NonLinearModelV0-FewShotDataset.ipynb

{
 "cells": [
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": [
    "# LinearModelV0 with FewShotDataset\n",
    "\n",
    "This notebook is an experimental approach on solving the ARC challenge. See https://arcprize.org/ for more details."
   ],
   "id": "2deab4fa4b404fd9"
  },
  {
   "cell_type": "code",
   "id": "469d911a-eb08-40db-9491-58582e0a5244",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-05T13:12:17.600899Z",
     "start_time": "2024-08-05T13:12:16.335485Z"
    }
   },
   "source": [
    "import os\n",
    "import random\n",
    "\n",
    "import torch\n",
    "\n",
    "from pathlib import Path\n",
    "\n",
    "from arc2024.data import FewShotDataset\n",
    "from arc2024.utils import plot\n",
    "from arc2024.utils.calc import accuracy_fn, norm_arc20204\n",
    "from arc2024.models.non_linear import NonLinearModelV0"
   ],
   "outputs": [],
   "execution_count": 1
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "# Setup Device Agnostic Environment",
   "id": "238e55441bbcf17a"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-05T13:12:17.892147Z",
     "start_time": "2024-08-05T13:12:17.601910Z"
    }
   },
   "cell_type": "code",
   "source": [
    "import arc2024.utils.amdgpu as amd\n",
    "\n",
    "if amd.is_amdgpu_on_linux():\n",
    "    # Use ROCm on rx6600/6600xt by setting the environment variable HSA_OVERRIDE_GFX_VERSION to \"10.3.0\"\n",
    "    # Source: https://github.com/ROCm/ROCm/issues/1698#issuecomment-1261646854\n",
    "    amd.setup_amdgpu_environment(\n",
    "        hsa_override_gfx_version=\"10.3.0\"\n",
    "    )\n",
    "\n",
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "device"
   ],
   "id": "564e0858a54fdba8",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "device(type='cuda')"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "execution_count": 2
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "## Prepare Datasets and DataLoaders",
   "id": "e1d56ce2d680401c"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-05T13:12:21.367293Z",
     "start_time": "2024-08-05T13:12:17.893Z"
    }
   },
   "cell_type": "code",
   "source": [
    "dataset_dir = Path('datasets')\n",
    "\n",
    "BATCH_SIZE = 32\n",
    "NUM_WORKERS = os.cpu_count()\n",
    "\n",
    "train_dataset = FewShotDataset(\n",
    "    dataset_dir=dataset_dir,\n",
    "    dataset_name='training',\n",
    "    permutations_enabled=True\n",
    ")\n",
    "train_dataloader = torch.utils.data.DataLoader(\n",
    "    dataset=train_dataset,\n",
    "    batch_size=BATCH_SIZE,\n",
    "    shuffle=True,\n",
    "    num_workers=NUM_WORKERS\n",
    ")\n",
    "\n",
    "test_dataset = FewShotDataset(\n",
    "    dataset_dir=dataset_dir,\n",
    "    dataset_name='evaluation'\n",
    ")\n",
    "test_dataloader = torch.utils.data.DataLoader(\n",
    "    dataset=test_dataset,\n",
    "    batch_size=BATCH_SIZE,\n",
    "    shuffle=True,\n",
    "    num_workers=NUM_WORKERS\n",
    ")"
   ],
   "id": "bb2cae609fdde826",
   "outputs": [],
   "execution_count": 3
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "## Train Model",
   "id": "2177ae9236464708"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-05T13:12:37.150005Z",
     "start_time": "2024-08-05T13:12:37.129010Z"
    }
   },
   "cell_type": "code",
   "source": [
    "model = NonLinearModelV0(\n",
    "    in_features=900, # 30x30\n",
    "    hidden_features=600, # do some compression\n",
    "    out_features=900\n",
    ").to(device)\n",
    "\n",
    "model"
   ],
   "id": "3f0a32b80655a8ce",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "NonLinearModelV0(\n",
       "  (flatten): Flatten(start_dim=1, end_dim=-1)\n",
       "  (linear1): Linear(in_features=900, out_features=600, bias=True)\n",
       "  (relu1): ReLU()\n",
       "  (linear2): Linear(in_features=600, out_features=900, bias=True)\n",
       "  (relu2): ReLU()\n",
       "  (unflatten): Unflatten(dim=1, unflattened_size=(1, 30, 30))\n",
       ")"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "execution_count": 5
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "### Verify Input and Output Shapes with Random Tensor",
   "id": "73ba81c9ca21ee99"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-04T12:21:22.808130Z",
     "start_time": "2024-08-04T12:21:22.665657Z"
    }
   },
   "cell_type": "code",
   "source": [
    "x = torch.rand([1, 30, 30]).to(device)\n",
    "model(x).shape"
   ],
   "id": "e1b7befceacdb2e",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([1, 1, 30, 30])"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "execution_count": 5
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-04T12:21:22.913519Z",
     "start_time": "2024-08-04T12:21:22.809018Z"
    }
   },
   "cell_type": "code",
   "source": [
    "X, y, _ = train_dataset[0]\n",
    "X = X.to(device) / 10\n",
    "y = y.to(device) / 10\n",
    "\n",
    "y_pred = model(X)\n",
    "y_pred, y_pred.shape"
   ],
   "id": "c490ced35ddd64a0",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(tensor([[[[0.1616, 0.5808, 0.0000, 0.1905, 0.0000, 0.0882, 0.1266, 0.1101,\n",
       "            0.0562, 0.0000, 0.0000, 0.0000, 0.0000, 0.2197, 0.3875, 0.0000,\n",
       "            0.0000, 0.0000, 0.0000, 0.0000, 0.0461, 0.0000, 0.0928, 0.0000,\n",
       "            0.0130, 0.0000, 0.3987, 0.0000, 0.1360, 0.5793],\n",
       "           [0.1371, 0.0393, 0.0000, 0.0151, 0.0000, 0.0000, 0.5231, 0.1874,\n",
       "            0.0000, 0.0000, 0.0569, 0.1457, 0.0000, 0.0000, 0.0000, 0.1606,\n",
       "            0.0507, 0.0000, 0.4667, 0.1733, 0.1434, 0.0156, 0.2047, 0.0000,\n",
       "            0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.6841],\n",
       "           [0.1206, 0.4218, 0.0000, 0.2076, 0.0000, 0.0770, 0.0000, 0.3432,\n",
       "            0.0000, 0.0000, 0.1939, 0.0000, 0.4347, 0.0197, 0.0000, 0.7384,\n",
       "            0.0000, 0.1315, 0.2661, 0.0030, 0.0110, 0.1159, 0.1432, 0.0768,\n",
       "            0.1300, 0.2811, 0.0000, 0.2150, 0.0000, 0.0000],\n",
       "           [0.0891, 0.0969, 0.0000, 0.0000, 0.0000, 0.1536, 0.4914, 0.0662,\n",
       "            0.0445, 0.0000, 0.0000, 0.0000, 0.1206, 0.2063, 0.0000, 0.0451,\n",
       "            0.1259, 0.0558, 0.0000, 0.2651, 0.2538, 0.0000, 0.2312, 0.2918,\n",
       "            0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.1484],\n",
       "           [0.5017, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0756, 0.0970,\n",
       "            0.0000, 0.2523, 0.1249, 0.0528, 0.1295, 0.2021, 0.0000, 0.0000,\n",
       "            0.2944, 0.1127, 0.0777, 0.0000, 0.0000, 0.0000, 0.0065, 0.0000,\n",
       "            0.0000, 0.1001, 0.0000, 0.0000, 0.0000, 0.0075],\n",
       "           [0.0000, 0.1449, 0.0000, 0.0000, 0.1810, 0.2316, 0.0747, 0.2301,\n",
       "            0.0000, 0.0000, 0.3516, 0.0070, 0.0545, 0.0000, 0.1351, 0.1461,\n",
       "            0.2829, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.3652, 0.3040,\n",
       "            0.0000, 0.0000, 0.3147, 0.0000, 0.0000, 0.0000],\n",
       "           [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.3087, 0.3163, 0.2828,\n",
       "            0.0000, 0.0000, 0.0000, 0.0157, 0.0000, 0.0746, 0.2381, 0.0000,\n",
       "            0.2125, 0.0000, 0.1244, 0.1622, 0.0000, 0.0000, 0.0000, 0.0000,\n",
       "            0.1596, 0.0000, 0.0574, 0.0000, 0.0000, 0.0000],\n",
       "           [0.0000, 0.0000, 0.5920, 0.1206, 0.0966, 0.0000, 0.0000, 0.0000,\n",
       "            0.0000, 0.0391, 0.0478, 0.0000, 0.3036, 0.0000, 0.0000, 0.0000,\n",
       "            0.1395, 0.0000, 0.0000, 0.1506, 0.0000, 0.0000, 0.0811, 0.0000,\n",
       "            0.3147, 0.0903, 0.0000, 0.0000, 0.0716, 0.1005],\n",
       "           [0.0000, 0.0000, 0.0000, 0.0000, 0.1554, 0.0489, 0.1732, 0.0000,\n",
       "            0.0000, 0.0000, 0.0000, 0.0000, 0.1436, 0.1363, 0.0000, 0.0000,\n",
       "            0.0000, 0.5013, 0.0597, 0.2919, 0.0000, 0.0000, 0.0000, 0.1549,\n",
       "            0.0000, 0.0925, 0.0000, 0.0000, 0.0000, 0.1878],\n",
       "           [0.1042, 0.0127, 0.0000, 0.0000, 0.2826, 0.0000, 0.1069, 0.0000,\n",
       "            0.1797, 0.2031, 0.1450, 0.0000, 0.0000, 0.0000, 0.0000, 0.0637,\n",
       "            0.2416, 0.0000, 0.0000, 0.1593, 0.0733, 0.0000, 0.2820, 0.0000,\n",
       "            0.1744, 0.0000, 0.0000, 0.0000, 0.2937, 0.1066],\n",
       "           [0.3076, 0.5712, 0.0000, 0.4075, 0.0000, 0.1773, 0.0000, 0.4792,\n",
       "            0.2123, 0.0000, 0.0000, 0.0000, 0.3296, 0.4324, 0.7023, 0.3216,\n",
       "            0.1778, 0.2209, 0.0666, 0.3499, 0.0000, 0.0000, 0.2956, 0.1087,\n",
       "            0.0000, 0.2222, 0.0000, 0.2212, 0.0000, 0.0000],\n",
       "           [0.4417, 0.0000, 0.0681, 0.4184, 0.0713, 0.3002, 0.1289, 0.0000,\n",
       "            0.0000, 0.0000, 0.1069, 0.3329, 0.0000, 0.0000, 0.0000, 0.0000,\n",
       "            0.0304, 0.1967, 0.0000, 0.0000, 0.0000, 0.3225, 0.0000, 0.0000,\n",
       "            0.0000, 0.0000, 0.0000, 0.0000, 0.3202, 0.0000],\n",
       "           [0.0000, 0.0000, 0.0000, 0.1473, 0.0000, 0.0000, 0.0000, 0.0000,\n",
       "            0.3566, 0.0000, 0.0229, 0.4523, 0.4347, 0.3050, 0.0000, 0.1427,\n",
       "            0.0000, 0.2846, 0.0000, 0.0598, 0.0000, 0.0000, 0.1294, 0.0000,\n",
       "            0.0000, 0.0000, 0.0430, 0.0000, 0.2718, 0.0000],\n",
       "           [0.2200, 0.0000, 0.0000, 0.0000, 0.0887, 0.0000, 0.1303, 0.0290,\n",
       "            0.2757, 0.0000, 0.2054, 0.3967, 0.0364, 0.2193, 0.2922, 0.5505,\n",
       "            0.0000, 0.0000, 0.0510, 0.0000, 0.0000, 0.0000, 0.3260, 0.0000,\n",
       "            0.0000, 0.2174, 0.0000, 0.0000, 0.0000, 0.0000],\n",
       "           [0.0482, 0.1743, 0.0000, 0.1720, 0.0000, 0.0000, 0.1967, 0.6557,\n",
       "            0.0882, 0.0675, 0.0000, 0.1958, 0.0000, 0.4624, 0.0525, 0.0000,\n",
       "            0.0000, 0.1815, 0.0000, 0.0000, 0.0000, 0.2138, 0.0000, 0.0000,\n",
       "            0.0987, 0.2127, 0.0000, 0.0000, 0.0000, 0.0000],\n",
       "           [0.1555, 0.0000, 0.0562, 0.5119, 0.4802, 0.0026, 0.1286, 0.1253,\n",
       "            0.0000, 0.0000, 0.0000, 0.0000, 0.0461, 0.0000, 0.0000, 0.1269,\n",
       "            0.0000, 0.0000, 0.0937, 0.0000, 0.3885, 0.0000, 0.0000, 0.1906,\n",
       "            0.0000, 0.2900, 0.0000, 0.0000, 0.0127, 0.2590],\n",
       "           [0.0000, 0.4256, 0.0000, 0.0000, 0.0000, 0.0000, 0.2599, 0.0000,\n",
       "            0.0000, 0.0000, 0.1382, 0.0000, 0.0000, 0.0000, 0.0000, 0.2319,\n",
       "            0.1547, 0.2089, 0.1801, 0.0000, 0.6606, 0.0000, 0.0000, 0.0000,\n",
       "            0.0000, 0.0000, 0.0500, 0.0000, 0.1846, 0.3407],\n",
       "           [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n",
       "            0.0000, 0.0000, 0.0115, 0.4736, 0.1281, 0.0888, 0.0000, 0.1901,\n",
       "            0.6243, 0.1132, 0.0000, 0.0000, 0.3254, 0.0622, 0.0000, 0.1552,\n",
       "            0.1534, 0.2107, 0.2824, 0.0000, 0.0000, 0.0000],\n",
       "           [0.0000, 0.0551, 0.0161, 0.1119, 0.0478, 0.0000, 0.0000, 0.0000,\n",
       "            0.0171, 0.0413, 0.0000, 0.6926, 0.0000, 0.0000, 0.0000, 0.3760,\n",
       "            0.0000, 0.0000, 0.1121, 0.0795, 0.0000, 0.4693, 0.5276, 0.0000,\n",
       "            0.4994, 0.0000, 0.0000, 0.1700, 0.0000, 0.0000],\n",
       "           [0.0000, 0.3804, 0.0000, 0.0315, 0.0000, 0.0000, 0.3833, 0.0000,\n",
       "            0.1488, 0.1954, 0.1912, 0.0775, 0.2031, 0.0000, 0.0000, 0.0092,\n",
       "            0.0000, 0.0000, 0.0000, 0.0000, 0.1554, 0.3644, 0.0749, 0.2117,\n",
       "            0.0000, 0.2360, 0.4196, 0.0000, 0.2435, 0.0000],\n",
       "           [0.2620, 0.0000, 0.4724, 0.1231, 0.0000, 0.3065, 0.0824, 0.2241,\n",
       "            0.0000, 0.0367, 0.3268, 0.3750, 0.4384, 0.0787, 0.0000, 0.2140,\n",
       "            0.0223, 0.2594, 0.0000, 0.1345, 0.0000, 0.1244, 0.0000, 0.0000,\n",
       "            0.0000, 0.0000, 0.0000, 0.1551, 0.0000, 0.0000],\n",
       "           [0.0443, 0.0000, 0.5665, 0.0353, 0.2579, 0.0000, 0.1826, 0.0000,\n",
       "            0.0959, 0.0374, 0.0000, 0.3511, 0.0000, 0.1984, 0.4071, 0.0000,\n",
       "            0.4278, 0.3067, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n",
       "            0.5185, 0.2257, 0.0000, 0.0272, 0.3182, 0.0625],\n",
       "           [0.0000, 0.0000, 0.0000, 0.1808, 0.0000, 0.0000, 0.0000, 0.7271,\n",
       "            0.0000, 0.2825, 0.0927, 0.0000, 0.0000, 0.0000, 0.0401, 0.0000,\n",
       "            0.0000, 0.2073, 0.0000, 0.0000, 0.2296, 0.3315, 0.0407, 0.0000,\n",
       "            0.0348, 0.4161, 0.0000, 0.0000, 0.0113, 0.0000],\n",
       "           [0.1240, 0.0000, 0.2055, 0.0170, 0.0000, 0.0000, 0.0000, 0.0000,\n",
       "            0.1324, 0.0000, 0.2051, 0.0000, 0.0000, 0.0443, 0.0000, 0.3472,\n",
       "            0.0000, 0.0410, 0.0000, 0.0000, 0.0495, 0.0000, 0.0000, 0.1231,\n",
       "            0.0000, 0.0000, 0.1329, 0.3721, 0.2257, 0.0000],\n",
       "           [0.0000, 0.1575, 0.0000, 0.0000, 0.0483, 0.0000, 0.2776, 0.2185,\n",
       "            0.0000, 0.0000, 0.0155, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n",
       "            0.0769, 0.4499, 0.3997, 0.0000, 0.0000, 0.0000, 0.0000, 0.0992,\n",
       "            0.0000, 0.0000, 0.3711, 0.0000, 0.1406, 0.0488],\n",
       "           [0.1686, 0.0000, 0.0000, 0.5925, 0.0849, 0.1313, 0.0416, 0.0000,\n",
       "            0.7244, 0.1893, 0.1070, 0.0000, 0.1485, 0.0000, 0.2356, 0.0000,\n",
       "            0.1278, 0.1311, 0.0000, 0.2496, 0.1596, 0.0000, 0.1801, 0.0000,\n",
       "            0.0000, 0.0000, 0.0000, 0.0000, 0.3381, 0.0000],\n",
       "           [0.0000, 0.4516, 0.0000, 0.4718, 0.3553, 0.0000, 0.0000, 0.0130,\n",
       "            0.0000, 0.0000, 0.2493, 0.0158, 0.0000, 0.0000, 0.0000, 0.0000,\n",
       "            0.0000, 0.1234, 0.5844, 0.0000, 0.2639, 0.0000, 0.4391, 0.0000,\n",
       "            0.0000, 0.0000, 0.0000, 0.0000, 0.0742, 0.0000],\n",
       "           [0.0000, 0.0000, 0.5569, 0.0000, 0.0000, 0.0342, 0.0344, 0.0000,\n",
       "            0.2470, 0.0000, 0.0475, 0.0000, 0.0000, 0.0000, 0.3215, 0.0000,\n",
       "            0.0613, 0.2041, 0.0000, 0.0753, 0.0000, 0.2267, 0.1462, 0.3660,\n",
       "            0.0000, 0.2279, 0.1495, 0.0000, 0.0000, 0.0000],\n",
       "           [0.0000, 0.0000, 0.0943, 0.0000, 0.8249, 0.0000, 0.0606, 0.0000,\n",
       "            0.1461, 0.3490, 0.0615, 0.0000, 0.0000, 0.0000, 0.1209, 0.0000,\n",
       "            0.0804, 0.0807, 0.0673, 0.0000, 0.2344, 0.0000, 0.1983, 0.0000,\n",
       "            0.2170, 0.0355, 0.0000, 0.0223, 0.0000, 0.0898],\n",
       "           [0.1629, 0.1108, 0.0000, 0.0000, 0.0731, 0.0000, 0.0000, 0.2853,\n",
       "            0.2521, 0.1357, 0.0000, 0.1114, 0.4758, 0.0000, 0.3664, 0.2652,\n",
       "            0.0000, 0.3186, 0.0000, 0.2143, 0.0000, 0.0000, 0.4188, 0.0000,\n",
       "            0.0000, 0.0000, 0.3508, 0.5329, 0.6666, 0.0000]]]], device='cuda:0',\n",
       "        grad_fn=<ViewBackward0>),\n",
       " torch.Size([1, 1, 30, 30]))"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "execution_count": 6
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "### Setup Loss, optimizer and evaluation metrics",
   "id": "eee172db70a38518"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-04T12:21:23.317161Z",
     "start_time": "2024-08-04T12:21:22.914383Z"
    }
   },
   "cell_type": "code",
   "source": [
    "loss_fn = torch.nn.MSELoss()\n",
    "optimizer = torch.optim.SGD(params=model.parameters(), lr=0.01)"
   ],
   "id": "1ce92989eac86aa2",
   "outputs": [],
   "execution_count": 7
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "### Run Training Loop",
   "id": "518300e154e22dd8"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-04T12:25:00.719963Z",
     "start_time": "2024-08-04T12:21:23.318425Z"
    }
   },
   "cell_type": "code",
   "source": [
    "from timeit import default_timer as timer \n",
    "\n",
    "epochs = 50\n",
    "\n",
    "time_start = timer()\n",
    "for epoch in range(epochs):\n",
    "    ### Training\n",
    "    train_loss = 0\n",
    "    train_accuracy = 0\n",
    "\n",
    "    model.train() # Set model into training mode\n",
    "    \n",
    "    for b, (X_train, y_train, challenge_ids) in enumerate(train_dataloader):\n",
    "        for challenge_id in challenge_ids:\n",
    "            X_shots, y_shots = train_dataset.get_samples(challenge_id)\n",
    "            X_shots, y_shots = X_shots.to(device) / 10, y_shots.to(device) / 10\n",
    "            \n",
    "            # Forward pass:\n",
    "            shots_pred = model(X_shots)\n",
    "            \n",
    "            # Calculate loss:\n",
    "            loss = loss_fn(shots_pred, y_shots)\n",
    "            \n",
    "            # Reset optimizer gradients:\n",
    "            optimizer.zero_grad()\n",
    "            \n",
    "            # Backward pass:\n",
    "            loss.backward()\n",
    "            \n",
    "            # Adjust parameters:\n",
    "            optimizer.step()\n",
    "        \n",
    "        # Send to device and normalize:\n",
    "        X_train, y_train = X_train.to(device) / 10, y_train.to(device) / 10\n",
    "        \n",
    "        # Forward pass:\n",
    "        train_pred = model(X_train)\n",
    "        \n",
    "        # Calculate loss:\n",
    "        loss = loss_fn(train_pred, y_train)\n",
    "        train_loss += loss\n",
    "        train_accuracy += accuracy_fn(y_train, norm_arc20204(train_pred))\n",
    "        \n",
    "        # Reset optimizer gradients:\n",
    "        optimizer.zero_grad()\n",
    "        \n",
    "        # Backward pass:\n",
    "        loss.backward()\n",
    "        \n",
    "        # Adjust parameters:\n",
    "        optimizer.step()\n",
    "\n",
    "    train_loss /= len(train_dataloader.dataset)\n",
    "    train_accuracy /= len(train_dataloader.dataset)\n",
    "    \n",
    "    ### Testing\n",
    "    test_loss = 0\n",
    "    test_accuracy = 0\n",
    "    model.eval() # set model into evaluation mode\n",
    "\n",
    "    with torch.inference_mode():\n",
    "        for X_test, y_test, _ in test_dataloader:\n",
    "            # Send to device and normalize:\n",
    "            X_test, y_test = X_test.to(device) / 10, y_test.to(device) / 10\n",
    "\n",
    "            # Forward pass:\n",
    "            test_pred = model(X_test)\n",
    "\n",
    "            # Calculate loss:\n",
    "            loss = loss_fn(test_pred, y_test)\n",
    "            test_loss += loss\n",
    "            test_accuracy += accuracy_fn(y_test, norm_arc20204(test_pred))\n",
    "    \n",
    "        test_loss /= len(test_dataloader.dataset)\n",
    "        test_accuracy /= len(test_dataloader.dataset)\n",
    "    \n",
    "    print(f\"Epoch {epoch + 1}/{epochs} | Train loss: {train_loss:.5f} | Train accuracy: {train_accuracy:.2f}% | Test loss: {test_loss:.5f} | Test accuracy: {test_accuracy:.2f}%\")\n",
    "time_end = timer()\n",
    "\n",
    "print(f\"Total time taken: {time_end - time_start}\")"
   ],
   "id": "96d6551b2f990249",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/50 | Train loss: 0.01075 | Train accuracy: 0.09% | Test loss: 0.01042 | Test accuracy: 0.23%\n",
      "Epoch 2/50 | Train loss: 0.00984 | Train accuracy: 0.07% | Test loss: 0.01018 | Test accuracy: 0.19%\n",
      "Epoch 3/50 | Train loss: 0.00954 | Train accuracy: 0.06% | Test loss: 0.00977 | Test accuracy: 0.20%\n",
      "Epoch 4/50 | Train loss: 0.00924 | Train accuracy: 0.06% | Test loss: 0.00945 | Test accuracy: 0.21%\n",
      "Epoch 5/50 | Train loss: 0.00907 | Train accuracy: 0.06% | Test loss: 0.00947 | Test accuracy: 0.19%\n",
      "Epoch 6/50 | Train loss: 0.00891 | Train accuracy: 0.06% | Test loss: 0.00924 | Test accuracy: 0.20%\n",
      "Epoch 7/50 | Train loss: 0.00867 | Train accuracy: 0.06% | Test loss: 0.00900 | Test accuracy: 0.19%\n",
      "Epoch 8/50 | Train loss: 0.00849 | Train accuracy: 0.07% | Test loss: 0.00888 | Test accuracy: 0.19%\n",
      "Epoch 9/50 | Train loss: 0.00827 | Train accuracy: 0.07% | Test loss: 0.00868 | Test accuracy: 0.20%\n",
      "Epoch 10/50 | Train loss: 0.00818 | Train accuracy: 0.07% | Test loss: 0.00872 | Test accuracy: 0.18%\n",
      "Epoch 11/50 | Train loss: 0.00812 | Train accuracy: 0.07% | Test loss: 0.00843 | Test accuracy: 0.21%\n",
      "Epoch 12/50 | Train loss: 0.00803 | Train accuracy: 0.07% | Test loss: 0.00843 | Test accuracy: 0.19%\n",
      "Epoch 13/50 | Train loss: 0.00791 | Train accuracy: 0.07% | Test loss: 0.00819 | Test accuracy: 0.25%\n",
      "Epoch 14/50 | Train loss: 0.00774 | Train accuracy: 0.07% | Test loss: 0.00829 | Test accuracy: 0.18%\n",
      "Epoch 15/50 | Train loss: 0.00766 | Train accuracy: 0.07% | Test loss: 0.00826 | Test accuracy: 0.18%\n",
      "Epoch 16/50 | Train loss: 0.00760 | Train accuracy: 0.07% | Test loss: 0.00817 | Test accuracy: 0.19%\n",
      "Epoch 17/50 | Train loss: 0.00754 | Train accuracy: 0.07% | Test loss: 0.00806 | Test accuracy: 0.18%\n",
      "Epoch 18/50 | Train loss: 0.00746 | Train accuracy: 0.07% | Test loss: 0.00802 | Test accuracy: 0.19%\n",
      "Epoch 19/50 | Train loss: 0.00743 | Train accuracy: 0.07% | Test loss: 0.00801 | Test accuracy: 0.19%\n",
      "Epoch 20/50 | Train loss: 0.00741 | Train accuracy: 0.07% | Test loss: 0.00802 | Test accuracy: 0.18%\n",
      "Epoch 21/50 | Train loss: 0.00739 | Train accuracy: 0.07% | Test loss: 0.00793 | Test accuracy: 0.18%\n",
      "Epoch 22/50 | Train loss: 0.00735 | Train accuracy: 0.07% | Test loss: 0.00798 | Test accuracy: 0.18%\n",
      "Epoch 23/50 | Train loss: 0.00727 | Train accuracy: 0.07% | Test loss: 0.00785 | Test accuracy: 0.17%\n",
      "Epoch 24/50 | Train loss: 0.00725 | Train accuracy: 0.07% | Test loss: 0.00790 | Test accuracy: 0.17%\n",
      "Epoch 25/50 | Train loss: 0.00722 | Train accuracy: 0.07% | Test loss: 0.00802 | Test accuracy: 0.18%\n",
      "Epoch 26/50 | Train loss: 0.00719 | Train accuracy: 0.07% | Test loss: 0.00774 | Test accuracy: 0.18%\n",
      "Epoch 27/50 | Train loss: 0.00715 | Train accuracy: 0.07% | Test loss: 0.00778 | Test accuracy: 0.18%\n",
      "Epoch 28/50 | Train loss: 0.00715 | Train accuracy: 0.07% | Test loss: 0.00777 | Test accuracy: 0.18%\n",
      "Epoch 29/50 | Train loss: 0.00714 | Train accuracy: 0.07% | Test loss: 0.00784 | Test accuracy: 0.17%\n",
      "Epoch 30/50 | Train loss: 0.00714 | Train accuracy: 0.07% | Test loss: 0.00759 | Test accuracy: 0.23%\n",
      "Epoch 31/50 | Train loss: 0.00714 | Train accuracy: 0.07% | Test loss: 0.00783 | Test accuracy: 0.18%\n",
      "Epoch 32/50 | Train loss: 0.00707 | Train accuracy: 0.07% | Test loss: 0.00756 | Test accuracy: 0.19%\n",
      "Epoch 33/50 | Train loss: 0.00690 | Train accuracy: 0.07% | Test loss: 0.00761 | Test accuracy: 0.17%\n",
      "Epoch 34/50 | Train loss: 0.00684 | Train accuracy: 0.07% | Test loss: 0.00751 | Test accuracy: 0.18%\n",
      "Epoch 35/50 | Train loss: 0.00678 | Train accuracy: 0.07% | Test loss: 0.00741 | Test accuracy: 0.18%\n",
      "Epoch 36/50 | Train loss: 0.00676 | Train accuracy: 0.07% | Test loss: 0.00751 | Test accuracy: 0.17%\n",
      "Epoch 37/50 | Train loss: 0.00675 | Train accuracy: 0.07% | Test loss: 0.00746 | Test accuracy: 0.17%\n",
      "Epoch 38/50 | Train loss: 0.00673 | Train accuracy: 0.07% | Test loss: 0.00751 | Test accuracy: 0.17%\n",
      "Epoch 39/50 | Train loss: 0.00673 | Train accuracy: 0.07% | Test loss: 0.00740 | Test accuracy: 0.18%\n",
      "Epoch 40/50 | Train loss: 0.00670 | Train accuracy: 0.07% | Test loss: 0.00739 | Test accuracy: 0.16%\n",
      "Epoch 41/50 | Train loss: 0.00666 | Train accuracy: 0.07% | Test loss: 0.00728 | Test accuracy: 0.17%\n",
      "Epoch 42/50 | Train loss: 0.00660 | Train accuracy: 0.07% | Test loss: 0.00729 | Test accuracy: 0.17%\n",
      "Epoch 43/50 | Train loss: 0.00656 | Train accuracy: 0.07% | Test loss: 0.00732 | Test accuracy: 0.16%\n",
      "Epoch 44/50 | Train loss: 0.00650 | Train accuracy: 0.07% | Test loss: 0.00729 | Test accuracy: 0.16%\n",
      "Epoch 45/50 | Train loss: 0.00648 | Train accuracy: 0.07% | Test loss: 0.00720 | Test accuracy: 0.17%\n",
      "Epoch 46/50 | Train loss: 0.00642 | Train accuracy: 0.07% | Test loss: 0.00722 | Test accuracy: 0.17%\n",
      "Epoch 47/50 | Train loss: 0.00635 | Train accuracy: 0.07% | Test loss: 0.00708 | Test accuracy: 0.17%\n",
      "Epoch 48/50 | Train loss: 0.00629 | Train accuracy: 0.07% | Test loss: 0.00707 | Test accuracy: 0.17%\n",
      "Epoch 49/50 | Train loss: 0.00627 | Train accuracy: 0.07% | Test loss: 0.00702 | Test accuracy: 0.16%\n",
      "Epoch 50/50 | Train loss: 0.00626 | Train accuracy: 0.07% | Test loss: 0.00711 | Test accuracy: 0.17%\n",
      "Total time taken: 217.3964193310003\n"
     ]
    }
   ],
   "execution_count": 8
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "# Evaluate Model",
   "id": "12e02627456a3a8b"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-08-04T12:25:57.746166Z",
     "start_time": "2024-08-04T12:25:57.185885Z"
    }
   },
   "cell_type": "code",
   "source": [
    "evaluations = 3\n",
    "\n",
    "evaluation_dataset = FewShotDataset(\n",
    "    dataset_dir=dataset_dir,\n",
    "    dataset_name='evaluation'\n",
    ")\n",
    "\n",
    "model.eval() # set model into evaluation mode\n",
    "with torch.inference_mode():\n",
    "    \n",
    "    for _ in range(evaluations):\n",
    "        X, y, _ = random.choice(evaluation_dataset)\n",
    "        X, y = X.to(device) / 10, y.to(device) / 10\n",
    "        \n",
    "        y_pred = model(X)[0]\n",
    "        y_pred_norm = norm_arc20204(y_pred)\n",
    "        \n",
    "        plot.input_and_output(X * 10, y * 10, y_pred_norm * 10)\n",
    "        accuracy_pred = accuracy_fn(y, y_pred_norm)\n",
    "        print(f\"Accuracy: {accuracy_pred:.2f}% ({int(accuracy_pred / 100 * 900)}/900)\")\n"
   ],
   "id": "14b6af5afe472c9e",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 1920x1440 with 3 Axes>"
      ],
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 11.33% (102/900)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 1920x1440 with 3 Axes>"
      ],
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABgwAAAIwCAYAAABa50BbAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjkuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/TGe4hAAAACXBIWXMAAC4jAAAuIwF4pT92AABQu0lEQVR4nO39eZRU1b0/7r8LWmZkVgaZURRQAecRUJxwDMTEaBziJTdxTrwmNxojcBM1+WiMV02MX4noNRqNxoizMYCzKOA8AKIMiqjMgsz0+f3hgh/dVd1dXVXdRcPzrNVrwanaZ++qc+qcXfU6e59UkiRJAAAAAAAA27V6xW4AAAAAAABQfAIDAAAAAABAYAAAAAAAAAgMAAAAAACAEBgAAAAAAAAhMAAAAAAAAEJgAAAAAAAAhMAAAAAAAAAIgQEAAAAAABACAwAAAAAAIAQGAAAAAABACAwAAAAAAIAQGAAAAAAAACEwAAAAAAAAQmAAAAAAAACEwAAAAAAAAAiBAQAAAAAAEAIDAAAAAAAgBAYAAAAAAEAIDAAAAAAAgBAYAAAAAAAAITAAAAAAAABCYAAAAAAAAITAAAAAAAAACIEBAAAAAAAQAgMAAAAAACAEBgAAAAAAQAgMAAAAAACAEBgAAAAAAAAhMAAAAACgCAYPHhypVGrz3+DBg7Mqt2WZVCoVo0ePrtF2Uhi2G9QNJcVuAADAtixJkpg5c2bMmjUrPvnkk1ixYkWsXbs2mjdvHq1atYrWrVtHv379olu3bsVuKtSqDz/8MGbMmLH5c7Fu3bpo2rRptGnTJrp37x5777137LjjjsVuJgCwnfj888/j3XffjTlz5sSyZcti9erV0bhx42jRokV069Yt+vbtGx07dix2M6HGCQzY7syZMye6d+9eZtmgQYPi2WefLU6DANjmfP311/Hggw/GQw89FM8991wsX768yjJt27aNAw44IEaMGBEjRozwQynbnCRJ4umnn4577rknnn766Vi4cGGlz69Xr14MGDAgRowYEWeddVZ06tSplloKbA8yfS+sTKNGjaJFixbRokWL2G233WKfffaJww47LIYMGRL16pm8Aeqqt956K/7v//4vHnnkkZg1a1aVz+/Ro0ecdNJJcdZZZ8WAAQNqoYVQ+5zVgK1ersNUAWrbypUr41e/+lV07tw5zjnnnHjkkUeyCgsiIhYtWhSPP/54nHvuudG+ffs466yz4uOPP67hFlPX3XnnnWnD++fMmVPsZqV55JFHYs8994zjjjsu/vrXv1YZFkRElJaWxrRp0+KKK66IHj16xI9//ONYtGhRLbS2ZnTr1q3MdjrnnHOK3aQ6S9+QYlizZk188cUXMXPmzHjsscdizJgxMXTo0OjZs2f8/ve/jw0bNhS7iWxlzjnnnDLHKqNJty7vvvtuHHfccdG/f/+44YYbsgoLIiI+/vjjuPHGG2PgwIFx7LHHxjvvvFPDLa059tHCGT16dFqfvC4TGAAAFMAzzzwTvXv3jt/85jexdOnSvNa1evXquPvuu2P33XePiy++ONasWVOgVkLt+uqrr+LUU0+Nk08+Od57772c17Nu3bq47bbbYvfdd4/x48cXsIUA+ZkzZ05cdtllceCBB8aHH35Y7OYAVUiSJH7961/HgAED4qmnnsprXU8//XQMHDgw/ud//ieSJClQC6H4TEkEAJCn3/72t3HFFVdU+EWhVatWMXTo0Nhjjz2iXbt20a5du0iSJJYtWxazZ8+OqVOnxiuvvBKrV68uU279+vVx8803x6WXXuqKH+qcefPmxdFHHx0zZszI+HiLFi1i2LBhsccee0SHDh2iefPm8fnnn8dnn30WkyZNiilTpqSVWbx4cXzrW9+Ka665Jn7xi1/U9EsAtjNNmzaNXr16ZXxs1apVsWjRogovCpg2bVoMHTo0Xnrppdhll11qsplAjtauXRunnXZaPPzwwxkf32GHHWLIkCGx//77R4cOHaJt27axePHiWLBgQUybNi3+/e9/x7p168qU2bBhQ4waNSreeOONuO+++6Jhw4a18EqgZgkMAADycOWVV8bVV1+d8bFjjz02fvnLX8ZBBx0U9evXr3Q9q1atikcffTRuvvnmeOmll2qiqVBr5s+fH4MGDco4PVL37t3juuuuixNPPDEaNGhQ4TrmzZsX119/ffzxj3+M0tLSzcuTJInLL788SktL44orrqiJ5gPbqX333bfKe9t99NFHcd9998WNN96YNk3avHnz4tRTT41XXnmlBltJRLiam2rbsGFDfOtb34onn3wy7bFmzZrF6NGj49xzz41WrVpVuI7ly5fHXXfdFb/61a/iq6++KvPYww8/HMOHD4/x48dHSYmfW6nbTEkEAJCjsWPHZgwLOnToEBMnTownn3wyDj300CrDgoiIJk2axHe/+9148cUXY8KECdGvX7+aaDLUuDVr1sQpp5ySMSy46KKL4r333osRI0ZUGhZERHTp0iVuuummmDx5csYRNldeeWX84x//KFCrAbLTs2fP+OUvfxnvvPNO7L///mmPT548OR544IEitAyozE9/+tOMYcHgwYPjgw8+iP/6r/+qNCyI+GZ05MUXXxzTp0+PoUOHpj3+xBNPxKWXXlqwNkOxCAwAAHLwwQcfxEUXXZS2vHfv3vHyyy/HkCFDcl73EUccEdOmTYvLLrssnyZCUfzqV7+KqVOnpi2//PLL46abborGjRtXa3377bdfTJo0Kbp06VJmeZIkMXLkyFiwYEFe7QXIRfv27eOxxx6L9u3bpz122223FaFFQEUef/zxuOWWW9KWH3nkkfHEE09UexqxDh06xKOPPhpHHXVU2mM333xzxmAC6hKBAQBADn70ox+l3Yy4devWMXHixILcb6BBgwZx3XXXxb333lvlldiwtXj33XfjD3/4Q9rys88+O6655pqc19utW7d45pln0sKGZcuWuZIPKJp27drFz3/+87TlL774YqxataoILQLKW7t2bZx//vlpy/fYY4945JFHqn0hwyaNGjWK8ePHR9++fdMeO++889LudQB1iUm1oMA++eSTmDJlSsydOzdWrVoVrVu3jp133jkOOeSQ2HnnnWu07jVr1sSrr74a06dPj6VLl8YOO+wQHTt2jN122y0GDhwYqVSqRusH2F488cQT8cILL6Qtv/XWW6Njx44Fret73/tezmVXrVoVr776anz22WexcOHCWLNmTbRr1y522mmnGDBgQK3dlPHDDz+Mt956Kz799NNYuXJlNGjQINq3bx9nnXVWtde1cePGePPNN+P999+PL774ItasWRNNmzaNvfbaK4488sis1zNv3rx4++23Y+HChbFw4cKoX79+tG3bNjp27BgHHXRQNGvWrNpty9aHH34YH3zwQSxatCgWLVoUpaWl0bx58+jYsWPsvvvusdtuu2U1jdXW6KqrroqNGzeWWbbzzjtnDBGqa7fddotRo0al3ez4/vvvj8svvzz22muvvOvY1ugbQs0bPnx4WnC5du3aeO+992K//fbLaZ11+Vy3ePHimDx5csyfPz8WLlwYjRs3ji5dusQ+++wT3bt3L3h9hbTpfZ8zZ04sWrQolixZEiUlJbHjjjtGt27dYo899kgb7ba1SJIk3nnnnfjoo49i4cKFsXjx4mjatGm0a9cuunXrFvvtt1+NzKv/0UcfxVtvvRXz58+Pr776Knbcccfo2bNnHHDAAdGmTZuC15eL2267LebNm1dmWSqVirFjx0aTJk3yWnfjxo3j9ttvj0MPPbTM/Zbmzp0b/9//9//FhRdemNf6t0VJksSbb74Zb7/9dnz55ZexcePG6NChQ3Tp0iUOPvhgN43eWiSwnZk9e3YSEWX+Bg0aVGW5rl27lilz9tlnl3n8oYceSg488MC0dW/6S6VSyf7775888cQT1W7zpEmT0tY3adKkzY9/9NFHyQ9+8IOkSZMmFdbfpUuX5Je//GWycuXKatc/aNCgar9fmYwbNy6tXbNnz8743IpeR3X+Klo3QL6GDh2adsw54ogjit2szf7+978nRx11VNKwYcNKj5N9+/ZNRo0alSxfvjynesqvb9SoUZsfW7lyZXLttdcmPXr0qLD+LVV1rps7d25ywQUXJK1atcq4rmzOTZ9++mnyX//1X8nuu+9e6fuyww47JIcddlhy3333JaWlpTm9N+W9+eabybnnnpt07ty5yvNXy5Ytk+HDhyf33Xdfsnbt2rR1nX322XmfI7fcVoUyZ86cpF69eml1jRs3rmB1rF+/Ptl1113T6viP//iPKstW1ZfL1qhRoyrdlzfJ1OfM5S8TfUN9Q/KX6/fCTJo2bZq2rscffzztedvyue7ZZ59Njj322KSkpKTCOvfcc8/kzjvvLFNfrseTQp7Xxo8fn5x88slJixYtqjyOdOnSJRk5cmTy/PPPZ1xX+XNNLn9b7hNVmTJlSnLmmWcmO++8c6XrbN68eTJ8+PDk1Vdfzfl92qS0tDQZO3ZsMnDgwArrq1+/fnLMMcekvU+F3G7ZtrVnz55p9ebaB6jIOeeck1bHrrvuWuVnq3yfrmvXrjnVX9WxZUs1uY9Wtn2/+uqrZNSoUUn79u0rXG+LFi2Ss88+O6fzdbb9s6pkOjdU1Jctf/zK5a+Q/eRCMiUR5Gn58uVx0kknxfDhw2Py5MkVPi9Jknjttddi2LBhcc4558SGDRsKUv+4ceOiX79+MW7cuEqHvc6bNy+uvvrq6NOnT0yYMKEgdQNsjz755JOMx9ELLrigCK0p64033oj9998/vvOd78QzzzwTa9eurfT57733XowZMyZ69uxZ0PmWX3311ejTp09cfvnl8fHHH+e9vr/85S+xxx57xB//+MdYunRptcuvXr06fvazn0WvXr3i97//fUyfPr3S569fvz5eeOGFOO2006J///7x7rvv5tr0mD17dnzrW9+KAQMGxB133BGffPJJlWWWLVsWDz30UJx22mmx77775lx3bbvrrrvKXF0XEdGqVas47bTTClZHSUlJjBw5Mm35fffdlzZF2PZK3xBqX4sWLdKWLVu2rFrrqKvnutWrV8fIkSNjyJAh8dRTT1X6Pfedd96Jc845J4YMGRKLFi3Kqb5CevbZZ2PgwIFx8sknx/jx42P58uVVlpk3b16MHTs2Dj/88Lj44otroZWZzZ07N0aMGBH77bdf3H333fHFF19U+vwVK1bEQw89FAcccECMGDGi2vvnJrNnz47DDz88Ro4cGa+//nqFz9u4cWM8/fTTm9+n8qMPa8sLL7wQH330Udry8847r6D1ZFrfhx9+GC+++GJB66mrpkyZEn379o0xY8bE559/XuHzli9fHnfddVf07ds3/vd//7cWW0h5AgPIw+LFi+PQQw+NRx99tFrl7rrrrjjnnHPyrv/WW2+Nc889N1avXp11mXnz5sWwYcPi8ccfz7t+gO3Ro48+GkmSlFnWvn37OOmkk4rUom889dRTcdhhh8WUKVOqXXbRokXx4x//OC655JK0H3yr6/nnn4/BgwenDf3O1XXXXRcjR47MeS7ozz//PAYPHhzXX399Tj8ov/3223HwwQfHE088Ue2ykyZNiv322y8efvjhtH0mW1999VVO5YohU3/ojDPOiEaNGhW0nrPPPjttWoWvv/46Jk2aVNB66iJ9QyiOTD80t2zZMuvydfVct3r16jjxxBPjL3/5S7XOc88991wcfvjhOQUjhXLjjTfG0KFD44033sh5HcU6R0+ePDn233//eOihh3Iq/9BDD8WBBx4Ys2bNqla5jz/+OA4//PBq/wh+8803x5lnnplzXygfmfom/fr1iwMOOKCg9ey///6x5557pi1/7LHHClpPXTR16tQYMmRIVhfNbLJq1ar4yU9+EldeeWUNtozKuIcB5GjDhg1xyimnlLkSY8CAAXHMMcdEjx49omXLlrFkyZJ49dVX46GHHkrrRN5zzz1xyimnxLe//e2c6p88eXKZg2dJSUkcccQRMXTo0OjUqVOsXbs25s6dG4888khaJ2jdunUxYsSIePbZZ+PAAw/Mqf6atvfee2/+96xZs+Lrr7/e/P+mTZtGr169qlyHm4QCNSHTj5KHH354jcwLm61JkybFiSeemPGqvv79+8eJJ54Y3bp1i8aNG8eCBQviueeei6effjptBMJNN90UGzdujFtuuSWndnz++ecxfPjwMj9W7L///nH00UdH165do3nz5rFgwYJ4//3344EHHqhyff/617/it7/97eb/N2zYMIYMGRKDBw+O9u3bR0lJSXz66afx6quvxsqVK9PKf/HFF3HggQfG3Llz0x7r169fDBo0KPr27bv5R50vv/wyXnnllXjiiSdixYoVm5+7YsWKGDFiRLz88ssxYMCArN6Lxx9/PE455ZSM26Rt27YxdOjQ2HfffaNdu3bRqFGjWLZsWcybNy+mTp0aL7/8cpn6y+vSpcvm8+SSJUvSvoDtscceVZ4D27dvn9XryNaKFSti2rRpacuPO+64gtYT8c09EQYMGJAWjk2aNKlG6stVgwYNyvRn3n///Vi/fv3m/7dq1aqgc2HrG+obUhxz584tsz9u0q5du6zK1+Vz3WmnnZZxhFKbNm3iW9/6Vuy9996x0047xeLFi+O9996Lhx56KBYsWBARER988EFO9zIqhCuuuCKuvfbajI/16tUrjj766Ojdu3e0a9cukiSJpUuXxsyZM2Pq1Knx2muvVTqKok+fPpvf63nz5pUJRXbYYYfo06dPle2r7N4Szz77bBx33HFpwVC9evXisMMOi4MPPji6d+8eLVu2jNWrV8enn34azz33XEyYMKHMVf4zZsyIYcOGxdSpU2PHHXessk1Lly6NIUOGxKeffpr2WK9evWL48OHRq1ev2HHHHePzzz+Pl19+OR5//PHNn42//e1vsc8++1RZT6E9++yzactqqq9w7LHHxjvvvFNm2dZ4MUNN76NbWrZsWZxyyilljpEDBgyIk046Kbp27RoNGzaM+fPnx4QJE2LChAlpn62rr7462rRpEz/96U+zqq+29erVa/Nonc8//zxtpM+WfZeKtG7duiaalr9izocExVCoexg0atRo87932223ZMKECRWW/fLLL5Njjjkmrd7dd989qzZnmo9uy/oPOuigZPr06RWWf+KJJ5JOnTplrH/NmjVV1l+MeWpron6AQsg0J//1119ftPYsWrQo4zG+S5cuyZNPPllhuU8++SQ5/vjjM86l+fDDD2dVd/ly9evX3/zvvfbaK3n55ZcrLLt69eoy/890rttyfSNGjEjmzZuX9fo2btyYHHHEEWnrPPjgg5PJkydX+rqWLl2aXHrppUkqlSpTtnv37slXX31V5fvy4YcfJi1btkyre+edd05uvfXWZP369ZWWX7NmTTJ+/Phk2LBhSbdu3Sp9bq7n1kJ77rnnMu5LX375ZY3Ud95556XVdeSRR1ZaprbvYVBT9SeJvqG+IYVQqHsY3HDDDWnradCgQcb7g2xL57pMn99UKpVcdtllyapVqzKWWb9+ffLrX/862WGHHTaXady4cU7boHzd2c6F/8ADD2Q8X/Xv3z95+umnqyy/aNGi5Pbbb0/22muvKo/jhZqffpMFCxZkvFfBD37wg2Tu3LmVlp01a1bG3yS+/e1vZ1V3pvsntWjRIvnLX/5SYZmFCxcm3/ve9yrc1tXZbrlYt25dxvt5/f3vf6+R+u6///6M5+bK+n3FuIdBTdS/SWV9k06dOmW8t8sm06dPTw466KCM66isT7NJMe5hUBP1by1MSQQ52pTo77fffvHKK6/EEUccUeFz27VrF+PHj49+/fqVWT59+vSc57TbVP/gwYNj4sSJ0bt37wqfe9xxx8Xzzz8fHTt2TKv///2//5dT/QDbo3Xr1sXs2bPTlhdznvnLL7885s+fX2ZZ9+7d46WXXopjjz22wnK77LJLPProo/H9738/7bEf/vCH1ZrSZJNNV64dcsgh8cILL8RBBx1U4XOzmaZm0/ouvvjiePDBB6Nz585Zr+/666+PiRMnlll2wQUXxIsvvljlMPSWLVvG73//+/jLX/5SZvns2bPj1ltvrbLd3//+99PmBu7bt29MmTIlfvzjH1c5GqVhw4Zx0kknxeOPP75VXpmWyYwZM9KWdenSJesrbKsr02cuUxu2J/qGUPsWLVqU8TNz6KGHRtOmTbNaR1081y1fvjwuvfTStOU33XRTXHfdddG4ceOM5UpKSuLKK6+Me++9N+rXrx8RkVN/I1cLFy6Mc889N235d77znZg8eXIcffTRVa6jTZs2MXLkyHjrrbdizJgxNdHMCp177rllrmCuX79+3HPPPXHHHXdUOWKtZ8+e8eSTT8YPfvCDMssffPDBeO211yot+/zzz8ddd91VZlmzZs3iqaeeyvh+btK2bdu49957N9/roTa3dUTEnDlzMt7Pq6b67ZnWu2bNmoyjf7YXm/omHTt2jOeffz6GDRtW4XN79+4dEydOjMGDB6et4/zzz6/JZpKBwADy0LJly3jwwQezGkLUsGHDjJ3Jp59+Ouf627RpEw8++GBWP7r06NEj7rnnnrTlf/rTn8oMjwegYvPnz884/2qhp3fJ1qJFi+Luu+8us6x+/frx0EMPxS677FJl+VQqFePGjUubc3XhwoVp681WixYt4v77789qeHs29t9//7jhhhuqVWbVqlVx3XXXlVl2/PHHxy233BKpVCrr9fzgBz9Iu8HuH/7wh1i3bl2FZf71r3/Fq6++WmZZ27Zt45lnnqn0R6CKdOvWrdpliiHTFAWdOnWqsfoyrXvBggVFu6ni1kLfEGrPF198ESeddFLGG3j+8Ic/rNa66tq57v/+7//S7j9w5plnxoUXXphVnd/+9rfj5z//edZtLJQ//OEPaVP+HXbYYfG3v/0tGjZsWO31de3atVBNq9KUKVPiySefLLPs2muvjdNPPz3rdaRSqbjttttijz32KLN8yymxMsl089n//d//zXoKuxtvvDEOPvjgrNtZKJn6JhE11z+paL3Vmbt/W3XPPfdEjx49qnxeo0aN4sEHH4w2bdqUWT5x4sScb8xObgQGkIdLLrmkWnPPHn300WlX2mWa7zdbo0ePTjuQVmbw4MExYsSIMss+//zzGD9+fM5tANieVHRzvurc2LCQxo4dmzaH7Y9+9KPo379/1usoKSmJm2++OW15rvcxuPTSSwv6Rey6667bfBVitu64445YtGjR5v/Xq1cv42vMxlVXXVXmh5fPP/88XnnllQqf/7vf/S5t2c033xwdOnTIqf66YsmSJWnLWrRoUWP1ZVr3xo0b69RNomuCviHUvI8//jh++9vfxl577ZXxfLDffvvFd7/73Wqts66d6/785z+X+X+jRo3SwouqXHnllbV6wcXKlSvjj3/8Y5llDRo0iLvvvjvq1dv6fxor37/o1atXxlEeVdlhhx3iiiuuKLPsySefzHglfsQ3YfwjjzxSZtnAgQPTRipUJpVK5bxv5iNT36RRo0Y1dj+bhg0bZgyeinlz763BiBEj0kYNVKZNmzYxevTotOXljzvUrK3/qAhbsepeOVK/fv20G/3kOny+UaNGceaZZ1a73H/+53+mLSt/pQIAmVU0lLomfxitzDPPPJO2LJchu5tuiLild955J+3GXVVJpVKVDk2vrl133TUOP/zwapd78MEHy/z/iCOOiO7du+fUhs6dO6eNwMh0A72IiK+++iqee+65Msu6desWp556ak511yWZPhs1GaRVtO7anu5ga6JvCPmbOnVq9O/fP+Nf7969o02bNtGzZ8+4/PLL48svv0wr36lTp3jggQeqdYV/XTvXzZs3L95///0yy0466aTYeeedq1VnkyZNMk6LWFOee+65tFD5u9/9bq2OEsjVmjVr4rHHHiuz7Jxzzql2yLRJ+Wlh1qxZE5MnT8743GeeeSbtRrQjR46s1j4e8U3IUNtTeNZ236Si9W/PfZOI6v9uFvHNiKXyoyX1TWqXwABy1LNnz5yuoOzZs2eZ/y9fvjyn+gcPHpzTD1RDhw5Nmyaios4BAGVlmo4oIqr9pakQNm7cmDb1ze677572w3+2Mv2o/dJLL1VrHb169cpqKqRsVedqpE3Wrl2b9r4ccsghebWj/A8wb7zxRsbnvfDCC2lT4px++uk5f6GnYhV95ir6jG4P9A0hf19//XW89dZbGf9mzpyZ8YrlTfr37x8TJkyo9g/Qde1cl+n4MHz48JzqzLVcLjIFILmErMXw6quvpo0AyGd7t27dOu18sa1t72LJ1D/ZnvsmzZo1i6FDh1a7XIsWLeLII48ss+zjjz+OhQsXFqppVKHyO64BFdp1111zKlf+xJxrYFB+pEK26tWrF3vvvXe88MILm5d98MEHsWrVqmjSpElO6wTYXlR0E79ly5ZV+8q6fM2YMSO+/vrrMsvyuXJrv/32S1v2+uuvV+vL3cCBA3Ouv1DrmzZtWto0TXfccUc8/PDDObdj3rx5Zf6/5RQQW8o0fUO+P+DUFZk+G7n2cbJR/qbSm2zPfRl9QyiOLl26xAUXXBA//elPY4cddqh2+bp2rss0pW6ux5/+/ftH/fr1a+X+M+XP0fXq1ct6Dv5iy3QBx/nnn5/X1DqrVq0q8/9st3enTp1y7vPmup/kqrb7JhGZ+yfb87l07733zvnCmYEDB8bjjz9eZtm0adPi2GOPLUTTqILAAHKUzY2OMynfiSw/vC9bvXv3zqlcxDdXoG75pTBJkli0aFG17scAsD2q6NhfjMAg0xe78jexq44+ffpkVUdldtppp5zrL9T6Mt3g7pNPPinoDecWL16ccXmmKZzKT/GwrWrVqlXasop+1C+ETOuuX79+wW62XRfpG0LNatiwYey4447RsmXL2G233WKfffaJww8/PIYMGZLXHPh17VxXfiqmkpKStFH02WrcuHF06dIlZs+enVP56ih/ju7WrVs0b968xusthEzb+4MPPihoHdlu73zPNbUpU99k9erVsW7duhq5j8HatWvTgryK2rG9KPT+kmkqOGqGwABylMvVI4WUz3zZmcouW7bMl0KAKnTq1ClSqVTa0OIvvvgirw5xLjLdQC2feVkzfZmpbOqFTAr9Y20u66voC28hVTQXbab3a3v5kphpKqrPPvusxupbsGBB2rL27dtv19M/6RtC/gYNGlTh3P01pa6d68oHts2bN89rasbaug9U+XN0XTo/b03bu9DnmppU0TSZn332WXTr1q3g9WXqm1TWju1BTfRNqB3uYQB1VNOmTQtadsWKFfk0B2C70KBBg4xfMKZOnVrrbcl03C72uaGkpLDXouSyvkxBSm0pfzPFiPy2SV2SKTCbO3dutUepZCvTZ662Q7utTbE//0Bu6tq5rvyxId/zXG2dJ8ufo5s1a1Yr9RbCtrK9a7tP1K1bt4wjCWqq355pvQ0bNqwTN9auKfomdZfAAOqo8vNW51u2rgzHBCi2TPOvvvbaa7XejkzHbeeGzPPV3nrrrZEkScH+5syZk7HuTFeJ5rNN6pL+/ftnvMJ0ypQpNVJfpvUW+h4adY3PP2w/inmuK39syPc8V1vnyfLn6JUrV9ZKvYWQaXt/8MEHBd3ed955Z8a6C7m9a7tP1KBBg+jXr1/a8trsm+y5555Fn52imPRN6i6BAdRR+dysJ1PZfKaxyNb69etrvA6AmjZ48OC0Zc8//3yt3LBvS4WeMz5T2Vzv11NMbdu2TVtW3amVctWmTZu0ZcW8KrA2tWjRIgYMGJC2/Kmnnip4XQsXLow33ngjbfmQIUMKXlcmW2t/Rt8Qth/FPNeVPzasWLEibarG6qjpm9BuUv4cXZfOz1vT9i70uaamZeq310TfJCLi6aefTlumb6JvUlcJDKCOmjlzZs5lZ8yYUeb/qVQqYydkk/KJeK4H0NqYexGgpp144olpV1IvWLAgHnnkkVptR7t27dKW5XMDvPfffz9tWWXnhq1VpptPz507t1bqbt++fdqyt99+u1bq3hqceOKJacv++te/xtq1awtaz1133ZXWF2natGmVX8q39f6MviFsP4p5rit/k+YNGzbERx99lNO6Vq9eHfPmzStEs6pU/hw9Z86cOjO9yda0vcufL6pj+vTp+Tan2jL1Td5+++2CjzKYOnVqvPXWW2nLTzjhhErLbevn00L2TSIqv0l8ppEcubyfW+t7WdsEBlBHTZs2LadypaWl8eabb5ZZtscee0STJk0qLFN++GamOZqzMWvWrJzKAWxNunTpEkceeWTa8j/96U+12o7ddtstbf7dfOZkzfTFKdP0S1u7fffdN+rVK9vFff7552ul7oMOOiht2UsvvVQrdW8NzjrrrLQwbcmSJXH//fcXrI6NGzfG2LFj05Z/97vfzThlw5a29f6MviFsP4p5rsvUN8j1+PPmm2/W2gjN8ufo0tLSeOWVV2ql7nwdcMABacuKtb3nz58fX3zxRU7rynU/ycegQYOie/fuactvvfXWgtbz5z//OW1Zr1694rDDDqu03LZ+Ps3nM55pf6nsu0mmqUFzeT+31veytgkMoI6aNGlSTsO7/v3vf6cdNA888MBKy5Qf9jVnzpychp0+99xz1S4TkX4jsNqe9gOgvJ/85Cdpy/7973/HP/7xj1prQ/369WP//fcvs2z69Onx3nvv5bS+Bx54IG3ZwQcfnNO6iql169ZpXyamT5+ecQRFoR166KFp56y//e1vNXreynSzzGKdJ3v06JHxSrpf/OIXBZuG4A9/+EPGq+EvvvjiKsuW7898/PHH1a5/3bp1MXny5GqXi6j5/oy+IWw/inmuy3R8+Oc//5nTuh566KF8m5O1TFPT3H333TVaZ6GOVUOGDElb12OPPVYrU6fU1e29SSqVigsvvDBt+V133ZXz+by81157LcaNG5e2/KKLLsp4f6ctlT+frly5MhYtWlTtNmyt59OVK1fGhAkTql3uq6++SivXo0ePjCOsN8k0XVEufb1CvZcRdbt/IjCAOmrNmjXx17/+tdrlbr/99rRlxx13XKVlevfuXeb/K1eurHZndNKkSTknteVvbFOXblAFbJuOP/74OPTQQ9OW/+hHP4oFCxYUtK6//e1v8dlnn2V87JhjjklblukKp6q88MIL8e6775ZZttdee2UcAl8XnHzyyWnLfvvb39Z4vc2bN08bfTJnzpz4+9//XqN1llfM8+Svf/3rtKteFyxYEJdeemne6541a1aMGjUqbfl3vvOd2HvvvassX74/M3PmzGr/wH7PPffkfAO/mu7P6BvC9qVY57ouXbpEnz59yiwbP358fPnll9Vaz+rVq3M6ZuVq0KBBafd/+vvf/16jU/sU6li14447pgUen376aY0HHhERRx11VNoPoWPHjq12SPzmm2/mNRI2H+eff3506tSpzLLS0tIYOXJkrFmzJq91r1mzJkaOHBmlpaVllnfp0iV+9KMfVVm+/Pk0IuLVV1+tVhtmzZoVEydOrFaZTWrjfJqpn1GVu+++O23bVLdvEvFNmFMdy5cvz3lk7NbWJ8+XwADqsNGjR1frZk3PP/98PPjgg2WWtW/fPmNnc0uZhn3de++9Wde7fv36+MUvfpH188sr37GbPXt2XjfWAiiE2267LRo1alRm2eLFi+PII4+MTz75JO/1r1u3Li677LI4/fTTY926dRmfc+6556a14c9//nO15s3fsGFDXHTRRWnLMy2rKy688MK0q4z++te/5nxFXHX893//d9qyiy++uOBB0iaZbn6dy9VUhbL33ntnvNr/jjvuiKuuuirn9c6bNy+OOuqoWLVqVZnlLVu2jBtuuCGrdZTvz2zYsKFaYc7y5cvj17/+ddbPL6/8tqqJ7aRvCNuPYp7rfvzjH5f5/5o1a+LnP/95tdZx9dVXx+eff17IZlWqSZMmaX2bdevWxZlnnpn2Y2+hlD9WLVu2LOebLV955ZVpyy677LIaP+d36NAhTjrppDLLpk2bFnfddVfW60iSpKj9ykaNGsUf//jHtOXvvfdenHzyyTmHBmvXro3hw4fHO++8k/bYrbfeGg0bNqxyHfmeTyO+6Xvmug8Xch+tyIMPPhgvvPBC1s9fsmRJjB49Om15+eNOeXvttVfafQyq+16OGTMm53ubbG198nwJDKAOW7RoUZx66qlZ3Uxwzpw5ccYZZ6QtP++88zLeHGZLBx98cLRo0aLMsptuuimrqzFKS0vjwgsvrHayu6U999yzzP+XL18eL7/8cs7rAyiEPn36xE033ZS2/IMPPoiDDjoo5+GsERETJkyIgQMHxu9///tKn9e2bds466yzyizbsGFDjBgxIqsfqJMkiZEjR6bdpG2nnXaK73//+9Vv+FaiRYsW8bOf/azMsiRJ4qyzzorx48fnvN4nn3wyzjvvvEqfM2TIkLT5ahctWhRHH310fPrpp9Wuc86cOZU+3q9fv7RlTzzxRLXrKaRrrrkmBgwYkLb817/+dVx66aXV/mL++uuvx+DBg9Pei1QqFbfffnt07Ngxq/UcffTRaaMf/ud//ierUQZr1qyJ73//+zF79uys211e+f7Mu+++W5BwcUv6hrD9KOa57qyzzkoLK+66666sRzn+85//jN/97ne5NjFnl1xySVq7X3jhhfje976X1XGzvKqOeeWPVRG5n6MHDRoURx11VJllS5cujWOPPTY++OCDnNa5Zs2auO2226oM3jNdCHDxxRdnfePgyy67LF588cWc2lgoJ598csYfnP/1r3/FCSecUOFo3op88cUXcfLJJ8eTTz6Z9thFF10Uw4YNy2o9PXv2jF133bXMsvvuuy/r9/Y3v/lNXlM9FXIfrczpp59eZZ824psQ5jvf+U7atExDhgzJ2OfdUqNGjdJG4rz88stZTxl79913x4033pjVczOprfey1iSwnZk9e3YSEWX+Bg0aVGW5rl27lilz9tln51T/qFGj0uqvyqRJk9LKNGrUaPO/DznkkGTGjBkVln/qqaeSzp07p61j9913T1avXp1Vu88777y08r169Uree++9Cst8+OGHybBhwzK2edPf7Nmzq6x78uTJaeV22223ZOLEicnGjRuzaj9ATbniiivSjlGb/oYNG5a89NJLWR2rVq1aldx///3JoYceWq1j5eLFi5NOnTqllenevXvyzDPPVFju008/TU488cSM7X744Yezeu3ly40aNSqrcplkOtdNmjQp5/WtX78+GTp0aNo6U6lU8sMf/jD56KOPslrPzJkzk6uvvjrp27dvEhFJ3759qywzZ86cpHXr1ml177zzzsltt92WrF+/vtLya9euTR599NHkhBNOSLp3715lfbvvvnuZeurXr5/ccMMNybJly7J6jTXhk08+Sbp06ZJx/+rZs2fy0EMPJevWrat0HfPmzUsuueSSpH79+hnXc/XVV1e7Xccdd1zaeg444IBk3rx5FZaZNm1acsABB1Tan8nGfffdl7HuKVOmVPt16BvqG5K/XL8X5mNbOteNGzcuY73//d//XeFxZP369ck111yTNGjQYHOZxo0b57QNcu2DjB8/PkmlUmnlBwwYUGm/aZMlS5YkY8eOTfbaa68qfxP47LPPknr16qX1BR5++OEqz4GZzJ8/P+nYsWNa25s1a5Zcc801WZ33S0tLk5deein5yU9+kuy8885JRCQXXHBBleXOPvvstHpbtmyZ3HnnnRWWWbRoUXLGGWdUuK3z7TtW17p165Kjjz46Y5+iefPmyQ033JAsXbq00nUsX748ufnmm5OWLVtmXM+wYcOq7OeV97vf/S5tPe3atUtefPHFCst89tlnyVlnnVXp+TSbY0uh99HK+ia77LJL8tRTT1VYdsaMGckhhxyScR3Tp0/Pqv77778/rXyTJk2Sf/7znxWWWbp0aXLZZZdtPi5kei/HjRtXZd3r1q1LdtxxxzLlmjZtmowbNy5ZtWpVVu3fmqSSxNhNti9z5syJ7t27l1k2aNCgePbZZyst161btzJXEJx99tlx5513Vrv+0aNHx5gxY8osq+pj+Oyzz8aQIUPKLLv22mvjyiuv3HwTlZKSkhg6dGgceeSR0alTp1i3bl3MmTMnHn300Yx3l2/YsGE8++yzVd7UbpN58+ZFv3790oZnlZSUxIknnhiDBw+O9u3bx5o1a2L+/PkxadKkmDRpUmzYsCEivrkK9pJLLolf/epXZcrPnj07unXrVmX9ffv2zTg3buPGjWOXXXaJJk2apD32xBNPZH3VIUA+rrnmmrjyyisrPJ63bt06hg4dGn369Im2bdtG27ZtI+KbYb8ff/xxTJs2LV555ZW06VY2qepYOWnSpDj66KM3H3O3NHDgwDjxxBOjW7du0ahRo1iwYEE8//zz8dRTT2W80vuCCy6IW265JYtXHWk3chs1alTGIcTZyHSumzRpUsabFGZr6dKlcfDBB8f06dPTHqtfv37su+++cfjhh0f37t2jdevWUVpaGsuWLYuFCxfG22+/HdOmTUu7Gqpv375p93vI5KmnnoqTTjop4w0J27VrF0cddVTss88+0a5du2jUqFEsW7YsPvnkk3j99dfjxRdf3HzVe9euXau8Iuu6667LOBVEKpWKXXbZJVq0aBH169cv89iPf/zjKod252vOnDlx1FFHVThPfcuWLeP444+PPfbYIzp06BDNmjWLL774Ij777LOYNGlSvPbaaxk/U6lUKq655pqcprSZOnVqHHjggWk3oWvcuHEMHz48DjnkkGjbtm2sXLky5s6dGxMmTIiXXnppczt23XXXOPHEE9OuxszmK9WaNWuiU6dOsWTJkrTHmjdvHh07dkybYizim3mfy9M31Dckf7l+L8zHtnauO/nkk+ORRx5JW962bdsYPnx47L333tG2bdtYunRpvPfee/GPf/yjzJXcJ5xwQqxYsaLMqMxst0E+fZBf/epX8Zvf/CbjY7vuumscffTR0bt372jXrl0kSRLLli2LmTNnxuuvvx6vvPLK5nN7Nr8JHH/88RmvMm7QoEF07tw5mjZtmvZaxo4dG/vuu2/G9U2dOjUGDx6c8X46TZs2jUMPPTQOPvjg6NChQ7Rq1SpWr14dy5Yti88++yxef/31mDZtWixevLhMuWz6fkuXLo3+/fvHvHnz0h7bddddY/jw4bHrrrtuPpe/8sor8dhjj5WZw/3666+Pyy67rEzZfPqOuVizZk2ceuqp8dhjj2V8fIcddogjjzwy9t9//2jfvn20adMmli5dGgsWLIhp06bFM888U+FolJNPPjnuv//+rKYi2tJXX30V/fr1yzjq8Oijj46jjjoqdtlll9i4cWMsWLAgXnjhhXjmmWdi9erVEfHNlfXXXntt/PSnPy1TNttjSyH30fLPu+SSS+KBBx4o87nfZ599Nn83adCgQcyfPz8mTpwY//73vzP2m2+44Ya011aRDRs2xIEHHpixn3PQQQfFCSecEN26dYtUKhVffPFFTJ48OZ566qnN/e5UKhV//OMf4/zzzy9Tdty4cXHOOedUWf8FF1wQf/rTn9KW169fPzp37hzNmzfPONq1/LRfW4XiZRVQHNvKCINJkyYlt9xyS8ZUu6q/Bg0aJI8++mi1237nnXfmVF+zZs2SyZMnZ7wKJZuryJIkSSZMmFDhFYYV/WW7boBCeOqppzJe9ZXPX+PGjZNf/OIXyZo1a6qs//HHH0+aNm2aV30XXXRRta7OLV9+axphsMmSJUuS448/vmDbJJurLrd8TZlGGlTnr2vXrlXW8/XXX2++KjTbv9q6om/p0qXJKaecUrD3v3Xr1pVeJZaNMWPG5FR3hw4dklmzZuXUl9skl75UJvqG+obkb1sYYbBJsc51q1atSo444oic6th9992TJUuWJIMGDcppG5RfX3XPazfccEPaldXV/cvmN4H33nuv2v2zqvaJd955J9ltt90Ktr2zGWGQJEkya9asZJdddsmpjtNOOy0pLS3Ne7sVwsaNG5OrrroqKSkpKcj7V1JSkowePTopLS3NuU3//ve/kx122CGnuh9++OG8ji2F3Eczbd/XXnst5+8oV1xxRbXfy3fffTdp3rx5TvXdcsstGc8N2YwwSJIkWbBgQdK+fftq1ZntumubexhAHXbBBRfE7bffnvFqtIp07tw5Hn/88TjhhBOqXd/ZZ58dd9xxR5SUlGRdpkePHvHiiy/GAQccUO36tnTEEUfEP//5z9h5553zWg9ATTnmmGNixowZ8ctf/jJtftzqatasWfznf/5nzJw5M6699tqsrlQaNmxYPP/88xVekVaZNm3axK233ho33XRT2lUvdV2rVq3i0UcfjRtvvDHvc0jXrl3jBz/4QdbPHzx4cLz22mtx/PHH51xnNm1u0qRJPP3002lzG28NWrZsGf/85z/jH//4R+yxxx45r6dBgwbxwx/+MKZPnx6nnHJKXm266qqrYsyYMWlXwVVm4MCB8corr0TPnj3zqvvss8+OsWPHRvPmzfNaT0X0DWH7VKxzXePGjePRRx+Nc889t1p1HHroofH8889nvElobfnpT38aTz/9dMZ5x7OVzXvdp0+feOaZZ6JXr14511Nev379YsqUKXHhhRdW63ifyX777Zd1P6Vnz57x3HPPxSGHHFKtOs4///z461//Wq3zbk2qV69ejBkzJqZNm5Z33+moo46KqVOnxqhRo/J6fUceeWSMHz8+mjZtmnWZnXbaKZ5++uk4+eSTc643omb20S3tt99+MWHChOjUqVPWZZo0aRJ/+MMf4uqrr652fX379o1JkybFTjvtlHWZZs2axb333hsXXHBBtevbUvv27WPixIkZb2Zd5xQ7sYDati2NMNhk5syZyZlnnplxTsBNf507d06uuOKKZMWKFTm1e0szZsxITj311KRhw4YV1texY8fkN7/5TfL1119vLpfPVWSbrF69Ovn73/+e/Md//Eey//77J+3bt0+aNm2acR5KV5EBxbJy5cpk3LhxyfHHH582l2VFfzvttFNy0kknJXfffXeZY2d1lZaWJvfff38ydOjQSo/TEZH06dMnueqqq3Ke6778+rbGEQZbWr16dfKnP/0pGTJkSMb5Scv/1atXLxk4cGDy85//PHn22WfzunJs8uTJyRlnnJG0a9cuq33hjDPOSB577LFqz8c+bdq05Be/+EVyzDHHJN26dUtatmyZ8Qq6YlzRV1pamjz++OPJ6aefnrRp0yar979///7Jb37zm+TTTz8teHumTJmSHHPMMZVeYdirV6/k5ptvLjMfcT4jDDb56quvknHjxiVnnnlmMmDAgGSnnXaqsB+Xib7h/5++IbnalkYYbKlY57pnn322ymNqnz59krFjx5Y5txVrhMEmpaWlyX333Zcce+yxSZMmTap8v3bbbbfk4osvTl5//fVq1bNhw4bkiSeeSM4///zk0EMPTTp27Jg0a9Ys4yiH6uwTn3/+eXLllVcmAwYMyGrEROPGjZMjjzwyufbaa5P333+/mu/WN0pLS5Pbb789GTBgQKX71ZFHHplMnDixTNmtoT9S3htvvJH85Cc/Sbp3717l+xfxzejPXPaBbMyfPz8ZOXJk0qxZswrrb926dXLZZZclixcv3lyuEMeWQuyjlW3fZcuWJVdccUWlfeEdd9wxOfvss5OPP/44h3evrKVLlyY/+9nPKh3t27Rp0+SHP/xhmXtZ5TPCYJPS0tLkueeeS37yk58kQ4YMSXbZZZdkxx13zDgycmsdYeAeBlAHZDvX5erVq+PVV1+N6dOnx5IlS6Jhw4bRoUOH2HXXXWPfffcteKK/atWqeOmll2Lu3LmxaNGiSKVSsfPOO8fee+8d/fv332quIAAoptLS0pgxY0bMmjUrPvnkk1i5cmWsXbs2mjdvHq1atYo2bdrEnnvuGV27di143V9//XVMnjw5FixYEF9++WWsW7cu2rZtGzvttFMMGDAgOnfuXPA664q1a9fG1KlTY/78+bF48eJYunRplJSURPPmzaNt27ax2267Re/evfO+cq+8JEnirbfeio8++igWLlwYS5Ys2Vxv586dY4899ogePXps8+fQJEli5syZMWPGjPjkk09ixYoVsX79+mjSpEm0adMmevToEXvvvXe0aNGixtuyfPnyeOGFFzbvCyUlJdGpU6cYOHBgXqMiapK+IZCNYpzrFi9eHC+//HJ89tlnsWjRomjUqFF07tw59tlnn7xHadW0tWvXxpQpU+LTTz+NhQsXxvLly6Nx48bRokWL6NGjR/Tp0yfat29f7GZWaOnSpTF16tT48ssvY/HixfHVV19FkyZNonnz5tGhQ4fo3bt39OjRI+2+RvmYNWtWvPnmm/HZZ5/FihUronnz5tGjR4848MADN9+zqy5ZsGBBvPPOOzF37txYtmxZrFmzJho2bBgtW7aMrl27Rr9+/ap1pXyu1q9fHy+//HJ89NFHsWjRoti4cWO0bds29txzz9h3332rNbKvNmVzb5HS0tJ444034p133okvvvgikiSJnXfeObp06RKHHnpote8BUZXS0tKYMmVKTJ8+PRYuXBjr1q2LVq1aRZ8+feLAAw8seH3bCoEB1AE1cXMsAADqJn1DAGBrk8/NyNm6bFuT1AIAAAAAADkRGAAAAAAAAAIDAAAAAABAYAAAAAAAAITAAAAAAAAACIEBAAAAAAAQAgMAAAAAACAEBgAAAAAAQESkkiRJit0IAAAAAACguIwwAAAAAAAABAYAAAAAAIDAAAAAAAAACIEBAAAAAAAQAgMAAAAAACAEBgAAAAAAQAgMAAAAAACAEBgAAAAAAAAhMAAAAAAAAEJgAAAAAAAAhMAAAAAAAAAIgQEAAAAAABACAwAAAAAAICJKit0AAAAAgLoolUrlXDZJkgK2hK1BrvvD9rQveI+2TsXYLo6fWy8jDAAAAAAAAIEBAAAAAAAgMAAAAAAAAEJgAAAAAAAAhMAAAAAAAAAIgQEAAAAAABACAwAAAAAAIAQGAAAAAABACAwAAAAAAIAQGAAAAAAAACEwAAAAAAAAQmAAAAAAAACEwAAAAAAAAAiBAQAAAAAAEBElxW4AAAAAANlJpVJFqTdJkqLUm4t83qN8XmeuZYvV3mKoa+3dXmwv26UYx8+6+N4aYQAAAAAAAAgMAAAAAAAAgQEAAAAAABACAwAAAAAAIAQGAAAAAABACAwAAAAAAIAQGAAAAAAAACEwAAAAAAAAQmAAAAAAAACEwAAAAAAAAAiBAQAAAAAAEAIDAAAAAAAgBAYAAAAAAEBEpJIkSYrdCAAAAAAAoLiMMAAAAAAAAAQGAAAAAACAwAAAAAAAAAiBAQAAAAAAEAIDAAAAAAAgBAYAAAAAAEAIDAAAAAAAgBAYAAAAAAAAITAAAAAAAABCYAAAAAAAAITAAAAAAAAACIEBAAAAAAAQAgMAAAAAACAEBgAAAAAAQESUFLsBAAAAANubVCqVU7kkSQrcEgrFNqUQ7EdV8x7VLCMMAAAAAAAAgQEAAAAAACAwAAAAAAAAQmAAAAAAAACEwAAAAAAAAAiBAQAAAAAAEAIDAAAAAAAgBAYAAAAAAEAIDAAAAAAAgBAYAAAAAAAAITAAAAAAAABCYAAAAAAAAITAAAAAAAAAiIiSYjcAAAAAoJhSqVSt15kkSa3XWdfkul2K9d7aptuWYhwXyE4xPmt17XiUDyMMAAAAAAAAgQEAAAAAACAwAAAAAAAAQmAAAAAAAACEwAAAAAAAAAiBAQAAAAAAEAIDAAAAAAAgBAYAAAAAAEAIDAAAAAAAgBAYAAAAAAAAITAAAAAAAABCYAAAAAAAAITAAAAAAAAACIEBAAAAAAAQESXFbgAAAABAMSVJklO5VCpV4JawpVy3C9ueYnzWirX/bS/HlXxeZzG2zfZ0PDLCAAAAAAAAEBgAAAAAAAACAwAAAAAAIAQGAAAAAABACAwAAAAAAIAQGAAAAAAAACEwAAAAAAAAQmAAAAAAAACEwAAAAAAAAAiBAQAAAAAAEAIDAAAAAAAgBAYAAAAAAEAIDAAAAAAAgBAYAAAAAAAAEVFS7AYAAAAAbG9SqVRO5ZIkKXBLtl7by3uU6+uMqHuvtRiK8R7ZpmxSF/cFIwwAAAAAAACBAQAAAAAAIDAAAAAAAABCYAAAAAAAAITAAAAAAAAACIEBAAAAAAAQAgMAAAAAACAEBgAAAAAAQAgMAAAAAACAEBgAAAAAAAAhMAAAAAAAAEJgAAAAAAAAhMAAAAAAAACIiFSSJEmxGwEAAACQj1QqlXPZ7eWnEe8R26Nc9/t89vl8Pmu5KkZ7HRe2TUYYAAAAAAAAAgMAAAAAAEBgAAAAAAAAhMAAAAAAAAAIgQEAAAAAABACAwAAAAAAIAQGAAAAAABACAwAAAAAAIAQGAAAAAAAACEwAAAAAAAAQmAAAAAAAACEwAAAAAAAAAiBAQAAAAAAEAIDAAAAAAAgIkqK3QAAAACAuiiVStV6nUmSFKVsMeT6/ubzOuvaNs1VPq+zru1Hda29uSrGNt2e9qPtiREGAAAAAACAwAAAAAAAABAYAAAAAAAAITAAAAAAAABCYAAAAAAAAITAAAAAAAAACIEBAAAAAAAQAgMAAAAAACAEBgAAAAAAQAgMAAAAAACAEBgAAAAAAAAhMAAAAAAAAEJgAAAAAAAAREQqSZKk2I0AAAAAgLoslUrVep1+1tv25Lof1bV9wedl62WEAQAAAAAAIDAAAAAAAAAEBgAAAAAAQAgMAAAAAACAEBgAAAAAAAAhMAAAAAAAAEJgAAAAAAAAhMAAAAAAAAAIgQEAAAAAABACAwAAAAAAIAQGAAAAAABACAwAAAAAAIAQGAAAAAAAACEwAAAAAAAAIiKVJElS7EYAAEDNSuVccsyY0YVrRg0bNWp0kWrO7SvFmDFjcq4x99fq6w9ALlKp3M6lfnYCCs3xqGYZYQAAAAAAAAgMAAAAAAAAgQEAAAAAABACAwAAAAAAIAQGAAAAAABACAwAAAAAAIAQGAAAAAAAACEwAAAAAAAAQmAAAAAAAACEwAAAAAAAAAiBAQAAAAAAEAIDAAAAAAAgBAYAAAAAAEAIDAAAAAAAgIgoKXYDAACgpo0ZM7rW6xw1alTOZceMGZNjudE51zlqVO5lAeq6VCqVU7kkSQrckq3X9vRaa1uu+19ExHUFbEe2LttO9oV8tks+cv2sFaO9xTouOB7VLCMMAAAAAAAAgQEAAAAAACAwAAAAAAAAQmAAAAAAAACEwAAAAAAAAAiBAQAAAAAAEAIDAAAAAAAgBAYAAAAAAEAIDAAAAAAAgBAYAAAAAAAAITAAAAAAAABCYAAAAAAAAITAAAAAAAAAiIiSYjcA6p5UziXHjBlduGbUglGjRudYMilkMwCgqEaNGlVn6hwzZkyBWwJAZVKp3L8fJsn28b1pe3qPrs/xteb1Ku/Jp3CO8timubo+j7I/y7Fcsfa/XD8zde3zwtbLCAMAAAAAAEBgAAAAAAAACAwAAAAAAIAQGAAAAAAAACEwAAAAAAAAQmAAAAAAAACEwAAAAAAAAAiBAQAAAAAAEAIDAAAAAAAgBAYAAAAAAEAIDAAAAAAAgBAYAAAAAAAAITAAAAAAAABCYAAAAAAAAERESbEbAHXNmDGja73OUaNG5Vx2zJgxeZQdnVO5PJoLAFudXM+lxTp/A0BFUqlUzmWTJKnVcmTp9LpVZ6674HW5V5nzPliMz0u+ZalcMbZpsfajfBhhAAAAAAAACAwAAAAAAACBAQAAAAAAEAIDAAAAAAAgBAYAAAAAAEAIDAAAAAAAgBAYAAAAAAAAITAAAAAAAABCYAAAAAAAAITAAAAAAAAACIEBAAAAAAAQAgMAAAAAACAEBgAAAAAAQESUFLsBsD0ZNWpUnapzzJgxBWwJABTPqFGjcy47ZkxuZYtxHs3ndQJsz5IkKXYT2EqkUqmcy16Xa8F7cq4yZ6l8uimjcy+a5Pharz8j9zpz5biw7clnm+ZzbKhrjDAAAAAAAAAEBgAAAAAAgMAAAAAAAAAIgQEAAAAAABACAwAAAAAAIAQGAAAAAABACAwAAAAAAIAQGAAAAAAAACEwAAAAAAAAQmAAAAAAAACEwAAAAAAAAAiBAQAAAAAAEAIDAAAAAAAgBAYAAAAAAEBElBS7AbA9GTNmTE7lRo0aVet1AgDfGDVqdLGbUEuSYjcAoGhSqVRO5ZKkbh0761p7iyGf9yjX/eiyM3KuMuL0HMuNzqPOfMqyTcl1n48ozvGorrW3WIwwAAAAAAAABAYAAAAAAIDAAAAAAAAACIEBAAAAAAAQAgMAAAAAACAEBgAAAAAAQAgMAAAAAACAEBgAAAAAAAAhMAAAAAAAAEJgAAAAAAAAhMAAAAAAAAAIgQEAAAAAABACAwAAAAAAIAQGAAAAAABARJQUuwFQ14waNTrnsmPG5FZ2zJgxOdeZj9xf66hCNgMACiApdgO2acXqqwAUSpLkdp5IpVLbRZ35yKe9uSrG64yIuK4Yld5b+1Umu9Z+ndSsYnxO69qxbHtihAEAAAAAACAwAAAAAAAABAYAAAAAAEAIDAAAAAAAgBAYAAAAAAAAITAAAAAAAABCYAAAAAAAAITAAAAAAAAACIEBAAAAAAAQAgMAAAAAACAEBgAAAAAAQAgMAAAAAACAEBgAAAAAAAARkUqSJCl2IwAAoCaNGTOm2E0gg1GjRhehVl9/gHSpVCqncnXtJ5VcX2dE3Xutdcn1eWyXXP0sj7LXFawVtSOf15orn5etUzGOgXXxuGuEAQAAAAAAIDAAAAAAAAAEBgAAAAAAQAgMAAAAAACAEBgAAAAAAAAhMAAAAAAAAEJgAAAAAAAAhMAAAAAAAAAIgQEAAAAAABACAwAAAAAAIAQGAAAAAABACAwAAAAAAIAQGAAAAAAAACEwAAAAAAAAIiKVJElS7EYAAEDNShW7ARScrzEAFF4qlXufoRg/seXa3mL9HJjP+5ururRdIupWe/Npa117j3JVF1+nEQYAAAAAAIDAAAAAAAAAEBgAAAAAAAAhMAAAAAAAAEJgAAAAAAAAhMAAAAAAAAAIgQEAAAAAABACAwAAAAAAIAQGAAAAAABACAwAAAAAAIAQGAAAAAAAACEwAAAAAAAAQmAAAAAAAABERCpJkqTYjQAAAKgNY8aMybnsqFGjCtgSoNBSqVTOZf00UnNsFwrBfgS1xwgDAAAAAABAYAAAAAAAAAgMAAAAAACAEBgAAAAAAAAhMAAAAAAAAEJgAAAAAAAAhMAAAAAAAAAIgQEAAAAAABACAwAAAAAAIAQGAAAAAABACAwAAAAAAIAQGAAAAAAAACEwAAAAAAAAQmAAAAAAAABERCpJkqTYjQAAAAAAakcqlcq5rJ8St051bZvm095c1bXXWazPmhEGAAAAAACAwAAAAAAAABAYAAAAAAAAITAAAAAAAABCYAAAAAAAAITAAAAAAAAACIEBAAAAAAAQAgMAAAAAACAEBgAAAAAAQAgMAAAAAACAEBgAAAAAAAAhMAAAAAAAAEJgAAAAAAAAhMAAAAAAAACIiFSSJEmxGwEAAABQLKlUKqdyde0nlVxfZ0Tde611ie3ClvLZH3JlP2JLRhgAAAAAAAACAwAAAAAAQGAAAAAAAACEwAAAAAAAAAiBAQAAAAAAEAIDAAAAAAAgBAYAAAAAAEAIDAAAAAAAgBAYAAAAAAAAITAAAAAAAABCYAAAAAAAAITAAAAAAAAACIEBAAAAAAAQESXFbgAAAABAvlKpVLGbsNVLkqTYTdim5boP1rXtks9nrRivtRjHhnxeZ116j+y72yYjDAAAAAAAAIEBAAAAAAAgMAAAAAAAAEJgAAAAAAAAhMAAAAAAAAAIgQEAAAAAABACAwAAAAAAIAQGAAAAAABACAwAAAAAAIAQGAAAAAAAACEwAAAAAAAAQmAAAAAAAACEwAAAAAAAAAiBAQAAAAAAEBGpJEmSYjcCAAAAoFhSqVRO5eraTyq5vs5iyfX9zed11rVtyranGJ/TYuz3PqdbLyMMAAAAAAAAgQEAAAAAACAwAAAAAAAAQmAAAAAAAACEwAAAAAAAAAiBAQAAAAAAEAIDAAAAAAAgBAYAAAAAAEAIDAAAAAAAgBAYAAAAAAAAITAAAAAAAABCYAAAAAAAAITAAAAAAAAAiIiSYjcAAAAAIF+pVCrnskmSFLAl26a69B4Vq6257oPFaG8+n5d81KX9qFjq0ntUrP2ImmWEAQAAAAAAIDAAAAAAAAAEBgAAAAAAQAgMAAAAAACAEBgAAAAAAAAhMAAAAAAAAEJgAAAAAAAAhMAAAAAAAAAIgQEAAAAAABACAwAAAAAAIAQGAAAAAABACAwAAAAAAIAQGAAAAAAAACEwAAAAAAAAIqKk2A0AAAAAyFeSJDmXTaVSBWxJdvJpb12qk61TXfu85CPX9hbr81LX2purutbe7YkRBgAAAAAAgMAAAAAAAAAQGAAAAAAAACEwAAAAAAAAQmAAAAAAAACEwAAAAAAAAAiBAQAAAAAAEAIDAAAAAAAgBAYAAAAAAEAIDAAAAAAAgBAYAAAAAAAAITAAAAAAAABCYAAAAAAAAITAAAAAAAAAiIhUkiRJsRsBAAAAUNekUqmcy/o5pmr5vL+5KsZ2sR9tnWyXqhXjM1osuW7TurgfGWEAAAAAAAAIDAAAAAAAAIEBAAAAAAAQAgMAAAAAACAEBgAAAAAAQAgMAAAAAACAEBgAAAAAAAAhMAAAAAAAAEJgAAAAAAAAhMAAAAAAAAAIgQEAAAAAABACAwAAAAAAIAQGAAAAAABARJQUuwEAAAAA+UqlUjmXTZKkVsuRHe8vhZDrsSGf/a8Yx6NiqEttzVc+27SuMcIAAAAAAAAQGAAAAAAAAAIDAAAAAAAgBAYAAAAAAEAIDAAAAAAAgBAYAAAAAAAAITAAAAAAAABCYAAAAAAAAITAAAAAAAAACIEBAAAAAAAQAgMAAAAAACAEBgAAAAAAQAgMAAAAAACAEBgAAAAAAAARkUqSJCl2IwAAAACoWalUKueyuf58VIw6qVm2KcWW6z5o/8uOEQYAAAAAAIDAAAAAAAAAEBgAAAAAAAAhMAAAAAAAAEJgAAAAAAAAhMAAAAAAAAAIgQEAAAAAABACAwAAAAAAIAQGAAAAAABACAwAAAAAAIAQGAAAAAAAACEwAAAAAAAAQmAAAAAAAABERCpJkqTYjQAAAACoa1KpVK3XWayfcXJ9rX52Ykv2I7Y3+ZwnirXfG2EAAAAAAAAIDAAAAAAAAIEBAAAAAAAQAgMAAAAAACAEBgAAAAAAQAgMAAAAAACAEBgAAAAAAAAhMAAAAAAAAEJgAAAAAAAAhMAAAAAAAAAIgQEAAAAAABACAwAAAAAAIAQGAAAAAABACAwAAAAAAICISCVJkhS7EQAAAAAAQHEZYQAAAAAAAAgMAAAAAAAAgQEAAAAAABACAwAAAAAAIAQGAAAAAABACAwAAAAAAIAQGAAAAAAAACEwAAAAAAAAQmAAAAAAAACEwAAAAAAAAAiBAQAAAAAAEAIDAAAAAAAgBAYAAAAAAEAIDAAAAAAAgBAYAAAAAAAAITAAAAAAAABCYAAAAAAAAITAAAAAAAAACIEBAAAAAAAQAgMAAAAAACAEBgAAAAAAQAgMAAAAAACAEBgAAAAAAAAhMAAAAAAAAEJgAAAAAAAAhMAAAAAAAAAIgQEAAAAAABACAwAAAAAAIAQGAAAAAABARPz/AODWWMrUq+f+AAAAAElFTkSuQmCC"
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 44.11% (397/900)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 1920x1440 with 3 Axes>"
      ],
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 10.00% (90/900)\n"
     ]
    }
   ],
   "execution_count": 13
  },
  {
   "metadata": {},
   "cell_type": "code",
   "outputs": [],
   "execution_count": null,
   "source": "",
   "id": "2db9dc3a233128c0"
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}