwcOHEBpaSkOHz48pF1QREQEbrnlFqhUKjvNnK8x3PYzUny1f40Wg8GA/fv3o6SkxOG1s2fPeqBUFH8lNDQUCxYsQHx8PKZPn+5TQtBgSCQSzJ8/HykpKZgzZ46NaUT//nX8+HGv31qz4pUC0nXXXYdXX30VgYGBdmrMoqIivPHGG+jo6BhRrilPcfHiRVy+fBlxcXFISUlxOsERQnD06FG89tpr0Ov1PvWdY4lWq8WWLVuwe/dulzz6oqKi8NRTTyE7O9st6nBPMpz2Mxp8tX+NFp1Oh3/84x/44osvHF6fSHVBGXvCw8PxxBNPYObMmT4/NlkJDAzET3/6U9x+++3gcrk2AlL//uVL3the+ctwuVwIhUKHebbCwsKQk5OD7u7uQZ83GAwoLS1FV1fXWBZzWFjdhXt6elBcXOzw2/pTVlZGhaMB8Pl8pKSk4Nprr0V9fT2qq6sdbqEoFArEx8dj8uTJCAkJGXb0Xmv7aW9vR21trVfEwLG2H3e61Pf09KCsrMwmiezly5fR29vrMyu8kVJXV2cTN6qnpwcNDQ0T2vtoLCCEoKqqCvX19SgpKfHJ+pXJZEhOTkZUVJSdR1Z3dzdKS0vR3NzMencOhMPhICEhAQqFgj0XGxsLuVzudGxiGAaxsbG4/vrr0dbWhtLSUq+uP+uWoaNvIoSgr6/Pq8vvCK8UkJyRl5eH5ORkp5NEfX091qxZ45X2E62trXjnnXeGjA3R3t5OhaMBBAcHY8WKFXjooYfw6aef4re//a3DiTw/Px8vvvgiZDKZzaDkKi0tLdiwYQMKCgr8Lrlmf2pra7F27VqbbSWNRuMTtgGjgRCCffv24b333rOJ6eSNyTJ9HaPRiG3btuHPf/4zuru7vd7I3xEZGRl44403EBsba+fwYM11WFZWNmj7CQgIwAMPPIB77rmHPScQCIYcmzgcDpYuXYrZs2fj+++/xyuvvDKhtr29Aa8RkBiGgVgshlAodOjyaCUoKGjITOMikQhKpRKhoaHsOaPRCK1W6/HJzmQyOXVJn+iYzWY2G7ZIJLIJe8DlchEVFQVCCGJiYiCXyx2uSKKjo5GSkuJQCO3r64NWqwWHw4FEInGo3jaZTKitrfVKw3hr/fS3S+Pz+ZBIJA6NI00mE7RarUOVdnNzM8rLy73yO92BTqdDe3u7XZgIQgjq6+tx+fJlv9eUjRdisRgBAQEIDAy0sachhKC5uRlXrlxxSRPLMAwkEgkEAgF0Op1HQytYxx+FQoHk5GSH4WZ6e3tRWVmJiooKiMVimznHikQiQVxcnMN8ZEMhFovB4XAQHBzstRG7reOPTCaz0x5Zx5+Ojg6f7GteIyCJxWI88MADmDVrFlJSUiAUCkf8rpCQEDz99NO466672HPFxcX4+OOPvTYnFOUq7e3teP/99/HVV1/hjjvuwJIlS+wGBoZhMG/ePISFhTkUeJ21n1OnTuGzzz6DTCbDY489hpSUlDH5jrHCWj/bt29nz2VnZ+PRRx91GJCtoqICmzZtQkNDg921zs5O1NfXj2l5PYXFYsHXX3+N6upqh9dLSkp8xg7C2xEKhVi+fDluuukmJCQkjCpyclBQEB555BHk5ORgz5492L59u0c06RwOB4sWLcKSJUugUqkcCj796T9/DYTH4yEnJ2fYZbBYLNizZw+bZ7Gjo2PY7xgPJk2ahMcee8xhvkfr+FNdXY3Tp097qISjwOo67OgAQMbrCAkJIZ9//jkZK/bt20diY2NHVDZvqB9vPsaifrhcLlm3bh0xmUxubQefffYZkclkJC0tjRw7dszhPZWVlWTevHleXT/9j5tvvplUV1c7/JajR4+S1NRUj7cRT9aPrx/eXj+BgYHkww8/dNj+9Ho9ef755wnDMC69S6FQkJ07dxKLxUJ+85vfEIFA4JH6cXX8sfavsZi/jEYjefXVVwmXy/Xq9uOv4w8hxHs0SGNNTEwM7r//ftTX1+PIkSOoqqrydJEoTiCEoLCwEB9++CFiY2ORn59vExV5OBiNRhw/fhwXL15ER0cH7r//foSHh9vZAFRWVuLIkSOora31Kc1KTU0NtmzZ4nCVW15ejs7OzvEvFMXvCQsLQ35+PqKjo5Genm5zrbOzE4cOHUJdXR2Ki4sdbq8xDINp06Zh2rRp7LZcUFAQ4uPjwTAMsrOz8fjjj6Ourg7fffedVzndWFEqlbj33nuh0Wjcpo3u7u7G4cOHUVNTg8LCQq9wEnGGX48/zqQnjKMUN9YaJJPJRHp7e0l5eTm59dZbvUoC9/VjrOqHz+cTsVhMFi9eTGpra0f823d3d5MVK1YQiURCHnroIVJfX096e3uJ2Wy2uW/nzp0kPj6eiEQiwuFwvL5+rAeXyyUikYiIxWK7IyAgwOXVu7+1H385vLV+srOzyfHjx4lWqyVGo9GmL126dInccMMNRCwWEx6P5/B5LpdLXnnlFdLV1UW0Wi3RarWkt7eX1dr09fURrVZLDhw4QJKTk8e1flzVIJnNZtLb20u0Wq3btN3V1dXk1ltvJWKxmPD5fK9vP/46/pCJpEHicrkQiUQQi8V+E3fC3zEajTAajdDr9aM2rrcaZ5vNZojFYohEIrt7zGazxw1DR4K13BTKeMIwDAICAhzaHBFCYDAYhvSItBr4Osq3yOfzwefzIRQKvdZAmcPhOBxLRoPFYoFer/cZb1J/Hn8Y4uXqOwqFQqFQKJTxxjvFcgqFQqFQKBQPQgUkCoVCoVAolAFQAYlCoVAoFAplAFRAolAoFAqFQhkAFZAoFAqFQqFQBuDU3/0/8QsmPIQQxtF5Wj9XofXjHFo/zqH14xxaP86h9eMcWj/OGax+AKpBolAoFAqFQrGDCkgUCoVCoVAoA6ACEoVCoVAoFMoAqIBEoVAoFAqFMgAqIFEoFAqFQqEMgApIFAqFQqFQKAOgAhKFQqFQKBTKAJzGQaJQKBSKe+Hz+eBw/rs2tVgsMJlMIISGpaFQvAkqIFEoFMo4IZVKcddddyEzM5M9V1lZia1bt6KlpcWDJaNQKAOhAhKFQqGMExKJBEuWLMHixYvZc8eOHcO+ffuogESheBleKSDFx8cjKysLXC7X7lpNTQ2Ki4thNBo9UDLvJzY2FllZWeDz+QCuqu8vXLiAsrIyD5eMMtYIhUJMnToVSqUSpaWluHjxIiwWC4Cr2zrZ2dmIiYlh7+/s7ERRURG6uro8VeQJC8MMmt1gQhIREYGcnByIRCL23MA27E0kJycjIyMDHR0dKCoqQk9Pj9veLRaLkZOTg/DwcLtrer0eZ86cgVqtdtvfGwkMw2DSpElIS0tz2JbH+7cbq/bjlQLS3LlzsXbtWojFYrtrf//73/HKK6/QQX0QZsyYgfXr1yM4OBgAYDAYsH79epSXl1MbBz8nODgYv/jFL7BgwQL84Q9/wOXLl9nBQSwW45FHHsGdd97J3n/u3Dk8++yztC9RPE56ejrefPNNREdHAwAIIXZt2FtgGAbz58/HmjVrcObMGTz33HNuFZDkcjmeffZZzJkzx+5ac3MzVq9e7XEBicPhYPHixXjuuefsFBme+O3Gqv14pYAkEokQERHhUEAKDg72m9WXTCZz+/eoVCpERkYiKCgIwFUByVE9UvyHgIAAhIWFQaFQQKlUIiIiAtHR0YiLi4PJZAJwtd9ERUUhIiKCfS4yMhKxsbHQarXsuZ6eHrS3t3vdpDSeSKVShISEsIbUFosF7e3tbp0EJyJisRhyudzhzkBMTAwiIyPZ9kkIQWBgoFeN9VwuF3K5HGKxmC2vTCZz+D0jwVo/cXFxUCgUNn3VisVigVAodMvfG0nZrN/K5XIRHR2NiIgIhwKSdfzp6elBa2srOw6NFQKBAHK53O3txysFpIkAh8PBkiVLcN9997mtgwGAUqmkAtEEIyMjAytXroRKpcLkyZPBMAwWLlyIlJQUVtDh8XiYNGmSzXNxcXFYu3Yturu72XP79+/HBx98MKGFgblz5+KJJ55g1fU6nQ5/+tOf8K9//cvDJfNtcnJy8OyzzyIkJMTuWmhoKORyuQdK5TpBQUF4+umnkZeXh7i4OPB47p0+rfWjUCiQkZHh1nePloG/HcMwSEhIcDh39R9/iouL8c4776Curm68i+wWvEpA4nA44HA4XrVqGCsYhkF8fDzmzp1r09FG+v0WiwWEEDAMY+NCTPFfrL91REQEZs2aZWNfFBsbi9jYWKfPS6VSXHfddTbn1Go1AgICoNPpnD5rbW/+QP9BnmEYxMTE4MYbb2QXGj09PdizZ49bFjJcLnfC9U/rmKZQKDBnzhyHmhFfQCgUIisrC/PmzXPL+waO9Y7qhxBio801mUzjqt0d6W8XExOD6OhocLlcuwX7cOcoZ2ON9V1cLndM5AavEZACAgJw6623IisrC9OnT2eNjP0Vi8WCQ4cOwWw2s40lMDAQixcvtlvpD0VfXx/27duH06dPIzs7G4sWLUJAQMBYFJviJTAMg/z8fMyZMwfJycmszdloycrKwurVq6HX6we9p7OzEzt37kRFRYVb/qYniY2NxdKlS220FwPHH4FAgEWLFkGhUIz67wUGBiI1NXXU7/EVuFwubrrpJsycORPp6elUu/0fwsLCsHTpUptFzeTJk23qhxCCQ4cO4ciRI6xQ1N3djStXroxLGUfz2x07dgzffvstqqqq0NbWZnNt2rRpWLhwoUtzvMlkwv79+3HixAm7a2M1BvbHawQkkUiEpUuX4t57750QWiRCCI4cOYLvv/+ePadQKDBp0qQRCUh79uzBRx99hJ/+9Ke48cYbqYDk53A4HMydOxcvvfQSeDye27QSmZmZuOaaa5xqh6qrq3HmzBm/EJBUKhWefPJJG6Fl4PgjEAhw66234pZbbhn135toGl4ej4f58+fj2WefZXcIKFcFpAcffBAzZ85kzw1sGxaLBUeOHMGbb74Js9kMwF6jNJaM9LcjhODkyZN4++23YTAY7Mo7bdo0/OpXv3JJ4DIYDOju7sbJkyftxqSxGgP743EBSSaTISMjAwqFglXJTRQIIWzDB4De3l78+OOPwxZudDodqqurYTabJ7Rx7USDYRi3b9kwDDPk4kQikeC6664Dj8dDeXm5TwtK1jocatyx3kcZPq5sgbS1teHChQus5pIQgitXrvjFeMblcjFp0iSoVCr2XGxsLEJDQ4dsUxaLBWaz2WaeGC8sFgtKS0tx8OBBp7+dXC5HRkaGzbxlLXdQUBAyMjIgkUjYaxkZGRAIBC71p6HazViMgf3xuICUnJyMN954A8nJyQgNDfV0cTxKV1cXNm7ciE8//XRYz1ksFnR2do5NoSiUAYSFheGXv/wltFotfve732Hjxo0eGcAp/sOFCxfwwgsvoKGhAcBVAam7u3vMvZ/GA6FQiJ/97Ge477772HN8Pt/rjdKNRiO2bduGPXv2OBVS5syZg1//+teIioqyu2ad3xMTE9lzUqkUAoFgTMrsbjwuIAmFQkRFRdlI16OFy+UiKCjIxh3SYDCgq6vLq1ckFosFra2tni4GheIULpeLiIgImM1mREdHQ6FQoLe31+v7lyMMBgOam5shk8kQFBREt6ZHSUBAAIKCgtgVvVAodLqV0tPTA61Wi4aGBtTW1qK+vn68iuoyAoEAwcHBiIiIGFb74PF4CA4OhkwmQ3R09JBOE1bMZjM0Gg20Wq2Nh6kn6OzsHHTxHRgYCIlE4tBwPDAwEAqFAlFRUS45jFjR6/XQaDTs+wwGA3p7e10ur8VigUajgU6nQ3d396gdSTwuII0FcrkcTz/9NLKysthzZ86cwcaNG6kAQqG4CWuwuLS0NBQVFflk/6qoqMDatWuhUCjw+OOPOwzOR3Gd3NxcPPbYY+yWCpfLZUNPDMRisWD37t3Yvn071Go12tvbx7u4LpGeno6nn34aKpUK2dnZLj8XGxuLFStWIDk52Sb33lC0tbVh48aNOHPmjNduM3I4HNx+++1YtmwZFAqFjTbMGkhTpVIhJCRkWF6LJ06cwKZNm1ihyGw2o6SkxGVBR6PR4A9/+ANOnjyJsrKyUWsgfUpAIoS4VFFisRh5eXk27pgikWjYW1cUijdjNdjsb1Rs7R/j4eRgTTcwadIkCIVCn+xf7e3tOHjwIEJCQrBw4UK3hy7wd2eT/jAMg9jYWCxatAgymczpvVb7y5KSEuzatcurt2gjIiIwf/58Oy2Is7bCMAxkMhluuOEGl4Sq/u/SarUoKCjAwYMHR17oMYZhGKSlpWHx4sUObYmSk5ORnJw8rHcSQlBTU4NvvvlmSJMRq0H7wP6l1+tx6tQp7Nq1a1h/ezB8QkAyGo04dOgQzp49ix9++AEGg8HTRaJQPAohBAUFBfjtb3+LxMRELFiwAAzD4MCBAygrK8O1116L2bNnU68hF9Hr9dizZ49bUzhwOBzk5eUhNzfXrwUlPp+P/Px8ZGdnY9q0aUNGetZoNNi7dy8qKytx/Phxr9SQDEVhYSGOHj2KsrIyO81XdnY2brjhBsTFxTnMp9Yfs9mMo0eP4vTp06yQ1NHRgerq6jEru7cx3Pk9KCgICxcuREJCAmbOnDmmY5xPCEgGgwFfffUVPvnkE1gsFr8w3KNQRoPFYsHBgwdx6NAhzJ8/HzNmzIBAIMCWLVvw9ddf4/nnn8fMmTN9xhjS0+h0Omzfvh3//Oc/3fZOHo+Hl156Cdddd51fe8AJBAIsXboUjz76KDgczpARpjs6OrB582Z8++23MJvNPhdwlBCCo0eP4rXXXoNer7ebj6677jq8+uqrCAwMHLIuTCYT9uzZg9///vdsPRBCJtQcN9z5PSQkBA8//DBuvPHGMQsQacXjAlJXVxd++OEHdHR0IDExcVCJ22KxoK+vb5xLR6F4L1b335aWFhQWFoLH46GpqQl9fX2orq5GQUEB5HI5UlJSwOPxUFFRgebmZvb5wMBAJCcnIzAw0INf4T24a1ISCoVITk6GXC5HdHS0X2uPgP+6Wg8ljLe1taG8vBw1NTVoaWnx+vE8NjYWKpUKGRkZdloxs9mMvr4+iEQipKSk2GSRT01NhUgkchgIsaWlBRUVFTAajQCuak9qamr8blekrq4ONTU1rHaQz+e7bX5nGAZ8Pn9cFn8eF5DKysqwZs0ahIWF4dVXX8Xtt9/u6SJRKD7F+fPnsXr1agBAU1MTCCHYu3cvCgsLce211+LNN99EaGgoNm3ahN27d7PPpaen46233kJaWpqniu6XhIWFYdWqVcjLy4NcLqfbnP+hqKgIa9euRUNDg42g7o1wuVwsXboUTz75JKRS6aAu+SkpKXjzzTeRkJDAnpPJZINuMxYUFOCNN96ARqMBcFUwGBhp2tchhGDfvn147733WKEnKCjIJ+d3jwtIJpMJHR0d4HA4o5ai+Xw+JBIJZDKZ36cqoVCs9Pb22gVr1Ol0aGtrQ2dnJxtAtLu7G62trdDr9ejt7YVAIEBzc7PLXiYCgQASicTvNSKjhc/nIyYmxu9TinC5XEgkEgQHBzt1f+/t7YVer4darUZpaSlaWlrGsZQjRy6XIzU11eH2qEgkQmhoKCIjI5GcnIykpCSn77LWQUNDA0pLS9HV1TVWxR5TOBwOJBIJAgICbLRmwFXBSKvVwmAwoL6+HpcvX2Y1ZSEhIWMassBoNEKr1aKzs9OtmkmPC0iJiYl4/PHHERcXh+nTp4/qXdnZ2XjkkUcQFRXl94MTheKMefPm4e6774ZSqURERASEQiEefPBBzJ07FwcOHMDf/vY31NfX4+233x7S48jKddddh4cffnhMch5RfA+VSoUnnngCSUlJyMnJcXiPyWTCl19+ib1796K6utrjcX3cAcMwmDdvHsLCwiCXy4dcYBgMBmzduhUHDhxAZWXlsOL6eBuhoaF47LHHkJmZiczMTBvtqEajwSeffILTp0+jpKRkXD0TL126hI8//hi1tbU4d+6c297rcQEpPDwct99++5ACjSuGfDExMfjJT36CyMhIdxWPQvE5GIZBeno67rnnHpt9+ry8POTl5UGj0WDbtm3o6OjAN9984/J79Xo97r33XiogUQBc1QrccsstmDJlyqD3mM1mFBUV4W9/+5vPGWM7Y/LkyZg8ebJL9xqNRpw8eRJffPHFGJdq7AkMDMS8efNsQuhY0el0OHz4sNtc7IeDWq3Gjh07UFNT49b3elxAckZraysOHTqE+vp6XLx40dPFoVAogxATE4P7778f9fX1OHLkCKqqqjxdJJdQKBTIz8+HSCRi3bZHSnx8PObOnQuVSoXo6GiH91gnkf71U15e7nepgjo7O3Ho0CHU1dWhuLjYr4Sjobh48SIKCgpYo3+DweDX81ddXR0OHTqEhoaGMe33rvQvd+PVAlJ9fT02bNiA4uJir/d4oFAmMqmpqfif//kfqNVqPPPMMz4jIMXHx+PFF1+EXC7HypUrRyUgZWVl4bXXXoNCoRjUSFej0WDz5s34+uuv2XMWi8XvvJiamprw+9//HidPnpxwY3dBQQHWrFkDrVbLnvPnOrh8+TLeeOMN1NTUjOl3utK/3I3HBKTIyEgolUqkpqYOauBnsVig1+uh1+uhUqkQFhbGXtPr9aiurkZvby+io6MRERGBpKQkO+NstVqNxsZGXLlyxa8b6UCs9dPR0YHm5uYJtYIbKTKZDCqVyqH7qEajQXV1NWt06O00Njbixx9/REhICOLi4mwGlPDwcEydOhVtbW2orq6GTqcb9D0MwzjtX1a4XC5EIhHEYvGQsV+8CZ1Oh4qKCnR0dLCeRSOFy+VCLBbbGa/2hxAy7PxS3kZYWBhUKhUmT55sl2etvb0dtbW1KCsrQ1tbm09/pyv09PSgpqYGer2ePVdRUYHu7m6bc/6MRCJBYmIi+Hw+amtrHRqgm0wmVFRUoKioCOHh4YiOjh62d6cr/cvdeGQkYxgGCxcuxFNPPYWQkJAhbYaEQiEeeOAB3Hnnney5yspKrFu3DiUlJVi+fDl++tOfsgknrZjNZuzYsQMff/wxOjs7/c6d0hlqtRpvvfUWfvzxR7dGB/Znpk6dildeeQWhoaF2144fP47XX38djY2NHijZ8LC6+Z89exY5OTlYu3Yt4uLi2Ov5+flITU3FhQsXsHbtWly+fHnQd/H5/EH7lz9gDTPC4/G8MlGqN5Kfn4+VK1ciNDTULv1GQUEBfv3rX6O1tdXt9iDeSFlZGdauXctGviaEoK2tbUItxjMyMvDOO+9ArVbjjTfewKFDh+zu6e3txUcffYR//vOfuPPOO7Fq1SqfSAztsaWeTCZDUlISRCLRoAGfrBJjcHAwEhISbIwBxWIxwsLCIJVKER8fj+zsbIfux42NjTh79qxX5/oZCwwGA0pLS1FcXOzpong9AoEAAoEACoUC11xzjUOvlLa2NshkMnZl6O3tqaWlBS0tLZBKpXbbN3K5HHK5HIQQyOVySCQSGAwGm0CJDMMgICAAYrHYaf+yYjabYTAYoNVqfSoKsFarxaVLl0b1Dmv7CQgIGLSOfLV+HBEWFoasrCw77RFwNUr2+fPn/cKmqq+vDz09PU6joLe2tuLChQuj2pr1JSwWC3Q6HXp6eth2L5VKkZ6ejtDQ0EEdOMxmM2pqalBTU4Pc3NxhpZdxpX+NFR4RkAgh+O6776DRaJCWloYHH3zQoRYpOjoaK1euhEajQW5urs21iIgIPPPMM2htbcW1115LY7NQRgSPx8OSJUtw0003ISEhAVKp1OF9kyZNwquvvor6+nps2bIFZ8+eHeeSuh9r/2poaMD27dtx5MgR9lpkZCQefPBBpKamutS/Ll++jM8++wx1dXU4f/78WBfda3C1/UzU+vFVLBYL9u7di6amJqdtv7Gx0WfiOrmD9vZ2/OEPf8DXwsZfBAAAZAJJREFUX3+N2267DYsWLRrbXGgu9q8x+/vj+tf6UVxcjOLiYlx//fVYunSpQwFJLpfjJz/5icPnZTIZFi9ePNbFpPg5XC4XM2fOxGOPPeZ0IFSpVLj33nvR2NiIo0eP+oWAZO1fPT09OH/+vI2AZO1feXl5Lr2rrq4OX3zxxYTYVumPq+1notaPr0IIwQ8//IAffvjB00XxKnp6evDNN9+Ay+UiKioKt9xyy5j+PVf711jhO9aUFJeoqKjA1q1b0dXV5fXh/J3BMAyysrKQmZk57I5x5coV/PDDD063MoKCgpCXlwelUonJkye7/DdEIhFuuOEGBAcH48KFCzhz5oxPZiPvD5/PR25uro1BbVRUlMsRtimOIYSguLgY586dQ3FxsY1Xk68hEAhw3XXXITExEbm5uT5liE9xP4QQnDlzBn/961+hUqmQm5sLkUiE/Px8GzvFlpYWFBQUQK/XD9p++o8/FRUVKCwsdMmGazz6F23lfkZhYSHOnz8Pi8Xi0x4kHA4HixYtwsqVK4edCf3TTz/FuXPnnApICoUCzz33HGbMmDEsr4igoCA8+uijMBqNeP/993H+/HmfN8gUCoVYvnw5li5dyp6z2v9RRo7FYsGePXuwYcMG1v7IVxGLxXjggQewbNkyCIXCcUkUSvFeLBYLvvnmG3z33Xe44YYbkJaWhujoaDz66KM24+6JEydQVlaG1tbWQdtP//Fn69atLo+p49G/PC4g6XQ6lJeXA7g6afWXPg0GAxoaGpzGCGEYBuHh4TaeR0ajEWq1GhqNBi0tLRPKxb2vr8/nJ2wrAQEBkMlkwxaQoqOjkZaWhvb2dqjVavT19SEyMtImpUZSUpLdOVdgGAaBgYEghIyru+lIcda/+iMSiUb0Pa2trazHkq+EQHAnhBA0NTXh0qVLdlpIk8mE2tpadHR0+LyWkWEYNs8lhQKADcHT3NyMK1euQKfTQalU2rSRiIgIpKSkQC6X213rj3X8seZ6FAgE7P2hoaE2fav//D7W/cvjAlJpaSnWrFmDiIgIrFy5EjfffDN7rbq6GuvXr8eVK1cGfV4gEOAXv/gF7rnnHvZcW1sb3nnnHRQWFqKurs7nByfK8MjPz0dCQgJKSkrwxhtvoL6+Hg888ADuuOMO9h6RSDRkgkl/wFn/Gi0WiwU7duzAn//8Z7S1tU2oMBpWjEYjtm7disOHDzu8Tscfir9z/vx5rF69GkqlEi+88AJmz57NXktJScGbb74Jo9GIhIQEl9+pVCqxZs0aZGdn24WSGM/53eMCUnd3N86cOQOZTAa1Wm2jLero6MCZM2ecuqoLhUIsXboUhBBWytTr9bh48SJOnDgx5uWneB9hYWGQyWRgGAYymQwdHR1ISUmx84QcLTweD0KhEIQQmEwmr9RUWvtXSEgIWltb3fZeo9EIg8GAiooKnDhxwuvDHowVhBBUV1ezcXAmKiaTCWazeUJqESc6XV1dKCoqQmRkJJqammAwGMDlcsHj8SCVSp3m6hsIwzAQCoWQyWTIysrCjBkz7O4Zz/nd4wKSFZ1Oh23bttl4BzU3N9Mgh5Rhc/LkSTZh4l133QWJROKwo40GhmEwe/Zs/O///i/Ky8uxdevWCePu29HRgW3btuHy5cs4ceIE1ZBMcPr6+rBr1y4UFBSgpKTEaWR2iv/S3d2Nv/71rygoKMANN9yARYsWDds8Ytq0aXj11VchkUhsgtt6Cq8RkAwGA7755hvs3buXPeeNK3KK91NcXIwPPvgAycnJ2Lx585BBDkfK9OnTMW3aNBw7dgz79u2bMAKSRqPBP//5Txw8eJD2UQr6+vpw4MABbNq0ibaHCUxvby927doFDocDHo+HBQsWDFtAmjx5MtLS0rwmrqHXCEhWaAejAFfbQUlJCb766iu2kwmFQkydOhVKpZK9r62tDUVFRejp6WGf+/HHH21U/WPZ2bylI48HDQ0NOHPmDGpra2l+Pwq6urrwww8/oLGxEZWVlbQ9UABctU28cuUKduzYMWjuRgAIDg7GtGnT7KJvDzam9h9/xmsx6nUCEoUCXO1ku3fvxrfffsuei4iIwG9+8xsbAam8vByvvvoqKioq2HM6nc7vsqN7A2fOnMGvfvUrNDY2jjqxK8X3qa2txRtvvIHz58+zCxQKBQD279+PgoICpwvIzMxMvPfee4OmJxmIJ8YfKiBRvBatVmsT24LD4dgJPn19fWhvbx9VUExrJndXVsBcLhdhYWE+4eIPXNW6hYWFISwsDBKJZFTvMhgMaG1tnZDeahMVZ+3HaDSio6NjwmwtU1ynt7d3yDh8jY2NqKmpQWBgIHsuMDAQoaGhDtOXeGL8oQISZUJjNBqxbds2/POf/3RJQAoNDcVzzz3ncgoOT5OcnIxVq1YhISEBkyZN8nRxKD4GbT+UsaKmpgbr1q2zya82f/58PPXUUzZCkyehAhLFp+FwOOBwOA6NAXk83pA2Qtb9clcNjiMjI3HvvffauLV7sxeXTCZDXl4eUlNTnd5nsVjYUBmDJZ9kGAZcLhdcLpe9n+LfuNp+KJTh0t3djVOnTgG4Oo4zDIOYmBi7DAjWscYToUSogETxaWJiYvDUU0+hvb3d7ppCoYBCoXDr3+vp6cG2bdtw5swZ9lxtba1Pbzv19fVh3759OH36NLKzs7Fo0SIEBATY3ZeWloaVK1eirq4OO3futLH7olAolJEgEAiwYMECTJ8+HVlZWTZjj9lsxoEDB3D8+HFcvHhx3G3dqIBE8WliYmLw//7f/3OozXCmDRkpWq0W27dvt9FMEUK8Wos0FH19fdizZw8++ugj/PSnP8WNN944qICUmpqKqqoqnDlzhgpIFApl1AgEAixatAiPPfYYq0myYjKZsH//fvzud7+DxWIZ93GWCkgUn8FgMODMmTMQiUSIj49HSkrKiIWgnp4eXLx4Ec3NzaiqqhrWdpEvC0P96e7uxvnz59Hc3Izq6mqYzeYhv43D4UAikeC6664Dj8dDeXk5FZQoLEqlEmlpaQ7du5uamlBSUuI3uSIp7sO6fe8Ii8UCs9nskS19KiBRfIauri5s3LgRmzdvxuOPP47Vq1c7jbPhjMbGRrz99ts4ffo0Ojs73VtQH6G6uhqvv/46zp8/P6w6CAsLwy9/+UtotVr87ne/w8aNGydsqhGKLbm5uXj99dcdJkXeu3cvXn75ZbemvKFQxhIqIFF8CpPJhL6+vlFPyCaTCU1NTaitrXVTyXwTo9EIk8mEwMBABAYGsjnshoLP54PP57t9C5PiO/D5fMjlchs7P5VKhZiYGJvYNt3d3dBqtS45TUxkhEIhgoODh92n+vr60NXVRRcpYwAVkCg+Q1BQEH7xi18gNzcXycnJ4PFo8x0NMTExeOWVV9DV1WVzTiwWO32ura0NGzduxJkzZ3DlyhW/2XKkDA9X2o/RaMT27duxa9cu1NfX0wCjTpg6dSqefPJJyGSyYT1XUlKC999/Hw0NDWNTsAmMX84wDMOwx0R0RR6LVZqn65FhGIhEIuTm5mLx4sWjepfVqHqiT+zBwcHIz88f9nO9vb0oKCjAwYMH3V8oH8WVPufpPuRuhmo/hBCYTCacP38eu3bt8rvvdycMwyA6OhoLFy5EZGTksJ6Vy+X485//7HfznTc4v/ilgBQUFIRly5YhOzsbx44dw8mTJ/2q4TgjPT0d8+bNc+iFNFLq6uqwd+9edHR0uO2dwyE+Ph4LFixAdHQ0kpOTR/Wuixcv4uDBg6ipqYFarXZTCSkTFS6Xi9mzZ2P69OlOhaTKykrs27cP3d3d41g6z9He3o69e/eiuroahYWFE2b8HS79209mZuaIot2rVCo88sgjqKurw/79+3Hp0qUxKOn44i3txy8FpJCQEDzyyCPo6+vDm2++icLCwgmzPzt16lS88sorw1bTOqOgoACFhYUeE5CSk5OxevVqxMbGjnpbraioCG+88QY6OjrsApJRKMOFx+Nh0aJFeOaZZ5wKSP/+979x8uTJCSMgtbS0YNOmTTh+/DjtZ07o3364XO6Ixre4uDg8//zzaG9vR0tLi18ISN7SfnxeQLJYLKisrMSxY8cQFhaGpKQk8Pl81iBwMNdBf8Wa3d5ZqPaAgACkpKS4nCRQLpfj2muvhVwuR0VFxbjlXoqNjYVKpUJmZiYCAwNtPNbq6upQU1PjUAWrVCqRkJDg0NiREIK+vj4YjcYxLbu3YM243tHRgcTERISHh3u6SF6PQqFAfHy8S2OHQCBAbGwshEKhw+s1NTWoq6tDaWnphHJvJ4TAaDROqG8eDiKRCElJSQgPD3ep/Tga51QqFeLi4sAwDPh8PgQCgU8ZwUskEqSkpCAsLMwuoK+3tB+fF5BMJhO2bt2Kffv24eabb8batWshl8s9XSyPUVBQgLKyMqeeENHR0Vi/fj1mzpzp0jtTUlLw5ptvoqWlBf/7v/+L3bt3u6u4g8LlcrF06VI8+eSTkEqlNr8pIQT79u3De++957AD3XvvvXjhhRd8JqHsWFJWVoY1a9YgLCwMr776Km6//XZPF8nryc/Px4svvuhS+2EYBmFhYQ6vmc1m7NixA3/84x/R3d3t09HWKe5FqVTi5ZdfxrRp01xqPwMFJIZh8Mgjj+D5558fcagTTxMTE4O1a9ciIyPDaxduPi8gEULQ0tKClpYWpKamorW1lZWizWYzdDqdh0s4vmg0miE9RXQ6HRobGx2m57AGAuzf6QICAhAfHw+ZTAalUonQ0FDo9fohszWPFrlcjtTUVHYlTwiBVquFwWBAfX09Ll++7FAT1NTURG0e/oNer0dVVRU6OzuhVqvR3t6OgICAIT3VHGE0GqHVatHZ2ek3GjgejweJRGKjLYqOjkZKSsqw68haP9bJzGw2s9qjibLF39fXx7YRurX2X8RisY1daEREBBITE5GSksKeG077YRgGra2trCG8VqtFR0eHT/XLgIAAxMXF2diVelv78XkBqT9nz57FSy+9ZNMQz50753FLeG+jvb0d77//PrZv3253TSaT4dFHH0VOTo7dNbFYjAcffBBz587FgQMH8Le//Q0Gg2E8igzgqvD3ySef4PTp0ygpKZkwk4476O3txV/+8hccOXIE8+bNw3333TeoWn8wzp49i08//RQNDQ24cuXKGJV0fElMTMTjjz8OpVLJnktJSRl23QD/rZ/+QTcn2vhz6tQpfPbZZ1Cr1TTC+n8QCoVYvnw5brrpJvZcUFAQEhMTbe4bafupqKjApk2bUF1djdOnT7u17OONt7UfvxKQamtrJ3zgP1fQarX47rvvHF5TKBRYuHChQwFJIBAgLy8PeXl50Gg02LZt25iV0dFeuk6nw5EjR7Bz584Rv3eiapb6+vpQUFCAgoICSKVSLFu2bNhCQE1NDb788ks0NTWNUSnHn/DwcNx+++1uyVbvj/UzXCoqKrB9+/YJG53eEXw+H7m5ubjvvvuc3jfS9tPc3Izdu3f7xaLF29qPXwlIlJGjUqmQn5+PqKgoxMfHO7zHYDDg+++/R2lpKY4cOTImKlCpVIq5c+ciJiZmSNdp4KpAl5+fb+O1N2fOHBtvEKPRyGaDPnHixLhqvbyR8+fP45NPPhm2gPTjjz+O+bbqWGBt244cF5KSkuw8Pi9evIiCgoJht29frZ/h4qx+Jmr/EgqFmDVrls2WWf9r6enpDp8zm804efIkiouLXWo/IpEIc+fORUJCAnJzc/3CCcmbx2cqIFEAAJMmTcIrr7yC2NhYCAQCh/fodDp8/vnn2LZtG5vyw92EhITgiSeewI033uhSKov4+Hi8+OKLNgMTj8ez+QaDwYB//OMf+PTTT73CM8LTnDhxAkVFRcN+zmw2e9Xg5SrWth0TE2N3jcPh2AmKBQUFWLNmDbRa7bD+jq/Wz3BxVj9jNS54OyKRCPfffz+WL1/u8PpgY6rJZMKuXbvwwQcfwGg0Dtl+goKC8PDDD+O2224Dn8/3CwHJm8dnKiBRAFz1GhOJRE49dwghY26czeFwIBKJXDaQtZZ7qPutxn+Uq4OyNxhAjheuthErVqPXiaANGgm0fuxhGAZCoXBEzg/WscmV7f/R/B1vxlvHZ2ai2mRQKBQKhUKhDAZNxU2hUCgUCoUyACogUSgUCoVCoQyACkgUCoVCoVAoA6ACEoVCoVAoFMoAqIBEoVAoFAqFMgCnbv4Mw1AXNwCEEIfRCmn9XIXWj3No/TiH1o9zaP04h9aPc2j9OGew+gGoBolCoVAoFArFDiogUSgUCoVCoQyACkgUCoVCoVAoA6ACEoVCoVAoFMoAqIBEoVAoFAqFMgAqIFEoFAqFQqEMgApIFAqFQqFQKANwGgfJXXA4HPB4PDDMf8MNmM1mmEym8fjzdvD5fHA4/5UNLRYLTCYTCJl4YSG4XC54PMfNwJO/0Whw9E0mkwlms9lDJaJQKP4IwzDg8/k2c9twoWOT9zIuAlJKSgruvvtuhISEAAAIITh06BD27Nkz7g1DKpXirrvuQmZmJnuusrISW7duRUtLy7iWxdNwOBzcdNNNmD9/vo3ACHj2NxotM2bMwOLFiyEQCAAABoMBO3fuxIkTJzxcMgqF4k/ExcXhnnvuQURExIieNxqN2LNnDw4fPuzmklHcwbgISHFxcfj5z3+O2NhYAFcnX7PZjH379o375CuRSLBkyRIsXryYPXfs2DHs27dvwglIDMNg5syZePbZZ8Hlcm2uefI3Gi1ZWVl46qmnIJFIAAA9PT2orKykAhKFQnEr0dHReOihh5Camjqi5w0GAxobG3HkyJEJuYPh7YyZgMQwDCZNmoS0tDRMnToVEolkVGpId+NNZfEkDMOwhy+RnJyMjIwMO80XAEydOtVG7c3n8zF16lTccccddvdaLBZcuHABZWVlY15mivfC5/ORnZ2NmJgYp/fV1NSguLgYRqNxnEpG8Uas4096ejqkUumIx08ul4usrCwsXboUDQ0NOHPmDAwGg5tLSxkpYyYgcTgcLF68GM899xxEIhGCgoLG6k9RJhgMw2D+/PlYs2YNu43WH5FIBKFQyP5fKBTi/vvvx09+8hO7ew0GA9avX4/y8nK6gpvAiMViPPLII7jzzjud3vf3v/8dr7zyCrq6usapZBRvo//4I5FIRjW38fl83HXXXVi0aBH+9a9/4YUXXphwOxnezJhusUkkEkRERNht31A8C4/Hg1wuR2BgIGQymaeLMyQBAQEICwuzMbyOiYlBZGQk+Hz+kM8zDAOpVAqpVGp3ra+vDyqVCnFxcdBqtWhra4PFYnFr+SneD4fDQXBw8JC2JFFRUYiLi4NGo3Hb39br9WhtbfVJh4iJgCvjj8ViQXt7O3p6euye53A4CA0NRWBgoMP3BwYGsmOxI404xRbr/CUSidhzOp0Ora2tbjcHGRcbJIp3IZfLsXLlSkybNg0JCQleL8BmZGRg5cqVNpNXXFzcoN53w4HP52P58uWYOXMmvv/+e/zud79DR0fHqN9L8U/y8vLw+9//3q3CzPnz5/HOO++grq7Obe+kuA9Xxp/e3l589NFHOHjwoN3zIpEITzzxBG677bZxKa+/Y52/cnJy2HOnT5/Gu+++i+bmZrf+LZ8TkBiGGVTKJoQMufq33tNf0vR2jYG7BZjAwEBMmzYNN954o1vf626sv3VERARmzZo1pH3ISP9GSkoKUlJSoNFoHG7ZUa7irO/1x2KxeP125cBv4XK5LtmRKBQKKBQKt5ZFKBRCLBa79Z2U0TOc8cdoNOLcuXMOBaTAwEAsWrTITrvh7QtTb4HD4dj0TUfzl9lsRkBAgNv/ts8JSFlZWVi0aJHDyigpKcHu3buh1WoHfb6npwfbtm3DmTNn2HO1tbVoa2sbi+KOmtjYWCxduhRyudxt75TJZEhISHDb+8YChmGQn5+POXPmIDk5GcHBwZ4u0oRn0qRJuP322wfdKgCu9q9du3bh8uXL41iy4ZOXl4d58+axk5RIJLIJ/UGZ2Lhz/Onr68O//vUvNDY2sufCw8OxZMmSMVn0+RNcLhc33XQTZs6cyQpJ4zl/+ZyAlJmZiZUrVzq0nfnqq6/w3XffORWQtFottm/fbiORuqJ58hQqlQpPPvnkiN1IHeGqJsCTcDgczJ07Fy+99BJ4PJ7Xl3cikJaWhueeew6RkZGD3qNWq3H58mWvFpAYhkFubi5efPFFVmPoi56clLHDneNPX18f9uzZg71797Ln0tLSMG3aNCogDQGPx8P8+fPx7LPPsv1zPOcvrxaQpFIpMjIybLwEsrKyIBQKHaonXVVZeqsw1J/4+HgkJycjMzMTUql03NSxfX19uHjxIpqamnDlyhWP1hXDMOByuePWGcLDw5Gfn4/6+npcvHgR7e3t4/J3vRGJRIKMjAybhcjUqVMhFosdtsW2tjZcuHABDQ0NbrcDcBcCgQDp6emIjIxEWloa+Hw+3eZwER6Ph7S0NERFRTm9r6GhAZcuXfJZg/OAgABkZGQgIiICSUlJLgtH1jARbW1tqKmpQWlpqc2WmjWunFKpRFpaGpKSkqhW/D9wuVxMmjQJKpXK7hqfz0dCQoJb7E1HglcLSLGxsXjttdeQnp7OnpNIJGwAQH+FYRgsWLAAq1evRmBgoFu314aiq6sLGzduxL///W90d3f77EA3EqZNm4Z33nkHNTU1+NWvfoVjx455ukgeQ6lU4sUXX8S0adPYc2KxeFCX5gsXLuCFF15AbW2t1wqWwcHBePrpp3HzzTdDKpV6bND1RSQSCR577DEsXbrU6X1ffvklXnvtNbd6+Y0nYWFhWLlyJWbNmoXg4GCXF2disRiPPvoo7r33XmzZsgVvvPEGdDqd3X25ubl4/fXXIZfLx3Vc92aEQiF+9rOf4b777rO7xjCMRz2tvXKEsMZNUiqVUKlUbATuoRAKhQgPD4fFYkFXVxf6+vrGuKTuQSAQIDg4mF3NMgyD6OhoxMbGuuTG7k7MZjNaW1tRW1s7rn93pBiNRnR1ddkIcmKxeETB2zgcDgQCwahzK/kyAQEBCAoKgkKhQExMzJB9r6enB1qtFg0NDaitrUV9ff04ldR1rP3LOp4Mta2h0+mg0WjcamguFArtJlxr3bW3t3vtQoTP5yM4OBhhYWEujcVyudynt8MZhgGPx4NAIHD6Hdbfzlo/XC6XFXpCQ0MHHT8CAwOhUql8IrzKWGGd3611JJFI2PluuPTvQ2OR8cErBaS5c+fi4YcfRmRk5LAqberUqawGYOPGjSguLh7DUrqP9PR0PP300wgLCwNwtZMmJyfTFa4LlJWV4f3330dDQwN7buHChXj44YdtgkW6wo8//og//elP7DbBRCQ3NxePPfYYFAoFkpKSnN5rsViwe/dubN++HWq12ms1R9b+pVKpkJ2dPeT9hw8fxubNm90a0XjKlCk2fRwADhw4gC1btqCpqQlNTU1u+1vuJCkpCc888wzi4+MnhBF7a2sr3nvvPXzxxRdYvnw5li1b5lBQsv52KSkpWLFixZBbj5T/Yp3freMzl8sdUdsaj/HHq2Zgq0SZkJCA2267bdiur0qlEkqlEtXV1fj73/9uJ8V7o+sxwzCIjIzEzTffPOTK1ln5Ha1YrPe7qg0hhLCHL0AIQVtbGw4ePIgrV66w55VKpUuriYHfWV9fj71793rtZDXWMAyD2NhYLFq0yG6FO7AtWW0qSkpKsGvXLq/O1xcREYH58+cPutga2A4qKyvx9ddfo7e3121l0Ol0ePjhh9ltFUIISktLsXv3bq/WdMvlcsybN29YTiJWt2xfGUf6o9PpUFBQAC6Xi+zsbJjNZofziPW3u/baa/Hwww/bXLcaEQ+sA+oIcJX4+PgRze8DIYTg8uXL+Prrr2EymcakvXmNgCQUCnHTTTchPT0dubm5o9paCgoKwrJlyzB16lT2XGVlJfbt24fu7m53FHfUMAyDvLw85ObmIi0tzWGU5/4YDAbs378fJSUldtcEAgHy8/NtVsft7e3sZD979mxMnz7d6furqqqwb98+1NfX+0RessLCQhw9ehRlZWUjWjlY66f/ltD58+edekD6K3w+n20/06ZNs9G8mc1mHD16FKdPn0ZKSgrmz58Pk8mEvXv3orKyEsePH/cJp4fBqKurw759+2za0PHjx92ea62qqgoffvghQkJC2HOHDx/2asFyJGRkZODpp59GbW0t9u3bZ6PZ9SUIISgoKMBvf/tbh0KNs98uOzsbK1asQH19Pb755ht0dnYO2r8oI8c6hz7//POoqKgYm/m9v9Zg4AGAjPTgcrlk3bp1xGQykYFYLBbyf//3f0QgELD3y2Qy8pe//IUYDAaHzwwXo9FIDAYDe+zevZvExMSM6FvGqn5effVV0tvbS4xG45Df09HRQR588EEiEAjsjtDQUPLRRx/Z3H/p0iUyd+5cIpVKyXvvvUcsFovT9+/fv58kJSURPp9PGIbxivpx1n42bNhAAgMDCY/Hs3v2ySefJFqt1un3Wuunfz06epc7jrGoH3ceEomEfPDBB8RgMNi1Rb1eT1avXk2EQiG5++67SVNTE6mqqiILFy4kAoGAcLlcr6+fm2++mVRXVztsB8eOHSPXXHONTTtwxzcNPBiGIXw+f0R/x5Pt5/rrryeXL1922pf6YzabicFgIGfPniW5ubnj0n7Hqn64XK7D8bb/b+eofqx1cPLkSZKVleW0f/Xnyy+/JJGRkT5TPyM9XBmfXcVkMo3Z/E4I8bwGSSqVIiUlBREREVAqlW6LZDzQficiIgLTp09HZGQkKioqPG4vQQhBfX09CgsLERoaitTUVJtvr6urQ01NDbs67+npQUNDg0N1vE6nw5UrV/D999+z51paWhAVFQUOhwOFQjGoarempgZ1dXU4d+4cenp6fCZLudlsRl9fn0Pj1sbGRhw/ftzpSq2mpgZtbW1evb0x1kgkEqSkpCAsLAzR0dEQCARoaWlBRUUF2w6MRiNqampgMBjQ2NiIkydPwmAwoKWlxevrLjY2FiqVChkZGXZtwdq/zp49Oy4OHYQQn+lbo6GtrQ0VFRUoLy93mJfMlzCbzUNq+Lq6uvDDDz/YhLZQKpVISEhASEgIcnJyEBERwfaviQqPx0NiYiLCw8ORlJRkY9dlsVhQWVkJtVqNyMhIJCYmuhx+g8vlgsvlQiAQjMn2pccFpJSUFLz55ptITk4eMlHkaLjmmmvwzjvvoKGhAWvXrnUYEn48sVgs2LFjB44ePYrrr78e69evh1KpBHB1MN23bx/ee+89duA2m82DxpfR6/X47LPPsHPnTvZcTEwMVqxYgfT0dBvD0P6YzWbs2LEDf/zjH9Hd3e210cSHy3fffYfz58879ULp6+uziWw7EYmJicHatWuRkZGB8PBwAEBBQQHeeOMN1k3bYrGw7aKoqAjPP/+807boLXC5XCxduhRPPvkkpFKpjUt1//7V1dU14duBO7G2n9bW1glRr2VlZVizZo2N8HPvvffihRdeQGxsLNauXQuj0cj2r4mKWCzG448/jttvvx0ymcxmwWKdv7Zu3YqlS5fi5ZdfHjScyHjjcQFJJBIhISFhSI8ZZ/T29kKv19udZxgGIpEIAQEBEIvFSExMRGBgIBQKBUJDQ6HT6RzGqhgvent7YbFYoNFoYDabQQiBVquFwWBAfX09Ll++7NKqkxDCesIIBAJIJBKEhoZCqVQiJSXF7n6LxQKtVgu9Xo+6ujq7oGbegk6nQ3t7OwICAiCRSGwEHpFIhNDQUGi1Wmi1Whs7mK6uLnR1dXmiyD5FQEAA4uLikJyczJ7TaDQoKytDZ2en3f09PT0+pRWQy+VITU11uBptb2/HlStXvF4L5itYx+CGhgaUlpZOmP6n1+tRVVVlc86qnQ4KCoJKpRpSG2Ktu+7ubp+253MEj8eDRCKBXC5HXFycjbG/2WyGVqtFV1cXqqurceXKFajVaq+qA48LSKPFZDLhyy+/xN69e+0qlsfj4Y477sCSJUvYyTUoKAiPP/44Fi5ciD179mD79u0eiUHC4XCwaNEiLFmyBCqVCqGhodBoNPjkk09w+vRplJSUjEhoue666/DAAw9AqVQiMTHR4T3t7e346KOPcO7cOZw7d86rGqQVi8WCr7/+GtXV1cjOzsajjz7KagEYhsG8efMQFhaGkpISbNq0aUKsVikUb8RgMGDr1q04cOAAKisr3er954scPXoUK1euRFJSEp588kmnoWr6z1/V1dVe40TkLhITE/H4448jLi7OzlGorq4OH374IcrLy1FUVOShEjrH5wUks9mMoqIi/O1vf7Nz8+NyuUhOTsbixYvZcwEBAZgzZw4IIVCr1fjnP/853kUGcHWSz8rKwn333ceuMBobG3H48GHs2rVrxO9NTEzEsmXLnAYi6+npwcGDBz2+zegMQgjOnz+P8+fPo62tDffee6/NNsnkyZMxefJkfP/999i6dSsVkCgUD2E0GnHq1Cl88cUXni6KV1BWVoaysjJMmTIFd999t1MBydn85Q+Eh4fj9ttvdxgmor29Hd98841N4nhvC4PgswJS5/9v78zjm6ry/v+5WZukTdM93dKW0rKUTUAsFaGKIDqsMw4gjAuOo84MPOowPIqDQ31eooOi8wzjMqPiKIOKw0ufgj6yDoIiiEChhZZudG+TdElLlmbP+f3BL3maNg1pm73n/XrdP3Lvzb3ffHPOvd9zznfp6cGJEyfQ0tKCsrKykG5cLS0tOHHiBNra2gZM13oCl8vF7NmzHSkSBnNOrq+vx7fffhu0GY+HQ3JyMh544AGn3EVXrlzBDz/8ELTZiSm+hxCCc+fO4e9//ztkMhkKCwsRGRkZaLFCDoVCgU8//RQymQxz5sxxuWQfDmRmZmLevHkQCAQAbrSf8+fPo6SkJKTfLaGAp++v/hBCcOHCBVy4cAHl5eU+mX0LWQNJqVRi586dOHv2bMj7EVRVVeGll15CU1PTsH4Lj8fD/fffj3Xr1oHH4w0aLXH58mUUFRVBqVR6NUtwIMnKysKzzz7reIjZbDbs2rULJSUl1EAaxdhsNhw+fBjHjx/H3XffjWnTplEDaRjU19dj+/btiI2NxRtvvBG2BtKUKVOwdetWJCUlAbix9PWnP/0Jly5dCkr/zHDC0/dXf2w2Gw4dOoTXXnsNRqPRJ3ZAwAwkqVSKW265Bbm5uYiIiPD4eyqVCs3NzaitrUVXV9dN17vlcjlKSkoQExODjIwMpwSUdhm6u7vR2NgYMKPBarUOy2FcIBAgIyPDEabt6gVgTyfQ3t6O6upqqNXqgDqmDwe1Wo0rV67g+vXrkMlkTlWwWSyWY9RnJy0tDbfccgu6urrQ2NgYcr+X4h3MZjPMZjMMBkNQ+tmFAjabDXq9Hr29vWE94NBoNKitrYVOp0NGRgZEIhFkMhluueUWR9sxmUxobm526YDOZrORnp6O2NhYx75x48YNmi3aYDCgsbER3d3daG9vH5WzVEKhEBMmTEBkZOSg76++9PT0oLm52RG4ZLVa0dTUBK1W67P+HRADiWEYLFq0CFOnToVQKHRY7Z5w+vRpbN++HZ2dnWhqanJ7rj2M/cyZMygoKMAf//hHSKXSATKUlJSgqKgIjY2NI/pd/iYjIwNFRUWYOHHioLWAzGYzPvvsM+zZswc9PT0hWWW7vLwcmzZtQnJyMrZs2YLCwkK35xcWFiI3Nxfl5eUoKipCVVWVfwSlUCghiT2FhT00f9asWVi+fDlmz57tMF66urrw0ksv4cSJEwO+LxKJ8Pjjj+Pee+917BMKhYP6H8nlcrzyyiu4ePEi5HK5T35TsCOTybB161aYTCaPatmVlJTgpZdeQnd3N4Abg3+FQuHTwY9PDSSTyQStVgs+nw8+n+/kgJWQkDCk3BBGoxFmsxlyuRyXL1/2OIxUoVBAoVAgKSlpwBScXQa1Wh0S6d/5fL7TDFhcXBwmTpzostAfIQQGgwG9vb1oaGhAaWlpyI5SNBoNKioq0NXVBaVSCa1W63Yq1l5VmxCCuLg4iEQiGI3GsB4BewsOh4PIyEiHrgghMJlMoyLJoR0ulzto4jmz2Rw2y9OU/8OeGkSj0aC9vR1arRaxsbGOATVww60jKSnJ5UyHWCxGdnY2pk2b5vY+FosFBoMBKpUKlZWVIVNQ3RcIBAKMGzfupueZTCZH3rry8nK/5mDzmYFkXx9UKpWYNGkSHnzwQafpx6FgNBqxb98+fPvtt6ipqRmVSyZCoRAPPPAAbrvtNse++Pj4QS1vpVKJjz76CFVVVTh37lzIGkd9UavV+OCDD3D8+HEsXLgQy5cvd5tjJDU1FRs3bkRbW5uj/VDcM3PmTGzbts1hBNjDkI8dOxZgyfxHQUEB7r//fpeDpjNnzuCzzz4b9aHs4YpKpcLbb7+Nr776CosXL8Z9993nlCLm0Ucfxfz58wd8j8/nY/r06Te9/vnz57F3795hB+SMNiwWC/bv3+9IIeHvNAg+M5D6epgvXLgQK1asGLaBZDabcerUKbz33nteljJ04PP5uOuuu7BmzRqPzu/p6cGBAwdw+vRpH0vmP/R6PY4cOQKGYRAdHY0lS5a4NZDi4uLw05/+FFqtFleuXKEGkgfk5OQ4OeIajUbU1NTg3//+d1gY2Z4wYcIEPPLIIy5nCiIiIvA///M/1EAKU7RaLQ4ePAg2m42UlBSnJTOBQICFCxeO6Po1NTX46KOPXCZipQzEarXizJkzeO+99wLy/AnqKDaVSoXTp09DLpejurra69evqKhASUkJqqqqAuqbk5KSgp/+9KduS32IRCJkZWXd9Fp1dXU4d+4cGhoagr4cxHCx50j6+OOPkZycjIKCgqBJTR9KdHd346uvvsLly5cd+7KysjBr1izH8iWbzcatt96KBx980O0DSi6X48yZM9DpdD6Xe6QwDIO8vDysXbt2wLIrwzDIz88fdPk2KysLq1evRltbG86cOYOOjg5/iBxwTCYTvvvuO1itVuTm5mLGjBmOepdcLhf5+fno7e11PH9CPbKYEIJLly7hn//8J9LS0pCfnz/sKEiLxYLz58+jpqYGp06dCnndDAWlUoni4mJkZGRg1qxZHr3DgBurBadPn0ZbWxuuXr0auMGZu0q28FL1XnfVtN1RWlpK5syZQ6KjowmPx/OqDDabjezcuZMkJiaSyMhIwmKx/FoNuW+1erPZTK5fv066u7sH3Xp6eojJZLqpzvbu3UsyMzNJVFSUT6qS+0s/N9v4fD6RSCRk0aJFpKamxq1ONBoNeeKJJ/yii2DRj6dtMCoqikgkEsf25JNPkp6eHif99fb2um2b3d3d5PPPPydpaWlBqR9Xzx+DwTDob+nt7R20LRmNRtLT00POnj1LZs2aNWraD8MwRCgUEolEQp555pkB1djtbeSdd94h0dHRfm3HvtJPREQEkUgkZMWKFaS5udntM8YdGo2GPP3000QikRChUEgYhgkL/XiysdlsIhaLSWZmJtm3b5/HOquqqiL33HMPkUgkhM/nB0Q/hJDgnkHicrlISEhwFHHti8lkglwuh8FgQEJCwk2X72QymZODM3Aj1LKnpyfgFj2HwxnWDIjRaERbW5uT02hDQwO6urrCLmV9f4xGI4xGY1jWL/IXVqt1QDtpa2tDVVUV4uPjkZycDIFA4Nhc0dnZic7OTqhUqqDNF6PT6VBbWwuj0Yjk5GRERkY6AkeGij04QCwWO2ZQRgOEEPT29qK3t9elD6i9jYhEoqDLhjxcDAYDDAaDI0XKcOsQ6nQ6tLW1jcplNavV6lidqa+vR2VlpUffq62thVKpDLjOgrqH28PYXa33NzQ0YNu2baiqqsLq1auxevVqtx0zOjraqVRFONDY2Iht27Y5LT92dnZS/wjKsDl9+jR++9vfYuzYsXj++eddRkjasdlsKC4uxocffoiuri63S8SB5MqVK440Ec8++yzuuOOOQItECSHs7cfTBIb9sdlso94hW6fT4d1338UXX3zh0fl6vR7Xrl3zsVQ3xy8Gks1mc4TqcTgcp6rs7hAKhZgyZYrLYzExMYiLi4NAIEB2djby8/M9HrmYzWZYrdaAh31bLBYYjUa3jsYsFgscDgcMw8BisTiN0ru7u3Hp0qVRGSrKZrPB4XAGDcf2Nvb/Abjxv4XrrJV9Rkij0UClUrkNaTebzairq8MPP/wQtLNHwI00Efaafp6kB/Hk2WA0GsO2DVCcuX79etAWUw0VrFaro0ZdKOEXA6mmpgbbt29Hamoq7r//fkyaNGnE10xISMCvf/1rKJVKzJkzx+OXpEajwb59+3DlyhWcO3cuYEaSzWbD0aNHodFo3MpudwqNjIzE559/7tRR29vbR22SsTvuuAP33XcfZDIZ4uPjfX6/nJwcrFy5Emw2G/v27UN5ebnP7xlIOjo68Le//Q379+8f9BxCCH744YegNxTs/11qaiomTJhw0/PPnj2LAwcOuF16V6lUIZdYlkKhDBF3DkrwoiMUwzAkOTmZ7N+/f9jObq6w2WxDOl8ul5Nly5YNyVHOl/phGMbtNmfOHFJVVUVUKhVZu3at0zFv/j8j2Xypn8F09vvf/54YDAaP/nNvOGkvXLiQNDQ0kLa2NrJ06dKg1o+3de1uC4X2Y//vPH1WvPPOOyQyMtLnvztY9DPU7cknnxzgpG1n9+7dRCKRjGr9BNtG9TM8/RB/Omn3+2O8hqczR+3t7SgpKUFbWxtaW1uDJqfLzeTo7OzEkSNHEB0djcbGxqCR258kJiZi+vTpTo7CkydPdrs06Q0YhsG4ceMwfvx4pKam4rvvvoPJZEJbW5tP7+uKlJQUTJ061VG3kBCCyspKVFVV+bRNhEt7YxjG5bOiq6sLJSUlDgdcQgguXrwIs9kcNr/dn8hkMixZssRRA1OlUgVaJApl2AS1k7Y3qaiowObNm9HS0hJS9ciuXbuGoqIiMAwT9pFpgzFx4kS8/PLLSE1NdewTCoU+jyJisVhYunQpnn76aZw9exZFRUVobW0NSPuZOnUqXn31VSQmJgK4sab/+uuvo6amJqj9f4Kda9eu4Y9//CPq6uoc+/R6PS0nMkxuu+025OXlobKyEs888ww1kCghjV8MpIiICMTHx0MqlQ5a3djXsNlsCIVCREZGIjIyEoQQqNVqR+G7YMVsNgdtdJCvkUgkiI6ORnp6OpKSkhzGgaeYTCZ0dnaOqEivSCRCYmIihEIhuru70dnZOazrDJeYmBiIxWLIZDJIpVKHv5XVaoVIJPKrLKGGvf0kJSUNMKbtfb+pqQkKhSJsk6r6Ao1Gg8bGRkgkEsTFxTlFd0VERDie98ON+qJQggW/GEh5eXnYuHEj0tLSPHKS9JUM27dvd4wMCSHYt28fPvjgg4BHs1EGwmKxsGzZMqxZswYJCQnDStHQ1NSE1157DdXV1aipqfGBlL6Fy+Vi1apV+NnPfoakpCSaLXwIuGs/hBAcOnQI//jHP9DR0UGNoyFy4sQJtLa2Yty4cdi0aROys7MDLRKF4hP8YiDFxcXh9ttvh0wm88n1bTYbCCFgGGbQFAKxsbGYM2eO47Pd18DTlAMU/8IwDMaMGYP58+cP2deIEAKbzYaenh6cPXsWpaWlQ76/Paw/kO2DxWIhNzcX8+fPD5vke/5isPZjs9lgtVpRX1+P48ePBzxJbCjS2tqK1tZW9PT0oKenx+USbzinwqCMHkLeB8lqteLYsWM4c+YMxo8fj8WLFw+7Zg4lPLhw4QIOHTqE5uZmKBSKIX9fIpFg2bJlGDt2LObOnUuN6DBBq9Xiyy+/xNWrV3HmzBnquzVCFAoF3n33XUil0gHHOjo60NLSEgCpKBTvEfIGksViwdGjR/GXv/wFy5cvx5133kkNpFHOhQsX8Oqrr0Kn0w1rFBsdHY21a9firrvuAovForM3YYJOp8O+fftw4MABx6wzZfgoFAq8//77LvuHfRaXQgllQtZA0mq1qKioQHt7O+rq6mCxWCCXy3HixAmkpKQgLy/PZX02k8mEiooKKJVKVFdX004cJlgsFly9ehVyuRzl5eUwmUwe/bcsFgs5OTnIyMhw7EtKSkJ8fLzP0wjcDJvNhqqqKhw5cgRSqRQTJkxwOL4yDIOxY8diwYIF6OjoQHl5OQwGQ0DlDXbsL206c+Q96POTEs6ErIGkUCjwpz/9CefPn3cUtCspKcHGjRuRkZGB7du3O/kc2bl+/TrefPNNHDlyBBqNhjpohwk6nQ7vv/8+iouLodFoPPYt4XA4WLlyJX75y186RsIcDico6vaZzWZ89tln+Prrr3Hvvfdi27Ztjig2ewqCwsJCnDx5Es8++2xA8jNRKBRKuBISBhIhBBqNxqkIq1wuR3NzM5qbmx379Ho9WltbwWKxIJfLXfqftLe3o6Wlxel7lOBEq9VCLpd7lO+op6cHzc3NaGpq8ujabDYb0dHRiIqKQmpqKmQyWVAupdkdYTs6OgbMfERHRzvSICQnJ3s8mu/t7YVGoxl1S0xsNhsSiQRSqRQ6nW7U5hWjUCieERIGktFoxMcff4zDhw879qnV6kGr/XZ2duK///u/8fHHHw84ZjAYhhXVRPEvNpsNBw4cQGVlpUdO0iaTCZcvX/b4+snJyVi/fj3y8vIwYcKEoDSOPGXixIl45ZVXnAYQ7rCHuI+2ZIhisRi/+c1vsGLFChQXF2PPnj10BplCoQyK30uNDOdFZLFYUFpa6rZwZl/0ej1Onz495PtQggdCCKqqqlBVVeX1azMMA7FYjLlz52L27NkD7ms/J9gYrA8lJCRgwYIFHl9HoVCAy+XCZDKF9SyS3efI7mjP5/Mxa9YsEEJQU1MDDofjNCsXzrqgUChDxy8GUkNDA/7+978jPT0dCxYswNixY/1xWwplABMnTsT8+fMhk8mQkpLidKympgZHjx4Fh8PBokWLfJa3azjU1NTgnXfeQWpq6ohlmzp1KjZs2IDW1lYcPHgQHR0dXpQ0OCCE4PTp0/jzn/+MMWPG4J577kFUVBSAG8bvrbfeiqefftphIBkMBhw9ehSVlZWBFJtCoQQT7irZwkvVchmGIVwul8hkMnLgwAGPKmr3xRvV2Eey0WrI4aOftWvXEqVSSUwm04Dq7sXFxSQ9PZ2MHTuWHD9+3OnY4cOHiUwmC5h+WCwW4XK5LmUbKlarlRiNRnL27FkyZcqUgP8nvmo/bDab8Hg88pOf/IQ0NTW51IF9UygUZPXq1QHXhT/1Ey4b1Q/Vjy/0QwjxzwwSIQRms9nj0GsKxZuwWCxkZWVBKpVi/PjxEAqF4HK5AG60zYaGBrS2tqK8vBxarRZsNjvo2qnNZoPNZoNWq0VZWRn4fD7S09ORnp4+5GuxWCzweDxwudywToJptVphtVrR0dGBc+fOQaFQIDs7G7GxsQ4d2BEKhZgwYQJuv/12xz6dToeamhrodLpAiE+hUAJMSDhpUygjISIiAg899BBWr14NsVjsVDDZHkr/4YcfQqPRQK1WIzo6OoDSuqezsxM7duxAZGQkfvvb3+I3v/lNWBs53uDKlSvYtGkTkpOT8eKLL2L+/PkDzhGJRHj88cexevVqx76Kigps3ryZLrtRKKMUvxpINpsNGo0GKpXK7XkRERFOLzHgxggvNjYWRqMRvb291KGSclPYbDZEIhGio6ORkZGB3NxcxzH7bIxOp0NTUxOqq6vB4XAQFRUFiUTimGHS6/UwGAzQaDRBkWDQYrGgpaUFbDYbra2t6OrqciS0ZLFYEIlEDtkpN+jt7UVdXZ0jbUTf5w+Xy4VIJAKLxYJUKnUqm2E0GpGYmOiymK3VaoVOp6NRcBSv464f22w26HQ6mM3mAEg2+vCrgaRWq/Hee+/h66+/HvQcFouF+fPnY82aNeDz+QBuGEyrV6/GzJkz8d1332H37t0ehzRTRi9paWl44oknkJ2djenTpzsdUygUeO+991BZWYlLly6BEIKpU6fi0UcfRUpKCnJzc2Gz2fD1119j//79aGlpQXd3d4B+yUBsNhu++uorNDY2OvZJJBI89thjA34r5Qaunj/jx4/H448/7rKeWFpaGp577jmX/3trayveffdd1NbW+lRmyugjNjYWv/rVrzBp0qQBx3p6erBr1y6UlJQEQLJRiDsHJQTIaerJJ58kOp3OpYPp7t27iUQiCQonrkDpJ9i2YNXPtGnTyKVLl1y2o4qKClJQUOB0/ooVK4hCoXCcY7FYyB//+EfCZrNDQj9SqZQUFxe7/L2uKCkpIdOmTRvV7ef2228nVVVVHuvMztWrVwe0n3DUTyhs4aafzMxMcuzYMZftTi6Xk2XLlo1q/fir/RB/OWl7Ap/Px5w5c5CTk4O5c+d6lD2ZQnHFxIkTUVBQgKysrAElQ6qrq/H999+jsbHRZab1UKa3txeHDh2CXC4fcKxv/6L8HwqFAp9++imSkpIc+/Ly8jB79my3zyCJRIIVK1YgLy8PZ8+eRVlZmT/E9QlpaWkoLCx0KvJ96dIl/Pjjj26DFeLj41FYWOgof+MpGo0GJ06cQGtr67BlDkcyMzMxb948pKWlITU1NdDiUBBETtoCgQBr167FqlWrwOFwnCJMKJShUFBQgJdffhlRUVED2tGFCxewZcsWdHd3h10maY1Gg3/84x8ui+zGxMTgjTfeoAZSP+rr67F9+3ZH8k2GYfDYY49hxowZbg2kpKQkrF+/Hr29vSgqKsLly5dD1i9y3Lhx2LJliyMi0mazYefOnSgpKXFb0zA1NRUbN27ElClThnS/xsZGtLa2UgOpH1OmTMHWrVshlUod7iWUwBI0BpI9021/52wKZahwOByIRCJEREQMOGaxWNDb2wu9Xh8AyXwLIWRQo4/P51OHYhfYbLYBbcETw5lhGERERIBhmJCf7Waz2RAIBI5nLyHEI0d/uw6G+swWCAQujfjRDpvNhlAohEAgCLQolP8PE6qjHgqFQqFQKBRfQROoUCgUCoVCofSDGkgUCoVCoVAo/aAGEoVCoVAoFEo/qIFEoVAoFAqF0g9qIFEoFAqFQqH0w218KsMwNMQNACGEcbWf6ucGVD/uofpxD9WPe6h+3EP14x6qH/cMph+AziBRKBQKhUKhDIAaSBQKhUKhUCj9oAYShUKhUCgUSj+ogUShUCgUCoXSD2ogUSgUCoVCofQjtKssUigUCoUS5ohEIiQnJ7ssjKzX6yGXy2EymQIgWXhDDSQKhUKhUIKYvLw8bN68GVKpdMCx0tJSvPzyy2hqagqAZOENNZAoFAqFQgkiuFwuWKz/84BJSEjAjBkzkJ6ePuBci8WCiIgIf4o3aqAGEoVCoVAoQUJMTAxWrVqF3Nxcx76MjAxIJJLACTVKoQYShUKhUChBglgsxs9+9jPMnz/fsY9hBk32TPEh1ECiUCgUCmUYxMXFYfr06YiMjPTo/KamJpSVlcFsNg84lpKSgqlTp0ImkyEpKemmRlFNTQ0qKipQXl4OrVY7LPn9BZfLxdSpU52WCHt6elBSUoLr168HUDL3UAOJQqFQKJRhkJ2djRdffBHZ2dkenf+vf/0LW7ZscWkUTJ06Fa+++iqSk5MhFovdXocQgqNHj+KVV16BTqeDWq0elvz+QigU4tFHH8XPfvYzx77Lly/jqaeeogaSK8RiMWJiYtxayUajEV1dXTR8kUKhjBg2m424uDgIhUKfXF+n06Grqws2m80n1w9WGIZBbGwsoqKiBhwjhKC7u9vpBc7hcBAXFweBQODR9Q0GAzo7O2GxWLwm83Bw1X5kMhmkUikSExM9ukZKSgoyMjLQ2dmJrq4uGI1Gx7GIiAgkJCQgLi5uwPf664AQgpaWFiiVSpezUd5GKBQiLi4ObDZ7wDG9Xo/Ozk5YrVbHvv7v9+joaKSkpDjpKTY2Flwud8iyuGo/rmTwBgExkBiGwaJFi7Bu3Tq3CqqtrcVrr72Ga9eu+VE6CoUSjojFYqxfvx4FBQU+uf6pU6fwl7/8Bd3d3T65frASERGBBx98EIsXLx5wzGQyYdeuXfj8888d++Li4rBx40ZMnz7do+tfuXIFO3bsQEtLi9dkHg6u2o9EIkFSUpLH1ygoKMDOnTtRV1eHHTt2oKKiwqPvlZeX4/XXX0d7e7tjX2Njo9+MxunTp+Opp55CTEzMgGPnz5/HG2+84ZDN1fudw+Fg3LhxXpHFVfvpL4O3CNgMUkZGBu666y7weDzHPpvNBkKI43N0dLTHa7uU8INhGKdQV1/Rv92FG65Gff0JZx3Y25FQKMS0adOcnF+9iUajgUAggFqtDmt99oXNZoPH42HixIku9Wo0GnH8+HFwuVzHzFpkZCRmzJiBu+66y6N7CAQCREZGgs1m+1Wv/Z8/N2s/7mRjsVhgGAZSqRRSqRRJSUkDotIIIbBarS5nQdrb23Hq1Ck0NzcP/weNgISEBNxxxx0ujUGbzQahUOh4zrBYLGRlZQ14v/eHYRiw2Wyn5xMh5KYzsAKBANOnT3f6H6xWq09SHQSND5LBYMD//u//oqyszLFPoVBAoVAEUCpKoGAYBoWFhZg7d65PjSSLxYKjR4/ihx9+8Nk9AolMJsPy5ctdTtvbMZlMOHz4MM6fP+9HyfzHmDFjsHTpUqSlpTmFTnub8ePHY+PGjWhpacH+/ftRV1fns3sFA1KpFCtWrIBMJsOMGTNcnsPhcLBgwQKIxWKH8SCRSJCVleXxfdLT0/Hb3/4Wra2t+Oqrr3DlyhWvyO8OV8+fyMjIQduPq/eXHS6Xi0WLFmHmzJlu71lZWYkdO3a4XKq8du1a0PrqZGdn4+mnn3bIxzAMZs+efdOBmVQqxeOPP+70jr906RIOHjwIg8HgU5k9hhAy6AaA+GJjGIZs2rSJGI1GYkelUpFf/OIXhM1mOzYWi+WT+w9187d+Qm3zhX7YbDYpKioiRqORWCwWn206nY48/fTThGGYkNKPp1tBQQGpqKhwq4Oenh7yq1/9KqzaT99twYIFpL6+nlitVuJrrFYruXbtGrnzzjtDRj/D3aZNm0YuXLhArFYrsdlsg+rEZrM5tbfh/A9Wq5V0dHSQBx54wC/6cfX8cSe3q/eXfYuOjibvvvuu0/lXr14lBQUFA+7LYrFcXmMk70Jv6GfFihVEoVC4/X/6/sfu2oO773344YdEIpG4lSUzM5McO3bM6TqHDx8mMpnMq/ohhATPDBKAQacXwxGGYTBmzBiMGTNm2Dku1Go1ysvLodFovCxd4CGE4Nq1azh+/DgSEhKQl5fnkylULpeLCRMmYOHChR5N3RuNRlRUVKCjo8PrsgyXuLi4QfWTl5eHqKgot6M5Pp+PSZMmYeHChWhra0NlZWXAHWJHSt/+NXPmTIhEIrczkRaLBVevXoVcLh9wjM1mY9y4cUhLS3N7z/b2dlRUVKC5uXnU+CGxWCwnvcrlclRWVnrVcVgikWDSpEkeLRV7E1dLQP3p6enBlStXoFQq0dra6vT+ioqKQl5eHpKSkpCamurRPYPJwZ/FYiEnJwcZGRmYNm0a+Hy+23NdYddPb2+vY196ejpyc3PBZrMHfM8fLhVDIagMpNEEi8XC0qVLsWHDhmF3/PLycvz+97/32NEvlLDZbDhw4ABOnDiBefPmYfv27UhJSfH6fbhcLlauXIl7773XIwOpo6MDmzdvxtGjR70uy3DJy8sbVD98Pt/t8pr9nAcffBDLly/HF198ga1btwZ92PDN6Nu/IiMjERsb6/Z8nU6H999/H8XFxQOOCYVCvPDCC1izZo3ba5SUlOAPf/gDFAoFurq6RiJ+yHLmzBmvt58ZM2Zgx44dLh2EA01tbS1eeOEF1NbWQqVSOR2TyWTYunUrJk2aFJJZsDkcDlauXIlf/vKXEIlEN0094Aq7fvouN69duxYvvPCCx1GMgYQaSH6GzWZDLBZDKBQiLS0NMpls2AaSWq1GcnKyU8fU6/VQq9Vh4SB6/fp1XL9+Hc3NzZDL5V4ZXQgEAojFYqdZO4lE4vEDjMfj+SxMfKhERkZCJBIhJSUFaWlpN53hGAyGYRATE4OYmBikpaUhOTkZPB4P169f90sIsTfxtH9ptVrodDpHP+np6UFzc7PLgp8ikQitra1OvhIcDgfR0dFOUbh6vR6tra1QKpU++GXBg70PxcfHg8PhgBACjUaD3t5etLa2orm5edj+MiwWC2KxGBEREdDpdNBoNEhMTERbWxtMJlPQ+KbY209bWxuamppcRthxuVxIpdJh98tgwO6Eb7FY0N7eDjabjejoaCfna5PJhOvXr7tc/bHrp2+/6u7uDpn3EzWQ/ExcXBzWr1+PadOmITc3d0Qv/fT09AFJx06cOIH33nsPOp3OG+IGBRUVFdi8ebNXDJPCwkI89thjIR8dyWKxsGTJEvz85z+HVCq96SyRp+Tn5+PPf/4zrl27hrfeeguVlZVeua6/8LR/HTt2DHv27HEsJZpMJly+fNnluQaDAR9//DFOnTrlMKzT0tKwfv16jB8/3jc/JIiZN28e1q1bh6SkJMhkMhiNRnz88cc4fPgwmpubnZZThopYLMavf/1r3HbbbSguLsaePXtQV1eHoqIi8Hi8Qf8jf2NvP0qlMmwNYovFgn379qGkpMTR7hMTE7F+/XpMmTLFcV5FRQXefPNNdHZ2DrhGV1dXSOuHGkh+hGEYCIVCFBQUeCXUODo6GoWFhU771Gr1sJJvBTMdHR1eW9ISi8UwmUwej2CCtQYSwzAYP348li5d6lXfDPtMVGVlJfbu3QuGYUJmtAfgpv3L7nxZU1ODL7/80qMktFarFaWlpSgtLQVwQ/cTJkzA2rVrvSp7sGPvC1lZWVi8eLFjwKLValFaWor9+/eP6LrAjdmp/Px8LFmyBNXV1WCxWFCpVPj3v/898h/gJYbafkIVm82G8vJylJeXO/ZlZGRg5cqVTs8EpVKJI0eOBCwFgS+hBpKfGDt2LBYsWID09HRkZGQEWpxRS3l5Od58802P1r9ZLBYKCgqQn58ftIaSr4iNjcXq1asxa9YsfPfdd2GRAkCr1eLYsWOoqanByZMnhxUQYs/3I5PJfOITF6zw+XzcfffdmDhxIvLz870yCGMYxtG/7DN9UVFRGDt27IivTfENarUa+/btw8WLFx37qqurwzJQCKAGkt/Iy8vD5s2bIZVKweFQtQeKS5cueZxHhcPhYPPmzZg1a5bfI2gCTUJCAp588kno9Xq88MILuHDhQkjNJLlCp9Nhz549+PLLL4cdMXvLLbdgy5YtiImJGVX9WCAQYOXKlVi9evVNI7s8hcViYf78+Xjuuecc12MYZlTpNdTo7u7GBx984LR0bbPZQj7qdTBoS/QTLBYLPB4v7Ja/Qg2bzXbTaXE+n4+xY8ciLi4OqampIT17pNFoUFNT4+QXkpycjKysLLf+b/YXFY/HC7rQW1fExcUhOzsbMplsgMN9T08Pamtr0dbWBqVSOaJlkdHaj/u2h5HSt39lZmaCz+dDrVajtrbWyQm7vr7e72HvfWVLT08fVt+XSqXIzMzEhAkTBvV1FAgEmDJlisfX7+npQU1NTVAs6YWrMeQKaiBRKP2Ij4/H73//exQUFCAuLi4kDITBqKmpwfPPP4/6+nrHvgceeADPPvtsSITZesr06dNRVFSE5OTkAeUQrly5ghdeeAFNTU0h7TAaLrjqX/b/qK2tzXGeSqXy+8vYG32/sLAQzz33HCQSCaRSqctzkpOT8fzzz0Ov13t0zVOnTmHLli0u83RRfAc1kMKE3t5eGAwGaLXakF8KCTRcLteRzMyO2WyGTqdDT09PQEdxLBYLIpEIERERAwwcQgh0Op2TfEqlErW1tU4Fn5VKZdi1EXu+p9jY2AGzHL29vairq3MZwu8pQqEQERERiIyMHDDqt/c9jUYTVIn+ggW77uwkJCRgzJgxyM3NhV6vR3d3N5RKJWpqatDa2hpASV33fTv2/mU0GqHX6536kD0Ah8/nIzU1FTk5OW6jbnk8HtLT0z2Wq6GhYdTNWg4FLpcLiUQCtVoNnU7ntfQk1EAKA4xGI/bu3Ytjx46hvr5+RGG2FNdUVlbi/fffR3Nzc0BDjWNjY/GrX/0KkydPxuTJk51GuGq1Grt27XJyqO7s7PR6hetgpLS0FJs3b4ZMJsNjjz2GSZMmee3afD4fq1atwt13342srCynF5/FYsEXX3yBQ4cOobGxMWydVYcLh8PBihUrsGjRIkdbFYlEyM3Nhc1mw9dff439+/ejpaUl6LOP9+1fV69edfJhEwqFeOihhzBnzhzk5OS4zTpN8T4TJkzAiy++iLa2NuzatQslJSVeuS41kMIAs9mMs2fP4tNPPw20KGGLXC5HcXHxiGYhvEFkZCTmz5/vMoxdr9fj22+/HXa4dSjT3NyM5uZmZGZmYsmSJV41kLhcLvLz811m0rZarSgpKcEnn3wSdrNy3oDNZmPGjBlYu3btgJk3i8WCsrIyfPLJJyFRYspd/+LxeJgzZ85Ns61TfINUKsXy5cuhUChw5MiR0DSQBAIB5s2bh6ysLOTn53s9MigzMxPz5s1z6VvR3NyMkydPQqvVevWegaSzsxMnTpxAa2trWJYb8Tf29pOWluZx7aRwpaenBydOnEBLSwvKyspC+uWfnp6OtWvXorW1Fd9++y0aGho8+l58fDwKCwuRmpqKiRMn+lbIIMdoNOLYsWNQq9XIy8vD7NmzHdFmXC4Xc+fOBcMwqKqqwvfffw+hUIjCwkKkp6cPcEbWaDQ4efIkmpqacO7cuZBuW1KpFIWFhUhOTkZOTo5Xr00IwYULF3DhwoWQr7nJ5/Mds2tz584NmUhFv0opFouxbt06LF68GFwu1+sG0pQpU7B169YBTprAjcynpaWlYWUgtba24vXXX0dZWVlQRDeEOvb2I5VKR/0UuVKpxM6dO3H27NmQb1u5ubl44YUXIJfL8R//8R8eG0ipqanYuHEjpkyZ4pXorVCmt7cXn3zyCf71r3/hsccew4wZMxwvOT6fj/vvvx/Lly/HJ598gpKSEiQlJWHDhg3Iz88foLvu7m787W9/wzfffAOz2RzSfluZmZl47rnnfLKsZrPZcOjQIbz22mswGo0h3Q8FAgHWrl2LVatWeS0a0h/41UBiGAZ8Pt8rJSMiIiKQkZEBkUjk2Jebm4vo6GiX109MTMTkyZMhkUhGVCsomLDZbDAYDNTnaAQwDIPU1FQkJiY62k8wR3cZjUbU1NQgNjYWycnJg0bJ9IfNZiM9PR2xsbGQyWQ3jc4hhMBoNIZF22Kz2RAIBIiOjkZubq7HkUDjxo1DbGysy+eJwWBAY2Mjuru70d7eHtKzIJ5iNBodW394PB54PB6kUimmTZuG+Ph4xMfHO+nOXu+usbERHR0dQdm23PUvLpeLMWPGYPr06Y59EyZMQExMjMfvNJVKhebmZo+WFK1WK5qamqDVakPaiATcv/ttNhtaWlrQ2dmJhoaGES239v2PlEol2traRtQ3Q2OeywVSqRSbN2/G1KlTHfskEsmgFYfz8vKwY8cOyOVyvPTSSzhx4oSfJKUEM1wuF6tWrcIvfvELt+0nWOjq6sKOHTsgkUjw+OOP47HHHvMoFFkkEuHxxx/Hfffdh7i4uJAZwXmT6OhobNiwAQ899JBH5wuFQshkMpfH5HI5XnnlFVy8eJGGXvfh9ttvR2ZmJjgczgDdlZSU4KWXXoJSqQy4L99guOtfrtpPZGSkyxWLwTh9+jS2b9/u0UoGIQQKhSLkjaObYTQasXv3bnz++edQqVQjMpzt/9GDDz6I3bt346233hpRRFvIGUgcDscR0jt+/HhMmzbNccxqtUKv14PFYoHP5zst4QmFQkdSsr6zTv7CarVCp9N5dYlPp9ONqqRd3oRhGERERDjaxdSpU10mbTOZTDCZTNDr9UHxoDKZTLh27RrYbDYUCsWAUGN7KHp/xGIxsrOznQYU4YjNZoNer4dWq3XMatjhcDjIysryyn3sMw1lZWVeuV4oYTabodVqHbMCfQ30mJgYxMTEgBDiSDtiR6FQoLy8PKijKt31r6G2H5vNBqPR6DQjIpfLcfny5bBYwfAEFouFiIgIiESiQf2OrFYrWlpacOnSpWHfx2q1wmg0wmazISEhAfHx8YiLixtxkt+QM5BmzpyJ1atXIyUlBZmZmU7HysrK8PHHH4PL5eLhhx92qrRdVVWF3bt3o6WlxeNSE96krKwMW7du9eryTWdnJx29DpOkpCQ8/PDDyM3Nxa233uqyI9l9AL7++ms0NTUFfRiyWCzGo48+6jLCjc/nOy0NhCsqlQpvv/02vvrqKyxevBj33XdfSCf6DEa+//57PPfccxgzZgweeeQRpKWlDThHqVTio48+csq/1dDQENKOxkOltbUVH374oVMR15qaGo+TQ4YDWVlZeOSRRyCTyXz6/LG/31UqFYAbs2+XL18e8QRCyBlIOTk5ePjhhweUEwBudMA9e/ZAIBDg7rvvdjKQWlpa8OmnnwZsarehocFj51CK75FIJFi6dCkKCgoGPcceRfL++++HRBiyQCDAwoULAy1GQNFqtTh48CDYbDZSUlJw7733BlqksKOyshKVlZWYNm0alixZ4tJA6unpwYEDB3D69OkASBgcdHZ24osvvhjRzEioI5VKsXLlSpeJN72Jr97vATeQOjo6cPr0acjlcqdyCH3hcDiYOXMmcnJyMGfOHKdpc5vNhpKSEly9ehXnz5+HXq+H1WrF0aNHndLWX758GTqdzue/hxI8xMbGoqCgADExMQOOpaSkIDEx0eX3tFotzpw5g9bW1qANcSeE4NKlS/jnP/+JtLQ05OfnD1r3aTRC9eMfBusb0dHRWLRoEbKysnDhwgVUVlb6WbLAUV1djQsXLqC2ttYxo0EZiEqlcrz7q6urAy2OSwJuINXV1eGll14aUFCzL/ZMto888gh4PJ7TMpXFYkFxcTHefvttmEwm9Pb2gmEYvPXWW05rnhaLhRpIo4y0tDT853/+JyZPnjzgGJvNHjTyRKVS4a233sLJkydhMBiCwveoPzabDQcPHsQ333yDO++8E+PHj6cGQB+ofgJLUlISnnnmGeh0Orz44ouoqqoKyoGGL/j++++xZcsWaDSaoIzUCxZaWlqwfft2XL58OWiXHQNuIJnNZqjVardOawzDQCAQuFxWI4Sgt7cXPT09jg5ICAmrfEeUoSGRSJCUlITs7GwkJCS4bDd2CCHo6OhwGum1trZCqVSip6fH98KOAIPBAIPBgPb2dlRXV3vc5u36GakDY7AzXP3Y4fP5SElJccpvo1aroVAoUFtbS19+bmCxWIiMjASPx0NGRgbGjRsHjUYDhUIREsvVwI3VjatXryI6OhrJyckeJzc0Go3o6ekZ1e0jNjYWCQkJkMlkg0bMWq1WaLVap3e/WCyGVCp16zeYmprqt0CrgBtIFIq3ufPOO7FhwwbEx8cPGqZtx2KxYO/evU5lWgwGw6DLvcHIlStXsGnTJo9D93/6059iw4YNTgVEw5mh6sfOuHHj8Ic//MEpQ/Lp06fxxhtvoL293ckBmeIaLpeL1atXY968eThx4gRee+21kFh2slqtKC4uxvnz53Hbbbdh8+bNQwrnH80wDIN7770XTzzxBGJiYjzO1QbcSBPxzDPPICoqatBz+Hw+xowZ4w1Rb0rIGkiEEFgslgFhlBRKfHw8pk6d6hhluEpsZ8doNOLatWs4e/ZsyC4BXL9+fUi1hyZPnoze3l6wWCxwudywn0kaqn7sGAyGAcvyHR0dOHfuXNDPLgYKm80Gi8Xi1JfS0tKQkZEBpVKJyMhIaLVamM3moO9vLS0taGlpQWRkJDQajdNMNJvNHjCjZLFYYLVaR3XqFXuFDJlMhttuu83toITFYoHH4znN0EqlUtx6661uZ/39ScgaSB0dHfjss89QW1uLU6dOBX1no/iPs2fPoqioyKMpcavVOuraj10/mZmZjpQZFIo3KCkpQXFxsWN5icvl4r777sO8efOQl5eH559/Hi0tLdi7dy9qa2sDLK1n1NTUYPv27U6zGnfeeSfuu+8+R649vV6Pzz//HCUlJaO29FNUVBTuv/9+TJ48GTNnzrzp8zclJQUbNmxAR0eHY9+ECROCqpJByBpIKpUKn332GU6fPj2qXm6Um1NWVobLly97fP5oaz92/dxyyy2YP38+NZAoXuPq1at4++23HTNsfD4fUqkUc+fOxdixY5GdnY2GhgacOXMmZAykxsZG7Nq1y/GZYRhwOBzcc889DgPJYDDg0KFD+OSTT0bd88SOSCTCsmXLsHTpUo9mpRMSEvCLX/xiQLLbYCJkDaTIyEjMnTsX8fHxKC8vD5nORvEPo/Uh5SmEENhsNqonitchhDgFzPSFYRhERkZizpw5iIqKQmVlZUhEuN1MPi6Xi5kzZ6K3txf19fW4fPnyqHH9SExMxPTp05GcnIyUlJQhGzneMIqSkpKwaNEiNDc349KlS15LoByyBlJycjI2bdoErVaLbdu24dq1a0HfySgUCmW0ExcXh6effhp6vR6vv/46ampqQt6YEIlEePTRR7FmzRp89NFHKCoqGjVRbBMnTsTLL7+M9PT0gNWynDBhAl5++WXI5XJs2rQpfAykiIgIpKamQq/XQ6VSeZyriM1mO6ooe1pJmUKhUCj+gxAClUqFhoYGx0wBh8NBfHw8oqKiAlIX0xtcv34djY2N4HK5A44ZjcZRMViXSCSIjo5Geno6kpKSEB8f79H3NBoNuru7h5xfTiQSIS4uzmUKAB6Ph7i4OJjNZien75EScAMpJycH27Ztg1KpxM6dO/HNN98EWiQKhUKheAGz2Yy9e/fizJkzjn1SqRQbN27ElClTAijZ8CGE4NChQ6ipqXH5sm5tbXUbORsOsFgsLFu2DGvWrEFCQgLi4uI8/u7Jkyfxt7/9DQaDYUj3nDNnDp566imXlRF8hV8NJLvfg9VqBcMwYLFYiI6OxuzZs9Hd3Y0vvvjC4fTWFzab7TZxlD3ksq9FSv0rQhcWi+W0Lm1vNxT/wmKxwGazaV+iDIrVanVaHuu/VEYIQW1trZOPaGZmJtatW+c3GX1BU1NTwOp6BgMMw2DMmDGYP3++y3d2X/q2CUIImpqa8M033wx5CVIsFvs9OtCvBpJWq8Vnn32G0tJSFBQUOClXIBBgxYoVGDt27IDv8Xg8zJgxw+U1ORwOFixYALFY7HiI9/T0YP/+/airq/Pdj6H4BIlEgmXLljklAqutrcWBAwfcZluneJf4+Hg8+uijuPPOO3H06FH88MMPgRaJEmQoFAq8++67TokAy8rKhjwzQAlfmpqaUFxcjK6uLse+c+fOwWw2B1Aqz/GrgaTT6bBv3z6wWCz87ne/Q2FhocNAioiIwPLly7F06dIB32MYZlBPdzabjQULFmD+/PmOfY2Njbh06RI1kEKQ6OhorF27FnfddZdj39GjR3HixAlqIPmR+Ph4PPzwwzAYDNBoNCGdSJPiGxQKBd5//32nZzOdbaT0paWlBe+88w5qamoc+0KpjfjdB8munLq6Ohw5cgSJiYmYNGkShEIhGIa56XSdK/p/TyQSYdasWYM60FVUVDglp6IEnsTEREycOBHp6emIj48Hi8VCXV0d6urqcPHiRToq9QEajQZnzpxBV1cXxo0bh7S0NKfj9iW2YMtNQgkehrL03bePx8bG+lAqSrBgs9kcbjWhSECctAkhOHz4MM6ePYuZM2dix44dXq2tEh8fj9/97ncuX6odHR3YvHkzjh496rX7UUbO9OnTsW3bNkilUsTFxcFms2H//v3461//Cp1OFxL1m0KNpqYm/Nd//RckEgleeOEFrFmzJtAiUcKY/n2cQgl2AhbFptFooNFokJSUhLa2NqdQfaFQiKioKJcjV4PBALVa7XbkwmazIZFIXNaBEQgESElJgVQqhU6ng0aj8c4PoowIgUCA1NRUR0FIq9WK69evo7m5ecijD4FAALFY7Hbmw2Kx4Pr16yGzFu4LzGYzFAoFtFrtgH5g179Goxk1+VwovsVVH6dQgpmAh/nX1dWhqKgIkZGRjn2LFi3CunXrXOYz+OGHH/Duu++6fWgnJiZi/fr1LsNIxWIxfvOb32DFihUoLi7Gnj17RnVxwXBk3rx5g7YfO83NzXjrrbdQWVnpR8lCB7lcjjfffBPl5eW4evVqyPgMUCgUircIuIGkUqnw73//22mfVCqF2Wx2OQPU1NSEgwcPuq2mnZGRgZUrV7pMc8/n8zFr1iwQQlBdXe02fQAlsNhTQQw1xD8rKwuLFy92m0C0srISe/fuBcMw9OX//+mrB7VajZMnT9LoNYpXsPdlO7TUTehi/y9v5pvoi//WXsbGX36RATeQXFFaWoqdO3e6dLIuKSm5aRIutVqNffv24eLFi459WVlZuOeee5wqMlOCF4ZhUFBQgGeeeWbIHS0/P99l2+lLbGwsVq9ejVmzZuG7777D+fPnRyJuSGMymXD06FGo1WrHvvb2dq+l66eMXthsNu644w7MnDkTkydPhkgkglqtxqFDh1BfX48zZ87QHGchhFgsxqJFi5CVlYXZs2cPOsFw7tw5fPfdd6itrfWa/2hNTQ3eeecdpKamYtGiRZDJZF65rlvsFpmrDQAJxMZisQiPx3O5cTgcj67B4XCcvveTn/yENDU1ETs2m428+uqrhMfj3fRawaafYNu8oZ8VK1YQhUJB+mKxWIjRaBzyZrFYyM2w2WzEbDYTtVpNnnrqKcIwTFDrx9cbm8126i9cLtenOgkF/UybNo1cvHjRqd3s3r2bSCQSv8oRrPrxZOPz+eTVV18lBoOBmM1mQgghDQ0NZNGiRYTH4xE2mz2q9RNq7ScjI4McPHjQ7XPWZrOR119/nURGRnr8vvZkY7FYhMvlkrFjx5Ljx4+7vLdcLifLli3zin4IIcE5g2Sz2UacMbO/X1FHRwfOnTuHxsZGx776+no6egkSurq68OOPPyI5ORk5OTmIjo4Gm80eVtoHT2AYBhwOBxEREcjOzsbtt9+Ozs5OXLt2bVQ6bvfPiEy5kdj20qVLMBgMyMrKcjgXU26OQCBAdnY2EhISIJPJnPwBCSEwm81+z4pMGTkMw4DL5bp0f+mL1WqFyWTyqn+vPWWAxWLx23s7KA0kX3DlyhVs2rQJHM7//WSVSkUdtIOEkpISPPPMM5DJZNi2bRtmz57tl/tyOBysXr0a99xzD44cOYKioiKnrK+U0UtzczOKiooQExOD5557DqtWrQq0SCFDcnIy/vCHP2DGjBkeFzGlUIKNUWMg9fb20szaQYxWq4VWq4XJZIJCoRjRunVERIRbB+2+MAyDhIQEJCQkoLq6GvHx8bBYLNDpdNR4HuUYjUY0Njais7MTbW1tUKlU0Gq11LHYDTweDyKRCImJiRgzZgxycnIcx8xmM3Q6HXp6ekblLG24QwiBTqeD0WiEXq8Pi34yagwkSmigUqnw17/+Ffv27RvW91ksFubPn481a9a4DfN3xdSpU/Hyyy+jubkZ7733HsrLy4clAyW8MBqN2Lt3L86dO4f6+nqaF8oNs2bNwkMPPYTk5OQByX9LS0vxwQcfoK2tDdXV1QGSkOIr1Go1du3ahfPnz+Pq1athsWRPDSRKUKHT6fDNN9+M6BpRUVHDWg5JT09Heno6Ghoa8OWXX1IDiQLghj/jjz/+iB9//DHQogQ9Y8aMwc9//nNIJJIBx5qamvDFF19AqVT6XzCKz9Hr9fj222+xf//+QIviNaiBFKZIJBIUFhY6Vdqur6/Ht99+C71eH0DJfAOfz8ecOXOQk5ODuXPnOvmaDRWxWIwlS5Y4LQ80Nzfj5MmT0Gq13hCXQgkbuFwuZs+ejYkTJyI/P99p5tZqteLs2bMoKyvDxYsX6ewbJaSgBlKYkpSUhA0bNiA/P9+xb//+/bh48WJYGkgCgQBr167FqlWrwOFwbhpl4Y6YmBg88cQTTpESx44dQ2lpKTWQKJR+8Hg83H///Vi3bh14PJ5T37NYLDhw4ADeeustmM3mm+awo1CCCWoghSkMwwxwVo6IiAjbyuz2LOmeOmff7FoRERFO+yIiImjWdQrFBQzDgMfjOZWL6ovJZIJOpwsLp13K6IKhjZZCoVAoFArFGTokplAoFAqFQukHNZAoFAqFQqFQ+kENJAqFQqFQKJR+UAOJQqFQKBQKpR/UQKJQKBQKhULpBzWQKBQKhUKhUPrx/wBuSn9mzGVW2wAAAABJRU5ErkJggg==\n", 310 | "text/plain": [ 311 | "
" 312 | ] 313 | }, 314 | "metadata": { 315 | "needs_background": "light" 316 | }, 317 | "output_type": "display_data" 318 | } 319 | ], 320 | "source": [ 321 | "## generate new images by PixelCNN\n", 322 | "n_cols, n_rows = 8, 8\n", 323 | "C = 1\n", 324 | "H = 28\n", 325 | "W = 28\n", 326 | "\n", 327 | "# Create an empty array of pixels.\n", 328 | "pixels = torch.zeros(n_cols * n_rows, C, H, W).cuda()\n", 329 | "\n", 330 | "model.eval()\n", 331 | "with torch.no_grad():\n", 332 | " # Iterate over the pixels because generation has to be done sequentially pixel by pixel.\n", 333 | " for h in range(H):\n", 334 | " for w in range(W):\n", 335 | " for c in range(C):\n", 336 | " # Feed the whole array and retrieving the pixel value probabilities for the next pixel.\n", 337 | " logits = model(pixels)[:, c, h, w]\n", 338 | " probs = logits.sigmoid()\n", 339 | " # Use the probabilities to pick pixel values and append the values to the image frame.\n", 340 | " pixels[:, c, h, w] = torch.bernoulli(probs)\n", 341 | " \n", 342 | "generated_imgs = pixels.cpu().numpy()\n", 343 | "generated_imgs = np.array(generated_imgs * 255, dtype=np.uint8).reshape(n_rows, n_cols, H, W)\n", 344 | " \n", 345 | "fig = plt.figure(figsize=(8, 8), constrained_layout=True)\n", 346 | "gs = fig.add_gridspec(n_rows, n_cols)\n", 347 | "for n_col in range(n_cols):\n", 348 | " for n_row in range(n_rows):\n", 349 | " f_ax = fig.add_subplot(gs[n_row, n_col])\n", 350 | " f_ax.imshow(generated_imgs[n_row, n_col], cmap=\"gray\")\n", 351 | " f_ax.axis(\"off\")" 352 | ] 353 | } 354 | ], 355 | "metadata": { 356 | "kernelspec": { 357 | "display_name": "Python 3 (ipykernel)", 358 | "language": "python", 359 | "name": "python3" 360 | }, 361 | "language_info": { 362 | "codemirror_mode": { 363 | "name": "ipython", 364 | "version": 3 365 | }, 366 | "file_extension": ".py", 367 | "mimetype": "text/x-python", 368 | "name": "python", 369 | "nbconvert_exporter": "python", 370 | "pygments_lexer": "ipython3", 371 | "version": "3.9.7" 372 | } 373 | }, 374 | "nbformat": 4, 375 | "nbformat_minor": 5 376 | } 377 | -------------------------------------------------------------------------------- /models/stable_diffusion/README.md: -------------------------------------------------------------------------------- 1 | 2 | -------------------------------------------------------------------------------- /models/stable_diffusion/assets/blend_sd.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/xiaohu2015/nngen/d6580dd07fdc6cc9e545dc6630ca7afbb52bafc8/models/stable_diffusion/assets/blend_sd.png -------------------------------------------------------------------------------- /models/stable_diffusion/assets/img2img/sketch-mountains-input.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/xiaohu2015/nngen/d6580dd07fdc6cc9e545dc6630ca7afbb52bafc8/models/stable_diffusion/assets/img2img/sketch-mountains-input.jpg -------------------------------------------------------------------------------- /models/stable_diffusion/assets/inpainting_examples/overture-creations-5sI6fQgYIuo.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/xiaohu2015/nngen/d6580dd07fdc6cc9e545dc6630ca7afbb52bafc8/models/stable_diffusion/assets/inpainting_examples/overture-creations-5sI6fQgYIuo.png -------------------------------------------------------------------------------- /models/stable_diffusion/assets/inpainting_examples/overture-creations-5sI6fQgYIuo_mask.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/xiaohu2015/nngen/d6580dd07fdc6cc9e545dc6630ca7afbb52bafc8/models/stable_diffusion/assets/inpainting_examples/overture-creations-5sI6fQgYIuo_mask.png -------------------------------------------------------------------------------- /models/stable_diffusion/assets/sd.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/xiaohu2015/nngen/d6580dd07fdc6cc9e545dc6630ca7afbb52bafc8/models/stable_diffusion/assets/sd.png -------------------------------------------------------------------------------- /models/stable_diffusion/random_walks_with_stable_diffusion-no-images.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "markdown", 5 | "id": "a0e1dd3d", 6 | "metadata": {}, 7 | "source": [ 8 | "# A walk through latent space with Stable Diffusion\n", 9 | "Latent space walking, or latent space exploration, is the process of sampling a point in latent space and incrementally changing the latent representation. Its most common application is generating animations where each sampled point is fed to the decoder and is stored as a frame in the final animation. For high-quality latent representations, this produces coherent-looking animations. These animations can provide insight into the feature map of the latent space, and can ultimately lead to improvements in the training process.\n", 10 | "\n", 11 | "### References\n", 12 | "- https://keras.io/examples/generative/random_walks_with_stable_diffusion/" 13 | ] 14 | }, 15 | { 16 | "cell_type": "code", 17 | "execution_count": null, 18 | "id": "7d59a04f", 19 | "metadata": {}, 20 | "outputs": [], 21 | "source": [ 22 | "import inspect\n", 23 | "import math\n", 24 | "\n", 25 | "import torch\n", 26 | "from diffusers import (\n", 27 | " StableDiffusionPipeline,\n", 28 | " AutoencoderKL,\n", 29 | " UNet2DConditionModel,\n", 30 | " DDIMScheduler\n", 31 | ")\n", 32 | "from transformers import CLIPTextModel, CLIPTokenizer\n", 33 | "import numpy as np\n", 34 | "from PIL import Image\n", 35 | "from tqdm.auto import tqdm\n", 36 | "from IPython.display import Image as IImage" 37 | ] 38 | }, 39 | { 40 | "cell_type": "code", 41 | "execution_count": null, 42 | "id": "e6e8c9da", 43 | "metadata": {}, 44 | "outputs": [], 45 | "source": [ 46 | "device = \"cuda\"\n", 47 | "model_path = \"CompVis/stable-diffusion-v1-4\" # you can download the model weights and save locally\n", 48 | "model_path = \"/data2/hy/model_weights/stable-diffusion-v1-5/\"" 49 | ] 50 | }, 51 | { 52 | "cell_type": "code", 53 | "execution_count": null, 54 | "id": "3fdd42e9", 55 | "metadata": {}, 56 | "outputs": [], 57 | "source": [ 58 | "# add new method to original StableDiffusionPipeline\n", 59 | "class SDPipeline(StableDiffusionPipeline):\n", 60 | " \n", 61 | " @torch.no_grad()\n", 62 | " def encode_text(self, prompt):\n", 63 | " \"\"\"Encodes prompt into latent text encoding.\"\"\"\n", 64 | " # get prompt text embeddings\n", 65 | " text_inputs = self.tokenizer(\n", 66 | " prompt,\n", 67 | " padding=\"max_length\",\n", 68 | " max_length=self.tokenizer.model_max_length,\n", 69 | " return_tensors=\"pt\",\n", 70 | " )\n", 71 | " text_input_ids = text_inputs.input_ids\n", 72 | "\n", 73 | " if text_input_ids.shape[-1] > self.tokenizer.model_max_length:\n", 74 | " removed_text = self.tokenizer.batch_decode(text_input_ids[:, self.tokenizer.model_max_length :])\n", 75 | " logger.warning(\n", 76 | " \"The following part of your input was truncated because CLIP can only handle sequences up to\"\n", 77 | " f\" {self.tokenizer.model_max_length} tokens: {removed_text}\"\n", 78 | " )\n", 79 | " text_input_ids = text_input_ids[:, : self.tokenizer.model_max_length]\n", 80 | " text_embeddings = self.text_encoder(text_input_ids.to(self.device))[0]\n", 81 | " return text_embeddings\n", 82 | " \n", 83 | " @torch.no_grad()\n", 84 | " def generate_image(\n", 85 | " self,\n", 86 | " text_embeddings,\n", 87 | " height=512,\n", 88 | " width=512,\n", 89 | " num_inference_steps=50,\n", 90 | " guidance_scale=7.5,\n", 91 | " eta=0.0,\n", 92 | " generator=None,\n", 93 | " latents=None,\n", 94 | " output_type=\"pil\"\n", 95 | " ):\n", 96 | " \"\"\"Generates an image based on text_embeddings.\"\"\"\n", 97 | " if height % 8 != 0 or width % 8 != 0:\n", 98 | " raise ValueError(f\"`height` and `width` have to be divisible by 8 but are {height} and {width}.\")\n", 99 | "\n", 100 | " batch_size, seq_len, _ = text_embeddings.shape\n", 101 | "\n", 102 | " # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)\n", 103 | " # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`\n", 104 | " # corresponds to doing no classifier free guidance.\n", 105 | " do_classifier_free_guidance = guidance_scale > 1.0\n", 106 | " # get unconditional embeddings for classifier free guidance\n", 107 | " if do_classifier_free_guidance:\n", 108 | " uncond_tokens = [\"\"] * batch_size\n", 109 | "\n", 110 | " uncond_input = self.tokenizer(\n", 111 | " uncond_tokens,\n", 112 | " padding=\"max_length\",\n", 113 | " max_length=self.tokenizer.model_max_length,\n", 114 | " truncation=True,\n", 115 | " return_tensors=\"pt\",\n", 116 | " )\n", 117 | " uncond_embeddings = self.text_encoder(uncond_input.input_ids.to(self.device))[0]\n", 118 | "\n", 119 | " # duplicate unconditional embeddings for each generation per prompt, using mps friendly method\n", 120 | " seq_len = uncond_embeddings.shape[1]\n", 121 | "\n", 122 | " # For classifier free guidance, we need to do two forward passes.\n", 123 | " # Here we concatenate the unconditional and text embeddings into a single batch\n", 124 | " # to avoid doing two forward passes\n", 125 | " text_embeddings = torch.cat([uncond_embeddings, text_embeddings])\n", 126 | "\n", 127 | " # get the initial random noise unless the user supplied it\n", 128 | "\n", 129 | " # Unlike in other pipelines, latents need to be generated in the target device\n", 130 | " # for 1-to-1 results reproducibility with the CompVis implementation.\n", 131 | " # However this currently doesn't work in `mps`.\n", 132 | " latents_shape = (batch_size, self.unet.in_channels, height // 8, width // 8)\n", 133 | " latents_dtype = text_embeddings.dtype\n", 134 | " if latents is None:\n", 135 | " if self.device.type == \"mps\":\n", 136 | " # randn does not exist on mps\n", 137 | " latents = torch.randn(latents_shape, generator=generator, device=\"cpu\", dtype=latents_dtype).to(\n", 138 | " self.device\n", 139 | " )\n", 140 | " else:\n", 141 | " latents = torch.randn(latents_shape, generator=generator, device=self.device, dtype=latents_dtype)\n", 142 | " else:\n", 143 | " if latents.dim() != len(latents_shape):\n", 144 | " raise ValueError(f\"Unexpected latents dimension, got {latents.shape}, expected {latents_shape}\")\n", 145 | " if latents.shape[0] != batch_size:\n", 146 | " latents = latents.repeat(batch_size, 1, 1, 1)\n", 147 | " latents = latents.to(self.device)\n", 148 | "\n", 149 | " # set timesteps\n", 150 | " self.scheduler.set_timesteps(num_inference_steps)\n", 151 | "\n", 152 | " # Some schedulers like PNDM have timesteps as arrays\n", 153 | " # It's more optimized to move all timesteps to correct device beforehand\n", 154 | " timesteps_tensor = self.scheduler.timesteps.to(self.device)\n", 155 | "\n", 156 | " # scale the initial noise by the standard deviation required by the scheduler\n", 157 | " latents = latents * self.scheduler.init_noise_sigma\n", 158 | "\n", 159 | " # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature\n", 160 | " # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.\n", 161 | " # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502\n", 162 | " # and should be between [0, 1]\n", 163 | " accepts_eta = \"eta\" in set(inspect.signature(self.scheduler.step).parameters.keys())\n", 164 | " extra_step_kwargs = {}\n", 165 | " if accepts_eta:\n", 166 | " extra_step_kwargs[\"eta\"] = eta\n", 167 | "\n", 168 | " for i, t in enumerate(self.progress_bar(timesteps_tensor)):\n", 169 | " # expand the latents if we are doing classifier free guidance\n", 170 | " latent_model_input = torch.cat([latents] * 2) if do_classifier_free_guidance else latents\n", 171 | " latent_model_input = self.scheduler.scale_model_input(latent_model_input, t)\n", 172 | "\n", 173 | " # predict the noise residual\n", 174 | " noise_pred = self.unet(latent_model_input, t, encoder_hidden_states=text_embeddings).sample\n", 175 | "\n", 176 | " # perform guidance\n", 177 | " if do_classifier_free_guidance:\n", 178 | " noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)\n", 179 | " noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)\n", 180 | "\n", 181 | " # compute the previous noisy sample x_t -> x_t-1\n", 182 | " latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs).prev_sample\n", 183 | "\n", 184 | " latents = 1 / 0.18215 * latents\n", 185 | " image = self.vae.decode(latents).sample\n", 186 | "\n", 187 | " image = (image / 2 + 0.5).clamp(0, 1)\n", 188 | "\n", 189 | " # we always cast to float32 as this does not cause significant overhead and is compatible with bfloa16\n", 190 | " image = image.cpu().permute(0, 2, 3, 1).float().numpy()\n", 191 | "\n", 192 | " if output_type == \"pil\":\n", 193 | " image = self.numpy_to_pil(image)\n", 194 | " \n", 195 | " return image" 196 | ] 197 | }, 198 | { 199 | "cell_type": "code", 200 | "execution_count": null, 201 | "id": "76eda94d", 202 | "metadata": {}, 203 | "outputs": [], 204 | "source": [ 205 | "# Define noise scheduler: the parameters must match the original stable diffusion\n", 206 | "noise_scheduler = DDIMScheduler(\n", 207 | " num_train_timesteps=1000,\n", 208 | " beta_start=0.00085,\n", 209 | " beta_end=0.012,\n", 210 | " beta_schedule=\"scaled_linear\",\n", 211 | " clip_sample=False, # don't clip sample, the x0 in stable diffusion not in range [-1, 1]\n", 212 | " set_alpha_to_one=False,\n", 213 | " steps_offset=1,\n", 214 | ")\n", 215 | "# load Stable Diffusion Pipeline\n", 216 | "pipe = SDPipeline.from_pretrained(model_path, torch_dtype=torch.float16).to(device)" 217 | ] 218 | }, 219 | { 220 | "cell_type": "code", 221 | "execution_count": null, 222 | "id": "bed43749", 223 | "metadata": {}, 224 | "outputs": [], 225 | "source": [ 226 | "def image_grid(imgs, rows, cols):\n", 227 | " assert len(imgs) == rows*cols\n", 228 | "\n", 229 | " w, h = imgs[0].size\n", 230 | " grid = Image.new('RGB', size=(cols*w, rows*h))\n", 231 | " grid_w, grid_h = grid.size\n", 232 | " \n", 233 | " for i, img in enumerate(imgs):\n", 234 | " grid.paste(img, box=(i%cols*w, i//cols*h))\n", 235 | " return grid" 236 | ] 237 | }, 238 | { 239 | "cell_type": "markdown", 240 | "id": "1b27bb82", 241 | "metadata": {}, 242 | "source": [ 243 | "## Interpolating between text prompts\n", 244 | "\n", 245 | "In Stable Diffusion, a text prompt is first encoded into a vector, and that encoding is used to guide the diffusion process. The latent encoding vector has shape 77x768 (that's huge!), and when we give Stable Diffusion a text prompt, we're generating images from just one such point on the latent manifold.\n", 246 | "\n", 247 | "To explore more of this manifold, we can interpolate between two text encodings and generate images at those interpolated points:" 248 | ] 249 | }, 250 | { 251 | "cell_type": "code", 252 | "execution_count": null, 253 | "id": "21082172", 254 | "metadata": {}, 255 | "outputs": [], 256 | "source": [ 257 | "prompt_1 = \"A watercolor painting of a Golden Retriever at the beach\"\n", 258 | "prompt_2 = \"A still life DSLR photo of a bowl of fruit\"\n", 259 | "interpolation_steps = 8\n", 260 | "height = 512\n", 261 | "width = 512\n", 262 | "num_inference_steps = 25\n", 263 | "\n", 264 | "frames_per_second = 2\n", 265 | "\n", 266 | "encoding_1 = pipe.encode_text(prompt_1)\n", 267 | "encoding_2 = pipe.encode_text(prompt_2)\n", 268 | "\n", 269 | "interpolated_encodings = torch.cat([\n", 270 | " torch.lerp(encoding_1, encoding_2, weight) for weight in np.linspace(0., 1., interpolation_steps)\n", 271 | "], dim=0)\n", 272 | "\n", 273 | "\n", 274 | "# we generate a latents (noise) to let all generated images have same start noise.\n", 275 | "generator = torch.Generator(device).manual_seed(12345)\n", 276 | "latents_shape = (1, pipe.unet.in_channels, height // 8, width // 8)\n", 277 | "latents = torch.randn(latents_shape, generator=generator, device=device, dtype=encoding_1.dtype)" 278 | ] 279 | }, 280 | { 281 | "cell_type": "code", 282 | "execution_count": null, 283 | "id": "603046aa", 284 | "metadata": {}, 285 | "outputs": [], 286 | "source": [ 287 | "image = pipe.generate_image(\n", 288 | " interpolated_encodings,\n", 289 | " height=height,\n", 290 | " width=width,\n", 291 | " num_inference_steps=num_inference_steps,\n", 292 | " latents=latents)\n", 293 | "image[0].save(\n", 294 | " \"doggo-and-fruit-8.gif\",\n", 295 | " save_all=True,\n", 296 | " append_images=image[1:],\n", 297 | " duration=1000 // frames_per_second,\n", 298 | " loop=0\n", 299 | ")" 300 | ] 301 | }, 302 | { 303 | "cell_type": "code", 304 | "execution_count": null, 305 | "id": "499c877b", 306 | "metadata": {}, 307 | "outputs": [], 308 | "source": [ 309 | "IImage(\"doggo-and-fruit-8.gif\")" 310 | ] 311 | }, 312 | { 313 | "cell_type": "code", 314 | "execution_count": null, 315 | "id": "aadb104c", 316 | "metadata": { 317 | "scrolled": true 318 | }, 319 | "outputs": [], 320 | "source": [ 321 | "grid = image_grid(image, 1, len(image))\n", 322 | "grid" 323 | ] 324 | }, 325 | { 326 | "cell_type": "markdown", 327 | "id": "dc2773e3", 328 | "metadata": {}, 329 | "source": [ 330 | "To best visualize this, we should do a much more fine-grained interpolation, using hundreds of steps. In order to keep batch size small (so that we don't OOM our GPU), this requires manually batching our interpolated encodings." 331 | ] 332 | }, 333 | { 334 | "cell_type": "code", 335 | "execution_count": null, 336 | "id": "4b411252", 337 | "metadata": {}, 338 | "outputs": [], 339 | "source": [ 340 | "prompt_1 = \"A watercolor painting of a Golden Retriever at the beach\"\n", 341 | "prompt_2 = \"A still life DSLR photo of a bowl of fruit\"\n", 342 | "interpolation_steps = 128\n", 343 | "height = 512\n", 344 | "width = 512\n", 345 | "batch_size = 8\n", 346 | "num_inference_steps = 25\n", 347 | "\n", 348 | "frames_per_second = 8\n", 349 | "\n", 350 | "encoding_1 = pipe.encode_text(prompt_1)\n", 351 | "encoding_2 = pipe.encode_text(prompt_2)\n", 352 | "\n", 353 | "interpolated_encodings = torch.cat([\n", 354 | " torch.lerp(encoding_1, encoding_2, weight) for weight in np.linspace(0., 1., interpolation_steps)\n", 355 | "], dim=0)\n", 356 | "\n", 357 | "generator = torch.Generator(device).manual_seed(12345)\n", 358 | "latents_shape = (1, pipe.unet.in_channels, height // 8, width // 8)\n", 359 | "latents = torch.randn(latents_shape, generator=generator, device=device, dtype=encoding_1.dtype)" 360 | ] 361 | }, 362 | { 363 | "cell_type": "code", 364 | "execution_count": null, 365 | "id": "126d002d", 366 | "metadata": {}, 367 | "outputs": [], 368 | "source": [ 369 | "generated_images = []\n", 370 | "for i in range(interpolation_steps // batch_size):\n", 371 | " image = pipe.generate_image(\n", 372 | " interpolated_encodings[i*batch_size:(i+1)*batch_size],\n", 373 | " height=height,\n", 374 | " width=width,\n", 375 | " num_inference_steps=num_inference_steps,\n", 376 | " latents=latents\n", 377 | " )\n", 378 | " generated_images.extend(image)\n", 379 | "\n", 380 | "generated_images[0].save(\n", 381 | " \"doggo-and-fruit-128.gif\",\n", 382 | " save_all=True,\n", 383 | " append_images=generated_images[1:],\n", 384 | " duration=1000 // frames_per_second,\n", 385 | " loop=0\n", 386 | ")" 387 | ] 388 | }, 389 | { 390 | "cell_type": "code", 391 | "execution_count": null, 392 | "id": "0153a0b1", 393 | "metadata": {}, 394 | "outputs": [], 395 | "source": [ 396 | "IImage(\"doggo-and-fruit-128.gif\")" 397 | ] 398 | }, 399 | { 400 | "cell_type": "markdown", 401 | "id": "e977306d", 402 | "metadata": {}, 403 | "source": [ 404 | "We can even extend this concept for more than one image. For example, we can interpolate between four prompts:" 405 | ] 406 | }, 407 | { 408 | "cell_type": "code", 409 | "execution_count": null, 410 | "id": "8958eadd", 411 | "metadata": {}, 412 | "outputs": [], 413 | "source": [ 414 | "prompt_1 = \"A watercolor painting of a Golden Retriever at the beach\"\n", 415 | "prompt_2 = \"A still life DSLR photo of a bowl of fruit\"\n", 416 | "prompt_3 = \"The eiffel tower in the style of starry night\"\n", 417 | "prompt_4 = \"An architectural sketch of a skyscraper\"\n", 418 | "\n", 419 | "height = 512\n", 420 | "width = 512\n", 421 | "num_inference_steps = 25\n", 422 | "\n", 423 | "interpolation_steps = 6\n", 424 | "batch_size = 4\n", 425 | "\n", 426 | "encoding_1 = pipe.encode_text(prompt_1)\n", 427 | "encoding_2 = pipe.encode_text(prompt_2)\n", 428 | "encoding_3 = pipe.encode_text(prompt_3)\n", 429 | "encoding_4 = pipe.encode_text(prompt_4)\n", 430 | "\n", 431 | "interpolated_encodings_12 = torch.cat([\n", 432 | " torch.lerp(encoding_1, encoding_2, weight) for weight in np.linspace(0., 1., interpolation_steps)\n", 433 | "], dim=0)\n", 434 | "interpolated_encodings_34 = torch.cat([\n", 435 | " torch.lerp(encoding_3, encoding_4, weight) for weight in np.linspace(0., 1., interpolation_steps)\n", 436 | "], dim=0)\n", 437 | "interpolated_encodings = torch.cat([\n", 438 | " torch.lerp(interpolated_encodings_12, interpolated_encodings_34, weight) for weight in np.linspace(0., 1., interpolation_steps)\n", 439 | "], dim=0)\n", 440 | "\n", 441 | "\n", 442 | "generator = torch.Generator(device).manual_seed(12345)\n", 443 | "latents_shape = (1, pipe.unet.in_channels, height // 8, width // 8)\n", 444 | "latents = torch.randn(latents_shape, generator=generator, device=device, dtype=encoding_1.dtype)" 445 | ] 446 | }, 447 | { 448 | "cell_type": "code", 449 | "execution_count": null, 450 | "id": "3fe81192", 451 | "metadata": {}, 452 | "outputs": [], 453 | "source": [ 454 | "generated_images = []\n", 455 | "for i in range(interpolation_steps**2 // batch_size):\n", 456 | " image = pipe.generate_image(\n", 457 | " interpolated_encodings[i*batch_size:(i+1)*batch_size],\n", 458 | " height=height,\n", 459 | " width=width,\n", 460 | " num_inference_steps=num_inference_steps,\n", 461 | " latents=latents\n", 462 | " )\n", 463 | " generated_images.extend(image)" 464 | ] 465 | }, 466 | { 467 | "cell_type": "code", 468 | "execution_count": null, 469 | "id": "0c5f4fbc", 470 | "metadata": {}, 471 | "outputs": [], 472 | "source": [ 473 | "grid = image_grid(generated_images, interpolation_steps, interpolation_steps)\n", 474 | "grid" 475 | ] 476 | }, 477 | { 478 | "cell_type": "markdown", 479 | "id": "e28c3216", 480 | "metadata": {}, 481 | "source": [ 482 | "We can also interpolate while allowing diffusion noise to vary by dropping the `latents` parameter:" 483 | ] 484 | }, 485 | { 486 | "cell_type": "code", 487 | "execution_count": null, 488 | "id": "499ca272", 489 | "metadata": {}, 490 | "outputs": [], 491 | "source": [ 492 | "generated_images = []\n", 493 | "for i in range(interpolation_steps**2 // batch_size):\n", 494 | " image = pipe.generate_image(\n", 495 | " interpolated_encodings[i*batch_size:(i+1)*batch_size],\n", 496 | " height=height,\n", 497 | " width=width,\n", 498 | " num_inference_steps=num_inference_steps,\n", 499 | " )\n", 500 | " generated_images.extend(image)" 501 | ] 502 | }, 503 | { 504 | "cell_type": "code", 505 | "execution_count": null, 506 | "id": "ead171d3", 507 | "metadata": {}, 508 | "outputs": [], 509 | "source": [ 510 | "grid = image_grid(generated_images, interpolation_steps, interpolation_steps)\n", 511 | "grid" 512 | ] 513 | }, 514 | { 515 | "cell_type": "markdown", 516 | "id": "fe1ff149", 517 | "metadata": {}, 518 | "source": [ 519 | "## A walk around a text prompt\n", 520 | "Our next experiment will be to go for a walk around the latent manifold starting from a point produced by a particular prompt." 521 | ] 522 | }, 523 | { 524 | "cell_type": "code", 525 | "execution_count": null, 526 | "id": "e87b914d", 527 | "metadata": {}, 528 | "outputs": [], 529 | "source": [ 530 | "prompt = \"The Eiffel Tower in the style of starry night\"\n", 531 | "\n", 532 | "height = 512\n", 533 | "width = 512\n", 534 | "num_inference_steps = 25\n", 535 | "\n", 536 | "walk_steps = 128\n", 537 | "step_size = 0.001\n", 538 | "batch_size = 8\n", 539 | "\n", 540 | "encoding = pipe.encode_text(prompt)\n", 541 | "\n", 542 | "delta = torch.ones_like(encoding) * step_size\n", 543 | "\n", 544 | "walked_encodings = []\n", 545 | "for step_index in range(walk_steps):\n", 546 | " walked_encodings.append(encoding)\n", 547 | " encoding = encoding + delta\n", 548 | "walked_encodings = torch.cat(walked_encodings, dim=0)\n", 549 | "\n", 550 | "generator = torch.Generator(device).manual_seed(0)\n", 551 | "latents_shape = (1, pipe.unet.in_channels, height // 8, width // 8)\n", 552 | "latents = torch.randn(latents_shape, generator=generator, device=device, dtype=encoding_1.dtype)" 553 | ] 554 | }, 555 | { 556 | "cell_type": "code", 557 | "execution_count": null, 558 | "id": "4b1b9350", 559 | "metadata": {}, 560 | "outputs": [], 561 | "source": [ 562 | "generated_images = []\n", 563 | "for i in range( walk_steps // batch_size):\n", 564 | " image = pipe.generate_image(\n", 565 | " walked_encodings[i*batch_size:(i+1)*batch_size],\n", 566 | " height=height,\n", 567 | " width=width,\n", 568 | " num_inference_steps=num_inference_steps,\n", 569 | " latents=latents\n", 570 | " )\n", 571 | " generated_images.extend(image)" 572 | ] 573 | }, 574 | { 575 | "cell_type": "code", 576 | "execution_count": null, 577 | "id": "1c6401b2", 578 | "metadata": {}, 579 | "outputs": [], 580 | "source": [ 581 | "frames_per_second = 8\n", 582 | "generated_images[0].save(\n", 583 | " \"eiffel-tower-starry-night.gif\",\n", 584 | " save_all=True,\n", 585 | " append_images=generated_images[1:],\n", 586 | " duration=1000 // frames_per_second,\n", 587 | " loop=0\n", 588 | ")" 589 | ] 590 | }, 591 | { 592 | "cell_type": "code", 593 | "execution_count": null, 594 | "id": "0de5e6fd", 595 | "metadata": {}, 596 | "outputs": [], 597 | "source": [ 598 | "IImage(\"eiffel-tower-starry-night.gif\")" 599 | ] 600 | }, 601 | { 602 | "cell_type": "markdown", 603 | "id": "9311037c", 604 | "metadata": {}, 605 | "source": [ 606 | "Perhaps unsurprisingly, walking too far from the encoder's latent manifold produces images that look incoherent. Try it for yourself by setting your own prompt, and adjusting step_size to increase or decrease the magnitude of the walk. Note that when the magnitude of the walk gets large, the walk often leads into areas which produce extremely noisy images." 607 | ] 608 | }, 609 | { 610 | "cell_type": "markdown", 611 | "id": "46f92884", 612 | "metadata": {}, 613 | "source": [ 614 | "## A circular walk through the diffusion noise space for a single prompt\n", 615 | "\n", 616 | "Our final experiment is to stick to one prompt and explore the variety of images that the diffusion model can produce from that prompt. We do this by controlling the noise that is used to seed the diffusion process.\n", 617 | "\n", 618 | "We create two noise components, `x` and `y`, and do a walk from 0 to 2π, summing the cosine of our `x` component and the sin of our `y `component to produce noise. Using this approach, the end of our walk arrives at the same noise inputs where we began our walk, so we get a \"loopable\" result!" 619 | ] 620 | }, 621 | { 622 | "cell_type": "code", 623 | "execution_count": null, 624 | "id": "9985b9ec", 625 | "metadata": {}, 626 | "outputs": [], 627 | "source": [ 628 | "prompt = \"An oil paintings of cows in a field next to a windmill in Holland\"\n", 629 | "\n", 630 | "height = 512\n", 631 | "width = 512\n", 632 | "num_inference_steps = 25\n", 633 | "\n", 634 | "walk_steps = 128\n", 635 | "batch_size = 8\n", 636 | "\n", 637 | "encoding = pipe.encode_text(prompt)\n", 638 | "\n", 639 | "torch.manual_seed(0)\n", 640 | "latents_shape = (1, pipe.unet.in_channels, height // 8, width // 8)\n", 641 | "walk_noise_x = torch.randn(latents_shape, device=device, dtype=encoding.dtype)\n", 642 | "walk_noise_y = torch.randn(latents_shape, device=device, dtype=encoding.dtype)\n", 643 | "\n", 644 | "walk_scale_x = torch.cos(torch.linspace(0, 2, walk_steps) * math.pi).to(device, dtype=encoding.dtype)\n", 645 | "walk_scale_y = torch.sin(torch.linspace(0, 2, walk_steps) * math.pi).to(device, dtype=encoding.dtype)\n", 646 | "latents_x = walk_scale_x[:, None, None, None] * walk_noise_x\n", 647 | "latents_y = walk_scale_y[:, None, None, None] * walk_noise_y\n", 648 | "latents = latents_x + latents_y\n", 649 | "\n", 650 | "walked_encodings = encoding.repeat(batch_size, 1, 1)" 651 | ] 652 | }, 653 | { 654 | "cell_type": "code", 655 | "execution_count": null, 656 | "id": "533f8d41", 657 | "metadata": {}, 658 | "outputs": [], 659 | "source": [ 660 | "generated_images = []\n", 661 | "for i in range( walk_steps // batch_size):\n", 662 | " image = pipe.generate_image(\n", 663 | " walked_encodings,\n", 664 | " height=height,\n", 665 | " width=width,\n", 666 | " num_inference_steps=num_inference_steps,\n", 667 | " latents=latents[i*batch_size:(i+1)*batch_size]\n", 668 | " )\n", 669 | " generated_images.extend(image)" 670 | ] 671 | }, 672 | { 673 | "cell_type": "code", 674 | "execution_count": null, 675 | "id": "fcf9519d", 676 | "metadata": {}, 677 | "outputs": [], 678 | "source": [ 679 | "frames_per_second = 8\n", 680 | "generated_images[0].save(\n", 681 | " \"cows.gif\",\n", 682 | " save_all=True,\n", 683 | " append_images=generated_images[1:],\n", 684 | " duration=1000 // frames_per_second,\n", 685 | " loop=0\n", 686 | ")" 687 | ] 688 | }, 689 | { 690 | "cell_type": "code", 691 | "execution_count": null, 692 | "id": "fd020c39", 693 | "metadata": {}, 694 | "outputs": [], 695 | "source": [ 696 | "IImage(\"cows.gif\")" 697 | ] 698 | } 699 | ], 700 | "metadata": { 701 | "kernelspec": { 702 | "display_name": "Python 3 (ipykernel)", 703 | "language": "python", 704 | "name": "python3" 705 | }, 706 | "language_info": { 707 | "codemirror_mode": { 708 | "name": "ipython", 709 | "version": 3 710 | }, 711 | "file_extension": ".py", 712 | "mimetype": "text/x-python", 713 | "name": "python", 714 | "nbconvert_exporter": "python", 715 | "pygments_lexer": "ipython3", 716 | "version": "3.9.7" 717 | } 718 | }, 719 | "nbformat": 4, 720 | "nbformat_minor": 5 721 | } 722 | --------------------------------------------------------------------------------