Nerf-reconstruction.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "257860de-71ad-4386-88bd-fc945e3bc256",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "from torch.utils.data import DataLoader\n",
    "\n",
    "from dataset import get_rays\n",
    "from model import Nerf\n",
    "from ml_helpers import training"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5922e325-8976-4a36-88aa-908cbe4d3953",
   "metadata": {},
   "source": [
    "# Camera / Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "66e62ab8-9524-48b6-a646-01b27c9a3fd6",
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_size = 1024\n",
    "height = 400\n",
    "width = 400\n",
    "imgs = 100\n",
    "\n",
    "o, d, target_px_values = get_rays('datasets/monkey_3', mode='train')\n",
    "dataloader = DataLoader(torch.cat((torch.from_numpy(o).reshape(-1, 3).type(torch.float),\n",
    "                                   torch.from_numpy(d).reshape(-1, 3).type(torch.float),\n",
    "                                   torch.from_numpy(target_px_values).reshape(-1, 3).type(torch.float)), dim=1),\n",
    "                       batch_size=batch_size, shuffle=True)\n",
    "\n",
    "\n",
    "dataloader_warmup = DataLoader(torch.cat((torch.from_numpy(o).reshape(imgs, 400, 400, 3)[:, 100:300, 100:300, :].reshape(-1, 3).type(torch.float),\n",
    "                               torch.from_numpy(d).reshape(imgs, 400, 400, 3)[:, 100:300, 100:300, :].reshape(-1, 3).type(torch.float),\n",
    "                               torch.from_numpy(target_px_values).reshape(imgs, 400, 400, 3)[:, 100:300, 100:300, :].reshape(-1, 3).type(torch.float)), dim=1),\n",
    "                       batch_size=batch_size, shuffle=True)\n",
    "\n",
    "\n",
    "# test_o, test_d, test_target_px_values = get_rays('datasets/monkey_3', mode='test')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bbcc6284-ff38-4a9d-a914-7cd8ff9163fb",
   "metadata": {},
   "source": [
    "# Training - Change model name"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "230bd02d",
   "metadata": {},
   "outputs": [],
   "source": [
    "pth_file = 'nerf_models/monkey_biz.pth'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "edd96eec-e258-469d-8781-da1dc2bd1ca6",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 3907/3907 [01:44<00:00, 37.53it/s]\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 15625/15625 [06:43<00:00, 38.70it/s]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkAAAAGdCAYAAAD60sxaAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8pXeV/AAAACXBIWXMAAA9hAAAPYQGoP6dpAABADUlEQVR4nO3dfXyU1Z3///eQmwkEEoFAQhRCorYgQYWJjYlNsVsaBKpLi1tkW2T9tm7T1iJkcbnT1eJNqGUti9wtiF2pFfh1QZfWKASViGaAEhJAiAgSCYSEmCgz3Ob2/P7AjI6ZQCYMhOF6PR+PeTzCmc91nXMCZN45c11nbMYYIwAAAAvp1NEDAAAAuNwIQAAAwHIIQAAAwHIIQAAAwHIIQAAAwHIIQAAAwHIIQAAAwHIIQAAAwHJCO3oAl1NTU5OOHj2qbt26yWazdfRwAABAGxhjdOLECcXHx6tTp8Cs3VgqAB09elR9+/bt6GEAAIB2OHz4sK677rqAnMtSAahbt26Szn0Do6KiOng0AACgLdxut/r27et5HQ8ESwWg5re9oqKiCEAAAASZQF6+wkXQAADAcghAAADAcghAAADAcghAAADAcghAAADAcghAAADAcghAAADAcghAAADAcghAAADAcghAAADAcghAAADAcghAAADAciz1YaiXynMb9ulEbYN+8Z3rFRcd0dHDAQAAF8AKUACs+vth/fH9T/TZqbqOHgoAAGgDAhAAALAcAhAAALAcAhAAALCcdgWgRYsWKTExUREREXI4HNq8efN56/Pz8+VwOBQREaGkpCQtWbKk1dpVq1bJZrNpzJgxF90vAACAL34HoNWrV2vy5MmaNWuWioqKlJGRoZEjR6qsrMxnfWlpqUaNGqWMjAwVFRVp5syZmjRpktasWdOi9tChQ5o6daoyMjIuul8AAIDW2Iwxxp8DUlNTNXToUC1evNjTNnDgQI0ZM0Y5OTkt6qdNm6Z169appKTE05aVlaWdO3fK6XR62hobGzVs2DA98MAD2rx5s44fP67XXnut3f364na7FR0dLZfLpaioKH+mfV7fenqjqk7UKndShm6KD9x5AQDApXn99msFqK6uToWFhcrMzPRqz8zMVEFBgc9jnE5ni/oRI0Zo+/btqq+v97TNnj1bvXr10s9+9rOA9CtJtbW1crvdXo9LycivLAkAADqIXwGourpajY2Nio2N9WqPjY1VZWWlz2MqKyt91jc0NKi6ulqS9P7772v58uVatmxZwPqVpJycHEVHR3seffv2veAcAQDA1a9dF0HbbDavPxtjWrRdqL65/cSJE/rpT3+qZcuWKSYmJqD9zpgxQy6Xy/M4fPjwec/fXucZAgAAuAL59VEYMTExCgkJabHqUlVV1WJ1pllcXJzP+tDQUPXs2VN79uzRJ598orvvvtvzfFNT07nBhYZq37596tu3r9/9SpLdbpfdbvdnigAAwAL8WgEKDw+Xw+FQXl6eV3teXp7S09N9HpOWltaifsOGDUpJSVFYWJgGDBig3bt3q7i42PO455579N3vflfFxcXq27dvu/oFAABojd8fhpqdna0JEyYoJSVFaWlpWrp0qcrKypSVlSXp3NtO5eXlWrFihaRzd3wtWLBA2dnZevDBB+V0OrV8+XKtXLlSkhQREaHk5GSvPq655hpJ8mq/UL8AAABt5XcAGjdunGpqajR79mxVVFQoOTlZubm5SkhIkCRVVFR47c2TmJio3NxcTZkyRQsXLlR8fLzmz5+vsWPHBrRfAACAtvJ7H6Bgdqn2AUp9ZqOOuWv1+qRva1B8dMDOCwAAroB9gAAAAK4GBKAAss5aGgAAwY0ABAAALIcAFAA2sRMiAADBhAAEAAAshwAEAAAshwAEAAAshwAEAAAshwAEAAAshwAEAAAshwAEAAAshwAUADa2AQIAIKgQgAAAgOUQgAAAgOUQgAAAgOUQgAAAgOUQgAAAgOUQgAAAgOUQgALImI4eAQAAaAsCEAAAsBwCUACwDyIAAMGFAAQAACyHAAQAACyHAAQAACyHAAQAACyHAAQAACyHAAQAACyHABRARuyECABAMCAAAQAAyyEABYDNxlaIAAAEk3YFoEWLFikxMVERERFyOBzavHnzeevz8/PlcDgUERGhpKQkLVmyxOv5tWvXKiUlRddcc40iIyN166236k9/+pNXzRNPPCGbzeb1iIuLa8/wAQCAxfkdgFavXq3Jkydr1qxZKioqUkZGhkaOHKmysjKf9aWlpRo1apQyMjJUVFSkmTNnatKkSVqzZo2npkePHpo1a5acTqd27dqlBx54QA888IDWr1/vda5BgwapoqLC89i9e7e/wwcAAFCovwc899xz+tnPfqaf//znkqR58+Zp/fr1Wrx4sXJyclrUL1myRP369dO8efMkSQMHDtT27ds1d+5cjR07VpJ05513eh3z8MMP66WXXtJ7772nESNGfDnY0FBWfQAAwEXzawWorq5OhYWFyszM9GrPzMxUQUGBz2OcTmeL+hEjRmj79u2qr69vUW+M0VtvvaV9+/bpO9/5jtdz+/fvV3x8vBITE3Xffffp4MGD5x1vbW2t3G631wMAAMCvAFRdXa3GxkbFxsZ6tcfGxqqystLnMZWVlT7rGxoaVF1d7WlzuVzq2rWrwsPDNXr0aD3//PP6/ve/73k+NTVVK1as0Pr167Vs2TJVVlYqPT1dNTU1rY43JydH0dHRnkffvn39mS4AALhKtesi6K/f9WSMOe+dUL7qv97erVs3FRcX6+9//7uefvppZWdna9OmTZ7nR44cqbFjx2rw4MEaPny4Xn/9dUnSSy+91Gq/M2bMkMvl8jwOHz7c5jm2h2EbIAAAgoJf1wDFxMQoJCSkxWpPVVVVi1WeZnFxcT7rQ0ND1bNnT09bp06ddMMNN0iSbr31VpWUlCgnJ6fF9UHNIiMjNXjwYO3fv7/V8drtdtnt9rZMDQAAWIhfK0Dh4eFyOBzKy8vzas/Ly1N6errPY9LS0lrUb9iwQSkpKQoLC2u1L2OMamtrW32+trZWJSUl6tOnjx8zAAAAaMddYNnZ2ZowYYJSUlKUlpampUuXqqysTFlZWZLOve1UXl6uFStWSJKysrK0YMECZWdn68EHH5TT6dTy5cu1cuVKzzlzcnKUkpKi66+/XnV1dcrNzdWKFSu0ePFiT83UqVN19913q1+/fqqqqtJTTz0lt9utiRMnXuz3AAAAWIzfAWjcuHGqqanR7NmzVVFRoeTkZOXm5iohIUGSVFFR4bUnUGJionJzczVlyhQtXLhQ8fHxmj9/vucWeEk6deqUfvWrX+nIkSPq3LmzBgwYoJdfflnjxo3z1Bw5ckTjx49XdXW1evXqpdtvv11btmzx9AsAANBWNmOsc+mu2+1WdHS0XC6XoqKiAnbeO+a8rfLjZ/R/v75Dt/S9JmDnBQAAl+b1m88CAwAAlkMAAgAAlkMAAgAAlkMACiDLXEwFAECQIwABAADLIQABAADLIQAFwHk+Bg0AAFyBCEAAAMByCEAAAMByCEAAAMByCEAAAMByCEABZKGPVQMAIKgRgAAAgOUQgAAAgOUQgAAAgOUQgAKAjRABAAguBCAAAGA5BCAAAGA5BCAAAGA5BCAAAGA5BKAAYhtEAACCAwEIAABYDgEIAABYDgEIAABYDgEoAGxiJ0QAAIIJAQgAAFgOAQgAAFgOAQgAAFgOASiADBsBAQAQFAhAAADActoVgBYtWqTExERFRETI4XBo8+bN563Pz8+Xw+FQRESEkpKStGTJEq/n165dq5SUFF1zzTWKjIzUrbfeqj/96U8X3S8AAIAvfgeg1atXa/LkyZo1a5aKioqUkZGhkSNHqqyszGd9aWmpRo0apYyMDBUVFWnmzJmaNGmS1qxZ46np0aOHZs2aJafTqV27dumBBx7QAw88oPXr17e7XwAAgNbYjPHvypXU1FQNHTpUixcv9rQNHDhQY8aMUU5OTov6adOmad26dSopKfG0ZWVlaefOnXI6na32M3ToUI0ePVpPPvlku/r1xe12Kzo6Wi6XS1FRUW06pi2G/f4dHao5rTW/TJcjoXvAzgsAAC7N67dfK0B1dXUqLCxUZmamV3tmZqYKCgp8HuN0OlvUjxgxQtu3b1d9fX2LemOM3nrrLe3bt0/f+c532t2vJNXW1srtdns9AAAA/ApA1dXVamxsVGxsrFd7bGysKisrfR5TWVnps76hoUHV1dWeNpfLpa5duyo8PFyjR4/W888/r+9///vt7leScnJyFB0d7Xn07dvXn+kCAICrVLsugrbZvD/6wRjTou1C9V9v79atm4qLi/X3v/9dTz/9tLKzs7Vp06aL6nfGjBlyuVyex+HDh887LwAAYA2h/hTHxMQoJCSkxapLVVVVi9WZZnFxcT7rQ0ND1bNnT09bp06ddMMNN0iSbr31VpWUlCgnJ0d33nlnu/qVJLvdLrvd7s8UAQCABfi1AhQeHi6Hw6G8vDyv9ry8PKWnp/s8Ji0trUX9hg0blJKSorCwsFb7Msaotra23f12DHZCBAAgGPi1AiRJ2dnZmjBhglJSUpSWlqalS5eqrKxMWVlZks697VReXq4VK1ZIOnfH14IFC5Sdna0HH3xQTqdTy5cv18qVKz3nzMnJUUpKiq6//nrV1dUpNzdXK1as8Lrj60L9AgAAtJXfAWjcuHGqqanR7NmzVVFRoeTkZOXm5iohIUGSVFFR4bU3T2JionJzczVlyhQtXLhQ8fHxmj9/vsaOHeupOXXqlH71q1/pyJEj6ty5swYMGKCXX35Z48aNa3O/AAAAbeX3PkDB7NLvA5QmR0KPgJ0XAABcAfsAwbfW70MDAABXIgIQAACwHAIQAACwHAIQAACwHAIQAACwHAJQAFnnfjoAAIIbAQgAAFgOAQgAAFgOAQgAAFgOASgAbDa2QgQAIJgQgAAAgOUQgAAAgOUQgAAAgOUQgAKIbYAAAAgOBCAAAGA5BCAAAGA5BCAAAGA5BCAAAGA5BKAAYBtEAACCCwEIAABYDgEIAABYDgEIAABYDgEogAw7IQIAEBQIQAAAwHIIQAAAwHIIQAAAwHIIQAAAwHIIQIHATogAAAQVAhAAALAcAhAAALCcdgWgRYsWKTExUREREXI4HNq8efN56/Pz8+VwOBQREaGkpCQtWbLE6/lly5YpIyND3bt3V/fu3TV8+HBt27bNq+aJJ56QzWbzesTFxbVn+JeMYSMgAACCgt8BaPXq1Zo8ebJmzZqloqIiZWRkaOTIkSorK/NZX1paqlGjRikjI0NFRUWaOXOmJk2apDVr1nhqNm3apPHjx+udd96R0+lUv379lJmZqfLycq9zDRo0SBUVFZ7H7t27/R0+AACAbMbPZYvU1FQNHTpUixcv9rQNHDhQY8aMUU5OTov6adOmad26dSopKfG0ZWVlaefOnXI6nT77aGxsVPfu3bVgwQLdf//9ks6tAL322msqLi72Z7he3G63oqOj5XK5FBUV1e7zfN0//OcmHfz0lFb/6+1KTeoZsPMCAIBL8/rt1wpQXV2dCgsLlZmZ6dWemZmpgoICn8c4nc4W9SNGjND27dtVX1/v85jTp0+rvr5ePXr08Grfv3+/4uPjlZiYqPvuu08HDx4873hra2vldru9HgAAAH4FoOrqajU2Nio2NtarPTY2VpWVlT6Pqays9Fnf0NCg6upqn8dMnz5d1157rYYPH+5pS01N1YoVK7R+/XotW7ZMlZWVSk9PV01NTavjzcnJUXR0tOfRt2/ftk4VAABcxdp1EbTN5r3xjTGmRduF6n21S9Kzzz6rlStXau3atYqIiPC0jxw5UmPHjtXgwYM1fPhwvf7665Kkl156qdV+Z8yYIZfL5XkcPnz4wpMDAABXvVB/imNiYhQSEtJitaeqqqrFKk+zuLg4n/WhoaHq2dP7epm5c+fqmWee0caNG3XzzTefdyyRkZEaPHiw9u/f32qN3W6X3W4/73kCgX0QAQAILn6tAIWHh8vhcCgvL8+rPS8vT+np6T6PSUtLa1G/YcMGpaSkKCwszNP2+9//Xk8++aTefPNNpaSkXHAstbW1KikpUZ8+ffyZAgAAgP9vgWVnZ+uFF17Qiy++qJKSEk2ZMkVlZWXKysqSdO5tp+Y7t6Rzd3wdOnRI2dnZKikp0Ysvvqjly5dr6tSpnppnn31Wjz76qF588UX1799flZWVqqys1MmTJz01U6dOVX5+vkpLS7V161bde++9crvdmjhx4sXMHwAAWJBfb4FJ0rhx41RTU6PZs2eroqJCycnJys3NVUJCgiSpoqLCa0+gxMRE5ebmasqUKVq4cKHi4+M1f/58jR071lOzaNEi1dXV6d577/Xq6/HHH9cTTzwhSTpy5IjGjx+v6upq9erVS7fffru2bNni6fdKwDaIAAAEB7/3AQpml2ofoO/95yZ9/OkprfrX23U7+wABABBQHb4PEAAAwNWAAAQAACyHAAQAACyHAAQAACyHABQA59sFGwAAXHkIQAAAwHIIQAFknQ0FAAAIbgQgAABgOQQgAABgOQQgAABgOQQgAABgOQQgAABgOQSgAGAXIAAAggsBCAAAWA4BCAAAWA4BKICM2AkRAIBgQAACAACWQwACAACWQwACAACWQwACAACWQwACAACWQwAKABs7IQIAEFQIQAAAwHIIQIHENkAAAAQFAhAAALAcAhAAALAcAhAAALAcAhAAALAcAhAAALAcAhAAALCcdgWgRYsWKTExUREREXI4HNq8efN56/Pz8+VwOBQREaGkpCQtWbLE6/lly5YpIyND3bt3V/fu3TV8+HBt27btovu9XGxiJ0QAAIKJ3wFo9erVmjx5smbNmqWioiJlZGRo5MiRKisr81lfWlqqUaNGKSMjQ0VFRZo5c6YmTZqkNWvWeGo2bdqk8ePH65133pHT6VS/fv2UmZmp8vLydvcLAADQGpsxxq/t+1JTUzV06FAtXrzY0zZw4ECNGTNGOTk5LeqnTZumdevWqaSkxNOWlZWlnTt3yul0+uyjsbFR3bt314IFC3T//fe3q19f3G63oqOj5XK5FBUV1aZj2mLEH97VvmMn9Oefp+qOG2ICdl4AAHBpXr/9WgGqq6tTYWGhMjMzvdozMzNVUFDg8xin09mifsSIEdq+fbvq6+t9HnP69GnV19erR48e7e5Xkmpra+V2u70eAAAAfgWg6upqNTY2KjY21qs9NjZWlZWVPo+prKz0Wd/Q0KDq6mqfx0yfPl3XXnuthg8f3u5+JSknJ0fR0dGeR9++fS84RwAAcPVr10XQtq99/LkxpkXbhep9tUvSs88+q5UrV2rt2rWKiIi4qH5nzJghl8vleRw+fLjVWgAAYB2h/hTHxMQoJCSkxapLVVVVi9WZZnFxcT7rQ0ND1bNnT6/2uXPn6plnntHGjRt18803X1S/kmS322W329s0NwAAYB1+rQCFh4fL4XAoLy/Pqz0vL0/p6ek+j0lLS2tRv2HDBqWkpCgsLMzT9vvf/15PPvmk3nzzTaWkpFx0vwAAAK3xawVIkrKzszVhwgSlpKQoLS1NS5cuVVlZmbKysiSde9upvLxcK1askHTujq8FCxYoOztbDz74oJxOp5YvX66VK1d6zvnss8/qscce0yuvvKL+/ft7Vnq6du2qrl27tqlfAACAtvI7AI0bN041NTWaPXu2KioqlJycrNzcXCUkJEiSKioqvPbmSUxMVG5urqZMmaKFCxcqPj5e8+fP19ixYz01ixYtUl1dne69916vvh5//HE98cQTbeq3I53nMiQAAHAF8nsfoGB2qfYBumveu/qwkn2AAAC4FDp8HyCcn3WiJAAAwY0ABAAALIcABAAALIcABAAALIcABAAALIcABAAALIcABAAALIcABAAALIcAFEBGbAQEAEAwIAABAADLIQABAADLIQABAADLIQABAADLIQABAADLIQABAADLIQABAADLIQAFgM1m6+ghAAAAPxCAAsiwDyIAAEGBAAQAACyHAAQAACyHAAQAACyHAAQAACyHAAQAACyHAAQAACyHABQA7AIEAEBwIQAFENsAAQAQHAhAAADAcghAAADAcghAAADAcghAAADActoVgBYtWqTExERFRETI4XBo8+bN563Pz8+Xw+FQRESEkpKStGTJEq/n9+zZo7Fjx6p///6y2WyaN29ei3M88cQTstlsXo+4uLj2DB8AAFic3wFo9erVmjx5smbNmqWioiJlZGRo5MiRKisr81lfWlqqUaNGKSMjQ0VFRZo5c6YmTZqkNWvWeGpOnz6tpKQkzZkz57yhZtCgQaqoqPA8du/e7e/wAQAAFOrvAc8995x+9rOf6ec//7kkad68eVq/fr0WL16snJycFvVLlixRv379PKs6AwcO1Pbt2zV37lyNHTtWknTbbbfptttukyRNnz699cGGhrLqAwAALppfK0B1dXUqLCxUZmamV3tmZqYKCgp8HuN0OlvUjxgxQtu3b1d9fb1fg92/f7/i4+OVmJio++67TwcPHjxvfW1trdxut9fjUrCxEyIAAEHFrwBUXV2txsZGxcbGerXHxsaqsrLS5zGVlZU+6xsaGlRdXd3mvlNTU7VixQqtX79ey5YtU2VlpdLT01VTU9PqMTk5OYqOjvY8+vbt2+b+2sMYtkIEACAYtOsiaNvXljyMMS3aLlTvq/18Ro4cqbFjx2rw4MEaPny4Xn/9dUnSSy+91OoxM2bMkMvl8jwOHz7c5v4AAMDVy69rgGJiYhQSEtJitaeqqqrFKk+zuLg4n/WhoaHq2bOnn8P9UmRkpAYPHqz9+/e3WmO322W329vdBwAAuDr5tQIUHh4uh8OhvLw8r/a8vDylp6f7PCYtLa1F/YYNG5SSkqKwsDA/h/ul2tpalZSUqE+fPu0+BwAAsCa/3wLLzs7WCy+8oBdffFElJSWaMmWKysrKlJWVJenc207333+/pz4rK0uHDh1Sdna2SkpK9OKLL2r58uWaOnWqp6aurk7FxcUqLi5WXV2dysvLVVxcrAMHDnhqpk6dqvz8fJWWlmrr1q2699575Xa7NXHixIuZPwAAsCC/b4MfN26campqNHv2bFVUVCg5OVm5ublKSEiQJFVUVHjtCZSYmKjc3FxNmTJFCxcuVHx8vObPn++5BV6Sjh49qiFDhnj+PHfuXM2dO1fDhg3Tpk2bJElHjhzR+PHjVV1drV69eun222/Xli1bPP0CAAC0lc1Y6NYlt9ut6OhouVwuRUVFBey8o+dv1p6jbv3PA7fpzm/2Dth5AQDApXn95rPAAACA5RCAAqD5bn7LLKUBABDkCEAAAMByCEAAAMByCEAAAMByCEAAAMByCEAAAMByCEAAAMByCEAAAMByCEAAAMByCEABYBM7IQIAEEwIQAAAwHIIQAAAwHIIQAAAwHIIQAAAwHIIQAAAwHIIQAAAwHIIQAAAwHIIQAFk2AgIAICgQAAKAJuto0cAAAD8QQACAACWQwACAACWQwACAACWQwACAACWQwACAACWQwACAACWQwACAACWQwAKIMM+iAAABAUCUACwDyIAAMGFAAQAACynXQFo0aJFSkxMVEREhBwOhzZv3nze+vz8fDkcDkVERCgpKUlLlizxen7Pnj0aO3as+vfvL5vNpnnz5gWkXwAAAF/8DkCrV6/W5MmTNWvWLBUVFSkjI0MjR45UWVmZz/rS0lKNGjVKGRkZKioq0syZMzVp0iStWbPGU3P69GklJSVpzpw5iouLC0i/AAAArbEZ49+lu6mpqRo6dKgWL17saRs4cKDGjBmjnJycFvXTpk3TunXrVFJS4mnLysrSzp075XQ6W9T3799fkydP1uTJky+qX1/cbreio6PlcrkUFRXVpmPa4h8XvKedR1xaPjFF3xsYG7DzAgCAS/P67dcKUF1dnQoLC5WZmenVnpmZqYKCAp/HOJ3OFvUjRozQ9u3bVV9ff8n6BQAAaE2oP8XV1dVqbGxUbKz3KkdsbKwqKyt9HlNZWemzvqGhQdXV1erTp88l6VeSamtrVVtb6/mz2+2+YF8AAODq166LoG027xu/jTEt2i5U76s90P3m5OQoOjra8+jbt69f/QEAgKuTXwEoJiZGISEhLVZdqqqqWqzONIuLi/NZHxoaqp49e16yfiVpxowZcrlcnsfhw4fb1J/fvghhTWyECABAUPArAIWHh8vhcCgvL8+rPS8vT+np6T6PSUtLa1G/YcMGpaSkKCws7JL1K0l2u11RUVFej0uheQ3Kz+vJAQBAB/HrGiBJys7O1oQJE5SSkqK0tDQtXbpUZWVlysrKknRu1aW8vFwrVqyQdO6OrwULFig7O1sPPvignE6nli9frpUrV3rOWVdXp71793q+Li8vV3Fxsbp27aobbrihTf12pOZ34Yg/AAAEB78D0Lhx41RTU6PZs2eroqJCycnJys3NVUJCgiSpoqLCa2+exMRE5ebmasqUKVq4cKHi4+M1f/58jR071lNz9OhRDRkyxPPnuXPnau7cuRo2bJg2bdrUpn470pcrQB06DAAA0EZ+7wMUzC7VPkBjFxeo8NDnWvLTobor+cJ3tQEAgLbr8H2A4BsrQAAABBcCUAB0+uIiIPIPAADBgQAUCM0XQZOAAAAICgSgAPC8BcYaEAAAQYEAFADNt8GzESIAAMGBABQAti/WgCx0Qx0AAEGNABQAfn6kGQAA6GAEoACwcRE0AABBhQAUAF/eBk8CAgAgGBCAAogVIAAAggMBKABszStABCAAAIICASgAmq+BbiIBAQAQFAhAAeC5CLpjhwEAANqIABQAnrvgSUAAAAQFAlAAcBcYAADBhQAUAOwDBABAcCEABUTzChAAAAgGBKAAYAUIAIDgQgAKgOaLoLkGCACA4EAACoDmFaAm8g8AAEGBABQAzo9rJEnbSj/r4JEAAIC2IAAFgPtsgyTprzuPdvBIAABAWxCAAACA5RCAAACA5RCAAACA5RCAAACA5RCAAACA5RCAAACA5RCAAACA5RCAAACA5bQrAC1atEiJiYmKiIiQw+HQ5s2bz1ufn58vh8OhiIgIJSUlacmSJS1q1qxZo5tuukl2u1033XSTXn31Va/nn3jiCdlsNq9HXFxce4YPAAAszu8AtHr1ak2ePFmzZs1SUVGRMjIyNHLkSJWVlfmsLy0t1ahRo5SRkaGioiLNnDlTkyZN0po1azw1TqdT48aN04QJE7Rz505NmDBBP/7xj7V161avcw0aNEgVFRWex+7du/0dPgAAgGzGGL8+wjM1NVVDhw7V4sWLPW0DBw7UmDFjlJOT06J+2rRpWrdunUpKSjxtWVlZ2rlzp5xOpyRp3LhxcrvdeuONNzw1d911l7p3766VK1dKOrcC9Nprr6m4uNivCX6V2+1WdHS0XC6XoqKi2n2er+s//XXP15/MGR2w8wIAgEvz+u3XClBdXZ0KCwuVmZnp1Z6ZmamCggKfxzidzhb1I0aM0Pbt21VfX3/emq+fc//+/YqPj1diYqLuu+8+HTx40J/hAwAASPIzAFVXV6uxsVGxsbFe7bGxsaqsrPR5TGVlpc/6hoYGVVdXn7fmq+dMTU3VihUrtH79ei1btkyVlZVKT09XTU1Nq+Otra2V2+32egAAALTrImibzeb1Z2NMi7YL1X+9/ULnHDlypMaOHavBgwdr+PDhev31c287vfTSS632m5OTo+joaM+jb9++F5hZ+1zTJczz9Zm6xkvSBwAACBy/AlBMTIxCQkJarPZUVVW1WMFpFhcX57M+NDRUPXv2PG9Na+eUpMjISA0ePFj79+9vtWbGjBlyuVyex+HDh887v/Y6frre8/Xfdh29JH0AAIDA8SsAhYeHy+FwKC8vz6s9Ly9P6enpPo9JS0trUb9hwwalpKQoLCzsvDWtnVM69/ZWSUmJ+vTp02qN3W5XVFSU1+NSa2zy65pyAADQAfx+Cyw7O1svvPCCXnzxRZWUlGjKlCkqKytTVlaWpHOrLvfff7+nPisrS4cOHVJ2drZKSkr04osvavny5Zo6daqn5uGHH9aGDRv0u9/9Th9++KF+97vfaePGjZo8ebKnZurUqcrPz1dpaam2bt2qe++9V263WxMnTryI6QfGHTf07OghAAAAP4T6e8C4ceNUU1Oj2bNnq6KiQsnJycrNzVVCQoIkqaKiwmtPoMTEROXm5mrKlClauHCh4uPjNX/+fI0dO9ZTk56erlWrVunRRx/VY489puuvv16rV69Wamqqp+bIkSMaP368qqur1atXL91+++3asmWLp9+O1DnM728jAADoQH7vAxTMLtU+QL98uVBvfHDuGqacHw3W+G/1C9i5AQCwug7fBwi+jRz85XVIn1Sf6sCRAACAtiAABcDdN38ZgPrHRHbgSAAAQFsQgALAZrNp+MDe577u4LEAAIALIwAFSERYiCTpNBshAgBwxSMABUiX8HMB6Ew9AQgAgCsdAShAztY3SZKKyj7v4JEAAIALIQAFyLqd5z4CY2NJVQePBAAAXAgBCAAAWA4BKEBuvi7a8/Vnp+o6cCQAAOBCCEAB4kjo7vn60xO1HTgSAABwIQSgAElkA0QAAIIGAShA7vxGb8/XWw7WdOBIAADAhRCAAiSmW7jn679+cUcYAAC4MhGAAqRLeKjn6+2H2AsIAIArGQEIAABYDgEIAABYDgEogEI6fflZ8P9beKQDRwIAAM6HABRAB54e6fl66l92duBIAADA+RCAAshm+3IFKPQrq0EAAODKQgAKsB6R526H/3lGUgePBAAAtIYAFGCxURGSpCX5H3fwSAAAQGsIQAFWUuHu6CEAAIALIAAF2DM/HCzJ+44wAABwZSEABVjzh6L279mlg0cCAABaQwAKsK72cx+JcbK2oYNHAgAAWkMACrDO4SGSpDN1jR08EgAA0BoCUIB1+SIAuc+yAgQAwJWKABRgzQFIkj4od3XgSAAAQGsIQAEW+cU1QJL02P990IEjAQBYiTGmo4cQVAhAARYW8uW3tKjseMcNBEHv+Ok61Tc2dfQw9MbuCv3zsi2qcp/t6KFckDFGdQ2+v2ettX/V56fqVH78zAXrGpv8e6ExxuiT6lO8QH1FU5PR/xWX61DNqYCds8J1RmfrA3/9pTEm4Nd1PvbaB5r56u6Ane+P75cqLedtHfz0pKettqH9Y/7sVJ32HzsRiKFdsQhAl1j/6a/7/R+yI39INjYZfkhfASpcZ3Tr7DzdNe/dVmsam1p/sW929PgZvflBpefvdIXzE72w+aBfY/nln3eo4OMazf7bXk/b2fpGv0OAJO0+4tKE5Vu152j73x42xujDSreWvvtxi/9bE5Zv09An83TibL12HTmuN3ZXSJLe/KBS33j0Df3Xxv2qOtF6kBvyZJ7umPO2ak7WetqamozXXF/ZWqab/uNNbTlY0+Yx/+7Nfbpz7iY9//YBr3m8srVMRWWft+kcVe6z+r/icr1VcswzHmOMjvkZTF1n6r1eJL+u4EC1Nuyp9Ouc53OytsHnz5R1O4/q4VXFGvb7TQHpZ/+xE0rLeVvf/0N+u47/+NOTKjz0mc/nHnqlSAP/402V1Zy+mCF6uM7U609bDumVrWX67FRdQM7527/uVaX7rB5ft0eStOVgjb756Jt6/q39bTq++PBxjV+6xXPpxtAn8/T9P7yrj7/4t/L51wLRgaqTWvjOAZ2uC97rXdsVgBYtWqTExERFRETI4XBo8+bN563Pz8+Xw+FQRESEkpKStGTJkhY1a9as0U033SS73a6bbrpJr7766kX3e6UY8Nib6j/9da/HfUudyl5drJ+/tF3Pvvmh3thdof7TX9ev/7xDN/92g8YuLpDrTL0aGpt0tr5Rxhj9z/ul+vxUneoamuQ6Xa89R106W9+oQzWndPT4GZ2pa9SuI8c1Y+0uVZ+s1aZ9Vfr0RK0+/vSkXKfr1dhkdKDqRKsBp76xSZl/yNdPXtgq6dxv/sOfy9eHlW7P859Un9L+Yyc8v1l8UO7S5v2fqvorLxjN9hx1afl7pVr27kHP88YYrXB+on9etkX5H32qh17ZodLqUyr4uFqz/7pXmX/I18efntThz05r3c6javrKC8+eoy79z/ulrb7w+prXym1luv/FbTr1xbYEdQ1Nmva/u/S3XUf9+o/75gcVmvXqbi1796AeXLG91eDx5gcVmr5mV6u/eTU0NulA1UmfYzXG6FRtg+obm7SxpEqS9PGnp1R14qxWbStr8Rvo6Pmb9a1nNqrSdVYL3zmgCteXKxeVrrP689ZDSp/ztrJeLtSrReWqbWjUf/zfHj31eoknBJRWn9KU1cX6qA2/6dWcPPeD+sTZet38xAaNWfi+1/Pus/UXDM/jljq1eX+17n7+PW3e/6nPFa5N+6qUvbrYaz5fdf+L23TXvM16JvdDLXzngNdz7x2o1snaBr37UbXuWfC+fvnnHdpz1KWslwslSX/Y+JG+9fRbFwxvH1ae+34YYzRm0fv6/nP5+scF7+nhVUWa+epu1TY06ecvbT9vkPiq5o/GeS7vI0nSqm1levr1Es18dbd+uKigTedIn/O2Hl5VrJ+9tF2vbCuTJD362gdKfeYtrdt51FO356hLv3y5sNWxpee8pX/4z3wdqPL9d/7PL2zVv/6p0K9g9f6BamX9qbBFuNx15LiSH1+vxBm5njB68NOTumPO25q8urjN529W29DoNa8DVSf0atERGWO0Ye8xSdLhz87ornnv6okvgsDX/euK7frxfzu9frZI0vf+M19jFzt1+LOWIef1L8b+8tZDnrbiw8c1540PL/hzxNfK0Vf7bs8vEufT0HjufLO+WF36zy/+zUnSqdqGVj+t4N7FBXIerNH4pVu82rd/ci4Upjy9Ud//w7uefzfDn8vX79fv039u+KjFuYKFzfj56/7q1as1YcIELVq0SHfccYf++7//Wy+88IL27t2rfv36tagvLS1VcnKyHnzwQf3iF7/Q+++/r1/96ldauXKlxo4dK0lyOp3KyMjQk08+qR/+8Id69dVX9R//8R967733lJqa2q5+fXG73YqOjpbL5VJUVJQ/0/ZLbUOjvvnom5fs/MD5/Nv3v+H1Q69ZN3uoTlxgf6qMG2O0eX/1pRqaT+O/1Vcrtx2+YN3wgbHaWHLM53P9enTRwD7dtH5Py+ev6RKm46frvdoG9jn3/z81sYf+p+ATDYjrpht6d9XfdlW0OG+ZjxfEr/r1d6/X4c/OaN3Oo5oxcoBy3vjwi/H21o6y4236DX/YN3op/6NPdVv/7kpL6qlruoQrunOYztQ3Km/vMfWIDNerReXnPccPh1yrB+7or3sWeAfT63tF6iepCWpoatIzuR96PZc17Hotyf9YP729n978oFLHT9er4YsX5F9/93qtcB7SibMN6mYP1W++d4Niutr18acn1dBk1LtbhCLCOulMXaOeer3Ec87/uu9W7T7iUkRYiBZ8LaA+OSZZj73W8trI5nFI0s+/naiozmGqb2zS828fUFKvSD01JllHj5/V1L/s9Bzz7L0369//d5ck6Tvf6KUPK9yqOuH9i9g7U+/U0eNn9NGxE3p5yyH9cMi1mvvFC/ZPUvtpaL/uCg2xacSgOA147NzP7LSknrr7lniNGhynt0qqdGNsV8/39Dvf6KUfDblW13QJ07/88e+efm7qE6Wf3p6gPtdEKCI0RNGdw1R9sla/+vMOnaxt0JNjktWra7i+GRelTfuqtP2Tzz2h6rEf3KRu9lD94JY+OlnboLN1TTp+pk6hnTop/6NPdfxMnW7s3U0D+3RTn+jOCgux6f0DNapwndGeo251skkRYSFa4fwynD1+90367V+/XLH97wkOvfNhlVb9/dz/s9/eM0hp1/dU/56ROlRzSgeqTuqXf97hqU9J6K7th86tTHbvEqZff/cGz99x1rDr9cth1+uW2RskSQPiuunFf7lNp2ob1D8mUiE2mzpdgk9CuBSv334HoNTUVA0dOlSLFy/2tA0cOFBjxoxRTk5Oi/pp06Zp3bp1Kin58j9IVlaWdu7cKafTKUkaN26c3G633njjDU/NXXfdpe7du2vlypXt6teXyxWAJOnwZ6eV8ew7l7QPAACuNKU5o2SzBTYEXYrXb7/eAqurq1NhYaEyMzO92jMzM1VQ4HsZ1+l0tqgfMWKEtm/frvr6+vPWNJ+zPf1KUm1trdxut9fjcunbo4s+mTNau57I1K/uvP6y9QsAQEfaFyQXT4deuORL1dXVamxsVGxsrFd7bGysKit9XzRXWVnps76hoUHV1dXq06dPqzXN52xPv5KUk5Oj3/72t22e36UQFRGmf79rgP79rgEBOV9jk/F80OpXF++a03ZDY5NCQzqpqcmoUyebqk/W6pPqU4qLjlCf6M465j6rU7UN6tnVrms6h6nRGH1+qk6R9lCdqW/UZ6fqFH9NZ31+qk7us/Xq16OLjKSPKk8oqnOYenW1a89Rt25L7C7XmXodPX5WSb0i9VHlCZUfP6Pv3xSrM3WN+vRkra7pHK7QEJuamozsYSH6pPqUztQ36kDVSd2e1FOdw0N09PgZVZ+o1cHqU/rR0GtV13DuGpnw0E7qGWlXzalaHXPXKjEmUhFhnXS2vlFHj5/VwU9P6ds3xuivO4/qxyl9dcx9VvuOndD9aQkq++y0Otls2lhyTO4zDYrpGq7062P0111Hdec3eqm+0WjTvir16Bqu7l3CNSg+SrUNTcrf96mSekWq5mSdYrqF6/jpeuV/9KlGJsfJHhoie2gnzX/7gB77wUBVuWtlD+2kXeUuHag6qRNn6zUyuY8+rHQrJaGHdh45rrCQTuoWEaozdY3aWvqZjh4/o3+8NV7fSuypdz/6VCn9u+v9A9V6+8Nz11CNS+mr6C5hcp2uV+fwEN3Qu6tS+nfXqm2HdfN10TpV16gDx05oUHy0ig4fV+9udhV8XK3ahib17mZXUq+ucn5co672UH0zrpu+ldhDf915VI6E7nrq9RLd1CdKeyvcemTENz1/F3/bVaHEmEgNio/yvP0THtpJ9tBOOnG2QWEhNmUOilO/Hl30//39sK7r3lnlx896rukafG20ztQ3ynWmXj9OuU5bDn6mkgq3Ttc1qlc3u3p0Cde+YyfkSOius/WN2n/spOoam9QjMtzz1tCQftfodG2j9h070epbTokxkTry+WkN6dtdn52u04GqkwrpZPNcP/GtxB7qZg/94jokafuhz2UP7aQH7kjUkvyPFRtl1y3XXaPjZ+oVERaiiuNntL/K9zUyI5Pj1Nj05TUlIZ1sGnxttIoPH7+o/7upiT20tdT3Rbb+iI2yq+pErZr/+/9wyLUXfHus2Ve/Z7h4iTGRKq0O3F1sV4NJ/3CDEnpEdvQw2sb4oby83EgyBQUFXu1PPfWU+eY3v+nzmBtvvNE888wzXm3vvfeekWQqKiqMMcaEhYWZV155xavm5ZdfNna7vd39GmPM2bNnjcvl8jwOHz5sJBmXy9W2CQMAgA7ncrkC/vrt1wpQTEyMQkJCWqy6VFVVtVidaRYXF+ezPjQ0VD179jxvTfM529OvJNntdtnt9rZNDgAAWIZf1wCFh4fL4XAoLy/Pqz0vL0/p6ek+j0lLS2tRv2HDBqWkpCgsLOy8Nc3nbE+/AAAArfJ3yWjVqlUmLCzMLF++3Ozdu9dMnjzZREZGmk8++cQYY8z06dPNhAkTPPUHDx40Xbp0MVOmTDF79+41y5cvN2FhYeZ///d/PTXvv/++CQkJMXPmzDElJSVmzpw5JjQ01GzZsqXN/bbFpVhCAwAAl9aleP32OwAZY8zChQtNQkKCCQ8PN0OHDjX5+fme5yZOnGiGDRvmVb9p0yYzZMgQEx4ebvr3728WL17c4px/+ctfzDe/+U0TFhZmBgwYYNasWeNXv21BAAIAIPhcitdvv/cBCmaXcx8gAAAQGB2+DxAAAMDVgAAEAAAshwAEAAAshwAEAAAshwAEAAAshwAEAAAshwAEAAAshwAEAAAsx68PQw12zXs+ut3uDh4JAABoq+bX7UDu3WypAHTixAlJUt++fTt4JAAAwF8nTpxQdHR0QM5lqY/CaGpq0tGjR9WtWzfZbLaAndftdqtv3746fPjwVf0RG8zz6mGFOUrM82pihTlKzLM1xhidOHFC8fHx6tQpMFfvWGoFqFOnTrruuusu2fmjoqKu6n+wzZjn1cMKc5SY59XECnOUmKcvgVr5acZF0AAAwHIIQAAAwHIIQAFgt9v1+OOPy263d/RQLinmefWwwhwl5nk1scIcJeZ5OVnqImgAAACJFSAAAGBBBCAAAGA5BCAAAGA5BCAAAGA5BKAAWLRokRITExURESGHw6HNmzd39JB8ysnJ0W233aZu3bqpd+/eGjNmjPbt2+dVY4zRE088ofj4eHXu3Fl33nmn9uzZ41VTW1ur3/zmN4qJiVFkZKTuueceHTlyxKvm888/14QJExQdHa3o6GhNmDBBx48fv9RT9CknJ0c2m02TJ0/2tF0t8ywvL9dPf/pT9ezZU126dNGtt96qwsJCz/PBPs+GhgY9+uijSkxMVOfOnZWUlKTZs2erqakpqOf47rvv6u6771Z8fLxsNptee+01r+cv55zKysp09913KzIyUjExMZo0aZLq6uou+Tzr6+s1bdo0DR48WJGRkYqPj9f999+vo0ePXlXz/Lpf/OIXstlsmjdvXlDNsy1zLCkp0T333KPo6Gh169ZNt99+u8rKyq7cORpclFWrVpmwsDCzbNkys3fvXvPwww+byMhIc+jQoY4eWgsjRowwf/zjH80HH3xgiouLzejRo02/fv3MyZMnPTVz5swx3bp1M2vWrDG7d+8248aNM3369DFut9tTk5WVZa699lqTl5dnduzYYb773e+aW265xTQ0NHhq7rrrLpOcnGwKCgpMQUGBSU5ONj/4wQ8u63yNMWbbtm2mf//+5uabbzYPP/ywp/1qmOdnn31mEhISzL/8y7+YrVu3mtLSUrNx40Zz4MCBq2aeTz31lOnZs6f529/+ZkpLS81f/vIX07VrVzNv3rygnmNubq6ZNWuWWbNmjZFkXn31Va/nL9ecGhoaTHJysvnud79rduzYYfLy8kx8fLx56KGHLvk8jx8/boYPH25Wr15tPvzwQ+N0Ok1qaqpxOBxe5wj2eX7Vq6++am655RYTHx9v/vCHPwTVPC80xwMHDpgePXqYRx55xOzYscN8/PHH5m9/+5s5duzYFTtHAtBF+ta3vmWysrK82gYMGGCmT5/eQSNqu6qqKiPJ5OfnG2OMaWpqMnFxcWbOnDmemrNnz5ro6GizZMkSY8y5H1phYWFm1apVnpry8nLTqVMn8+abbxpjjNm7d6+RZLZs2eKpcTqdRpL58MMPL8fUjDHGnDhxwtx4440mLy/PDBs2zBOArpZ5Tps2zXz7299u9fmrYZ6jR482/+///T+vth/96Efmpz/9qTHm6pjj119MLueccnNzTadOnUx5ebmnZuXKlcZutxuXy3VJ5+nLtm3bjCTPL5BX0zyPHDlirr32WvPBBx+YhIQErwAUbPP0Ncdx48Z5/l/6ciXOkbfALkJdXZ0KCwuVmZnp1Z6ZmamCgoIOGlXbuVwuSVKPHj0kSaWlpaqsrPSaj91u17BhwzzzKSwsVH19vVdNfHy8kpOTPTVOp1PR0dFKTU311Nx+++2Kjo6+rN+XX//61xo9erSGDx/u1X61zHPdunVKSUnRP/3TP6l3794aMmSIli1b5nn+apjnt7/9bb311lv66KOPJEk7d+7Ue++9p1GjRl01c/y6yzknp9Op5ORkxcfHe2pGjBih2tpar7dSLxeXyyWbzaZrrrlG0tUzz6amJk2YMEGPPPKIBg0a1OL5YJ9nU1OTXn/9dX3jG9/QiBEj1Lt3b6Wmpnq9TXYlzpEAdBGqq6vV2Nio2NhYr/bY2FhVVlZ20Kjaxhij7Oxsffvb31ZycrIkecZ8vvlUVlYqPDxc3bt3P29N7969W/TZu3fvy/Z9WbVqlXbs2KGcnJwWz10t8zx48KAWL16sG2+8UevXr1dWVpYmTZqkFStWeMbXPOavCqZ5Tps2TePHj9eAAQMUFhamIUOGaPLkyRo/frxnbM3jPd/4r+Q5ft3lnFNlZWWLfrp3767w8PDLPu+zZ89q+vTp+ud//mfPh2NeLfP83e9+p9DQUE2aNMnn88E+z6qqKp08eVJz5szRXXfdpQ0bNuiHP/yhfvSjHyk/P98ztittjpb6NPhLxWazef3ZGNOi7Urz0EMPadeuXXrvvfdaPNee+Xy9xlf95fq+HD58WA8//LA2bNigiIiIVuuCfZ5NTU1KSUnRM888I0kaMmSI9uzZo8WLF+v+++9vdYzBNM/Vq1fr5Zdf1iuvvKJBgwapuLhYkydPVnx8vCZOnNjq+IJpjq25XHO6EuZdX1+v++67T01NTVq0aNEF64NpnoWFhfqv//ov7dixw+++gmWezTcl/OM//qOmTJkiSbr11ltVUFCgJUuWaNiwYa0e25FzZAXoIsTExCgkJKRF6qyqqmqRUK8kv/nNb7Ru3Tq98847uu666zztcXFxknTe+cTFxamurk6ff/75eWuOHTvWot9PP/30snxfCgsLVVVVJYfDodDQUIWGhio/P1/z589XaGioZwzBPs8+ffropptu8mobOHCg566Lq+Hv85FHHtH06dN13333afDgwZowYYKmTJniWdm7Gub4dZdzTnFxcS36+fzzz1VfX3/Z5l1fX68f//jHKi0tVV5enmf1p3l8wT7PzZs3q6qqSv369fP8PDp06JD+7d/+Tf379/eML5jnGRMTo9DQ0Av+PLrS5kgAugjh4eFyOBzKy8vzas/Ly1N6enoHjap1xhg99NBDWrt2rd5++20lJiZ6PZ+YmKi4uDiv+dTV1Sk/P98zH4fDobCwMK+aiooKffDBB56atLQ0uVwubdu2zVOzdetWuVyuy/J9+d73vqfdu3eruLjY80hJSdFPfvITFRcXKykp6aqY5x133NFiG4OPPvpICQkJkq6Ov8/Tp0+rUyfvH1MhISGe3zivhjl+3eWcU1pamj744ANVVFR4ajZs2CC73S6Hw3FJ5yl9GX7279+vjRs3qmfPnl7PXw3znDBhgnbt2uX18yg+Pl6PPPKI1q9ff1XMMzw8XLfddtt5fx5dkXP065JptNB8G/zy5cvN3r17zeTJk01kZKT55JNPOnpoLfzyl7800dHRZtOmTaaiosLzOH36tKdmzpw5Jjo62qxdu9bs3r3bjB8/3uftt9ddd53ZuHGj2bFjh/mHf/gHn7cy3nzzzcbpdBqn02kGDx7cIbfBN/vqXWDGXB3z3LZtmwkNDTVPP/202b9/v/nzn/9sunTpYl5++eWrZp4TJ0401157rec2+LVr15qYmBjz7//+70E9xxMnTpiioiJTVFRkJJnnnnvOFBUVee5+ulxzar6l+Hvf+57ZsWOH2bhxo7nuuusCdnv4+eZZX19v7rnnHnPdddeZ4uJir59JtbW1V808ffn6XWDBMM8LzXHt2rUmLCzMLF261Ozfv988//zzJiQkxGzevPmKnSMBKAAWLlxoEhISTHh4uBk6dKjntvIrjSSfjz/+8Y+emqamJvP444+buLg4Y7fbzXe+8x2ze/dur/OcOXPGPPTQQ6ZHjx6mc+fO5gc/+IEpKyvzqqmpqTE/+clPTLdu3Uy3bt3MT37yE/P5559fhln69vUAdLXM869//atJTk42drvdDBgwwCxdutTr+WCfp9vtNg8//LDp16+fiYiIMElJSWbWrFleL5DBOMd33nnH5//FiRMnXvY5HTp0yIwePdp07tzZ9OjRwzz00EPm7Nmzl3yepaWlrf5Meuedd66aefriKwBd6fNsyxyXL19ubrjhBhMREWFuueUW89prr13Rc7QZY4x/a0YAAADBjWuAAACA5RCAAACA5RCAAACA5RCAAACA5RCAAACA5RCAAACA5RCAAACA5RCAAACA5RCAAACA5RCAAACA5RCAAACA5RCAAACA5fz/KbQu7eNBO4EAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "device = 'cuda'\n",
    "\n",
    "tn = 8.\n",
    "tf = 12.\n",
    "nb_epochs = 1\n",
    "lr = 1e-3\n",
    "gamma = .5\n",
    "nb_bins = 100\n",
    "\n",
    "model = Nerf(hidden_dim=128).to(device)\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr=lr)\n",
    "scheduler = torch.optim.lr_scheduler.MultiStepLR(optimizer, milestones=[5, 10], gamma=gamma)\n",
    "\n",
    "\n",
    "training_loss = training(model, optimizer, scheduler, tn, tf, nb_bins, 1, dataloader_warmup, model_name=pth_file, device=device)\n",
    "plt.plot(training_loss)\n",
    "plt.show()\n",
    "training_loss = training(model, optimizer, scheduler, tn, tf, nb_bins, nb_epochs, dataloader, model_name=pth_file, device=device)\n",
    "plt.plot(training_loss)\n",
    "plt.show()"
   ]
  }
 ],
 "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": "3.11.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}