771 lines
221 KiB
Plaintext
771 lines
221 KiB
Plaintext
{
|
|
"cells": [
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"# MNIST digits dataset\n",
|
|
"\n",
|
|
"- First load and view the MNIST digits dataset\n",
|
|
"- There are 60000 images in this dataset, but we will only view the first 25 of them:\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 2,
|
|
"metadata": {
|
|
"slideshow": {
|
|
"slide_type": "slide"
|
|
}
|
|
},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"test_images shape (10000, 28, 28) train_images shape (60000, 28, 28)\n"
|
|
]
|
|
},
|
|
{
|
|
"data": {
|
|
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAxoAAAMpCAYAAACDrkVRAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/H5lhTAAAACXBIWXMAAA9hAAAPYQGoP6dpAABrtklEQVR4nO3debyN1d//8c9BODgOx3zCQaYQMlPRgL6SocEUZWpWSCQNKESKjBGKokiEJiSFRGVIXxIh8zFlOI4h09m/P+7f3f299mf5nsu29r72Pvv1fDx6PO71fqx97U/fe9nbp2uva8X4fD6fAAAAAIBFmbwuAAAAAEDGQ6MBAAAAwDoaDQAAAADW0WgAAAAAsI5GAwAAAIB1NBoAAAAArKPRAAAAAGBdFjeT0tLSJDk5WeLi4iQmJibYNSEC+Hw+SU1NlcTERMmUKbj9KusP/kK5/kRYg3Bi/cFrfAfDS1ey/lw1GsnJyVKsWDErxSFj2bt3rxQtWjSo78H6w+WEYv2JsAZhxvqD1/gOhpfcrD9XbXBcXJyVgpDxhGJtsP5wOaFaG6xBmLD+4DW+g+ElN2vDVaPBrTJcTijWBusPlxOqtcEahAnrD17jOxhecrM22AwOAAAAwDoaDQAAAADW0WgAAAAAsI5GAwAAAIB1NBoAAAAArKPRAAAAAGAdjQYAAAAA62g0AAAAAFhHowEAAADAOhoNAAAAANbRaAAAAACwjkYDAAAAgHU0GgAAAACso9EAAAAAYB2NBgAAAADraDQAAAAAWJfF6wIAXJ3q1aur7KmnnlLZQw895Bh/8MEHas7YsWNVtn79+quoDgAARCvuaAAAAACwjkYDAAAAgHU0GgAAAACso9EAAAAAYB2bwf9D5syZVRYfHx/QtUybcXPkyOEYlytXTs3p1q2byt58802VtWvXzjH++++/1Zxhw4ap7JVXXtHFImJUrVpVZUuWLFFZ7ty5Vebz+RzjBx98UM1p3ry5yvLly3cFFQL23XHHHSr78MMPHeMGDRqoOVu3bg1aTYh8L730kspM35GZMun/Jnvrrbc6xsuXL7dWF5CRcEcDAAAAgHU0GgAAAACso9EAAAAAYF3E79EoXry4yrJmzeoY16tXT825+eabVZYnTx6V3XfffYEXl459+/apbMyYMSq75557VJaamuoY//rrr2oOvxmNfLVq1XKM586dq+aY9hH578cQ0Wvm/Pnzao5pP0adOnVU5n+In+lauLz69eurzP9/+3nz5oWqnLBXs2ZNla1Zs8aDShDJOnXq5Bj37dtXzUlLS3N1LdNnLACNOxoAAAAArKPRAAAAAGAdjQYAAAAA62g0AAAAAFgXUZvBTYeVffvttyoL9JC9YPPfZGY6LOjUqVMq8z+YSkTkwIEDjvHx48fVHA6rCl/+hzeKiFSrVk1lM2bMcIyLFCkS8Htu27bNMR4+fLiaM2vWLJX98MMPKvNfu0OHDg24rmjkf9iXiEiZMmUc42jdDG46HK1kyZIqS0pKcoxjYmKCVhMyBv81kz17do8qQTiqXbu2yjp06OAYmw4GrVixoqvr9+7d2zFOTk5Wc0wPKvL/e4CIyE8//eTqPcMBdzQAAAAAWEejAQAAAMA6Gg0AAAAA1tFoAAAAALAuojaD79mzR2VHjx5VWTA3g5s24Jw4cUJlt912m8r8T0+ePn26tboQWd555x2VtWvXLqjv6b/ZPFeuXGqO6TR508blypUrW6srGj300EMqW716tQeVhB/TAw8eeeQRlflvkNyyZUvQakLkadiwocqefvrpdF9nWkd33323yg4dOhRYYQgLbdq0Udno0aNVlj9/fsfY9NCJZcuWqaxAgQIqe+ONN9Kty3R907Xatm2b7rXCBXc0AAAAAFhHowEAAADAOhoNAAAAANbRaAAAAACwLqI2gx87dkxlffr0UZn/xq1ffvlFzRkzZoyr99ywYYNj3KhRIzXn9OnTKjOdFNmjRw9X74mMpXr16ipr2rSpytycbGzarP3555+r7M0331SZ/ymkpj8XphPmb7/9dpVxCvPVMZ1+jf8xZcoUV/P8T7pH9DKdpjx16lSVuXlQjGnD7u7duwMrDJ7IksX5V9saNWqoOZMnT1ZZjhw5VLZixQrHeNCgQWrOypUrVZYtWzaVzZ492zFu3LixmmOydu1aV/PCFd92AAAAAKyj0QAAAABgHY0GAAAAAOsiao+Gyfz581X27bffOsapqalqTpUqVVTWtWtXlfn/1t20H8Pkt99+U9mjjz7q6rWIbFWrVnWMlyxZoubkzp1bZT6fT2ULFy50jE2H+jVo0EBlL730ksr8f/t+5MgRNefXX39VWVpamsr895j4HwYoIrJ+/XqVRSPT4YaFChXyoJLI4PbAVdOfK0Snjh07qiwxMTHd15kOWvvggw9slAQPdejQwTF2u+/L9Jnif7DfyZMnXV3LdCCgmz0Z+/btU9n777/v6j3DFXc0AAAAAFhHowEAAADAOhoNAAAAANbRaAAAAACwLuI3g5u42ayTkpLi6lqPPPKIY/zxxx+rOabNsogOZcuWVZn/IZKmza1//fWXyg4cOKAy/01gp06dUnO+/PJLV5lNsbGxjvGzzz6r5rRv3z6oNUSKu+66S2X+//tFK9Om+JIlS7p67f79+22XgwiQP39+lXXp0kVlpu/lEydOOMaDBw+2Vhe8YTpA74UXXnCMTQ9aefvtt1VmeoiK283f/l588cWAXte9e3eVmR7cEkm4owEAAADAOhoNAAAAANbRaAAAAACwjkYDAAAAgHUZcjO4GwMHDlRZ9erVVeZ/6nLDhg3VnK+//tpaXQhf2bJlU5n/yfEievOv6WT6hx56SGVr165VWaRsGi5evLjXJYStcuXKuZr322+/BbmS8GP682PaIP7HH3+ozPTnChlPiRIlHOO5c+cGfK2xY8c6xt99913A10Lo9e/fX2X+G79FRM6fP+8YL168WM3p27evys6ePZtuDdmzZ1eZ6cRv03diTEyMY2x6GMGCBQvSrSHScEcDAAAAgHU0GgAAAACso9EAAAAAYB2NBgAAAADronYz+OnTp1Xmfwq4iMj69esd48mTJ6s5pg1lpo2948ePd4xNp1UifN14440qM5367K9FixYqW758uZWakHGsWbPG6xICljt3bpX961//UlmHDh0cY9MmShPT6b/+pzwjY/JfR5UrV3b1uqVLl6ps9OjRVmpC8OXJk0dlTz75pMpMf4/y3/zdsmXLgOsoXbq0Y/zhhx+qOaYHCZnMmTPHMR4+fHjAdUUS7mgAAAAAsI5GAwAAAIB1NBoAAAAArIvaPRomO3bsUFmnTp0c46lTp6o5Dz74oKssZ86cjvEHH3yg5hw4cCC9MuGRkSNHqsz/AB4Rvf8i0vdjZMqk/3tEWlqaB5VkbAkJCdauVaVKFcfYtE5Nh48WLVpUZVmzZnWM27dvr+aY1ojp8KuffvrJMT537pyakyWL/lpat26dypDxmH5LP2zYsHRft3LlSpV17NhRZSkpKQHVhdDz/9wREcmfP7+r13bv3t0xLliwoJrTuXNnlTVv3lxllSpVcoxz5cql5pj2iZiyGTNmOMamvcIZEXc0AAAAAFhHowEAAADAOhoNAAAAANbRaAAAAACwjs3g6Zg3b55jvG3bNjXHtEn4jjvuUNlrr73mGCclJak5Q4YMUdn+/fvTrRN23X333SqrWrWqykwbvj777LNglOQZ08Zv/3/vDRs2hKiayGPaFG1aNxMnTnSMX3jhhYDf0/9QM9Nm8IsXL6rszJkzKtu8ebNj/N5776k5pgNKTQ9BOHTokGO8b98+NSc2NlZlW7ZsURkiW4kSJVQ2d+7cgK71559/qsx/rSGynD9/XmVHjhxRWYECBVS2c+dOx/hqDkdOTk52jE+ePKnmFClSRGV//fWXyj7//POA64hk3NEAAAAAYB2NBgAAAADraDQAAAAAWEejAQAAAMA6NoNfoU2bNqmsdevWKmvWrJnK/E8Vf+yxx9ScMmXKqKxRo0ZXUiIsMG1INZ1UevjwYZV9/PHHQanJtmzZsqls4MCBrl777bffOsb9+vWzUVKG9OSTT6ps9+7dKqtXr56199yzZ49jPH/+fDXn999/V9mPP/5orQaTRx991DE2beQ0bexFxtO3b1+VmR484Yab08MRWU6cOKEy08nxX3zxhcoSEhIc4x07dqg5CxYsUNm0adNUduzYMcd41qxZao5pM7hpXrTijgYAAAAA62g0AAAAAFhHowEAAADAOhoNAAAAANaxGdwC06al6dOnq2zKlCmOcZYs+n/++vXrq+zWW29V2bJly1zXh+A5d+6cyg4cOOBBJenz3/z90ksvqTl9+vRRmen05hEjRjjGp06dusrqosvrr7/udQmeuOOOO9KdE+jp0AhfVatWVVnjxo0DupZpE+/WrVsDuhYiy08//aQy0wMlbPL/O1mDBg3UHNNDDHioxf/hjgYAAAAA62g0AAAAAFhHowEAAADAOvZoXKHKlSur7P7771dZzZo1VWbak+Fv8+bNKluxYoXL6hBqn332mdclGJl+E+2//6JNmzZqjun3z/fdd5+1uoD0zJs3z+sSYNnXX3+tsrx587p6rf8hkp06dbJREuCK/+G9pv0YPp9PZRzY93+4owEAAADAOhoNAAAAANbRaAAAAACwjkYDAAAAgHVsBv8P5cqVU9lTTz3lGN97771qTuHChQN6v0uXLqnMdNibafMRgismJsZV1rJlS5X16NEjGCVd1jPPPKOyl19+WWXx8fGO8YcffqjmPPTQQ/YKAwARyZcvn8rcfq+9/fbbjjGHgyKUFi9e7HUJEY87GgAAAACso9EAAAAAYB2NBgAAAADraDQAAAAAWBcVm8FNm7XbtWunMv+N3yIiJUqUsFbH2rVrHeMhQ4aoOeF60nS0MZ30acpMa2vMmDGO8XvvvafmHD16VGV16tRR2YMPPugYV6lSRc0pWrSoyvbs2aMy/01t/pssgVAzPWChbNmyKvM/HRrhberUqY5xpkyB/zfNVatWXW05QMDuvPNOr0uIeNzRAAAAAGAdjQYAAAAA62g0AAAAAFhHowEAAADAuojfDF6oUCGVVahQwTEeN26cmlO+fHlrNfz0008qe+ONN1S2YMECx5gTvyNf5syZVfbkk086xvfdd5+ac/LkSZWVKVMmoBpMmyW/++47lfXv3z+g6wPBYnrAwtVsHEboVa1aVWUNGzZ0jE3fdefPn1fZ+PHjVXbo0KHAiwOuUqlSpbwuIeLxiQ4AAADAOhoNAAAAANbRaAAAAACwLmz3aCQkJKjsnXfeUZnp96E2f1Pn//v3ESNGqDn+B6GJiJw9e9ZaDQi91atXq2zNmjUqq1mzZrrXMh3qZ9pbZOJ/sN+sWbPUnB49eri6FhAJ6tatq7Jp06aFvhC4kidPHpWZPvP87d+/X2W9e/e2URJgzffff+8Ym/aQsd/2v+OOBgAAAADraDQAAAAAWEejAQAAAMA6Gg0AAAAA1nmyGbx27doq69Onj2Ncq1YtNefaa6+1VsOZM2dUNmbMGJW99tprjvHp06et1YDwtW/fPpXde++9KnvsscdU9tJLLwX0nqNHj1bZhAkTHOPt27cHdG0gHMXExHhdAgBc1qZNmxzjbdu2qTmmBxBdd911Kjty5Ii9wiIIdzQAAAAAWEejAQAAAMA6Gg0AAAAA1tFoAAAAALDOk83g99xzj6vMjc2bN6vsiy++cIwvXryo5phO+D5x4kRANSA6HDhwQGUDBw50lQEQWbhwoWPcqlUrjyqBLVu2bFHZqlWrHOObb745VOUAQeX/gCARkSlTpqhsyJAhKnv66acdY9PfXzMi7mgAAAAAsI5GAwAAAIB1NBoAAAAArKPRAAAAAGBdjM/n86U36eTJkxIfHx+KehBhUlJSJHfu3EF9D9YfLicU60+ENQgz1h+8xndwaJn+t549e7bKGjZsqLJPP/3UMe7cubOac/r06auoLvTcrD/uaAAAAACwjkYDAAAAgHU0GgAAAACs8+TAPgAAACCSnDx5UmWtW7dWmenAvieeeMIxNh3umxEP8eOOBgAAAADraDQAAAAAWEejAQAAAMA6Gg0AAAAA1rEZHAAAAAiAaYP4008/7SqLBtzRAAAAAGAdjQYAAAAA62g0AAAAAFjnqtHw+XzBrgMRKhRrg/WHywnV2mANwoT1B6/xHQwvuVkbrhqN1NTUqy4GGVMo1gbrD5cTqrXBGoQJ6w9e4zsYXnKzNmJ8LtqRtLQ0SU5Olri4OImJibFSHCKbz+eT1NRUSUxMlEyZgvsLPNYf/IVy/YmwBuHE+oPX+A6Gl65k/blqNAAAAADgSrAZHAAAAIB1NBoAAAAArKPRAAAAAGAdjQYAAAAA62g0XBg4cKDExMQ4/ilfvrzXZSHKjB8/XkqUKCHZs2eX2rVry88//+x1SYhCw4YNk5iYGOnZs6fXpSCKrFixQpo1ayaJiYkSExMj8+fP97okRJHU1FTp2bOnJCUlSWxsrNSrV0/WrFnjdVkRgUbDpYoVK8qBAwf++WflypVel4Qo8vHHH0uvXr1kwIABsn79eqlSpYrceeedcvjwYa9LQxRZs2aNvPPOO1K5cmWvS0GUOX36tFSpUkXGjx/vdSmIQg8//LAsWbJEpk+fLhs3bpTGjRtLw4YNZf/+/V6XFvZoNFzKkiWLFC5c+J9/8ufP73VJiCIjR46URx55RDp37iwVKlSQiRMnSo4cOeS9997zujREiVOnTkn79u1l8uTJkjdvXq/LQZRp0qSJDB48WO655x6vS0GUOXv2rMydO1eGDx8u9evXl9KlS8vAgQOldOnSMmHCBK/LC3s0Gi5t27ZNEhMTpVSpUtK+fXvZs2eP1yUhSpw/f17WrVsnDRs2/CfLlCmTNGzYUFavXu1hZYgm3bp1k6ZNmzrWIQBkdBcvXpRLly5J9uzZHXlsbCy/bnGBRsOF2rVry7Rp02TRokUyYcIE2blzp9xyyy2ujl4HrtZff/0lly5dkkKFCjnyQoUKycGDBz2qCtFk1qxZsn79ehk6dKjXpQBASMXFxUndunVl0KBBkpycLJcuXZIZM2bI6tWr5cCBA16XF/ayeF1AJGjSpMk//3flypWldu3akpSUJLNnz5auXbt6WBkABNfevXulR48esmTJEvVf9AAgGkyfPl26dOki1157rWTOnFmqVasm7dq1k3Xr1nldWtjjjkYA8uTJI2XLlpXt27d7XQqiQP78+SVz5sxy6NAhR37o0CEpXLiwR1UhWqxbt04OHz4s1apVkyxZskiWLFlk+fLlMmbMGMmSJYtcunTJ6xIBIKiuu+46Wb58uZw6dUr27t0rP//8s1y4cEFKlSrldWlhj0YjAKdOnZIdO3ZIkSJFvC4FUSBr1qxSvXp1Wbp06T9ZWlqaLF26VOrWrethZYgGd9xxh2zcuFE2bNjwzz81atSQ9u3by4YNGyRz5sxelwgAIZEzZ04pUqSIHD9+XBYvXiwtWrTwuqSwx0+nXOjdu7c0a9ZMkpKSJDk5WQYMGCCZM2eWdu3aeV0aokSvXr2kY8eOUqNGDalVq5aMGjVKTp8+LZ07d/a6NGRwcXFxUqlSJUeWM2dOyZcvn8qBYDl16pTjVwQ7d+6UDRs2SEJCghQvXtzDyhANFi9eLD6fT8qVKyfbt2+XPn36SPny5fkOdoFGw4V9+/ZJu3bt5OjRo1KgQAG5+eab5ccff5QCBQp4XRqiRJs2beTIkSPSv39/OXjwoFStWlUWLVqkNogDQEa0du1aue222/4Z9+rVS0REOnbsKNOmTfOoKkSLlJQU6devn+zbt08SEhLkvvvukyFDhsg111zjdWlhL8bn8/m8LgIAAABAxsIeDQAAAADW0WgAAAAAsI5GAwAAAIB1NBoAAAAArKPRAAAAAGAdjQYAAAAA62g0AAAAAFhHowEAAADAOlcng6elpUlycrLExcVJTExMsGtCBPD5fJKamiqJiYmSKVNw+1XWH/yFcv2JsAbhxPqD1/gOhpeuZP25ajSSk5OlWLFiVopDxrJ3714pWrRoUN+D9YfLCcX6E2ENwoz1B6/xHQwvuVl/rtrguLg4KwUh4wnF2mD94XJCtTZYgzBh/cFrfAfDS27WhqtGg1tluJxQrA3WHy4nVGuDNQgT1h+8xncwvORmbbAZHAAAAIB1NBoAAAAArKPRAAAAAGAdjQYAAAAA62g0AAAAAFhHowEAAADAOhoNAAAAANbRaAAAAACwjkYDAAAAgHU0GgAAAACso9EAAAAAYB2NBgAAAADraDQAAAAAWEejAQAAAMA6Gg0AAAAA1tFoAAAAALAui9cFAPgfo0ePVln37t0d402bNqk5d999t8p2795trzAAABC2li5dqrKYmBiV3X777aEox4E7GgAAAACso9EAAAAAYB2NBgAAAADraDQAAAAAWMdmcAvi4uJUlitXLpU1bdrUMS5QoICaM3LkSJWdO3fuKqpDOCpRooTKOnTooLK0tDTH+Prrr1dzypcvrzI2gyM9ZcuWVdk111yjsvr16zvGb7/9tprjv05tW7Bggcratm3rGJ8/fz6oNSD4TOuvXr16jvFrr72m5tx0001BqwkIR2+99ZZj7P/nRETkgw8+CFU5/xV3NAAAAABYR6MBAAAAwDoaDQAAAADWsUcjHf6/pe/bt6+aU7duXZVVqlQpoPcrUqSIyvwPbUPkO3LkiMpWrFihsubNm4eiHGQwFStWdIw7deqk5rRq1UplmTLp//aUmJjoGJv2Y/h8vius8MqY/hxMnDjRMe7Zs6eac/LkyWCVhCCIj49X2XfffecYHzx4UM0pXLiwykzzgEg0bNgwlT3++OOO8YULF9Qc0yF+XuCOBgAAAADraDQAAAAAWEejAQAAAMA6Gg0AAAAA1kXtZnDTIWemzYTt27d3jGNjY9WcmJgYle3du1dlqampjrHp8LXWrVurzHRA1pYtW1SGyHH69GmVccgebBk6dKhjfNddd3lUSfA89NBDjvG7776r5vzwww+hKgchYtr4zWZwZGR16tRRmf/hlitXrlRzZs+eHbSargR3NAAAAABYR6MBAAAAwDoaDQAAAADW0WgAAAAAsC5Dbgb3P1309ddfV3PatGmjsri4uIDeb9u2bSq78847Vea/ece0oTt//vyuMkS2PHnyqKxKlSqhLwQZ0pIlSxxjt5vBDx8+rDL/Tdam08NNp4Wb1KtXzzFu0KCBq9cB/8v08BXAhvr166vsxRdfdIzbtWun5hw7dsxaDabrV6pUSWU7duxwjHv37m2tBtu4owEAAADAOhoNAAAAANbRaAAAAACwjkYDAAAAgHUZcjP4Pffc4xg//PDD1q7tvwFHRKRRo0YqM50MXrp0aWt1ILLlyJFDZcWLFw/oWjVr1lSZ6UEDnDwePSZMmOAYz58/39XrLly4oDKbJyznzp3bMd60aZOak5iY6Opa/v9Oa9euDbguRA6fz6ey7Nmze1AJMppJkyaprEyZMo5xhQoV1BzTqdyBeuGFF1SWL18+lT3yyCOO8a+//mqtBtu4owEAAADAOhoNAAAAANbRaAAAAACwLkPu0WjVqlVAr9u1a5fK1qxZ4xj37dtXzTHtxzC5/vrrA6oLGU9ycrLKpk2bprKBAwemey3TnBMnTqhs3LhxLipDRnDx4kXH2O1nVLD5H2SaN2/egK+1b98+x/jcuXMBXwuRrUaNGir78ccfPagEkezMmTMq898TZHM/UNWqVVWWlJSkMtOBqJG0L4k7GgAAAACso9EAAAAAYB2NBgAAAADraDQAAAAAWJchN4P7H2Ty6KOPqjlff/21yrZv366yw4cPW6urUKFC1q6FjGfQoEEqc7MZHAhHbdu2VZn/Z3NsbGzA1+/fv3/Ar0V48n+IgYhISkqKYxwfH6/mXHfddUGrCRmT6fv2hhtuUNnvv//uGF/NwXg5c+Z0jE0PFzId5mt6sMGcOXMCriPUuKMBAAAAwDoaDQAAAADW0WgAAAAAsI5GAwAAAIB1GXIzuP+py+GyobZu3bpel4AIkymT878FmE4IBUKpffv2Knv++edVVrp0aZVdc801Ab3nhg0bVHbhwoWAroXwdeLECZV9//33jvHdd98domqQURQrVkxl/g+mEDE/jOCpp55yjI8cORJwHSNHjnSMW7Vqpeb4//1VROSmm24K+D3DAXc0AAAAAFhHowEAAADAOhoNAAAAANbRaAAAAACwLkNuBrepe/fujrH/yY5XwnTqpL9Vq1apbPXq1QG/JyKb/+Zvn8/nUSWIJCVKlHCMH3zwQTWnYcOGAV375ptvVlmg6/LkyZMqM20s/+qrr1R29uzZgN4TQMZWqVIlx3jevHlqTv78+VU2duxYlS1fvjygGnr37q2yTp06pfu6IUOGBPR+4Yw7GgAAAACso9EAAAAAYB2NBgAAAADromKPRo4cOVRWoUIFlQ0YMEBld911V7rX9z9UTcTdwWqmg1k6d+6sskuXLqV7LQDRyf/3yCIin332mWNcvHjxUJVzRfwPYxMRmTRpkgeVIJLly5fP6xIQAlmy6L+ydujQQWXvvvuuY+z272imQ5X79evnGPsfuicikpCQoDLTYXwxMTGO8QcffKDmvPPOOyqLdNzRAAAAAGAdjQYAAAAA62g0AAAAAFhHowEAAADAuojfDH7NNdeo7MYbb3SM586dq+YUKVJEZaYDoPw3bJsOz/vXv/6lMtMGdH+mjU333nuvykaPHq2y8+fPp3t9ANHJf9Oh//hqBPrwC5O7775bZU2aNFHZwoULA7o+okPz5s29LgEh0LZtW5VNmTJFZf4HiJo+n7Zv366yGjVqpJu1aNFCzbn22mtVZvo75pEjRxzjLl26qDkZEXc0AAAAAFhHowEAAADAOhoNAAAAANbRaAAAAACwLqI2g2fNmlVlpo3Yn376abrXeuWVV1T27bffquyHH35wjE0nQJpeZzqt11+BAgVUNnToUJXt2bNHZfPnz3eMz507l+77IfL4b7x1u+m2fv36Khs3bpyVmhBeNm3apLJbb73VMTadnrt48WKV/f3339bq6tq1q8qefvppa9dHdPjuu+8cY9MDBJDxtGnTRmVTp05V2YULF1R24sQJx/iBBx5Qc44fP66yESNGqKxBgwaOsWnDuOlhG/4b0kVE8ufP7xjv3btXzfH/7BYR2bFjh8oiCXc0AAAAAFhHowEAAADAOhoNAAAAANbRaAAAAACwLmw3g5tO/DZt4O7Tp0+61zKdKjt27FiV+W8gEtEbtr/66is154YbblCZ6eTu4cOHO8amDeOmUyc//PBDlX3zzTeO8euvv67mmDY7mWzYsMHVPISe/+Zv0wYzE9MJ8xUqVHCMN2/eHHhhCGu7d+92jIcMGRLyGgYOHKgyNoPjSpkehuLP9PeFpKQklfn/uUD4euyxx1RmWguDBw9WmWnTuBumz6d33nnHMa5bt25A1xbRm8b9H3QgEvkbv024owEAAADAOhoNAAAAANbRaAAAAACwLmz2aGTOnNkxHjRokJrTu3dvlZ0+fVplzz//vGM8a9YsNce0H8N0EIv/IWc33nijmrNt2zaVPfHEEyrz/z1e7ty51Zx69eqprH379ipr3ry5Y7xkyRI1x8R0QEzJkiVdvRahN3HiRMfY9LtVtx599FHHuGfPngFfC0jPnXfe6XUJyAAuXryY7hzTgWnZsmULRjkIkQULFqjMdBiz6e80gfI/UE/E3eHL7dq1U5npIFV/+/btc1dYhOOOBgAAAADraDQAAAAAWEejAQAAAMA6Gg0AAAAA1oXNZnD/jaqmjd9nzpxRmWlz7Ndff+0Y16lTR83p3Lmzypo0aaKy2NhYx/jVV19Vc0yHw7jZoHTy5EmVLVq0yFXmv/nogQceSPf9RESeeeYZV/MQHrZs2eJ1CfCI6RCyxo0bq+zbb79V2dmzZ4NS0+WYPk9Hjx4d0hqQMflvCjZ9JpYvX15lpoddPPnkk9bqQnAF+/MjPj5eZa1atVKZ/0N7TAfqzZ49215hGRB3NAAAAABYR6MBAAAAwDoaDQAAAADW0WgAAAAAsC7G5/P50pt08uRJ48YZmw4cOOAYFyhQQM05d+6cykwbw3LmzOkYly5dOuC6Bg4c6BgPHTpUzbl06VLA1490KSkpxhPObQrF+osUf/zxh8quu+46V6/NlMn53xVMfy5MG93CWSjWn0ho1uDNN9/sGL/44otqTqNGjVRWsmRJldk8LTchIcExvuuuu9ScsWPHqiwuLi7da5s2rTdv3lxl3333XbrX8kJGWn+RYtSoUSozPYygUKFCKvv777+DUZKn+A4OTL9+/VQ2aNAglR05csQxrlmzppoTLSd8m7hZf9zRAAAAAGAdjQYAAAAA62g0AAAAAFhHowEAAADAurA5GfzgwYOOsWkzeLZs2VRWpUqVdK/91VdfqWzFihUqmz9/vsp27drlGEfzxm9477ffflNZqVKlXL02LS3NdjmwaNy4cY5xpUqVXL3uueeeU1lqaqqVmkT0BvRq1aqpOS6eKSIiIsuWLXOMJ0yYoOaE68ZvhC/T+jt//rwHlSAcJSUlqezhhx9WmWkdTZo0yTGO5o3fgeKOBgAAAADraDQAAAAAWEejAQAAAMC6sNmjUb9+fce4ZcuWao7pt8GHDx9W2XvvvecYHz9+XM3h95uIRP6/FxURadasmQeVIFw88cQTXpdg/Bz+/PPPVdajRw/HOCMeoIbQMx0Y1qJFC5XNmzcvFOUgzCxZskRlpn0bM2bMUNmAAQOCUlM04Y4GAAAAAOtoNAAAAABYR6MBAAAAwDoaDQAAAADWhc1mcP8DpqZPn67mmDIgmmzevFllv//+u8quv/76UJQDizp16uQYP/3002pOx44dg1rDjh07VHbmzBnH+Pvvv1dzTA8p2LRpk73CgP+vdevWKjt37pzKTJ+LiE5Tp05V2aBBg1S2YMGCUJQTdbijAQAAAMA6Gg0AAAAA1tFoAAAAALCORgMAAACAdTE+n8+X3qSTJ09KfHx8KOpBhElJSTGeymoT6w+XE4r1J+LNGsyWLZvK/DeMi4gMHjxYZXnz5nWM58+fr+aYTss1bYY8ePDgf6kyumXk9ReuZs2apTLTwy+aN2+ust27dwelJi/xHQwvuVl/3NEAAAAAYB2NBgAAAADraDQAAAAAWEejAQAAAMA6NoPjqrARDV5iMy68xPqD1/gOhpfYDA4AAADAEzQaAAAAAKyj0QAAAABgHY0GAAAAAOtoNAAAAABYR6MBAAAAwDoaDQAAAADW0WgAAAAAsI5GAwAAAIB1NBoAAAAArKPRAAAAAGAdjQYAAAAA61w1Gj6fL9h1IEKFYm2w/nA5oVobrEGYsP7gNb6D4SU3a8NVo5GamnrVxSBjCsXaYP3hckK1NliDMGH9wWt8B8NLbtZGjM9FO5KWlibJyckSFxcnMTExVopDZPP5fJKamiqJiYmSKVNwf4HH+oO/UK4/EdYgnFh/8BrfwfDSlaw/V40GAAAAAFwJNoMDAAAAsI5GAwAAAIB1NBoAAAAArKPRAAAAAGAdjYYLQ4cOlZo1a0pcXJwULFhQWrZsKVu3bvW6LESRFStWSLNmzSQxMVFiYmJk/vz5XpeEKDJhwgSpXLmy5M6dW3Lnzi1169aVhQsXel0WogifgQgXw4YNk5iYGOnZs6fXpUQEGg0Xli9fLt26dZMff/xRlixZIhcuXJDGjRvL6dOnvS4NUeL06dNSpUoVGT9+vNelIAoVLVpUhg0bJuvWrZO1a9fK7bffLi1atJDffvvN69IQJfgMRDhYs2aNvPPOO1K5cmWvS4kYPN42AEeOHJGCBQvK8uXLpX79+l6XgygTExMj8+bNk5YtW3pdCqJYQkKCvPHGG9K1a1evS0GU4TMQXjh16pRUq1ZN3n77bRk8eLBUrVpVRo0a5XVZYY87GgFISUkRkf/5ogWAaHLp0iWZNWuWnD59WurWret1OQAQEt26dZOmTZtKw4YNvS4lomTxuoBIk5aWJj179pSbbrpJKlWq5HU5ABASGzdulLp168rff/8tuXLlknnz5kmFChW8LgsAgm7WrFmyfv16WbNmjdelRBwajSvUrVs32bRpk6xcudLrUgAgZMqVKycbNmyQlJQUmTNnjnTs2FGWL19OswEgQ9u7d6/06NFDlixZItmzZ/e6nIhDo3EFnnrqKfniiy9kxYoVUrRoUa/LAYCQyZo1q5QuXVpERKpXry5r1qyR0aNHyzvvvONxZQAQPOvWrZPDhw9LtWrV/skuXbokK1askHHjxsm5c+ckc+bMHlYY3mg0XPD5fPL000/LvHnzZNmyZVKyZEmvSwIAT6Wlpcm5c+e8LgMAguqOO+6QjRs3OrLOnTtL+fLlpW/fvjQZ6aDRcKFbt27y0UcfyYIFCyQuLk4OHjwoIiLx8fESGxvrcXWIBqdOnZLt27f/M965c6ds2LBBEhISpHjx4h5WhmjQr18/adKkiRQvXlxSU1Plo48+kmXLlsnixYu9Lg1Rgs9AeCUuLk7tyc2ZM6fky5ePvbou8HhbF2JiYoz51KlTpVOnTqEtBlFp2bJlctttt6m8Y8eOMm3atNAXhKjStWtXWbp0qRw4cEDi4+OlcuXK0rdvX2nUqJHXpSFK8BmIcHLrrbfyeFuXaDQAAAAAWMc5GgAAAACso9EAAAAAYB2NBgAAAADraDQAAAAAWEejAQAAAMA6Gg0AAAAA1tFoAAAAALDO1cngaWlpkpycLHFxcZc9vA7RxefzSWpqqiQmJkqmTMHtV1l/8BfK9SfCGoQT6w9e4zsYXrqS9eeq0UhOTpZixYpZKQ4Zy969e6Vo0aJBfQ/WHy4nFOtPhDUIM9YfvMZ3MLzkZv25aoPj4uKsFISMJxRrg/WHywnV2mANwoT1B6/xHQwvuVkbrhoNbpXhckKxNlh/uJxQrQ3WIExYf/Aa38Hwkpu1wWZwAAAAANbRaAAAAACwjkYDAAAAgHU0GgAAAACso9EAAAAAYB2NBgAAAADrXB3YBwAAECnKli2rskWLFjnGmTNnVnOSkpKCVhMQjbijAQAAAMA6Gg0AAAAA1tFoAAAAALCORgMAAACAdWwGBwAAEWvs2LEqa9OmjcoSEhIc4y+++CJoNQH4H9zRAAAAAGAdjQYAAAAA62g0AAAAAFhHowEAAADAuqjdDF6hQgWV3X333Sp79NFHHeM1a9aoOb/88our9xw1apRjfP78eVevAwAgGhUqVMgx/vTTT9WcOnXqqMzn86ls06ZNjnHXrl2vsjoA6eGOBgAAAADraDQAAAAAWEejAQAAAMA6Gg0AAAAA1kXFZvDHHntMZW+++abKcuXKle61rrvuOpW1bdvWVR3+G8m/++47V68DELlMnyumU4v//vtvx7h69epqTlxcnMrat2+vsmXLljnG+/fvT69M1w4ePKiyBQsWqGzt2rXW3hPRoWzZsirz/66uXbu2q2v169dPZf5r8ujRo1dQHTKSmJgYlc2cOVNld911l2NsepDQvn377BWWAXFHAwAAAIB1NBoAAAAArKPRAAAAAGBdVOzR+OSTT1T26quvqszNHo2r4X/QkOl32l9//XVQawAQWv3791dZ7969g/qe//rXv4J6fX+m38Nv3rxZZf6/gTb9JnrXrl3W6kJkSUhIUJn/b+TdMv1unn2R+F+xsbEqu+mmm1Tm//dC02frlClT7BWWAXFHAwAAAIB1NBoAAAAArKPRAAAAAGAdjQYAAAAA66JiM/ixY8dUNmDAAJWNGDFCZTly5HCM9+zZo+YUL17cVR158uRxjE2bitgMjnCTlJSkMv+NdO3atVNznnjiCVfX//LLLx3jzp07X0F14e/ee++1di3TAWP//ve/rV1/69atKitXrpxj7P85JiJy4403qqxSpUoqGzJkiGNsqp3N4NHBdDjfRx99pDLTwWr+TH/GTIdIAv/rzJkzKtu2bZvKrr32Wse4QIECQaspo+KOBgAAAADraDQAAAAAWEejAQAAAMA6Gg0AAAAA1kXFZnCTiRMnquzxxx9XWZUqVRzjkydPWqth3Lhx1q4FXKmGDRuqzLSp0rTROz4+3jH2+XwB11GnTp2AXxsJ7rzzTpWZNsL+8ccf6V7LtIHxwIEDgRUWoLi4OJVt3LhRZW4ektG8eXOV+T8cABnTgw8+qDLTmvnqq68cY9P39P79++0Vhqg1fvx4ld16662O8fXXXx+iajIO7mgAAAAAsI5GAwAAAIB1NBoAAAAArKPRAAAAAGBd1G4GNxk8eLDKXnzxRce4atWq1t4va9as1q4F/KcpU6ao7IYbbnCMa9asGfD1U1NTHeMPP/xQzVmzZo3KZs6cqbK///474DoiwY4dO1xlkeLuu+9WmZuN3yIi586dc4wnT55spSaEt1WrVqnM9F1qOhX+mWeecYzZ+I1g+fnnn9Od07p1a5X17dtXZaF+SEc4444GAAAAAOtoNAAAAABYR6MBAAAAwDr2aPyHOXPmqGzlypWO8ddff63m+P/23S3TnpD7778/oGshOuTLl09lQ4cOVVmXLl1UduzYMcd43bp1as6wYcNUtmnTJpWdPXvWMd6zZ48uFhHHtG9szJgxjvFDDz0U8PXr1q3rGG/YsCHgayF8tWjRwjGuXbu2mmM65POTTz5RWUbfw4XwFhMT4xibPiNNB4++8847Qasp0nBHAwAAAIB1NBoAAAAArKPRAAAAAGAdjQYAAAAA69gM/h/at2+vsipVqjjGlSpVsvZ+/hvNgfS8/PLLKuvatavKxo4dqzL/wydPnTplrzBEnNtuu01lDz74oMo6deqU7rUuXLigsu7du6tsy5Yt7opDxMiTJ4/KbrnlloCudfz4cZXt27cvoGuZ9OjRwzEuVqyYq9f17t3bWg2ILKaHFvjj8OX/jjsaAAAAAKyj0QAAAABgHY0GAAAAAOtoNAAAAABYFxWbwcuXL6+yefPmqax06dIqy5IleP8TffbZZ0G7NsJbjhw5VNa3b1+V+W/O7dmzp5rz3XffqWzx4sUq44Td6FWrVi2Vff311yrLnDlzQNc3bZg0nRZ/6dKlgK6P8GX6/2n16tUd40yZ9H/TTEtLU9mKFSsCquGZZ55xNe/pp592jJOSkly97tlnn1VZ0aJFHeP9+/e7uhYQbbijAQAAAMA6Gg0AAAAA1tFoAAAAALCORgMAAACAdVGxGfz6669XWcmSJVUWzI3fJqYNbP6b1ZAxvfTSSyozbQafPXu2Y2zawMsmb6SndevWKgt047eJ6WTcL7/8UmVr1651jD///HM1x/Sgjk2bNl1FdQimBg0aqMz/ZHDTxm/TwwL++uuvdN+vatWq6b6fiEjz5s3Tvdbp06dVZjqJvFy5ciqbM2eOY9y2bVs1Z/fu3enWAGR03NEAAAAAYB2NBgAAAADraDQAAAAAWBcVezRMv/l97rnnVPb666+rLHv27EGpSUSkSJEiQbs2wlu/fv1UZjr0bObMmY4x+zEQiE8//VRlpr1rNWvWVFn+/Pmt1VGjRo3/OhYRGTBggMpGjRqlsuHDhzvGhw8fvrrikK64uDiVmfY7+ktOTlbZ9OnTVbZ9+3aVlS1b1jHu06ePmtOiRQuVmfZ7+O9xGzFihJoTHx+vsm+//dbVPGQ8MTExjrHpexr/HXc0AAAAAFhHowEAAADAOhoNAAAAANbRaAAAAACwLio2g5uMGTNGZdu2bVNZnjx50r2W6aC/cePGqSx37tzuikOG9/PPP6vMtDHWfx2dPXtWzVmyZIm9wpAhrVq1SmVNmzZVWfHixVXmvxm8UKFCas69996rsi5duqjMf2OlSaZM+r9/9erVS2XVq1d3jO+44w41x3RQHAJ38803q+ytt95K93WTJ09W2auvvqoy09p68803HeO77rpLzUlNTVWZ/2GnIiK9e/d2jMuUKaPmTJw40dX1ly5d6hhzOF/GxObvq8cdDQAAAADW0WgAAAAAsI5GAwAAAIB1NBoAAAAArIvazeAmCxcuDOh1pg2OpUuXVln//v0d46pVq6o5SUlJKmOTWfiqXbu2yn755RfH+Pz582pOkyZNVNa9e3eVvfzyy47xnDlzXNWwZcsWXSyQjj179rjK/Jk+O5ctW6ayp59+2jGuVauW++L8NGjQwDH23+grok8Px9WpXLlyQK8zbfw2MZ1gb/p882c6GXz58uUqq1OnjmO8cuVKV3WZTqY3rTdEp3//+99elxDWuKMBAAAAwDoaDQAAAADW0WgAAAAAsI5GAwAAAIB1bAa3IGvWrCrz3/htcuHCBZVdunTJSk24OkWKFFHZF198oTLTScrPPPOMYzxjxgw159ixYyoznSbvvxk8V65cak5CQoLKAK99+OGHKvv4448d42+++UbNqV+/fkDvZ3oAB+zKkyePykwPQ1mwYEG61zI9DKVEiRLpXv/ZZ59Vc0wbv8uWLauyjz766L9e+3LXN20GB/7Xjh07vC4hrHFHAwAAAIB1NBoAAAAArKPRAAAAAGAdezQsGDx4cECve/fdd1W2b9++qy0HFqxfv15luXPnVlnfvn1VZtqT4UaPHj3SnWP6TfumTZsCej8g1C5evOgYr1u3Ts0JdI/GH3/8EdDrcHV8Pp+rzI20tLR0r2U6NNB0qGT27NlVtnPnTsf4lltuUXNSUlLSrROAe9zRAAAAAGAdjQYAAAAA62g0AAAAAFhHowEAAADAurDdDJ4vXz6VTZ06VWUzZ850ldliOsjt0UcfDehan3766dWWgyAZM2aMyl566SVX80yZv23btqmsTJkyKtu9e7dj3K9fPzXn5MmT6b4fMi7TZ9IjjzziGG/ZskXNmT17dtBqupzMmTM7xlWqVAn4Wv4by3/88ceArwV3TAfx9enTR2UtWrRwjOvUqaPmmA7si4uLS7eGhx56SGWmg/f++usvlQ0cONAx3r9/f7rvB6QnW7ZsXpcQ1rijAQAAAMA6Gg0AAAAA1tFoAAAAALCORgMAAACAdWG7Gdy0obZZs2YqK1u2rMqSk5MdY9OGr+3bt6usevXq6V7/ueeeU3NMJ0abjBgx4r/WifAxdOhQlV24cEFlN954o8oaNmyY7vXz5s2rsi+//FJlvXv3doxN6xbRo3DhwipbtGiRym644QbH2LTegq1QoUIq69Wrl2N8++23B3z933//3TFeuXJlwNeCO6bPwDNnzqgsR44cjvEPP/yg5gR6erhJamqqykwPO1i4cKG19wT+11133aWysWPHelBJeOKOBgAAAADraDQAAAAAWEejAQAAAMA6Gg0AAAAA1oXtZnDTRpqSJUuqrG7duipbtmyZY7xr1y41Z/PmzSq75ZZbVObmpFLTpjbTSbwDBgxwjP/+++90r43w8eabb3pdAqLcqFGjVOa/8dvE9Nm5detWlZ09ezbda8XGxqrM9JAM/43fIu4+T02nPJs2+3bv3j3da8GudevWqaxdu3Yq8////a233hrwe77//vuO8caNG9WcX375RWXLly8P+D0RnQ4dOqSy3377zTGuWLFiqMrJMLijAQAAAMA6Gg0AAAAA1tFoAAAAALAubPdo/PjjjypbvXq1yqZPn66yt99+2zEuUaKEmmPKAnX8+HGVVahQwdr1AUBEZOnSpSpr3bp1uq9bv369yky/a09JSUn3WvHx8SozHVwZKNN+jHvuuUdl/AY/PJgOGjVlQLg7f/68ytzspW3UqJHKOLDv/3BHAwAAAIB1NBoAAAAArKPRAAAAAGAdjQYAAAAA68J2M7jJs88+q7Js2bKpLFeuXOley7R50XTwkD/TZknTRiAAsG3JkiUqmzVrlsratm2b7rVsbuB26+LFi46x6QDCuXPnquynn34KVkkAcFkbNmxwjKtXr67muPk7ZzTjjgYAAAAA62g0AAAAAFhHowEAAADAOhoNAAAAANZF1GZwk3PnzqnsjTfeCOhaDzzwwNWWAwBBs2vXLpV17txZZZ999pljfPvtt6s5f/zxh8qaN2+ebg1btmxJd46IyLfffpvua/03WgJAOBkyZIhjXKlSJTVn9uzZoSonInFHAwAAAIB1NBoAAAAArKPRAAAAAGAdjQYAAAAA62J8Pp8vvUknT56U+Pj4UNSDCJOSkiK5c+cO6nuw/nA5oVh/IqxBmLH+4DW+g+ElN+uPOxoAAAAArKPRAAAAAGAdjQYAAAAA62g0AAAAAFhHowEAAADAOhoNAAAAANbRaAAAAACwjkYDAAAAgHU0GgAAAACso9EAAAAAYB2NBgAAAADraDQAAAAAWOeq0fD5fMGuAxEqFGuD9YfLCdXaYA3ChPUHr/EdDC+5WRuuGo3U1NSrLgYZUyjWBusPlxOqtcEahAnrD17jOxhecrM2Ynwu2pG0tDRJTk6WuLg4iYmJsVIcIpvP55PU1FRJTEyUTJmC+ws81h/8hXL9ibAG4cT6g9f4DoaXrmT9uWo0AAAAAOBKsBkcAAAAgHU0GgAAAACso9EAAAAAYB2NBgAAAADraDRcmDBhglSuXFly584tuXPnlrp168rChQu9LgtRYuDAgRITE+P4p3z58l6XhSjCZyC8tn//funQoYPky5dPYmNj5YYbbpC1a9d6XRaixIoVK6RZs2aSmJgoMTExMn/+fK9LihhZvC4gEhQtWlSGDRsmZcqUEZ/PJ++//760aNFCfvnlF6lYsaLX5SEKVKxYUb755pt/xlmy8EcXocNnILx0/Phxuemmm+S2226ThQsXSoECBWTbtm2SN29er0tDlDh9+rRUqVJFunTpIvfee6/X5UQUHm8boISEBHnjjTeka9euXpeCDG7gwIEyf/582bBhg9elAP/gMxCh8vzzz8sPP/wg33//vdelABITEyPz5s2Tli1bel1KROCnU1fo0qVLMmvWLDl9+rTUrVvX63IQJbZt2yaJiYlSqlQpad++vezZs8frkhCl+AxEqH322WdSo0YNadWqlRQsWFBuvPFGmTx5stdlAXCBRsOljRs3Sq5cuSRbtmzy+OOPy7x586RChQpel4UoULt2bZk2bZosWrRIJkyYIDt37pRbbrlFUlNTvS4NUYTPQHjlzz//lAkTJkiZMmVk8eLF8sQTT0j37t3l/fff97o0AOngp1MunT9/Xvbs2SMpKSkyZ84cmTJliixfvpwvWoTciRMnJCkpSUaOHMnPVhAyfAbCK1mzZpUaNWrIqlWr/sm6d+8ua9askdWrV3tYGaIRP526MtzRcClr1qxSunRpqV69ugwdOlSqVKkio0eP9rosRKE8efJI2bJlZfv27V6XgijCZyC8UqRIEdXQXn/99fyEFIgANBoBSktLk3PnznldBqLQqVOnZMeOHVKkSBGvS0EU4zMQoXLTTTfJ1q1bHdkff/whSUlJHlUEwC2ekelCv379pEmTJlK8eHFJTU2Vjz76SJYtWyaLFy/2ujREgd69e0uzZs0kKSlJkpOTZcCAAZI5c2Zp166d16UhSvAZCC8988wzUq9ePXnttdekdevW8vPPP8ukSZNk0qRJXpeGKHHq1CnHrwh27twpGzZskISEBClevLiHlYU/Gg0XDh8+LA899JAcOHBA4uPjpXLlyrJ48WJp1KiR16UhCuzbt0/atWsnR48elQIFCsjNN98sP/74oxQoUMDr0hAl+AyEl2rWrCnz5s2Tfv36yauvviolS5aUUaNGSfv27b0uDVFi7dq1ctttt/0z7tWrl4iIdOzYUaZNm+ZRVZGBzeAAAAAArGOPBgAAAADraDQAAAAAWEejAQAAAMA6Gg0AAAAA1tFoAAAAALCORgMAAACAdTQaAAAAAKxzdWBfWlqaJCcnS1xcnMTExAS7JkQAn88nqampkpiYKJkyBbdfZf3BXyjXnwhrEE6sP3iN72B46UrWn6tGIzk5WYoVK2alOGQse/fulaJFiwb1PVh/uJxQrD8R1iDMWH/wGt/B8JKb9eeqDY6Li7NSEDKeUKwN1h8uJ1RrgzUIE9YfvMZ3MLzkZm24ajS4VYbLCcXaYP3hckK1NliDMGH9wWt8B8NLbtYGm8EBAAAAWEejAQAAAMA6Gg0AAAAA1tFoAAAAALCORgMAAACAdTQaAAAAAKyj0QAAAABgHY0GAAAAAOtoNAAAAABYR6MBAAAAwDoaDQAAAADWZfG6AAAAAJtKlSqlsqFDhzrG99xzj5pTuXJllW3ZssVeYUCU4Y4GAAAAAOtoNAAAAABYR6MBAAAAwDoaDQAAAADWsRkcAABErHr16qls0aJFKjty5IhjPH78eDXn0KFD9goDwB0NAAAAAPbRaAAAAACwjkYDAAAAgHU0GgAAAACsYzM44IEHH3xQZY0bN1ZZ1apVHeNy5cq5uv6PP/6osmbNmjnGKSkprq4FhFLOnDlVtmzZMpUlJiY6xjfddJOas2vXLltlIUw0bdpUZXPmzFHZxIkTVfbiiy86xmfOnLFXGAAj7mgAAAAAsI5GAwAAAIB1NBoAAAAArGOPBmBZ/vz5HeMpU6aoOf77JURETpw4obJVq1Y5xqbfnN96660qu/nmm1W2evVqx7hChQpqDhAI//0SIiIFChRI93XHjx9X2W233aay6tWrq2zr1q2O8dGjR9N9P0Se0qVLO8azZ89Wc5YvX66yZ599VmVpaWn2CgPgCnc0AAAAAFhHowEAAADAOhoNAAAAANbRaAAAAACwjs3gFpg2nWXNmlVl119/vWPcvn17V9ffsmWLyipWrOiyOoTaokWLHOMSJUqoOcOHD1fZG2+8obJjx46l+37ly5dX2c8//6yysmXLOsb9+/dXc1599dV03w8ZQ6VKlVTWvXt3lSUlJaV7Lf+1JSJSvHjxdF83bNgwlZkeUhATE6Oy/fv3O8amz1xEluzZs6vM/2EaGzduVHNat26tMjZ+w4aEhATHuE2bNmrOCy+8oDLTAzL8vfTSSyobOnToFVQXGbijAQAAAMA6Gg0AAAAA1tFoAAAAALCORgMAAACAdWwG/w8NGjRQmf+GSdOce+65R2WmzYv+fD6fq7rKlCmjss2bNzvGnPLsjUaNGqnsxhtvdIxNJ9n269fPWg2mhwWMGjVKZf4bzzp37qzmsBk8etx+++0q69q1a0DXOnfunMpmzJiR7ns+//zzrq5v+qycNm2aY8zJ4JFv0KBBKqtdu7ZjbPo+PHnyZNBqQvSoU6eOyt566y3HuFatWmqO6fPJzd/vTOvd9GAN03d1JOGOBgAAAADraDQAAAAAWEejAQAAAMA6Gg0AAAAA1kX8ZvAiRYqobObMmY5xqVKlXF0rPj5eZTlz5nSMTZu8161bp7Jq1aq5ek83MmXS/aB/XfBGliz6j9D27dsd41mzZoWqnH/MmTNHZf6bwU2n8ObOnVtlbLSMfAMHDlRZnz59XL32/fffd4yPHDmi5rz55psqM82rWrWqY7x48WI1J3/+/K6uZVrjiBzZsmVTWYcOHVS2bNkyx3jfvn3BKglRxPQ5M3nyZJVdf/31jrHps2j+/PkqW7Bggcoeeughx7hVq1ZqjmlDetasWVV2/vx5lYUr7mgAAAAAsI5GAwAAAIB1NBoAAAAArIuoPRoNGzZUmek3dcWKFQtaDaaD8f766y+VmX7/l5iY6BhPnTpVzSlatKirOvwP7IM3vvvuO5X5H9h35syZUJXzD9MBav4KFSqksgceeEBlEydOtFITvGPa0xUbG6uy3bt3q+zFF190jA8cOODqPUuXLq2yF154wTEuUKCAmnP69GmVmfaY/P33367qQHh67rnnVJYrVy6V+a8/wAbTHgr//RgiIl9//bVjfNdddwX8ntu2bXOMTX+nNf0d0FTXr7/+GnAdocYdDQAAAADW0WgAAAAAsI5GAwAAAIB1NBoAAAAArIuozeCmzWOBbvw2bZbt27evyn788UfHeOvWra6uf/ToUZX16NHDMXa78XvXrl0qe/DBB129FsEVrhtS//zzT5X99ttvjnHFihXVnDJlygStJnjHdLjdv/71L5WZHnYxbNgwx/jJJ59Uc0yHnY4cOVJlTZs2dYyPHTum5gwZMkRlEyZMUBkiW+PGjVX2ww8/qGz9+vWhKAdR5uzZs67mmTaNB5PpgFzTA4ciCXc0AAAAAFhHowEAAADAOhoNAAAAANbRaAAAAACwLmw3g5s2itWpUyega+3Zs0dlps3Upo1oNrnd/O3PtBkp0jcHIbguXLigsosXL3pQCcLBhg0bVOb/oAsR82bw22+/3TFu1KiRmvPWW2+prHjx4unW9corr6hs7Nix6b4OkeXmm29Wmen7/IYbbrD2nrfeeqvKjhw54hj7PyAD0SMmJsZVdvz4ccc4e/bsas51112nsk6dOqmsevXqjvHBgwfVnHbt2qls//79Kosk3NEAAAAAYB2NBgAAAADraDQAAAAAWEejAQAAAMC6sN0M/uyzz6osR44crl67atUqx9i04dDmxu+8efOqzHTqbv369dO9ln/tIiJfffVVYIUhamXLlk1lpk1s/lJTU4NRDjx27tw5lZlOoDVJTEx0jOfOnavmmDZR+nw+lb377ruO8fz5813VgMjWoUMHlf3+++8q27lzZ7rXMm2yHTFihMpM38v+fw569+6t5owfPz7dGhD5KlasqDLTZ1avXr0cY9PfTf03eV9O27ZtHeM5c+a4el2k444GAAAAAOtoNAAAAABYR6MBAAAAwLqw3aMxadIkleXPn19lKSkpKnvggQccY9OhKDY9/vjjKhs0aFC6rzMdFtS6dWuVBbt+ZDwlSpRQWbly5dJ93aJFiwJ6P9OfzSpVqqisbt26Kvvkk08c461btwZUA67M7t27g3p9096yN9980zHeu3dvUGtAeOjSpYvK/L+nRcx7ibJmzeoYDxgwQM157LHHVLZ48WKV3XXXXY7x1KlT1ZwdO3aoLNDPRYSvo0ePqiwuLk5lNWrUcIzd7kc7c+aMyjZv3nwlJWYY3NEAAAAAYB2NBgAAAADraDQAAAAAWEejAQAAAMC6sN0MbjoUypSFWrNmzVTWv39/V6+9ePGiYzxx4kQ1h43f+G9MB/EVLVpUZfXq1Qvo+qY1uW7dOpVVq1bNMU5ISFBzihUrpjLTgYClS5d2jE0HcuHqZM6cWWW33HKLykwbHd348ssvVWb6rER08D8MLUsW/VcN/+/Dy/H/rDFtzHZ78NnHH3/sGN98881qTr9+/VTGZvCMx3RgX506dVTm//3qv4Yu59NPP1UZm8EBAAAAwBIaDQAAAADW0WgAAAAAsI5GAwAAAIB1YbsZPFzNnz9fZaZTIU26d+/uGJtOP0dkiY2NVVnBggUdY//NjCLmTWe33357uu+XPXt2lZk2tQXKdK34+Ph0X/fee++pzLRB+K+//lLZrl273BWHgM2aNUtl9957r8rcfpbZeh0ypsKFC6c7Z8uWLa6u9dtvvznGL730UkA1mUyYMEFlGzdutHZ9RJYff/xRZZUqVQroWq+99trVlpNhcEcDAAAAgHU0GgAAAACso9EAAAAAYB2NBgAAAADr2AyeDv8NPZky6d4sLS3N1bWWL19upSYEn2mT98CBA1VmOv24fPny1uo4efKkY2w6Wdt0wq7pJF5/U6ZMUZnpZPD169eney14JzExUWWdO3d2jO+77z41x7SB2/T/619//fW/XltEPwABSM/+/ftdzTN95tmyb9++oF0bGcMNN9zgGF/N3wGjFXc0AAAAAFhHowEAAADAOhoNAAAAANaxR+M/ZM2aVWU33nijY2z6LZ7pt849evRQ2bZt266iOoSS6WDGRo0aqezcuXMq8z+obufOnWrOggULXF3L/zA702+KTQdflS1bVmV//vmnY9yrVy8159SpUypDeLvjjjtU9uqrr6b7OtPBZ+PGjVNZy5YtHWPTHo3Nmzen+36IHjExMf91HC4aNGigsmDuCUHkOXv2rGNs+jvgsmXLVHb+/PlglRRxuKMBAAAAwDoaDQAAAADW0WgAAAAAsI5GAwAAAIB1UbsZPEeOHCrr0KGDykwbgP3NnDlTZR9++KHKONQlcjRu3Fhlpk3d9957r8o2bNhgrQ7/g/def/11Nefaa69V2eHDh1XWunVrx5iN35Hn1ltvVdmYMWPSfV3z5s1V9s0336iscOHCKuvfv3+61/d/aAGim/8DUkwPTPHCNddc4xg//vjjas706dNDVQ7CjOmw3a5duzrGR44cUXMmTJigMj4T/w93NAAAAABYR6MBAAAAwDoaDQAAAADW0WgAAAAAsC4qNoPHxcWpbPLkySq7//77073WM888ozLTabps/I5sps2LJ06cUNmmTZusvWf27NlV9sknnzjGTZs2VXNMJ4q3bdtWZevXr7+K6hAOTA+niI+PV9ny5csd4y+++ELN8d8YKyJy9913p3t90ynPpg2SiF7+J8UfOHBAzTE9fMW0qTZQpvXtf/0SJUqoOR07drRWA8KX6XNz8eLFKvN/2Erfvn3VnDlz5tgrLAPijgYAAAAA62g0AAAAAFhHowEAAADAOhoNAAAAANZFxWZw08nJbjZ+i4js2LHDMXZzCi8i3x9//KGyqlWrqmzSpEkqy5cvn2P866+/qjl//vmnyvr06aOycuXKOcY//fSTmvPEE0+ozObp5AgfpodMmB5c4J+ZNsa2bNlSZaNHj1bZ8ePHHeMpU6aoOTY38SLy+W/+fu2119ScESNGuLrWhx9+6BiXKlVKzalSpYrKXnjhBZX9/fffjnHjxo3VnL/++stVXYhsw4cPV5np74ozZ850jN2uW/wf7mgAAAAAsI5GAwAAAIB1NBoAAAAArMuQezTKly/vGD/77LOuXmf6XX6TJk2s1ITI4r+GREQGDRqkst69e6ssUyZn//6vf/3L1Xt+9tlnKvNfu4sWLXJ1LWRMBQsWdDXP/wC9JUuWqDm33HKLq2t17tzZMf78889dvQ74X+PHj3c1z/T7d9OBuP5SU1NVZtpPOXjwYMf4/PnzrupCZGvYsKHKTAdGnj17VmUcxnf1uKMBAAAAwDoaDQAAAADW0WgAAAAAsI5GAwAAAIB1GXIz+Msvv+wYt2nTxtXrxo4dq7Ldu3dbqQmRz39dXS4DguX33393Nc//QNKYmBg159ixYyozbdr95ptvXFYHuGdaa243jQP/TYkSJRzjjz/+2NXrHnroIZUtWLDARklRjTsaAAAAAKyj0QAAAABgHY0GAAAAAOtoNAAAAABYF/GbwStWrKiy3Llzp/u6SZMmqezbb7+1UhMABMP777+vsqxZs6rM/yEFa9euVXNMJ9G/9dZbV1EdAIRWbGysyp599lnHOD4+Xs2ZO3euyubNm2evMPyDOxoAAAAArKPRAAAAAGAdjQYAAAAA62g0AAAAAFgX8ZvBTSc5NmnSxDE2ne49evRolW3dutVeYQBg2fHjx1U2fPhwVxkAZDSdOnVS2ZNPPukYr1q1Ss0x/d0RwcEdDQAAAADW0WgAAAAAsI5GAwAAAIB1Eb9H4+uvv1aZ/2EtvXr1UnPYjwEAABAZatWqpbIXXnhBZYMHD3aMJ0+erOacO3fOXmH4r7ijAQAAAMA6Gg0AAAAA1tFoAAAAALCORgMAAACAdRG/GXzp0qUqy5Il4v+1AAAA8P/9/PPPKitWrJgHleBKcEcDAAAAgHU0GgAAAACso9EAAAAAYJ2rRsPn8wW7DkSoUKwN1h8uJ1RrgzUIE9YfvMZ3MLzkZm24ajRSU1OvuhhkTKFYG6w/XE6o1gZrECasP3iN72B4yc3aiPG5aEfS0tIkOTlZ4uLiJCYmxkpxiGw+n09SU1MlMTFRMmUK7i/wWH/wF8r1J8IahBPrD17jOxheupL156rRAAAAAIArwWZwAAAAANbRaAAAAACwjkYDAAAAgHU0GgAAAACso9FwoUSJEhITE6P+6datm9elIQoMHTpUatasKXFxcVKwYEFp2bKlbN261euyEEUuXbokL7/8spQsWVJiY2Pluuuuk0GDBvF8fYTM/v37pUOHDpIvXz6JjY2VG264QdauXet1WYgSqamp0rNnT0lKSpLY2FipV6+erFmzxuuyIkIWrwuIBGvWrJFLly79M960aZM0atRIWrVq5WFViBbLly+Xbt26Sc2aNeXixYvywgsvSOPGjWXz5s2SM2dOr8tDFHj99ddlwoQJ8v7770vFihVl7dq10rlzZ4mPj5fu3bt7XR4yuOPHj8tNN90kt912myxcuFAKFCgg27Ztk7x583pdGqLEww8/LJs2bZLp06dLYmKizJgxQxo2bCibN2+Wa6+91uvywhqPtw1Az5495YsvvpBt27bxTGmE3JEjR6RgwYKyfPlyqV+/vtflIArcfffdUqhQIXn33Xf/ye677z6JjY2VGTNmeFgZosHzzz8vP/zwg3z//fdel4IodPbsWYmLi5MFCxZI06ZN/8mrV68uTZo0kcGDB3tYXfjjp1NX6Pz58zJjxgzp0qULTQY8kZKSIiIiCQkJHleCaFGvXj1ZunSp/PHHHyIi8uuvv8rKlSulSZMmHleGaPDZZ59JjRo1pFWrVlKwYEG58cYbZfLkyV6XhShx8eJFuXTpkmTPnt2Rx8bGysqVKz2qKnLQaFyh+fPny4kTJ6RTp05el4IolJaWJj179pSbbrpJKlWq5HU5iBLPP/+8tG3bVsqXLy/XXHON3HjjjdKzZ09p376916UhCvz5558yYcIEKVOmjCxevFieeOIJ6d69u7z//vtel4YoEBcXJ3Xr1pVBgwZJcnKyXLp0SWbMmCGrV6+WAwcOeF1e2OOnU1fozjvvlKxZs8rnn3/udSmIQk888YQsXLhQVq5cKUWLFvW6HESJWbNmSZ8+feSNN96QihUryoYNG6Rnz54ycuRI6dixo9flIYPLmjWr1KhRQ1atWvVP1r17d1mzZo2sXr3aw8oQLXbs2CFdunSRFStWSObMmaVatWpStmxZWbdunfz+++9elxfW2Ax+BXbv3i3ffPONfPrpp16Xgij01FNPyRdffCErVqygyUBI9enT55+7GiIiN9xwg+zevVuGDh1Ko4GgK1KkiFSoUMGRXX/99TJ37lyPKkK0ue6662T58uVy+vRpOXnypBQpUkTatGkjpUqV8rq0sMdPp67A1KlTpWDBgo7NQECw+Xw+eeqpp2TevHny7bffSsmSJb0uCVHmzJkzkimT8+sic+bMkpaW5lFFiCY33XSTeqT3H3/8IUlJSR5VhGiVM2dOKVKkiBw/flwWL14sLVq08LqksMcdDZfS0tJk6tSp0rFjR8mShf/ZEDrdunWTjz76SBYsWCBxcXFy8OBBERGJj4+X2NhYj6tDNGjWrJkMGTJEihcvLhUrVpRffvlFRo4cKV26dPG6NESBZ555RurVqyevvfaatG7dWn7++WeZNGmSTJo0yevSECUWL14sPp9PypUrJ9u3b5c+ffpI+fLlpXPnzl6XFvbYo+HS119/LXfeeads3bpVypYt63U5iCKXe7rZ1KlTeSgBQiI1NVVefvllmTdvnhw+fFgSExOlXbt20r9/f8maNavX5SEKfPHFF9KvXz/Ztm2blCxZUnr16iWPPPKI12UhSsyePVv69esn+/btk4SEBLnvvvtkyJAhEh8f73VpYY9GAwAAAIB17NEAAAAAYB2NBgAAAADraDQAAAAAWEejAQAAAMA6Gg0AAAAA1tFoAAAAALCORgMAAACAdTQaAAAAAKzL4mZSWlqaJCcnS1xc3GVPKUZ08fl8kpqaKomJiZIpU3D7VdYf/IVy/YmwBuHE+oPX+A6Gl65k/blqNJKTk6VYsWJWikPGsnfvXilatGhQ34P1h8sJxfoTYQ3CjPUHr/EdDC+5WX+u2uC4uDgrBSHjCcXaYP3hckK1NliDMGH9wWt8B8NLbtaGq0aDW2W4nFCsDdYfLidUa4M1CBPWH7zGdzC85GZtsBkcAAAAgHU0GgAAAACso9EAAAAAYB2NBgAAAADraDQAAAAAWEejAQAAAMA6Gg0AAAAA1tFoAAAAALCORgMAAACAdTQaAAAAAKzL4nUBAAAAQCSaOXOmyurUqaOytm3bOsY//fRT0GoKJ9zRAAAAAGAdjQYAAAAA62g0AAAAAFhHowEAAADAOjaDh1DZsmUd44kTJ6o57du3V9mBAweCVhOix6233uoYL126VM3JlEn/twf/14mILF++3FZZAABErKSkJJWVKFFCZTNmzHCMK1SooOZcuHDBWl3hgjsaAAAAAKyj0QAAAABgHY0GAAAAAOtoNAAAAABYF/TN4HFxcSrLlSuXylJSUhzjM2fOBK0mr9x1112Ocf369dWchx9+WGVDhw5V2cWLF+0VhgynU6dOKnv66acd47S0NFfXGjlypMo++OADx3j8+PFqDmsUQDjr16+fyoYMGaKy4cOHq+z5558PSk0Ib8WKFVNZjRo1XL22dOnSjnGWLPqv4GwGBwAAAAAXaDQAAAAAWEejAQAAAMC6oO/ReO6551Rm+l1knz59HOO33noraDV5Ze3atenOGTBggMpmzpypsu3bt1upCZHPtB/jwQcfVFnlypUDur7pdW+++aZjPH/+fDVn9+7dAb0fIo/pwKpnnnlGZU8++aRjbPqN8qxZs1T2wAMPXEV1wP/w3zPqv29NRMTn86msZ8+eKtu2bZtj/O67715dcYgI8fHxKrvmmmtcvdb/e/LcuXM2Sgp73NEAAAAAYB2NBgAAAADraDQAAAAAWEejAQAAAMC6oG8Gd8t/E/Sff/6p5ixYsCBU5QRF4cKFvS4BYSxPnjwqq1q1qmM8depUNSd//vwqy549e7rvt2XLFpVlyqT/20PZsmXTvRaiR+fOnVU2atQolflvlhUReeyxxxxj0+FXpgdivPrqqyozrV/gf5keNPDEE084xoUKFXJ1rUOHDqls9erVgRWGiOK/jkwPM3Lro48+cozdHpob6bijAQAAAMA6Gg0AAAAA1tFoAAAAALCORgMAAACAdWGzGTxXrlyOsWnTa+PGjVXm5rRtL/j/+4iI9OrVK6BrtWrVSmVDhw4N6FoIDy1btlTZI488ojL/NW/arB3ohrI33nhDZabrT548OaDrI/JkzZpVZc8++6xj3L9/fzVn5MiRKjOtrxMnTjjG1apVU3NMm8FTU1NVBvw3derUUVmg35uPP/64yjZv3hzQtRBZ3nrrLcf4gQce8KiSyMUdDQAAAADW0WgAAAAAsI5GAwAAAIB1NBoAAAAArAv6ZvBdu3YF9LrcuXOr7JVXXlFZhw4dVHb8+PGA3tOm0qVLq6xWrVoeVAKvmdbo+++/H9C1TJu1AxUTExPy90R4M536PXjwYMe4Z8+eas7YsWMDej/TAz4OHz6ssv379wd0fUSHEiVKqGzMmDEBXWvp0qUqW7ZsWUDXQmQxPZCla9euHlSSsfA3CAAAAADW0WgAAAAAsI5GAwAAAIB1Qd+jMW3aNJUlJiaqzHRIk78777xTZffdd5/KpkyZ4q64IDL9zvjPP/90jEuVKuXqWp988omVmhAa/nsyRo0apeaYDtn7+++/VXbo0CHHOC4uTs1JSEhwVZf/9U+ePKnmxMfHqyzQAwER3kzrZtCgQSqbM2eOYzxhwoSA3zMpKckxfvjhhwO+FvC/Pv/8c5VVqFAh3deZPgNNB02ePXs2sMIQtkz70caNG6cy/0NM169fr+aYDh7F/+GOBgAAAADraDQAAAAAWEejAQAAAMA6Gg0AAAAA1gV9M/ilS5dUZjpIp3379o6x6cA7k27duqls3rx5jvHRo0ddXcumggULqszt5m9EjpYtW6rM/zA+t5upf/rpJ5U1bNjQMe7UqZOaM3nyZFfXf+GFFxxj/z8nl7s+Il+WLPqj/ocfflCZ/8MHRESeeOIJx/jixYsB1zFjxgzH2PSZOGLEiICvj+hUsWJFlfl8vnRf9/bbb6tsyZIlVmrC1cmVK5fKqlSporKyZcuqrHbt2o5x69at1Zy8efO6qqN79+6O8VdffaXmbN++3dW1ohV3NAAAAABYR6MBAAAAwDoaDQAAAADW0WgAAAAAsC7om8FNUlJSVOa/MdHtZvAbbrhBZcWKFXOMr2YzuP+pkI899pir17Vq1Srg90R4Mm2UNp367c904rdp47f/pjO3fv31V5X5b0gXcXeis/8p0CIijzzyiMpq1arlsjqEg/vvv19lpk2Ut99+u8qOHTsW0Hu2a9dOZXXq1HGMT506pea8+eabAb0fosPIkSNVFhMTozLTZvClS5c6xoMGDbJXGKwqWrSoyt577z2VmT7H/Jn+zml6iMrw4cNVtmvXrnTrwn/HHQ0AAAAA1tFoAAAAALCORgMAAACAdTQaAAAAAKzzZDO4yerVqx3jjh07BnytunXrOsYbNmxQc+rVq+cq8z+d8qWXXgq4Ljd+//13lR0/fjyo7wl3Xn75ZZXlzJkz3de99tprKhs6dGhANaxcuVJlCxcuVJnphGc3TJtzz507F9C1ED5Mn6dbt25V2apVqwK6fuHChVVmelBCpkzO/7Y1duxYNSfQtYuMafz48Y5xy5Yt1RzTxu9///vfKmvfvr1jbHpQB8LDli1bVFa5cmWVlSlTJt1rnTx5UmV79uwJrLCr4ObvCxkRdzQAAAAAWEejAQAAAMA6Gg0AAAAA1oXNHo0pU6Y4xg0aNFBzHnjgAVfXGjdu3H8dXwn/3xSnpaUFfC03KlSooDLTb1LffffdoNYR7apWraqyuLg4lfmvDxGRzJkzB6MkERHZvn170K59OabDsEz/3ghfd955p8r69++vsgsXLqR7rdy5c6ts7ty5KsufP7/KJk6c6Bi//vrr6b4foofpIFD/7z/TfiCTSZMmqezIkSMB1YXwYNovuGnTppDWkJqaqrKDBw+qzLROW7Ro4RhPmzbNWl3hjL8tAAAAALCORgMAAACAdTQaAAAAAKyj0QAAAABgXdhsBvc3YsQIlbVr1y7kdfhv/jYdDBRsderUURmbwe2qVKmSY2za3Jo3b16VBfvhAKHmf0CliEjWrFlVltH+vTOaO+64I9058+fPd3Ut/43k77zzjppTvHhxlZkeXPDCCy84xqaDtBC9unTporIiRYqk+zrTQbcLFiywUhPwn44ePaqynTt3qsy0Gfy7774LSk3hjjsaAAAAAKyj0QAAAABgHY0GAAAAAOtoNAAAAABYF7abwcOF/4ZG02bwL7/8UmUpKSkqM53Ei/AwZswYx9i0uTUa3H///SozndaL8Hbo0CHH+O+//1ZzZs+erbK4uDiVFShQwDE2nc5rOj1+/PjxKjN9LiI69ezZU2Vdu3ZVmZsHsDRq1EhlycnJAdUFBMuBAwe8LsET3NEAAAAAYB2NBgAAAADraDQAAAAAWEejAQAAAMC6qNgMfuzYMZXt2bNHZabTyGfOnBnQe1atWlVlbAbPeJ577jmvSwhY+fLlVTZ8+HBXr921a5djbNpsDO9s2rTJMX788cfVHNPG219//VVl/p+B48aNU3PWrl2rMtMJ4ohOxYoVU5lp/WXKpP/b56VLlxzjyZMnqzls/Ea4MT3E4PDhwx5U4j3uaAAAAACwjkYDAAAAgHU0GgAAAACsC9s9Gn/++afKPvjgA5WVKlVKZb///rtjbDo4yv83zOGscePGKsubN69jfPz48VCVg/9w9OhRr0twzX9PxoIFC9ScfPnyqcz0u1L/g/38D4hDeDF9dpoy08F7o0aNcowLFSqk5tx7770qY99O9CpdurRj/Nlnn6k55cqVc3Wtt956yzHu27dv4IUhKvmvRxGRhIQEV689c+aMY2za8zty5EiVmfY7+h9+6j8WEcmRI4fKBg8erLJPPvnEMTb9GQsX3NEAAAAAYB2NBgAAAADraDQAAAAAWEejAQAAAMC6sN0MfvLkSZV16dLFg0q8d+2116osa9asHlSScflvgjUdHGUydepUlZk22QZTrly5XNXQokWLdK9legjD3XffrbKtW7e6rA6RpEGDBip76qmnHOMhQ4aoOaYD+xC9/Dd6u934bRLOm1wRWqa/95geCPToo486xo899piaY9p0bXL+/HnH+NSpU2qO243l/hu4jxw5ouaY/h3j4+NVdvDgQcc4nP+ccEcDAAAAgHU0GgAAAACso9EAAAAAYB2NBgAAAADrwnYzeKQ7ceKEyg4cOOAYFylSJODrv/baa46xabPTxYsXA75+tPE/efPjjz9Wc0wbsky+++47x9jn86k5plO5TRusn3vuOcfYdHKzafNYrVq1VOZ/wqn/GhIR+fTTT13VhYzpo48+UllycrJjbDrxFvhPbjfH+lu2bJnKNm/efJXVIBIVKlRIZaNHj1ZZmzZtrL2n/9/RRPT392+//abm/Prrr9ZqcOv9998P+XsGijsaAAAAAKyj0QAAAABgHY0GAAAAAOtoNAAAAABYx2bwINm1a5fK7r//fsfYtPHWtAHKpGPHjo5x9+7d1Rw2g7u3dOlSx/i+++5Tc+bOnasy0wbx+vXrO8ZpaWlqzi233HKlJYqI+cRy0/WXL1+uMv/TwkN9gjnCS40aNVSWP39+lfl/tphOxgX+06BBgwJ63YQJE1R2/Pjxqy0HEeiBBx5QWaAbv7/44guVjRgxQmU//PCDyi5cuBDQe+L/cEcDAAAAgHU0GgAAAACso9EAAAAAYB17NELop59+coxbtGih5ph+S2j63bQ/0++tTb/Thzum/+2qVKmiskcffVRlL730UlBqEhE5ePCgyr7//nuVmQ5wTElJCUpNCH/Zs2dX2aRJk1S2f/9+lU2fPj0oNSFjqFixospy5syZ7uteeeUVlZn2wSE6zZs3T2WdO3dWmf+BoiL6wN2pU6faKwxXjDsaAAAAAKyj0QAAAABgHY0GAAAAAOtoNAAAAABYx2ZwD61du1ZlzzzzjMr69Omjsi+//DLda8Eu00bZAQMGqOzPP/90jHv37q3mlC9fXmVbtmxR2RtvvOEY79ixQ80xHTIE/CfTJkrTww1M2enTp4NSEzKGOnXqqCwuLi7d1507d05lPp/PSk2IfKZDjytXrhz6QnDVuKMBAAAAwDoaDQAAAADW0WgAAAAAsI5GAwAAAIB1MT4Xu69Onjwp8fHxoagHESYlJUVy584d1Pdg/eFyQrH+RCJ/DW7evFllps24NWvWVNnFixeDUlNGwPoz2717t2OcI0cONadRo0Yq27BhQ7BKyrD4DoaX3Kw/7mgAAAAAsI5GAwAAAIB1NBoAAAAArKPRAAAAAGAdJ4MDQAaXkJCgsldeeUVlbPyGDUlJSV6XACBMcEcDAAAAgHU0GgAAAACso9EAAAAAYB17NAAggytcuLDXJQAAohB3NAAAAABYR6MBAAAAwDoaDQAAAADW0WgAAAAAsI5GAwAAAIB1NBoAAAAArKPRAAAAAGAdjQYAAAAA61w1Gj6fL9h1IEKFYm2w/nA5oVobrEGYsP7gNb6D4SU3a8NVo5GamnrVxSBjCsXaYP3hckK1NliDMGH9wWt8B8NLbtZGjM9FO5KWlibJyckSFxcnMTExVopDZPP5fJKamiqJiYmSKVNwf4HH+oO/UK4/EdYgnFh/8BrfwfDSlaw/V40GAAAAAFwJNoMDAAAAsI5GAwAAAIB1NBoAAAAArKPRAAAAAGAdjQYAAAAA62g0AjBs2DCJiYmRnj17el0Kosj48eOlRIkSkj17dqldu7b8/PPPXpeEKJKamio9e/aUpKQkiY2NlXr16smaNWu8LgtRYsWKFdKsWTNJTEyUmJgYmT9/vtclIYqw/gJHo3GF1qxZI++8845UrlzZ61IQRT7++GPp1auXDBgwQNavXy9VqlSRO++8Uw4fPux1aYgSDz/8sCxZskSmT58uGzdulMaNG0vDhg1l//79XpeGKHD69GmpUqWKjB8/3utSEIVYf4HjHI0rcOrUKalWrZq8/fbbMnjwYKlataqMGjXK67IQBWrXri01a9aUcePGicj/HKBUrFgxefrpp+X555/3uDpkdGfPnpW4uDhZsGCBNG3a9J+8evXq0qRJExk8eLCH1SHaxMTEyLx586Rly5Zel4IoxPq7MtzRuALdunWTpk2bSsOGDb0uBVHk/Pnzsm7dOse6y5QpkzRs2FBWr17tYWWIFhcvXpRLly5J9uzZHXlsbKysXLnSo6oAAOEui9cFRIpZs2bJ+vXr+U0yQu6vv/6SS5cuSaFChRx5oUKFZMuWLR5VhWgSFxcndevWlUGDBsn1118vhQoVkpkzZ8rq1auldOnSXpcHAAhT3NFwYe/evdKjRw/58MMP1X/RA4BoMH36dPH5fHLttddKtmzZZMyYMdKuXTvJlImvEQCAGd8QLqxbt04OHz4s1apVkyxZskiWLFlk+fLlMmbMGMmSJYtcunTJ6xKRgeXPn18yZ84shw4dcuSHDh2SwoULe1QVos11110ny5cvl1OnTsnevXvl559/lgsXLkipUqW8Lg0AEKZoNFy44447ZOPGjbJhw4Z//qlRo4a0b99eNmzYIJkzZ/a6RGRgWbNmlerVq8vSpUv/ydLS0mTp0qVSt25dDytDNMqZM6cUKVJEjh8/LosXL5YWLVp4XRIAIEyxR8OFuLg4qVSpkiPLmTOn5MuXT+VAMPTq1Us6duwoNWrUkFq1asmoUaPk9OnT0rlzZ69LQ5RYvHix+Hw+KVeunGzfvl369Okj5cuXZw0iJE6dOiXbt2//Z7xz507ZsGGDJCQkSPHixT2sDNGA9Rc4Gg0gArRp00aOHDki/fv3l4MHD0rVqlVl0aJFaoM4ECwpKSnSr18/2bdvnyQkJMh9990nQ4YMkWuuucbr0hAF1q5dK7fddts/4169eomISMeOHWXatGkeVYVowfoLHOdoAAAAALCOPRoAAAAArKPRAAAAAGAdjQYAAAAA62g0AAAAAFhHowEAAADAOhoNAAAAANbRaAAAAACwjkYDAAAAgHU0GgAAAACso9EAAAAAYB2NBgAAAADr/h/8zI0z49J3gwAAAABJRU5ErkJggg==",
|
|
"text/plain": [
|
|
"<Figure size 1000x1000 with 25 Axes>"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
}
|
|
],
|
|
"source": [
|
|
"# Load and visualise the MNIST digits\n",
|
|
"import tensorflow as tf\n",
|
|
"tf.config.experimental.set_visible_devices([], \"GPU\")\n",
|
|
"\n",
|
|
"mnist = tf.keras.datasets.mnist\n",
|
|
"(train_images0, train_labels0),(test_images0, test_labels0) = mnist.load_data()\n",
|
|
"\n",
|
|
"print(\"test_images shape\",test_images0.shape,\"train_images shape\",train_images0.shape)\n",
|
|
"class_names=[\"0\",\"1\",\"2\",\"3\",\"4\",\"5\",\"6\",\"7\",\"8\",\"9\"]\n",
|
|
"import matplotlib.pyplot as plt\n",
|
|
"# plot first few images\n",
|
|
"plt.figure(figsize=(10,10))\n",
|
|
"for i in range(25):\n",
|
|
" # define subplot\n",
|
|
" plt.subplot(5,5,i+1)\n",
|
|
" # plot raw pixel data\n",
|
|
" plt.imshow(train_images0[i], cmap=plt.get_cmap('gray'))\n",
|
|
" plt.xticks([])\n",
|
|
" plt.yticks([])\n",
|
|
" plt.grid(False)\n",
|
|
" # Add a label underneath...\n",
|
|
" plt.xlabel(class_names[train_labels0[i]])\n",
|
|
"plt.show()"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"### Next build a neural-network classifier for these digits.\n",
|
|
"\n",
|
|
"- We will build a keras model, with the higher-level API concepts"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 3,
|
|
"metadata": {
|
|
"slideshow": {
|
|
"slide_type": "slide"
|
|
}
|
|
},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"WARNING:tensorflow:From f:\\KTU\\Neuroninių tinklų metodai\\venv\\Lib\\site-packages\\keras\\src\\backend.py:873: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.\n",
|
|
"\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"from tensorflow import keras\n",
|
|
"# Each MNIST images are 28*28. Therefore if there are N images, then the\n",
|
|
"# shape of the numpy array holding the images is N*28*28\n",
|
|
"# We will reshape that here to be N*784, using a numpy reshape.\n",
|
|
"# Note that this flattens each image into a single vector length 784.\n",
|
|
"test_images=test_images0.reshape(10000,784) # 10000 test patterns\n",
|
|
"train_images=train_images0.reshape(60000,784) # 60000 train patterns\n",
|
|
"\n",
|
|
"# Also rescale greyscale from 8 bit to floating point (by dividing by 255)\n",
|
|
"test_images=test_images/255.0\n",
|
|
"train_images=train_images/255.0\n",
|
|
"\n",
|
|
"# Create the model\n",
|
|
"\n",
|
|
"keras_model = tf.keras.models.Sequential([\n",
|
|
" tf.keras.layers.Dense(10, activation='softmax')\n",
|
|
"])\n",
|
|
"\n",
|
|
"keras_model.build(input_shape=[None,784])"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"## View the keras model summary information\n",
|
|
"\n",
|
|
"- This shows you how many layers your neural network has, and how many weights, etc."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 8,
|
|
"metadata": {
|
|
"slideshow": {
|
|
"slide_type": "slide"
|
|
}
|
|
},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"Model: \"sequential_2\"\n",
|
|
"_________________________________________________________________\n",
|
|
" Layer (type) Output Shape Param # \n",
|
|
"=================================================================\n",
|
|
" dense_6 (Dense) (None, 10) 7850 \n",
|
|
" \n",
|
|
"=================================================================\n",
|
|
"Total params: 7,850\n",
|
|
"Trainable params: 7,850\n",
|
|
"Non-trainable params: 0\n",
|
|
"_________________________________________________________________\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"# View the model summary information...\n",
|
|
"keras_model.summary()"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"## Train the Keras model\n",
|
|
"\n",
|
|
"- We will use SGD optimiser (ordinary gradient descent)\n",
|
|
"- We will use Cross Entropy loss (\"SparseCategoricalCrossentropy\")\n",
|
|
"- We will run 200 training iterations (epochs)..."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 4,
|
|
"metadata": {
|
|
"slideshow": {
|
|
"slide_type": "slide"
|
|
}
|
|
},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"Epoch 1/200\n",
|
|
"WARNING:tensorflow:From f:\\KTU\\Neuroninių tinklų metodai\\venv\\Lib\\site-packages\\keras\\src\\utils\\tf_utils.py:492: The name tf.ragged.RaggedTensorValue is deprecated. Please use tf.compat.v1.ragged.RaggedTensorValue instead.\n",
|
|
"\n"
|
|
]
|
|
},
|
|
{
|
|
"name": "stderr",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"f:\\KTU\\Neuroninių tinklų metodai\\venv\\Lib\\site-packages\\keras\\src\\backend.py:5727: UserWarning: \"`sparse_categorical_crossentropy` received `from_logits=True`, but the `output` argument was produced by a Softmax activation and thus does not represent logits. Was this intended?\n",
|
|
" output, from_logits = _get_logits(\n"
|
|
]
|
|
},
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"1/1 [==============================] - 3s 3s/step - loss: 2.3981 - sparse_categorical_accuracy: 0.1408 - val_loss: 2.3445 - val_sparse_categorical_accuracy: 0.1587\n",
|
|
"Epoch 2/200\n",
|
|
"1/1 [==============================] - 0s 104ms/step - loss: 2.3408 - sparse_categorical_accuracy: 0.1616 - val_loss: 2.2886 - val_sparse_categorical_accuracy: 0.1820\n",
|
|
"Epoch 3/200\n",
|
|
"1/1 [==============================] - 0s 90ms/step - loss: 2.2859 - sparse_categorical_accuracy: 0.1848 - val_loss: 2.2350 - val_sparse_categorical_accuracy: 0.2089\n",
|
|
"Epoch 4/200\n",
|
|
"1/1 [==============================] - 0s 95ms/step - loss: 2.2332 - sparse_categorical_accuracy: 0.2104 - val_loss: 2.1834 - val_sparse_categorical_accuracy: 0.2361\n",
|
|
"Epoch 5/200\n",
|
|
"1/1 [==============================] - 0s 90ms/step - loss: 2.1824 - sparse_categorical_accuracy: 0.2393 - val_loss: 2.1336 - val_sparse_categorical_accuracy: 0.2635\n",
|
|
"Epoch 6/200\n",
|
|
"1/1 [==============================] - 0s 95ms/step - loss: 2.1335 - sparse_categorical_accuracy: 0.2682 - val_loss: 2.0856 - val_sparse_categorical_accuracy: 0.2916\n",
|
|
"Epoch 7/200\n",
|
|
"1/1 [==============================] - 0s 93ms/step - loss: 2.0862 - sparse_categorical_accuracy: 0.2971 - val_loss: 2.0391 - val_sparse_categorical_accuracy: 0.3199\n",
|
|
"Epoch 8/200\n",
|
|
"1/1 [==============================] - 0s 86ms/step - loss: 2.0405 - sparse_categorical_accuracy: 0.3268 - val_loss: 1.9941 - val_sparse_categorical_accuracy: 0.3460\n",
|
|
"Epoch 9/200\n",
|
|
"1/1 [==============================] - 0s 86ms/step - loss: 1.9963 - sparse_categorical_accuracy: 0.3544 - val_loss: 1.9504 - val_sparse_categorical_accuracy: 0.3704\n",
|
|
"Epoch 10/200\n",
|
|
"1/1 [==============================] - 0s 85ms/step - loss: 1.9534 - sparse_categorical_accuracy: 0.3800 - val_loss: 1.9081 - val_sparse_categorical_accuracy: 0.3965\n",
|
|
"Epoch 11/200\n",
|
|
"1/1 [==============================] - 0s 113ms/step - loss: 1.9119 - sparse_categorical_accuracy: 0.4030 - val_loss: 1.8671 - val_sparse_categorical_accuracy: 0.4210\n",
|
|
"Epoch 12/200\n",
|
|
"1/1 [==============================] - 0s 127ms/step - loss: 1.8716 - sparse_categorical_accuracy: 0.4265 - val_loss: 1.8272 - val_sparse_categorical_accuracy: 0.4426\n",
|
|
"Epoch 13/200\n",
|
|
"1/1 [==============================] - 0s 120ms/step - loss: 1.8325 - sparse_categorical_accuracy: 0.4490 - val_loss: 1.7885 - val_sparse_categorical_accuracy: 0.4659\n",
|
|
"Epoch 14/200\n",
|
|
"1/1 [==============================] - 0s 114ms/step - loss: 1.7946 - sparse_categorical_accuracy: 0.4730 - val_loss: 1.7509 - val_sparse_categorical_accuracy: 0.4950\n",
|
|
"Epoch 15/200\n",
|
|
"1/1 [==============================] - 0s 120ms/step - loss: 1.7578 - sparse_categorical_accuracy: 0.4976 - val_loss: 1.7144 - val_sparse_categorical_accuracy: 0.5188\n",
|
|
"Epoch 16/200\n",
|
|
"1/1 [==============================] - 0s 121ms/step - loss: 1.7220 - sparse_categorical_accuracy: 0.5228 - val_loss: 1.6788 - val_sparse_categorical_accuracy: 0.5444\n",
|
|
"Epoch 17/200\n",
|
|
"1/1 [==============================] - 0s 121ms/step - loss: 1.6872 - sparse_categorical_accuracy: 0.5471 - val_loss: 1.6442 - val_sparse_categorical_accuracy: 0.5684\n",
|
|
"Epoch 18/200\n",
|
|
"1/1 [==============================] - 0s 126ms/step - loss: 1.6533 - sparse_categorical_accuracy: 0.5695 - val_loss: 1.6105 - val_sparse_categorical_accuracy: 0.5938\n",
|
|
"Epoch 19/200\n",
|
|
"1/1 [==============================] - 0s 120ms/step - loss: 1.6204 - sparse_categorical_accuracy: 0.5916 - val_loss: 1.5778 - val_sparse_categorical_accuracy: 0.6146\n",
|
|
"Epoch 20/200\n",
|
|
"1/1 [==============================] - 0s 129ms/step - loss: 1.5884 - sparse_categorical_accuracy: 0.6111 - val_loss: 1.5458 - val_sparse_categorical_accuracy: 0.6345\n",
|
|
"Epoch 21/200\n",
|
|
"1/1 [==============================] - 0s 138ms/step - loss: 1.5572 - sparse_categorical_accuracy: 0.6291 - val_loss: 1.5148 - val_sparse_categorical_accuracy: 0.6516\n",
|
|
"Epoch 22/200\n",
|
|
"1/1 [==============================] - 0s 124ms/step - loss: 1.5269 - sparse_categorical_accuracy: 0.6466 - val_loss: 1.4846 - val_sparse_categorical_accuracy: 0.6677\n",
|
|
"Epoch 23/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 1.4974 - sparse_categorical_accuracy: 0.6613 - val_loss: 1.4553 - val_sparse_categorical_accuracy: 0.6809\n",
|
|
"Epoch 24/200\n",
|
|
"1/1 [==============================] - 0s 115ms/step - loss: 1.4688 - sparse_categorical_accuracy: 0.6744 - val_loss: 1.4268 - val_sparse_categorical_accuracy: 0.6935\n",
|
|
"Epoch 25/200\n",
|
|
"1/1 [==============================] - 0s 123ms/step - loss: 1.4410 - sparse_categorical_accuracy: 0.6866 - val_loss: 1.3992 - val_sparse_categorical_accuracy: 0.7060\n",
|
|
"Epoch 26/200\n",
|
|
"1/1 [==============================] - 0s 114ms/step - loss: 1.4141 - sparse_categorical_accuracy: 0.6972 - val_loss: 1.3724 - val_sparse_categorical_accuracy: 0.7174\n",
|
|
"Epoch 27/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 1.3880 - sparse_categorical_accuracy: 0.7064 - val_loss: 1.3465 - val_sparse_categorical_accuracy: 0.7265\n",
|
|
"Epoch 28/200\n",
|
|
"1/1 [==============================] - 0s 116ms/step - loss: 1.3627 - sparse_categorical_accuracy: 0.7157 - val_loss: 1.3215 - val_sparse_categorical_accuracy: 0.7355\n",
|
|
"Epoch 29/200\n",
|
|
"1/1 [==============================] - 0s 121ms/step - loss: 1.3382 - sparse_categorical_accuracy: 0.7234 - val_loss: 1.2972 - val_sparse_categorical_accuracy: 0.7442\n",
|
|
"Epoch 30/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 1.3146 - sparse_categorical_accuracy: 0.7301 - val_loss: 1.2738 - val_sparse_categorical_accuracy: 0.7500\n",
|
|
"Epoch 31/200\n",
|
|
"1/1 [==============================] - 0s 120ms/step - loss: 1.2917 - sparse_categorical_accuracy: 0.7363 - val_loss: 1.2512 - val_sparse_categorical_accuracy: 0.7567\n",
|
|
"Epoch 32/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 1.2696 - sparse_categorical_accuracy: 0.7422 - val_loss: 1.2294 - val_sparse_categorical_accuracy: 0.7617\n",
|
|
"Epoch 33/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 1.2483 - sparse_categorical_accuracy: 0.7472 - val_loss: 1.2083 - val_sparse_categorical_accuracy: 0.7674\n",
|
|
"Epoch 34/200\n",
|
|
"1/1 [==============================] - 0s 123ms/step - loss: 1.2276 - sparse_categorical_accuracy: 0.7525 - val_loss: 1.1880 - val_sparse_categorical_accuracy: 0.7722\n",
|
|
"Epoch 35/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 1.2077 - sparse_categorical_accuracy: 0.7570 - val_loss: 1.1683 - val_sparse_categorical_accuracy: 0.7759\n",
|
|
"Epoch 36/200\n",
|
|
"1/1 [==============================] - 0s 115ms/step - loss: 1.1885 - sparse_categorical_accuracy: 0.7608 - val_loss: 1.1494 - val_sparse_categorical_accuracy: 0.7793\n",
|
|
"Epoch 37/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 1.1699 - sparse_categorical_accuracy: 0.7648 - val_loss: 1.1311 - val_sparse_categorical_accuracy: 0.7843\n",
|
|
"Epoch 38/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 1.1520 - sparse_categorical_accuracy: 0.7684 - val_loss: 1.1134 - val_sparse_categorical_accuracy: 0.7881\n",
|
|
"Epoch 39/200\n",
|
|
"1/1 [==============================] - 0s 139ms/step - loss: 1.1346 - sparse_categorical_accuracy: 0.7716 - val_loss: 1.0963 - val_sparse_categorical_accuracy: 0.7904\n",
|
|
"Epoch 40/200\n",
|
|
"1/1 [==============================] - 0s 127ms/step - loss: 1.1178 - sparse_categorical_accuracy: 0.7745 - val_loss: 1.0797 - val_sparse_categorical_accuracy: 0.7930\n",
|
|
"Epoch 41/200\n",
|
|
"1/1 [==============================] - 0s 131ms/step - loss: 1.1016 - sparse_categorical_accuracy: 0.7776 - val_loss: 1.0638 - val_sparse_categorical_accuracy: 0.7955\n",
|
|
"Epoch 42/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 1.0860 - sparse_categorical_accuracy: 0.7805 - val_loss: 1.0483 - val_sparse_categorical_accuracy: 0.7971\n",
|
|
"Epoch 43/200\n",
|
|
"1/1 [==============================] - 0s 130ms/step - loss: 1.0708 - sparse_categorical_accuracy: 0.7832 - val_loss: 1.0334 - val_sparse_categorical_accuracy: 0.7984\n",
|
|
"Epoch 44/200\n",
|
|
"1/1 [==============================] - 0s 126ms/step - loss: 1.0561 - sparse_categorical_accuracy: 0.7859 - val_loss: 1.0189 - val_sparse_categorical_accuracy: 0.7997\n",
|
|
"Epoch 45/200\n",
|
|
"1/1 [==============================] - 0s 123ms/step - loss: 1.0420 - sparse_categorical_accuracy: 0.7886 - val_loss: 1.0049 - val_sparse_categorical_accuracy: 0.8026\n",
|
|
"Epoch 46/200\n",
|
|
"1/1 [==============================] - 0s 126ms/step - loss: 1.0282 - sparse_categorical_accuracy: 0.7909 - val_loss: 0.9914 - val_sparse_categorical_accuracy: 0.8045\n",
|
|
"Epoch 47/200\n",
|
|
"1/1 [==============================] - 0s 134ms/step - loss: 1.0150 - sparse_categorical_accuracy: 0.7932 - val_loss: 0.9783 - val_sparse_categorical_accuracy: 0.8065\n",
|
|
"Epoch 48/200\n",
|
|
"1/1 [==============================] - 0s 133ms/step - loss: 1.0021 - sparse_categorical_accuracy: 0.7953 - val_loss: 0.9657 - val_sparse_categorical_accuracy: 0.8093\n",
|
|
"Epoch 49/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 0.9896 - sparse_categorical_accuracy: 0.7975 - val_loss: 0.9534 - val_sparse_categorical_accuracy: 0.8111\n",
|
|
"Epoch 50/200\n",
|
|
"1/1 [==============================] - 0s 121ms/step - loss: 0.9776 - sparse_categorical_accuracy: 0.7994 - val_loss: 0.9416 - val_sparse_categorical_accuracy: 0.8124\n",
|
|
"Epoch 51/200\n",
|
|
"1/1 [==============================] - 0s 112ms/step - loss: 0.9659 - sparse_categorical_accuracy: 0.8015 - val_loss: 0.9301 - val_sparse_categorical_accuracy: 0.8144\n",
|
|
"Epoch 52/200\n",
|
|
"1/1 [==============================] - 0s 116ms/step - loss: 0.9546 - sparse_categorical_accuracy: 0.8034 - val_loss: 0.9190 - val_sparse_categorical_accuracy: 0.8153\n",
|
|
"Epoch 53/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.9437 - sparse_categorical_accuracy: 0.8054 - val_loss: 0.9082 - val_sparse_categorical_accuracy: 0.8168\n",
|
|
"Epoch 54/200\n",
|
|
"1/1 [==============================] - 0s 121ms/step - loss: 0.9330 - sparse_categorical_accuracy: 0.8072 - val_loss: 0.8978 - val_sparse_categorical_accuracy: 0.8190\n",
|
|
"Epoch 55/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.9227 - sparse_categorical_accuracy: 0.8092 - val_loss: 0.8877 - val_sparse_categorical_accuracy: 0.8216\n",
|
|
"Epoch 56/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.9127 - sparse_categorical_accuracy: 0.8106 - val_loss: 0.8779 - val_sparse_categorical_accuracy: 0.8230\n",
|
|
"Epoch 57/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 0.9031 - sparse_categorical_accuracy: 0.8123 - val_loss: 0.8684 - val_sparse_categorical_accuracy: 0.8245\n",
|
|
"Epoch 58/200\n",
|
|
"1/1 [==============================] - 0s 121ms/step - loss: 0.8937 - sparse_categorical_accuracy: 0.8135 - val_loss: 0.8592 - val_sparse_categorical_accuracy: 0.8269\n",
|
|
"Epoch 59/200\n",
|
|
"1/1 [==============================] - 0s 130ms/step - loss: 0.8845 - sparse_categorical_accuracy: 0.8151 - val_loss: 0.8503 - val_sparse_categorical_accuracy: 0.8294\n",
|
|
"Epoch 60/200\n",
|
|
"1/1 [==============================] - 0s 123ms/step - loss: 0.8757 - sparse_categorical_accuracy: 0.8163 - val_loss: 0.8416 - val_sparse_categorical_accuracy: 0.8309\n",
|
|
"Epoch 61/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.8671 - sparse_categorical_accuracy: 0.8177 - val_loss: 0.8332 - val_sparse_categorical_accuracy: 0.8324\n",
|
|
"Epoch 62/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.8587 - sparse_categorical_accuracy: 0.8189 - val_loss: 0.8250 - val_sparse_categorical_accuracy: 0.8336\n",
|
|
"Epoch 63/200\n",
|
|
"1/1 [==============================] - 0s 156ms/step - loss: 0.8505 - sparse_categorical_accuracy: 0.8201 - val_loss: 0.8171 - val_sparse_categorical_accuracy: 0.8346\n",
|
|
"Epoch 64/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.8426 - sparse_categorical_accuracy: 0.8214 - val_loss: 0.8093 - val_sparse_categorical_accuracy: 0.8364\n",
|
|
"Epoch 65/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.8349 - sparse_categorical_accuracy: 0.8228 - val_loss: 0.8018 - val_sparse_categorical_accuracy: 0.8373\n",
|
|
"Epoch 66/200\n",
|
|
"1/1 [==============================] - 0s 126ms/step - loss: 0.8274 - sparse_categorical_accuracy: 0.8242 - val_loss: 0.7945 - val_sparse_categorical_accuracy: 0.8389\n",
|
|
"Epoch 67/200\n",
|
|
"1/1 [==============================] - 0s 124ms/step - loss: 0.8202 - sparse_categorical_accuracy: 0.8255 - val_loss: 0.7874 - val_sparse_categorical_accuracy: 0.8395\n",
|
|
"Epoch 68/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 0.8131 - sparse_categorical_accuracy: 0.8264 - val_loss: 0.7804 - val_sparse_categorical_accuracy: 0.8404\n",
|
|
"Epoch 69/200\n",
|
|
"1/1 [==============================] - 0s 120ms/step - loss: 0.8061 - sparse_categorical_accuracy: 0.8274 - val_loss: 0.7737 - val_sparse_categorical_accuracy: 0.8418\n",
|
|
"Epoch 70/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 0.7994 - sparse_categorical_accuracy: 0.8286 - val_loss: 0.7671 - val_sparse_categorical_accuracy: 0.8428\n",
|
|
"Epoch 71/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 0.7928 - sparse_categorical_accuracy: 0.8294 - val_loss: 0.7607 - val_sparse_categorical_accuracy: 0.8441\n",
|
|
"Epoch 72/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.7864 - sparse_categorical_accuracy: 0.8305 - val_loss: 0.7544 - val_sparse_categorical_accuracy: 0.8451\n",
|
|
"Epoch 73/200\n",
|
|
"1/1 [==============================] - 0s 116ms/step - loss: 0.7802 - sparse_categorical_accuracy: 0.8316 - val_loss: 0.7483 - val_sparse_categorical_accuracy: 0.8459\n",
|
|
"Epoch 74/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.7741 - sparse_categorical_accuracy: 0.8325 - val_loss: 0.7424 - val_sparse_categorical_accuracy: 0.8467\n",
|
|
"Epoch 75/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.7681 - sparse_categorical_accuracy: 0.8335 - val_loss: 0.7365 - val_sparse_categorical_accuracy: 0.8476\n",
|
|
"Epoch 76/200\n",
|
|
"1/1 [==============================] - 0s 115ms/step - loss: 0.7623 - sparse_categorical_accuracy: 0.8343 - val_loss: 0.7309 - val_sparse_categorical_accuracy: 0.8484\n",
|
|
"Epoch 77/200\n",
|
|
"1/1 [==============================] - 0s 132ms/step - loss: 0.7566 - sparse_categorical_accuracy: 0.8350 - val_loss: 0.7253 - val_sparse_categorical_accuracy: 0.8493\n",
|
|
"Epoch 78/200\n",
|
|
"1/1 [==============================] - 0s 124ms/step - loss: 0.7511 - sparse_categorical_accuracy: 0.8359 - val_loss: 0.7199 - val_sparse_categorical_accuracy: 0.8500\n",
|
|
"Epoch 79/200\n",
|
|
"1/1 [==============================] - 0s 132ms/step - loss: 0.7457 - sparse_categorical_accuracy: 0.8368 - val_loss: 0.7146 - val_sparse_categorical_accuracy: 0.8516\n",
|
|
"Epoch 80/200\n",
|
|
"1/1 [==============================] - 0s 125ms/step - loss: 0.7404 - sparse_categorical_accuracy: 0.8378 - val_loss: 0.7095 - val_sparse_categorical_accuracy: 0.8526\n",
|
|
"Epoch 81/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.7352 - sparse_categorical_accuracy: 0.8387 - val_loss: 0.7044 - val_sparse_categorical_accuracy: 0.8534\n",
|
|
"Epoch 82/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.7301 - sparse_categorical_accuracy: 0.8393 - val_loss: 0.6995 - val_sparse_categorical_accuracy: 0.8545\n",
|
|
"Epoch 83/200\n",
|
|
"1/1 [==============================] - 0s 124ms/step - loss: 0.7252 - sparse_categorical_accuracy: 0.8401 - val_loss: 0.6947 - val_sparse_categorical_accuracy: 0.8553\n",
|
|
"Epoch 84/200\n",
|
|
"1/1 [==============================] - 0s 122ms/step - loss: 0.7203 - sparse_categorical_accuracy: 0.8409 - val_loss: 0.6900 - val_sparse_categorical_accuracy: 0.8559\n",
|
|
"Epoch 85/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.7156 - sparse_categorical_accuracy: 0.8415 - val_loss: 0.6853 - val_sparse_categorical_accuracy: 0.8560\n",
|
|
"Epoch 86/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.7110 - sparse_categorical_accuracy: 0.8425 - val_loss: 0.6808 - val_sparse_categorical_accuracy: 0.8572\n",
|
|
"Epoch 87/200\n",
|
|
"1/1 [==============================] - 0s 116ms/step - loss: 0.7064 - sparse_categorical_accuracy: 0.8435 - val_loss: 0.6764 - val_sparse_categorical_accuracy: 0.8575\n",
|
|
"Epoch 88/200\n",
|
|
"1/1 [==============================] - 0s 120ms/step - loss: 0.7020 - sparse_categorical_accuracy: 0.8442 - val_loss: 0.6721 - val_sparse_categorical_accuracy: 0.8575\n",
|
|
"Epoch 89/200\n",
|
|
"1/1 [==============================] - 0s 114ms/step - loss: 0.6976 - sparse_categorical_accuracy: 0.8448 - val_loss: 0.6678 - val_sparse_categorical_accuracy: 0.8590\n",
|
|
"Epoch 90/200\n",
|
|
"1/1 [==============================] - 0s 114ms/step - loss: 0.6933 - sparse_categorical_accuracy: 0.8457 - val_loss: 0.6637 - val_sparse_categorical_accuracy: 0.8594\n",
|
|
"Epoch 91/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.6891 - sparse_categorical_accuracy: 0.8466 - val_loss: 0.6596 - val_sparse_categorical_accuracy: 0.8598\n",
|
|
"Epoch 92/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.6850 - sparse_categorical_accuracy: 0.8472 - val_loss: 0.6556 - val_sparse_categorical_accuracy: 0.8605\n",
|
|
"Epoch 93/200\n",
|
|
"1/1 [==============================] - 0s 137ms/step - loss: 0.6810 - sparse_categorical_accuracy: 0.8478 - val_loss: 0.6517 - val_sparse_categorical_accuracy: 0.8610\n",
|
|
"Epoch 94/200\n",
|
|
"1/1 [==============================] - 0s 129ms/step - loss: 0.6771 - sparse_categorical_accuracy: 0.8486 - val_loss: 0.6478 - val_sparse_categorical_accuracy: 0.8618\n",
|
|
"Epoch 95/200\n",
|
|
"1/1 [==============================] - 0s 122ms/step - loss: 0.6732 - sparse_categorical_accuracy: 0.8491 - val_loss: 0.6441 - val_sparse_categorical_accuracy: 0.8623\n",
|
|
"Epoch 96/200\n",
|
|
"1/1 [==============================] - 0s 126ms/step - loss: 0.6694 - sparse_categorical_accuracy: 0.8496 - val_loss: 0.6404 - val_sparse_categorical_accuracy: 0.8626\n",
|
|
"Epoch 97/200\n",
|
|
"1/1 [==============================] - 0s 137ms/step - loss: 0.6657 - sparse_categorical_accuracy: 0.8501 - val_loss: 0.6368 - val_sparse_categorical_accuracy: 0.8628\n",
|
|
"Epoch 98/200\n",
|
|
"1/1 [==============================] - 0s 126ms/step - loss: 0.6620 - sparse_categorical_accuracy: 0.8505 - val_loss: 0.6332 - val_sparse_categorical_accuracy: 0.8634\n",
|
|
"Epoch 99/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.6584 - sparse_categorical_accuracy: 0.8511 - val_loss: 0.6297 - val_sparse_categorical_accuracy: 0.8642\n",
|
|
"Epoch 100/200\n",
|
|
"1/1 [==============================] - 0s 120ms/step - loss: 0.6549 - sparse_categorical_accuracy: 0.8521 - val_loss: 0.6263 - val_sparse_categorical_accuracy: 0.8649\n",
|
|
"Epoch 101/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.6514 - sparse_categorical_accuracy: 0.8526 - val_loss: 0.6229 - val_sparse_categorical_accuracy: 0.8650\n",
|
|
"Epoch 102/200\n",
|
|
"1/1 [==============================] - 0s 114ms/step - loss: 0.6480 - sparse_categorical_accuracy: 0.8532 - val_loss: 0.6196 - val_sparse_categorical_accuracy: 0.8659\n",
|
|
"Epoch 103/200\n",
|
|
"1/1 [==============================] - 0s 120ms/step - loss: 0.6446 - sparse_categorical_accuracy: 0.8538 - val_loss: 0.6164 - val_sparse_categorical_accuracy: 0.8669\n",
|
|
"Epoch 104/200\n",
|
|
"1/1 [==============================] - 0s 116ms/step - loss: 0.6414 - sparse_categorical_accuracy: 0.8543 - val_loss: 0.6132 - val_sparse_categorical_accuracy: 0.8671\n",
|
|
"Epoch 105/200\n",
|
|
"1/1 [==============================] - 0s 115ms/step - loss: 0.6381 - sparse_categorical_accuracy: 0.8550 - val_loss: 0.6101 - val_sparse_categorical_accuracy: 0.8677\n",
|
|
"Epoch 106/200\n",
|
|
"1/1 [==============================] - 0s 114ms/step - loss: 0.6349 - sparse_categorical_accuracy: 0.8556 - val_loss: 0.6070 - val_sparse_categorical_accuracy: 0.8684\n",
|
|
"Epoch 107/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.6318 - sparse_categorical_accuracy: 0.8561 - val_loss: 0.6040 - val_sparse_categorical_accuracy: 0.8684\n",
|
|
"Epoch 108/200\n",
|
|
"1/1 [==============================] - 0s 114ms/step - loss: 0.6288 - sparse_categorical_accuracy: 0.8568 - val_loss: 0.6010 - val_sparse_categorical_accuracy: 0.8693\n",
|
|
"Epoch 109/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.6257 - sparse_categorical_accuracy: 0.8573 - val_loss: 0.5981 - val_sparse_categorical_accuracy: 0.8697\n",
|
|
"Epoch 110/200\n",
|
|
"1/1 [==============================] - 0s 115ms/step - loss: 0.6228 - sparse_categorical_accuracy: 0.8578 - val_loss: 0.5952 - val_sparse_categorical_accuracy: 0.8696\n",
|
|
"Epoch 111/200\n",
|
|
"1/1 [==============================] - 0s 116ms/step - loss: 0.6198 - sparse_categorical_accuracy: 0.8583 - val_loss: 0.5924 - val_sparse_categorical_accuracy: 0.8701\n",
|
|
"Epoch 112/200\n",
|
|
"1/1 [==============================] - 0s 120ms/step - loss: 0.6170 - sparse_categorical_accuracy: 0.8587 - val_loss: 0.5896 - val_sparse_categorical_accuracy: 0.8706\n",
|
|
"Epoch 113/200\n",
|
|
"1/1 [==============================] - 0s 115ms/step - loss: 0.6141 - sparse_categorical_accuracy: 0.8594 - val_loss: 0.5868 - val_sparse_categorical_accuracy: 0.8714\n",
|
|
"Epoch 114/200\n",
|
|
"1/1 [==============================] - 0s 114ms/step - loss: 0.6113 - sparse_categorical_accuracy: 0.8599 - val_loss: 0.5841 - val_sparse_categorical_accuracy: 0.8720\n",
|
|
"Epoch 115/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.6086 - sparse_categorical_accuracy: 0.8603 - val_loss: 0.5815 - val_sparse_categorical_accuracy: 0.8726\n",
|
|
"Epoch 116/200\n",
|
|
"1/1 [==============================] - 0s 124ms/step - loss: 0.6059 - sparse_categorical_accuracy: 0.8609 - val_loss: 0.5789 - val_sparse_categorical_accuracy: 0.8734\n",
|
|
"Epoch 117/200\n",
|
|
"1/1 [==============================] - 0s 127ms/step - loss: 0.6033 - sparse_categorical_accuracy: 0.8613 - val_loss: 0.5763 - val_sparse_categorical_accuracy: 0.8744\n",
|
|
"Epoch 118/200\n",
|
|
"1/1 [==============================] - 0s 124ms/step - loss: 0.6006 - sparse_categorical_accuracy: 0.8616 - val_loss: 0.5738 - val_sparse_categorical_accuracy: 0.8747\n",
|
|
"Epoch 119/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.5981 - sparse_categorical_accuracy: 0.8620 - val_loss: 0.5713 - val_sparse_categorical_accuracy: 0.8753\n",
|
|
"Epoch 120/200\n",
|
|
"1/1 [==============================] - 0s 120ms/step - loss: 0.5955 - sparse_categorical_accuracy: 0.8624 - val_loss: 0.5688 - val_sparse_categorical_accuracy: 0.8757\n",
|
|
"Epoch 121/200\n",
|
|
"1/1 [==============================] - 0s 115ms/step - loss: 0.5930 - sparse_categorical_accuracy: 0.8629 - val_loss: 0.5664 - val_sparse_categorical_accuracy: 0.8759\n",
|
|
"Epoch 122/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 0.5906 - sparse_categorical_accuracy: 0.8634 - val_loss: 0.5640 - val_sparse_categorical_accuracy: 0.8762\n",
|
|
"Epoch 123/200\n",
|
|
"1/1 [==============================] - 0s 114ms/step - loss: 0.5881 - sparse_categorical_accuracy: 0.8637 - val_loss: 0.5617 - val_sparse_categorical_accuracy: 0.8761\n",
|
|
"Epoch 124/200\n",
|
|
"1/1 [==============================] - 0s 122ms/step - loss: 0.5857 - sparse_categorical_accuracy: 0.8641 - val_loss: 0.5594 - val_sparse_categorical_accuracy: 0.8763\n",
|
|
"Epoch 125/200\n",
|
|
"1/1 [==============================] - 0s 123ms/step - loss: 0.5834 - sparse_categorical_accuracy: 0.8646 - val_loss: 0.5571 - val_sparse_categorical_accuracy: 0.8764\n",
|
|
"Epoch 126/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 0.5810 - sparse_categorical_accuracy: 0.8650 - val_loss: 0.5548 - val_sparse_categorical_accuracy: 0.8766\n",
|
|
"Epoch 127/200\n",
|
|
"1/1 [==============================] - 0s 116ms/step - loss: 0.5788 - sparse_categorical_accuracy: 0.8655 - val_loss: 0.5526 - val_sparse_categorical_accuracy: 0.8771\n",
|
|
"Epoch 128/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.5765 - sparse_categorical_accuracy: 0.8659 - val_loss: 0.5504 - val_sparse_categorical_accuracy: 0.8774\n",
|
|
"Epoch 129/200\n",
|
|
"1/1 [==============================] - 0s 125ms/step - loss: 0.5743 - sparse_categorical_accuracy: 0.8664 - val_loss: 0.5483 - val_sparse_categorical_accuracy: 0.8780\n",
|
|
"Epoch 130/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.5721 - sparse_categorical_accuracy: 0.8666 - val_loss: 0.5462 - val_sparse_categorical_accuracy: 0.8782\n",
|
|
"Epoch 131/200\n",
|
|
"1/1 [==============================] - 0s 140ms/step - loss: 0.5699 - sparse_categorical_accuracy: 0.8671 - val_loss: 0.5441 - val_sparse_categorical_accuracy: 0.8785\n",
|
|
"Epoch 132/200\n",
|
|
"1/1 [==============================] - 0s 128ms/step - loss: 0.5678 - sparse_categorical_accuracy: 0.8674 - val_loss: 0.5420 - val_sparse_categorical_accuracy: 0.8787\n",
|
|
"Epoch 133/200\n",
|
|
"1/1 [==============================] - 0s 116ms/step - loss: 0.5657 - sparse_categorical_accuracy: 0.8677 - val_loss: 0.5400 - val_sparse_categorical_accuracy: 0.8790\n",
|
|
"Epoch 134/200\n",
|
|
"1/1 [==============================] - 0s 116ms/step - loss: 0.5636 - sparse_categorical_accuracy: 0.8683 - val_loss: 0.5380 - val_sparse_categorical_accuracy: 0.8795\n",
|
|
"Epoch 135/200\n",
|
|
"1/1 [==============================] - 0s 159ms/step - loss: 0.5615 - sparse_categorical_accuracy: 0.8687 - val_loss: 0.5360 - val_sparse_categorical_accuracy: 0.8797\n",
|
|
"Epoch 136/200\n",
|
|
"1/1 [==============================] - 0s 125ms/step - loss: 0.5595 - sparse_categorical_accuracy: 0.8690 - val_loss: 0.5340 - val_sparse_categorical_accuracy: 0.8796\n",
|
|
"Epoch 137/200\n",
|
|
"1/1 [==============================] - 0s 128ms/step - loss: 0.5575 - sparse_categorical_accuracy: 0.8694 - val_loss: 0.5321 - val_sparse_categorical_accuracy: 0.8803\n",
|
|
"Epoch 138/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.5555 - sparse_categorical_accuracy: 0.8697 - val_loss: 0.5302 - val_sparse_categorical_accuracy: 0.8806\n",
|
|
"Epoch 139/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.5536 - sparse_categorical_accuracy: 0.8701 - val_loss: 0.5283 - val_sparse_categorical_accuracy: 0.8812\n",
|
|
"Epoch 140/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 0.5517 - sparse_categorical_accuracy: 0.8703 - val_loss: 0.5265 - val_sparse_categorical_accuracy: 0.8814\n",
|
|
"Epoch 141/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 0.5498 - sparse_categorical_accuracy: 0.8706 - val_loss: 0.5247 - val_sparse_categorical_accuracy: 0.8820\n",
|
|
"Epoch 142/200\n",
|
|
"1/1 [==============================] - 0s 121ms/step - loss: 0.5479 - sparse_categorical_accuracy: 0.8710 - val_loss: 0.5229 - val_sparse_categorical_accuracy: 0.8824\n",
|
|
"Epoch 143/200\n",
|
|
"1/1 [==============================] - 0s 114ms/step - loss: 0.5460 - sparse_categorical_accuracy: 0.8713 - val_loss: 0.5211 - val_sparse_categorical_accuracy: 0.8825\n",
|
|
"Epoch 144/200\n",
|
|
"1/1 [==============================] - 0s 123ms/step - loss: 0.5442 - sparse_categorical_accuracy: 0.8717 - val_loss: 0.5193 - val_sparse_categorical_accuracy: 0.8825\n",
|
|
"Epoch 145/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.5424 - sparse_categorical_accuracy: 0.8722 - val_loss: 0.5176 - val_sparse_categorical_accuracy: 0.8829\n",
|
|
"Epoch 146/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 0.5406 - sparse_categorical_accuracy: 0.8725 - val_loss: 0.5159 - val_sparse_categorical_accuracy: 0.8831\n",
|
|
"Epoch 147/200\n",
|
|
"1/1 [==============================] - 0s 122ms/step - loss: 0.5389 - sparse_categorical_accuracy: 0.8728 - val_loss: 0.5142 - val_sparse_categorical_accuracy: 0.8833\n",
|
|
"Epoch 148/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 0.5371 - sparse_categorical_accuracy: 0.8731 - val_loss: 0.5125 - val_sparse_categorical_accuracy: 0.8839\n",
|
|
"Epoch 149/200\n",
|
|
"1/1 [==============================] - 0s 115ms/step - loss: 0.5354 - sparse_categorical_accuracy: 0.8734 - val_loss: 0.5109 - val_sparse_categorical_accuracy: 0.8841\n",
|
|
"Epoch 150/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 0.5337 - sparse_categorical_accuracy: 0.8736 - val_loss: 0.5092 - val_sparse_categorical_accuracy: 0.8847\n",
|
|
"Epoch 151/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.5320 - sparse_categorical_accuracy: 0.8737 - val_loss: 0.5076 - val_sparse_categorical_accuracy: 0.8848\n",
|
|
"Epoch 152/200\n",
|
|
"1/1 [==============================] - 0s 123ms/step - loss: 0.5304 - sparse_categorical_accuracy: 0.8739 - val_loss: 0.5060 - val_sparse_categorical_accuracy: 0.8850\n",
|
|
"Epoch 153/200\n",
|
|
"1/1 [==============================] - 0s 120ms/step - loss: 0.5287 - sparse_categorical_accuracy: 0.8740 - val_loss: 0.5045 - val_sparse_categorical_accuracy: 0.8853\n",
|
|
"Epoch 154/200\n",
|
|
"1/1 [==============================] - 0s 123ms/step - loss: 0.5271 - sparse_categorical_accuracy: 0.8744 - val_loss: 0.5029 - val_sparse_categorical_accuracy: 0.8856\n",
|
|
"Epoch 155/200\n",
|
|
"1/1 [==============================] - 0s 134ms/step - loss: 0.5255 - sparse_categorical_accuracy: 0.8747 - val_loss: 0.5014 - val_sparse_categorical_accuracy: 0.8863\n",
|
|
"Epoch 156/200\n",
|
|
"1/1 [==============================] - 0s 131ms/step - loss: 0.5239 - sparse_categorical_accuracy: 0.8749 - val_loss: 0.4998 - val_sparse_categorical_accuracy: 0.8861\n",
|
|
"Epoch 157/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.5224 - sparse_categorical_accuracy: 0.8752 - val_loss: 0.4983 - val_sparse_categorical_accuracy: 0.8864\n",
|
|
"Epoch 158/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.5208 - sparse_categorical_accuracy: 0.8755 - val_loss: 0.4969 - val_sparse_categorical_accuracy: 0.8864\n",
|
|
"Epoch 159/200\n",
|
|
"1/1 [==============================] - 0s 121ms/step - loss: 0.5193 - sparse_categorical_accuracy: 0.8756 - val_loss: 0.4954 - val_sparse_categorical_accuracy: 0.8868\n",
|
|
"Epoch 160/200\n",
|
|
"1/1 [==============================] - 0s 113ms/step - loss: 0.5178 - sparse_categorical_accuracy: 0.8759 - val_loss: 0.4940 - val_sparse_categorical_accuracy: 0.8872\n",
|
|
"Epoch 161/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.5163 - sparse_categorical_accuracy: 0.8761 - val_loss: 0.4925 - val_sparse_categorical_accuracy: 0.8873\n",
|
|
"Epoch 162/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.5148 - sparse_categorical_accuracy: 0.8764 - val_loss: 0.4911 - val_sparse_categorical_accuracy: 0.8875\n",
|
|
"Epoch 163/200\n",
|
|
"1/1 [==============================] - 0s 120ms/step - loss: 0.5134 - sparse_categorical_accuracy: 0.8766 - val_loss: 0.4897 - val_sparse_categorical_accuracy: 0.8878\n",
|
|
"Epoch 164/200\n",
|
|
"1/1 [==============================] - 0s 147ms/step - loss: 0.5119 - sparse_categorical_accuracy: 0.8767 - val_loss: 0.4883 - val_sparse_categorical_accuracy: 0.8878\n",
|
|
"Epoch 165/200\n",
|
|
"1/1 [==============================] - 0s 130ms/step - loss: 0.5105 - sparse_categorical_accuracy: 0.8770 - val_loss: 0.4870 - val_sparse_categorical_accuracy: 0.8879\n",
|
|
"Epoch 166/200\n",
|
|
"1/1 [==============================] - 0s 114ms/step - loss: 0.5091 - sparse_categorical_accuracy: 0.8772 - val_loss: 0.4856 - val_sparse_categorical_accuracy: 0.8881\n",
|
|
"Epoch 167/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.5077 - sparse_categorical_accuracy: 0.8775 - val_loss: 0.4843 - val_sparse_categorical_accuracy: 0.8882\n",
|
|
"Epoch 168/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.5063 - sparse_categorical_accuracy: 0.8776 - val_loss: 0.4829 - val_sparse_categorical_accuracy: 0.8887\n",
|
|
"Epoch 169/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.5049 - sparse_categorical_accuracy: 0.8777 - val_loss: 0.4816 - val_sparse_categorical_accuracy: 0.8888\n",
|
|
"Epoch 170/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 0.5036 - sparse_categorical_accuracy: 0.8779 - val_loss: 0.4803 - val_sparse_categorical_accuracy: 0.8891\n",
|
|
"Epoch 171/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 0.5022 - sparse_categorical_accuracy: 0.8781 - val_loss: 0.4791 - val_sparse_categorical_accuracy: 0.8892\n",
|
|
"Epoch 172/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 0.5009 - sparse_categorical_accuracy: 0.8785 - val_loss: 0.4778 - val_sparse_categorical_accuracy: 0.8893\n",
|
|
"Epoch 173/200\n",
|
|
"1/1 [==============================] - 0s 121ms/step - loss: 0.4996 - sparse_categorical_accuracy: 0.8787 - val_loss: 0.4765 - val_sparse_categorical_accuracy: 0.8895\n",
|
|
"Epoch 174/200\n",
|
|
"1/1 [==============================] - 0s 127ms/step - loss: 0.4983 - sparse_categorical_accuracy: 0.8790 - val_loss: 0.4753 - val_sparse_categorical_accuracy: 0.8897\n",
|
|
"Epoch 175/200\n",
|
|
"1/1 [==============================] - 0s 134ms/step - loss: 0.4970 - sparse_categorical_accuracy: 0.8792 - val_loss: 0.4741 - val_sparse_categorical_accuracy: 0.8899\n",
|
|
"Epoch 176/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.4957 - sparse_categorical_accuracy: 0.8794 - val_loss: 0.4729 - val_sparse_categorical_accuracy: 0.8903\n",
|
|
"Epoch 177/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.4945 - sparse_categorical_accuracy: 0.8795 - val_loss: 0.4717 - val_sparse_categorical_accuracy: 0.8904\n",
|
|
"Epoch 178/200\n",
|
|
"1/1 [==============================] - 0s 116ms/step - loss: 0.4932 - sparse_categorical_accuracy: 0.8798 - val_loss: 0.4705 - val_sparse_categorical_accuracy: 0.8905\n",
|
|
"Epoch 179/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.4920 - sparse_categorical_accuracy: 0.8801 - val_loss: 0.4693 - val_sparse_categorical_accuracy: 0.8908\n",
|
|
"Epoch 180/200\n",
|
|
"1/1 [==============================] - 0s 132ms/step - loss: 0.4908 - sparse_categorical_accuracy: 0.8803 - val_loss: 0.4681 - val_sparse_categorical_accuracy: 0.8913\n",
|
|
"Epoch 181/200\n",
|
|
"1/1 [==============================] - 0s 115ms/step - loss: 0.4896 - sparse_categorical_accuracy: 0.8806 - val_loss: 0.4670 - val_sparse_categorical_accuracy: 0.8916\n",
|
|
"Epoch 182/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.4884 - sparse_categorical_accuracy: 0.8807 - val_loss: 0.4658 - val_sparse_categorical_accuracy: 0.8915\n",
|
|
"Epoch 183/200\n",
|
|
"1/1 [==============================] - 0s 117ms/step - loss: 0.4872 - sparse_categorical_accuracy: 0.8809 - val_loss: 0.4647 - val_sparse_categorical_accuracy: 0.8919\n",
|
|
"Epoch 184/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 0.4860 - sparse_categorical_accuracy: 0.8812 - val_loss: 0.4636 - val_sparse_categorical_accuracy: 0.8921\n",
|
|
"Epoch 185/200\n",
|
|
"1/1 [==============================] - 0s 118ms/step - loss: 0.4848 - sparse_categorical_accuracy: 0.8814 - val_loss: 0.4625 - val_sparse_categorical_accuracy: 0.8925\n",
|
|
"Epoch 186/200\n",
|
|
"1/1 [==============================] - 0s 119ms/step - loss: 0.4837 - sparse_categorical_accuracy: 0.8816 - val_loss: 0.4614 - val_sparse_categorical_accuracy: 0.8927\n",
|
|
"Epoch 187/200\n",
|
|
"1/1 [==============================] - 0s 120ms/step - loss: 0.4826 - sparse_categorical_accuracy: 0.8818 - val_loss: 0.4603 - val_sparse_categorical_accuracy: 0.8927\n",
|
|
"Epoch 188/200\n",
|
|
"1/1 [==============================] - 0s 116ms/step - loss: 0.4814 - sparse_categorical_accuracy: 0.8820 - val_loss: 0.4592 - val_sparse_categorical_accuracy: 0.8927\n",
|
|
"Epoch 189/200\n",
|
|
"1/1 [==============================] - 0s 114ms/step - loss: 0.4803 - sparse_categorical_accuracy: 0.8821 - val_loss: 0.4582 - val_sparse_categorical_accuracy: 0.8934\n",
|
|
"Epoch 190/200\n",
|
|
"1/1 [==============================] - 0s 116ms/step - loss: 0.4792 - sparse_categorical_accuracy: 0.8823 - val_loss: 0.4571 - val_sparse_categorical_accuracy: 0.8934\n",
|
|
"Epoch 191/200\n",
|
|
"1/1 [==============================] - 0s 116ms/step - loss: 0.4781 - sparse_categorical_accuracy: 0.8826 - val_loss: 0.4561 - val_sparse_categorical_accuracy: 0.8935\n",
|
|
"Epoch 192/200\n",
|
|
"1/1 [==============================] - 0s 135ms/step - loss: 0.4770 - sparse_categorical_accuracy: 0.8827 - val_loss: 0.4550 - val_sparse_categorical_accuracy: 0.8935\n",
|
|
"Epoch 193/200\n",
|
|
"1/1 [==============================] - 0s 143ms/step - loss: 0.4759 - sparse_categorical_accuracy: 0.8828 - val_loss: 0.4540 - val_sparse_categorical_accuracy: 0.8934\n",
|
|
"Epoch 194/200\n",
|
|
"1/1 [==============================] - 0s 127ms/step - loss: 0.4749 - sparse_categorical_accuracy: 0.8829 - val_loss: 0.4530 - val_sparse_categorical_accuracy: 0.8936\n",
|
|
"Epoch 195/200\n",
|
|
"1/1 [==============================] - 0s 126ms/step - loss: 0.4738 - sparse_categorical_accuracy: 0.8832 - val_loss: 0.4520 - val_sparse_categorical_accuracy: 0.8937\n",
|
|
"Epoch 196/200\n",
|
|
"1/1 [==============================] - 0s 115ms/step - loss: 0.4728 - sparse_categorical_accuracy: 0.8834 - val_loss: 0.4510 - val_sparse_categorical_accuracy: 0.8939\n",
|
|
"Epoch 197/200\n",
|
|
"1/1 [==============================] - 0s 87ms/step - loss: 0.4717 - sparse_categorical_accuracy: 0.8835 - val_loss: 0.4500 - val_sparse_categorical_accuracy: 0.8940\n",
|
|
"Epoch 198/200\n",
|
|
"1/1 [==============================] - 0s 90ms/step - loss: 0.4707 - sparse_categorical_accuracy: 0.8837 - val_loss: 0.4490 - val_sparse_categorical_accuracy: 0.8941\n",
|
|
"Epoch 199/200\n",
|
|
"1/1 [==============================] - 0s 86ms/step - loss: 0.4697 - sparse_categorical_accuracy: 0.8839 - val_loss: 0.4481 - val_sparse_categorical_accuracy: 0.8942\n",
|
|
"Epoch 200/200\n",
|
|
"1/1 [==============================] - 0s 88ms/step - loss: 0.4687 - sparse_categorical_accuracy: 0.8842 - val_loss: 0.4471 - val_sparse_categorical_accuracy: 0.8946\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"keras_model.compile(\n",
|
|
" optimizer=tf.keras.optimizers.Adam(0.001),\n",
|
|
" loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),\n",
|
|
" metrics=[tf.keras.metrics.SparseCategoricalAccuracy()],\n",
|
|
")\n",
|
|
"\n",
|
|
"# Train loop\n",
|
|
"history = keras_model.fit(\n",
|
|
" train_images,\n",
|
|
" train_labels0,\n",
|
|
" batch_size=len(train_images),\n",
|
|
" epochs=200,\n",
|
|
" validation_data=(test_images, test_labels0),\n",
|
|
")"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"## View the training performance\n",
|
|
"\n",
|
|
"- When the Keras fit loop runs, it returns a \"history\" object, which includes a dictionary of the trianing history.\n",
|
|
"\n",
|
|
"- Hence we can plot graphs of the training performance (Accuracy, Loss), for both the \"Training\" and \"Validation\" sets...."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 6,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"Key: loss\n",
|
|
"Key: sparse_categorical_accuracy\n",
|
|
"Key: val_loss\n",
|
|
"Key: val_sparse_categorical_accuracy\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"# first show keys for data series recorded by fit loop:\n",
|
|
"for item in history.history:\n",
|
|
" print(\"Key:\",item)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 5,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"image/png": "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",
|
|
"text/plain": [
|
|
"<Figure size 640x480 with 1 Axes>"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
},
|
|
{
|
|
"data": {
|
|
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAjcAAAHHCAYAAABDUnkqAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/H5lhTAAAACXBIWXMAAA9hAAAPYQGoP6dpAABwwElEQVR4nO3dd3wUdf7H8dfuZrPpCSE9BEIvShMEwbNTbFjOgqBSVDwLd3o5PY+fJ6CnYjvkvLPceSIWVGynd4pIDCAiVRAVBKSF0BIIIb3sZnd+f2yyEBIggU022byfj0ceuzs7M/v9ZELy5vv9zozJMAwDERERET9h9nUDRERERLxJ4UZERET8isKNiIiI+BWFGxEREfErCjciIiLiVxRuRERExK8o3IiIiIhfUbgRERERv6JwIyIiIn5F4UZEvMZkMjF9+vQGb5eZmYnJZGLOnDleb5OItD4KNyJ+Zs6cOZhMJkwmE8uWLav1vmEYpKSkYDKZuPLKK33QQu+YP38+JpOJpKQkXC6Xr5sjIs2Iwo2InwoKCuKdd96ptfzrr79mz5492Gw2H7TKe+bOnUtqair79+9n0aJFvm6OiDQjCjcifuryyy/ngw8+oLKyssbyd955hwEDBpCQkOCjlp2+kpISPv30U9LS0ujfvz9z5871dZOOq6SkxNdNEGl1FG5E/NSYMWM4dOgQ6enpnmV2u50PP/yQsWPH1rlNSUkJf/jDH0hJScFms9G9e3eee+45DMOosV5FRQW///3viY2NJTw8nKuuuoo9e/bUuc+9e/dy2223ER8fj81m44wzzmD27NmnVdt//vMfysrKuOGGG7jpppv4+OOPKS8vr7VeeXk506dPp1u3bgQFBZGYmMivf/1rtm/f7lnH5XLxt7/9jd69exMUFERsbCyXXnop3333HXDi+UDHzjGaPn06JpOJn3/+mbFjx9KmTRt+9atfAfDjjz8yYcIEOnXqRFBQEAkJCdx2220cOnSozu/Z7bffTlJSEjabjY4dO3L33Xdjt9vZsWMHJpOJ559/vtZ2y5cvx2Qy8e677zb0WyriVwJ83QARaRypqakMGTKEd999l8suuwyAL774goKCAm666SZeeOGFGusbhsFVV13F4sWLuf322+nXrx9ffvklDz74IHv37q3xx/SOO+7g7bffZuzYsQwdOpRFixZxxRVX1GpDTk4O55xzDiaTicmTJxMbG8sXX3zB7bffTmFhIffff/8p1TZ37lwuuugiEhISuOmmm/jTn/7E//73P2644QbPOk6nkyuvvJKMjAxuuukm7rvvPoqKikhPT2fDhg107twZgNtvv505c+Zw2WWXcccdd1BZWck333zDypUrGThw4Cm174YbbqBr1648+eSTnmCYnp7Ojh07mDhxIgkJCWzcuJF//etfbNy4kZUrV2IymQDYt28fgwYNIj8/nzvvvJMePXqwd+9ePvzwQ0pLS+nUqRPnnnsuc+fO5fe//32t70t4eDhXX331KbVbxG8YIuJXXn/9dQMw1qxZY/zjH/8wwsPDjdLSUsMwDOOGG24wLrroIsMwDKNDhw7GFVdc4dnuk08+MQDj8ccfr7G/66+/3jCZTMa2bdsMwzCM9evXG4Bxzz331Fhv7NixBmBMmzbNs+z22283EhMTjdzc3Brr3nTTTUZkZKSnXTt37jQA4/XXXz9pfTk5OUZAQIDx6quvepYNHTrUuPrqq2usN3v2bAMwZs6cWWsfLpfLMAzDWLRokQEYv/vd7467zonadmy906ZNMwBjzJgxtdatrvVo7777rgEYS5cu9SwbN26cYTabjTVr1hy3Tf/85z8NwNi0aZPnPbvdbsTExBjjx4+vtZ1Ia6NhKRE/duONN1JWVsZnn31GUVERn3322XGHpObPn4/FYuF3v/tdjeV/+MMfMAyDL774wrMeUGu9Y3thDMPgo48+YtSoURiGQW5urudr5MiRFBQUsG7dugbX9N5772E2m7nuuus8y8aMGcMXX3zB4cOHPcs++ugjYmJi+O1vf1trH9W9JB999BEmk4lp06Ydd51Tcdddd9VaFhwc7HleXl5Obm4u55xzDoDn++Byufjkk08YNWpUnb1G1W268cYbCQoKqjHX6MsvvyQ3N5dbbrnllNst4i8UbkT8WGxsLMOGDeOdd97h448/xul0cv3119e57q5du0hKSiI8PLzG8p49e3rer340m82eYZ1q3bt3r/H64MGD5Ofn869//YvY2NgaXxMnTgTgwIEDDa7p7bffZtCgQRw6dIht27axbds2+vfvj91u54MPPvCst337drp3705AwPFH37dv305SUhLR0dENbseJdOzYsdayvLw87rvvPuLj4wkODiY2NtazXkFBAeD+nhUWFnLmmWeecP9RUVGMGjWqxtlwc+fOJTk5mYsvvtiLlYi0TJpzI+Lnxo4dy6RJk8jOzuayyy4jKiqqST63+tozt9xyC+PHj69znT59+jRon1u3bmXNmjUAdO3atdb7c+fO5c4772xgS0/seD04TqfzuNsc3UtT7cYbb2T58uU8+OCD9OvXj7CwMFwuF5deeukpXadn3LhxfPDBByxfvpzevXvz3//+l3vuuQezWf9nFVG4EfFz1157Lb/5zW9YuXIl8+bNO+56HTp04KuvvqKoqKhG783mzZs971c/ulwuT89ItS1bttTYX/WZVE6nk2HDhnmllrlz52K1WnnrrbewWCw13lu2bBkvvPACWVlZtG/fns6dO7Nq1SocDgdWq7XO/XXu3Jkvv/ySvLy84/betGnTBoD8/Pway6t7surj8OHDZGRk8OijjzJ16lTP8q1bt9ZYLzY2loiICDZs2HDSfV566aXExsYyd+5cBg8eTGlpKbfeemu92yTizxTxRfxcWFgYL7/8MtOnT2fUqFHHXe/yyy/H6XTyj3/8o8by559/HpPJ5Dnjqvrx2LOtZs2aVeO1xWLhuuuu46OPPqrzj/XBgwcbXMvcuXM577zzGD16NNdff32NrwcffBDAcxr0ddddR25ubq16AM8ZTNdddx2GYfDoo48ed52IiAhiYmJYunRpjfdfeumlere7OogZx5xSf+z3zGw2c8011/C///3Pcyp6XW0CCAgIYMyYMbz//vvMmTOH3r17N7gnTMRfqedGpBU43rDQ0UaNGsVFF13Eww8/TGZmJn379mXhwoV8+umn3H///Z45Nv369WPMmDG89NJLFBQUMHToUDIyMti2bVutfT711FMsXryYwYMHM2nSJHr16kVeXh7r1q3jq6++Ii8vr941rFq1im3btjF58uQ6309OTuass85i7ty5PPTQQ4wbN44333yTtLQ0Vq9ezXnnnUdJSQlfffUV99xzD1dffTUXXXQRt956Ky+88AJbt271DBF98803XHTRRZ7PuuOOO3jqqae44447GDhwIEuXLuWXX36pd9sjIiI4//zzeeaZZ3A4HCQnJ7Nw4UJ27txZa90nn3yShQsXcsEFF3DnnXfSs2dP9u/fzwcffMCyZctqDCuOGzeOF154gcWLF/P000/Xuz0ifs93J2qJSGM4+lTwEzn2VHDDMIyioiLj97//vZGUlGRYrVaja9euxrPPPus5BblaWVmZ8bvf/c5o27atERoaaowaNcrYvXt3rVOjDcN96va9995rpKSkGFar1UhISDAuueQS41//+pdnnfqcCv7b3/7WAIzt27cfd53p06cbgPHDDz8YhuE+/frhhx82Onbs6Pns66+/vsY+KisrjWeffdbo0aOHERgYaMTGxhqXXXaZsXbtWs86paWlxu23325ERkYa4eHhxo033mgcOHDguKeCHzx4sFbb9uzZY1x77bVGVFSUERkZadxwww3Gvn376vye7dq1yxg3bpwRGxtr2Gw2o1OnTsa9995rVFRU1NrvGWecYZjNZmPPnj3H/b6ItDYmwzimn1RERFqM/v37Ex0dTUZGhq+bItJsaM6NiEgL9d1337F+/XrGjRvn66aINCvquRERaWE2bNjA2rVr+etf/0pubi47duwgKCjI180SaTbUcyMi0sJ8+OGHTJw4EYfDwbvvvqtgI3IM9dyIiIiIX1HPjYiIiPgVhRsRERHxK63uIn4ul4t9+/YRHh5+Wnf9FRERkaZjGAZFRUUkJSWd9B5qrS7c7Nu3j5SUFF83Q0RERE7B7t27adeu3QnXaXXhpvqGgLt37yYiIsKr+3Y4HCxcuJARI0Yc90Z9LZm/1weq0R/4e32gGv2Bv9cH3q+xsLCQlJSUGjf2PZ5WF26qh6IiIiIaJdyEhIQQERHhlz+s/l4fqEZ/4O/1gWr0B/5eHzRejfWZUqIJxSIiIuJXFG5ERETEryjciIiIiF9pdXNu6svpdOJwOBq0jcPhICAggPLycpxOZyO1zHdaan1WqxWLxeLrZoiISBNRuDmGYRhkZ2eTn59/StsmJCSwe/duv7yGTkuuLyoqioSEhBbXbhERaTiFm2NUB5u4uDhCQkIa9MfQ5XJRXFxMWFjYSS8w1BK1xPoMw6C0tJQDBw4AkJiY6OMWiYhIY1O4OYrT6fQEm7Zt2zZ4e5fLhd1uJygoqMX88W+IllpfcHAwAAcOHCAuLk5DVCIifs7nf6FefPFFUlNTCQoKYvDgwaxevfq46zocDh577DE6d+5MUFAQffv2ZcGCBV5rS/Ucm5CQEK/tU5qH6mPa0HlUIiLS8vg03MybN4+0tDSmTZvGunXr6Nu3LyNHjvQMIRzrz3/+M//85z/5+9//zs8//8xdd93Ftddey/fff+/Vdmlehv/RMRURaT18Gm5mzpzJpEmTmDhxIr169eKVV14hJCSE2bNn17n+W2+9xf/93/9x+eWX06lTJ+6++24uv/xy/vrXvzZxy0VERKS58lm4sdvtrF27lmHDhh1pjNnMsGHDWLFiRZ3bVFRUEBQUVGNZcHAwy5Yta9S2tjapqanMmjXL180QERE5JT6bUJybm4vT6SQ+Pr7G8vj4eDZv3lznNiNHjmTmzJmcf/75dO7cmYyMDD7++OMTXnOloqKCiooKz+vCwkLAPffi2PkXDocDwzBwuVy4XK4G12QYhufxVLY/HRdffDF9+/bl+eefP+19rVq1itDQ0Fo1+LK+0+VyuTAMA4fDccIJxdU/E/48N8ffa/T3+kA1+gN/rw+8X2ND9tOizpb629/+xqRJk+jRowcmk4nOnTszceLE4w5jAcyYMYNHH3201vKFCxfWmjgcEBBAQkICxcXF2O32U25nUVHRKW97qiorK7Hb7Z7wdizDMHA6nQQEnPyQ22w2Kisrj7svX9R3uux2O2VlZSxdupTKysqTrp+ent4ErfItf6/R3+sD1egP/K0+s8tOYGUxJlyUBcYA3quxtLS03uuajOr/jjcxu91OSEgIH374Iddcc41n+fjx48nPz+fTTz897rbl5eUcOnSIpKQk/vSnP/HZZ5+xcePGOtetq+cmJSWF3NzcWncFLy8vZ/fu3Z6ztxrKMAyKiooIDw9v0gmsEydO5M0336yx7LXXXuP222/ns88+Y+rUqfz0008sWLCAlJQU/vCHP7Bq1SpKSkro2bMnTzzxRI3hwU6dOnHfffdx3333AWCxWPjnP//J/PnzWbhwIcnJyTz77LNcddVVTVbj6SovLyczM5OUlJQTHluHw0F6ejrDhw/36zv1+nON/l4fqEZ/4LX6DAMcpVBeAOUFmCrLwVUJLgc4K8GoGtlw2jEV7IH8TLCXYjJcYLjc21P9vOr10c9dleC0V+3PDk4HJqe96rn7teexshyTwx1AXB1+RfnoD7x6DAsLC4mJiaGgoKDW3+9j+aznJjAwkAEDBpCRkeEJNy6Xi4yMDCZPnnzCbYOCgkhOTsbhcPDRRx9x4403Hnddm82GzWartdxqtdb6ZjudTkwmE2az2XMdF8MwKHPU71YDLpeLMruTAIfTK9eBCbZa6hWSXnjhBbZu3cqZZ57JY489BuAJe//3f//Hc889R6dOnWjTpg27d+/miiuu4Mknn8Rms/Hmm29y9dVXs2XLFtq3b+/ZZ/X3odpf/vIXnnrqKaZOncqcOXO49dZb2bVrF9HR0addZ1Mwm82YTKY6j3td6rteS+bvNfp7faAaWxzDAEeZ+6usgLDyvVhzN2J12auWl7gf7SVQUQSlh6A0z/1YdhgwwGyFyvKqMJMPZfnu4NGcmCyYq37fgveOYUP24dNhqbS0NMaPH8/AgQMZNGgQs2bNoqSkhIkTJwIwbtw4kpOTmTFjBuCeC7J371769evH3r17mT59Oi6Xiz/+8Y+N1sYyh5NeU79stP2fyM+PjSQk8OSHKDIyksDAQEJCQkhISADwzFt67LHHGD58uGfd6Oho+vbt63n9l7/8hf/85z/897//PWGonDBhAmPGjKGwsJAnnniCv//976xevZpLL730VMsTEfE9ZyVUFFb1Tjireiyc7ueOMneviKMU7FWPdS4rPRJKqsOLJ6gcvc6RYRUrcAnAJi/VYbJAcBQEBIPZAharOwiZLYAJzGaIaAdtUiEoEkxmMJmqHo/3Zaralw0sge59WgKP/zwgEIKjq/ZvAh/OJ/JpuBk9ejQHDx5k6tSpZGdn069fPxYsWOCZZJyVlVWj96C8vJw///nP7Nixg7CwMC6//HLeeustoqKifFRB8zdw4MAar4uLi5k+fTqff/45+/fvp7KykrKyMrKysk64nz59+nieh4aGEhERcdzrEYmINAqXC+zFUJYHhzMhP8sdRqwh7sfyQneoAHc4Kc8/0vNRkuvuDaksdweZynKorKgROJqSYbHhIABrSCSmwBB3DdYQsAZDYCgEhkFIWwht634MbuMOHE67O2wER0FQVNVjpHt9Xc/Lw+cTiidPnnzcHoMlS5bUeH3BBRfw888/N0Grjgi2Wvj5sZH1WtflclFUWER4RLjXhqVOV2hoaI3XDzzwAOnp6Tz33HN06dKF4OBgrr/++pNOoD62O9BkMrW4M6ZExMdcLqis6uEo3AcFe9zDK45S97BLwW4oygZHGRZHORfkHSRg9xNVc0oK3T0sNNI0UU9vhcXdW2GyVAWNo0JHdQA5NozUtSwwtOY2xyyrdLr4Yv58Lr/88mY/7OZyGZRXOil3uCh3OCl3OClzHHldZne/Lqt6Xmp3Ehtu4+o+8SffeSPxebhp7kwmU72GhsAdbioDLYQEBjT5vZcCAwNPeEp8tW+//ZYJEyZw7bXXAu6enMzMzEZunYi0ePZSKNpfNfRSCiUH3UGkOMf9WJrnDi6O8iO9IpVlVb0jVUM1lWX1/jgzEAVQ1yaWQIhq7x5isQS622QyQ1CEuwcDE5hw92yEtD3yFRQBAUEQYHP3fgTYwBYOtgj3kEpTcnrnP4eVThelDiflVaGizFH16HleSbnjyHtl9prPy+xOSh1OyuyVNQJLucNFhcNJRaUL+ym0dUCHNgo3cvpSU1NZtWoVmZmZhIWFHbdXpWvXrnz88ceMGjUKk8nEI488oh4YkdbA5XIP2VQUu4d2Sg+5w0ppXlXwKAdzgLvXonA/HN4JxQfcvSWlh9xf3hQWDxHJEBrj7s2wRUBkCkQkuns2CGDN9z9y9pBzCQiOdL8fFOF+tAa3iCEYwzBwOI2qAFHpCRblDieFZRX8mGfC8cN+7E6qwkalJ5xUB5Kjn5fVCCjudR3Opj3h2WoxEWS1EGy1HHkMtBBsNRNktRASaCHYGkCn2NCT76wRKdz4iQceeIDx48fTq1cvysrKeP311+tcb+bMmdx2220MHTqUmJgYHnrooeNez0ZEmhHDqDrttgLKigktz8a0eyWUHXL3ohTnuMNI8QH3UI+9KsRUFFf1tpScfhusoUd6P0LaQniC+yssoSqkhLh7Q6p7R6zBR15bQ44MywQEuye4nqhch4MDO0wYqedDIw3bGIaB3emq1ZtRWh0ejuoJqd37UUmZw1UjkNRYv6pHxOk6UfiwwJafvFKL2eSeyhAcGFAVMCwEVz2GBNZ8HhRoIcQaQHCg2b3+UesGWS3YrGaCAiwEVQUW95cZW4AFi7n+odKXFyhUuPET3bp1q3XbigkTJtRaLzU1lUWLFtVYdu+999Z4fewwVfWlkI7u4cnPzz/1xoq0JobhDh8lue5ekJJcd49J9XBOSW7V8I39mMmu1a8rqoZ4Kqieb2IFhsGpnWljMkNgOIS0OSaUBLp7d1yVEBbnHvKJSHL3lAS3gch27smrPlbucFJUXklRuYPiikrP88LySorLK2u8V7Pno7JWL8jJw4f3BJhNNQNGgJnykiKS4tsSEmitGUgCLUcCR6DlqPcC6g4rVgu2ALNuEHwUhRsREW9wOtxn7+TthLwd7mGd3F9g/w/ucONlleYgLJGJmMLi3WEkLM491BMa6w4hgeFgCzty5k1gmPt1QFCTDumUO5wcLrWTV2Inv9TB4VI7h0sdFJdXUlJRSXHVV/XzUruTikonFQ4nhwstPPXzUhxOFxWVLiocpzb/oz6sFlNVWAg4bo9HcNWQy9HLq8NH9ZDMkecBNd63Wmr2VDkcDubPn8/llw9s9hOKWyKFGxGR43E5j1zvo6LI3dNSPYm2cK87yByuCjMFe9ynI9fFZHb3ftjCq4ZzEt1BJDzBHUYCQ6uuExLk7kEJCDoy4TWg6hoj1mDPOg7DxPwvFjTJmTbVFzItLKukoMxBQZmDwqrHgjIHheVHPS+r9LxXWO4OMuWO0wkjJigvr/OdcFsAYUEBhAcFEB5kJcx25HlEUABhtgBCbAE1ejdCagWRgOOGD2nZFG5EpPWxl7h7WQ7vcp9+7LS7h4+K9kPORndYKc0De9V91Ezm4weXo1lDoE1HiO7oHtZp2xkS+kD8Ge5w4i2nOJfB5TIotrsDyOESB7klFRwqtnOouIJDJXZyi92vjw4wheWO0560GmA2ERUSSHSolaiQQKKCrYQHWQkPCiDUZiHU5g4joYHu17YAC2ZcrF2zigt+dS4hQYHYAszYrBbCgwIICwzA3IC5H9L6KNyIiP8xXHBoK+RuhuyfIGeDe6Kto8x9AbiGDhNVB5vA8COTaMMTjgoyVY9h8U025GMYUFReSWGBO5TkFruHfg6X2jlUXPVYYudw1bLCMgdFFZWc6t0ELWYTkcFWIoPdPSMR1c+rHt3Lj3oeHEBUcCBtQt29Kg2dD+JwOMjfAn3aRWrYRhpM4UZEWhaXyz0kdHhnzfkteTuhJJcAp50rywqwrD9J74YtEtq0h6gO7h4XcA8dxfeCmO7uOSxBUUDVWUpBEe7ho0bidBnkV81N8XyVusNJdUjJK3WQV1JBXrGdg4UWHCsXnXzHdQgMMBMdEkh0aCBtwwKJCbPRNjSQtlWPUSG1g0tIYP3udSfSHCjciIjvuVxQtM89TFR9Wf3S3CPXVynNq7q0fumR+wAdhwmwAEZAMKb4MyCht/srMsV9FVlbhPsCcI185k+5w8nBoooaYeXo3pSjA0xeiXsoqGG9Ku6gERpooW2YjZiwQKJDbUSHWj2PbULc4SUqJNDTsxIeFECQF65+LtKcKdyISOMrynEPD3nCyiF3eCnYcyTMnCCw1GK2ugOKZ0iok/t5eAIOw8ySb1dx4dXjsNqCGqWcUnsl+wvKySkoZ39BOdmF5ewvKCO7+nVBOYdKGlDPUaJCrESHBNIm1N2zEh0SSHRYoGdZ29BAwm1mflz9LddfOZKI0MapUaQlU7gRkdNXUQx7v4P9P7pvVlhRVHV3ZRfsWw/71p18H+YAd+9Kmw7u4BIWf8yl8yPdw0e2cPf1V8zH6X1wOCi1ZR7//RMwDIOiisqjQkqZJ6wceSyjsLyyXvsLDDATE3pUUAkNdPemhB4JKke/FxVsJaAeZ+04HA72/gjBgeqBEamLwo2InFj1WUQHfq6a41J1+vPhTPeVcJ0Od0+McZJ7m8X2qLoOS8yRwBKR7A4zbVIhPAksjfcryTAM8ksdVT0tdYeW7IJySuwnv0cbuIeDEqOCSYwMIiEiyP0YWfU60v06MtiqeSoiPqBwI4D7ysX3338/999/P+C+Yeh//vMfrrnmmjrXz8zMpGPHjnz//ff069fvlD/XW/sRL3G5IG+7+8Jz2T+6e2Kyf3IPIZ1MZAq0Gwihce7eFXPVr5eoFOgyHMIb9yZ6heUOMnNL2HOohKX7TWxc+AsHiuzu4aNCd4ipqKzfNVcig601QkpCRM3QkhAZRHiQzuARaa4UbqRO+/fvp02bNl7d54QJE8jPz+eTTz7xLEtJSWH//v3ExMR49bOknlwuKMiCrFWw+TPYvsh9P6JjmczQtiu07XLkGi7RHd29L2YrhES7h4oaib3SxcHiCg4UlnOgqIIDRRUcrHqeU1jOtoPF7M47+vbRFjjO3e5jwgJJOE5gSYhwP4YE6lejSEumf8FSp4SEhCb5HIvF0mSf1ao5He5Tpg9uxpyzibMylxDw72fh0Db3fY2OFhDsvuhcYh/3BegS+rhPj/bmReiOUVxRyf78MnYdKuWXA0XsOFhCTmE5BworOFBUzuHS+l20LjbcRmKkDUrz6dctleQ2IVXhxR1k4iJs2AI0T0XE3ync+IF//etfTJ8+nT179mA+6k67V199NW3btuXhhx8mLS2NlStXUlJSQs+ePZkxYwbDhg077j6PHZZavXo1v/nNb9i0aRNnnnkmDz/8cI31nU4nd955J4sWLSI7O5v27dtzzz33cN999wEwffp03njjDc++ARYvXkxqamqtYamvv/6aBx98kB9++IHo6GjGjx/P448/TkCA+8f1wgsvpE+fPgQFBfHvf/+bwMBA7rrrLqZPn+6Nb2fL5nRA7lb3VXYPbobcLXBwizvYuNyTYC1AytHbWALd82G6jYTul0Ni31OajHsypfZKMnNL2ZFbzOb9RWzOLmTP4TL25pdRVI8JulaLidgwG7ERQcSF26q+3IGlQ3QIvZIiiAoJPOqePT108TeRVkrh5mQMw31tjfpwudzr2i1g9sJ9Sqwh9bra6Q033MBvf/tbFi9ezCWXXAJAXl4eCxYsYP78+RQXF3P55ZfzxBNPYLPZePPNNxk1ahRbtmyhffv2J91/cXExV155JcOGDeOll17i4MGD/P73v6+xjsvlol27dnzwwQe0bduW5cuXc+edd5KYmMiNN97IAw88wKZNmygsLOT1118HIDo6mn379tXYz969e7n88suZMGECb775Jps3b2bSpEkEBQXVCC9vvPEGaWlprFq1ihUrVjBhwgTOPfdchg8fftJ6/IZhuINM5lL3sFLORveNGl3H6eUIDIOYbrjadmVzrotuv7qKgIQz3Bex8+JE3vxSO5uzi9iSXcTm7CJ2HCwm81AJOYUVJ9wuIiiA5DYhdI0Lo2tcGIlRwe4AE+EOMVHBVl1yX0TqReHmZByl8GT95hKYgShvfvb/7avXFVHbtGnDZZddxjvvvOMJNx9++CExMTFcdNFFmM1m+vbt61n/L3/5C//5z3/473//y+TJk0+6/3feeQeXy8W///1v7HY7gwcPZt++fdx9992edaxWK48++qjndceOHVmxYgXvv/8+N954I2FhYQQHB1NRUXHCYaiXXnqJlJQU/vGPf2AymejRowf79u3joYceYurUqZ6eqT59+jBt2jQAunbtyj/+8Q8yMjL8O9w4HbB7NexYDHu+c0/6LcurvV5guHsYKa6n+0q7sVVfEclgMuF0ONg6fz5du10Gp9iz4XIZHCiqIPNQCbsOlbDtQDFbcorZkl14whATFWIltW0oPRLC6ZkYQWpMKEmRQSRGBRNm068jEfEO/TbxEzfffDOTJk3ipZdewmazMXfuXG666SbMZjPFxcVMnz6dzz//nP3791NZWUlZWRlZWVn12vemTZs8w0B2u/vCZEOGDKm13osvvsjs2bPJysqirKwMu93e4DOgNm3axJAhQ2qcPnvuuedSXFzMnj17PD1Nffr0qbFdYmIiBw4caNBnNWu5W93hJT/Lfdp1zs/uU7GP7UW02CBlEKT+yj2cFH+G+6wlL5x+bBgG2YXlbD9Q4gkxmYdK2XWohKy80hPe7bldm2B6JITTLT6crvFhpLYNpWNMKFEhgafdLhGRk1G4ORlriLsHpR5cLheFRUVEhIfXmPtyWp9dT6NGjcIwDD7//HPOPvtsvvnmG55//nkAHnjgAdLT03nuuefo0qULwcHBXH/99Z6g4g3vvfceDzzwAH/9618ZMmQI4eHhPPvss6xatcprn3G0Y+dSmEwmXK76nebb7BiGO7hk/+T++uVL900f6xLSFjpfDB3OhaR+ENcLAmyn3YT8Uju/5BTzS04RW3OK2JLjHlLKP8FEXovZRLs2wXRoG0rHtiF0T4ige0I43eLDdJq0iPiUws3JmEz1v1meywVWp3t9b4SbBggKCuLXv/41c+fOZdu2bXTv3p2zzjoLgG+//ZYJEyZw7bXXAu45NJnHOU22Lj179uStt96ivLzcs2zlypU11vn2228ZOnQo99xzj2fZ9u3ba6wTGBiI03niC6T17NmTjz76CMMwPL033377LeHh4bRr167ebW72HGXunpltX8FPH7gviHc0s9V9zZg2qe6emLgeEHcGxHQ7rZ+twnIHm/bmszzHxNrPN7M9t4Rfcoo5WFT3UJLFbKJD2xA6tg2lQ9tQUmNC6NA2lA7RISS3CcZaj6vpiog0NYUbP3LzzTdz5ZVXsnHjRm655RbP8q5du/Lxxx8zatQoTCYTjzzySIN6OcaOHcvDDz/MnXfeyeTJk8nNzeW5556rsU7Xrl158803+fLLL+nYsSNvvfUWa9asoWPHjp51UlNT+fLLL9myZQtt27YlMjKy1mfdc889zJo1i9/+9rdMnjyZLVu2MG3aNNLS0rzTG+YrJYfc15DZvQr2rIGcDZ6zlwB3L11iP/ewUvtzoOtw9+0GTpFhGOzNL2PD3kJ+3lfAhn2FbNpfyP6C6oBqgR01hyWTo4LpFh9WNZQUTo+EcLrEhekmiyLS4ijc+JGLL76Y6OhotmzZwtixYz3LZ86cyW233cbQoUOJiYnhoYceorCwsN77DQsL43//+x933XUXF1xwAb169eLpp5/muuuu86zzm9/8hu+//57Ro0djMpkYM2YM99xzD1988YVnnUmTJrFkyRIGDhxIcXGx51TwoyUnJzN//nwefPBB+vbtS3R0NLfffjt//vOfT/0b4wuG4T576ZcFsHWheyIwx9zyOTTOPV/mjGuh+2X17yE8htNlsONgMT/vL2TjvkI27itg477C4w4pJUTYiDKVMfSMjvRIjKRrfBhd48M1oVdE/IZ+m/kRs9lc69RqcPeYLFq0qMaye++9t8brY4epDKPmH+JzzjmHdevWUVhYSEREBGazucY6NpuN119/3XOad7UZM2Z4nsfGxrJw4cJa7Tv2sy644AJWr15dR4VuS5YsqbXs6Kse+5TLCT9/At/MdPfOHC2+N3Q83z3c1G7gKU38re6RWZeVz3eZefywp4At2YV1Tu4NMJvoFh/OGUkRnJkcSa+kCLrFhxMSgPs6MJd113VgRMQvKdyIeIPTAT++D8tmuq/6C+4r/Xa6ELqNgK4jILLhc4Z2HSph9c48ft5fyM9VQ0t13ZE6JNBCz8QIeiVGeMJM1/iwOq/G63DU72q/IiItlcKNyKmyl0LmN7A1HbbMh8K97uVBUXDO3TDoTvc9lxogv9TO8u2H+GZrLsu2HTzmfkluAWYTPRMjGJjahrPat+GMpAhS24bqAnciIlUUbkQawuWEnz6EH+dB5jJwHnWWUWgsDJkMZ9/uvit2PZTaK1m3K5+VOw7xzbZcftqTj+uoUTqrxUT/lDb0bhfp6ZnpEhdGYEALnlwtItLIFG5E6sPpgI2fwNdP17wGTWQKdBnmPrup88UnvblkQamDNZl5rM7MY/XOPDbsLaDSVXPOUde4MH7VNYbzusYwuGNbQjXRV0SkQfRbsw7HTnCVlu+UjqlhuOfP/PwJrJkNRVWTtYPbwDn3QM+r3Lc1OMGk4HKHk3VZh/n6l4Ms/SWXzdmFHNuUpMggBnWM5twuMfyqawyJkY13920RkdZA4eYo1WeOlJaWEhysPzD+pLTUfduCE54dZBiY9q2jS87nWD79H+z9zn3rg2qhsTDoNzD4NxAUcZxdGGzcV8iCDdl8s/UgG/cV1uqZ6RQbyuCO0ZydGs2gjtG0a1P/K1GLiMjJKdwcxWKxEBUV5blHUUhISI17HJ2My+XCbrdTXl7esi84dxwtsT7DMCgtLeXAgQNERUVhsViOXQEObHJfYG/9OwQc2MgZANVn1JutkHou9LkJzvx1nbc6OFBUTsamA6zccYjVO/OOulCeW1y4jV91ieGC7rEM6dyWuPCgRqlVRETcFG6OUX3H6lO5CaNhGJSVlREcHNygUNRStOT6oqKijtyN3DDcVwr+8X3Y9F8oOehZzwgIYn/omcT3G4Elqa872BwzOdjlcvfOfLs9l8WbD7A6M6/GUFOQ1cwF3WIZ0Suhqmem5X2/RERaMoWbY5hMJhITE4mLi2vw9UAcDgdLly7l/PPP98uLo7XU+qxW65Eem53fwMI/w/71R1YICHbf8qD75VT2vJY1i5dz+a8ux3JMjVmHSvlg7W4+XLunVu9M35QoLuwWy6CO0fRvH0VIoP5piYj4is9/A7/44os8++yzZGdn07dvX/7+978zaNCg464/a9YsXn75ZbKysoiJieH6669nxowZBAV5t6vfYrHUHsKoxzaVlZUEBQW1qD/+9dVi66usgI3/hbVzYMcS9zJrKPS6CnpfD6nnQ0Cge/kxgbbM7uSLDft5/7vdrNyR51keZgvgnE5tObdLW4b3ite8GRGRZsSn4WbevHmkpaXxyiuvMHjwYGbNmsXIkSPZsmULcXFxtdZ/5513+NOf/sTs2bMZOnQov/zyCxMmTMBkMjFz5kwfVCDNWvYG+P4t9zVpyg67l5ksMPA2uOAhCIutczPDgPW78/l4fTaf/bCPogr3FYFNJjivayw3DmzHsJ7xuqGkiEgz5dNwM3PmTCZNmsTEiRMBeOWVV/j888+ZPXs2f/rTn2qtv3z5cs4991zPTSFTU1MZM2YMq1atatJ2SzN3aLt76GnL/CPLIpKh31jofwu0Sa17s+IKPvwui9d/sJC98si9rdpHh3DDgHZcN6AdSVE6i05EpLnzWbix2+2sXbuWKVOmeJaZzWaGDRvGihUr6txm6NChvP3226xevZpBgwaxY8cO5s+fz6233nrcz6moqKCi4shVZKvvhu1wOLx+j53q/fnrvXuabX2GAXnbMGd+gylzGaZfvsDkcmCYLBg9rsTVdyxGxwvBXNXTclT7XS6Db3cc4oPv9vLV5gM4nAZgIijAzKVnxHP9gGTO7tDGc2uDZlf7KWi2x9FL/L0+UI3+wN/rA+/X2JD9mAwfXbFu3759JCcns3z5coYMGeJZ/sc//pGvv/76uL0xL7zwAg888ACGYVBZWcldd93Fyy+/fNzPmT59Oo8++mit5e+88w4hIZon0VIFOMuILfyJxILviSn+mWDH4RrvZ0f0ZWPyTRQHJde5fWklrMgxsSzHTF7FkTOZ2ocanBPv4qy2BsE+n5EmIiLVSktLGTt2LAUFBURE1H2tsWot6tf3kiVLePLJJ3nppZcYPHgw27Zt47777uMvf/kLjzzySJ3bTJkyhbS0NM/rwsJCUlJSGDFixEm/OQ3lcDhIT09n+PDhLWvCbT01i/oKdmP+dhbmje9hOuq+TobFhtFuIEaH8zA6XUTb5AGcX8fm2w+W8ObKXfxn/T7KHC4AIoICuLpvIjcMaEeXmCDf19jImsVxbET+Xh+oRn/g7/WB92usHnmpD5+Fm5iYGCwWCzk5OTWW5+TkHLkeyTEeeeQRbr31Vu644w4AevfuTUlJCXfeeScPP/xwnReWs9ls2Gy1L7xmtVob7QeqMffdHPikvvzd8M1f4fu3wVXVNdmmI/S8EroMx5QyCNNx7utkGAZLt+Yye9lOvv7lyDVteiSEc9u5HbmqX5JncnB1t6e/H0Pw/xr9vT5Qjf7A3+sD79XYkH34LNwEBgYyYMAAMjIyuOaaawD3FXAzMjKYPHlynduUlpbWCjDVp2vrflB+6tB2WP73mqGm4/lwwZ+gw9AT3tep1F7Jx+v28vq3O9l+sARwrz6sZzwTz01lSKe2urieiIgf8umwVFpaGuPHj2fgwIEMGjSIWbNmUVJS4jl7aty4cSQnJzNjxgwARo0axcyZM+nfv79nWOqRRx5h1KhRDb4mjTRz+VmwYAps/hyoCq7VoSb13BNuWmqv5PVvM/nX0h0UlLkDUZgtgBsHpjB+aAc6tA1t5MaLiIgv+TTcjB49moMHDzJ16lSys7Pp168fCxYsID4+HoCsrKwaPTV//vOfMZlM/PnPf2bv3r3ExsYyatQonnjiCV+VII1h83z45G4oz3e/7nYpnHufu6fmBBxOF++t2c0LGVs5WOSej9OhbQgThqZy/YB2hAf5d9eviIi4+XxC8eTJk487DLVkyZIarwMCApg2bRrTpk1rgpZJkzMM+PoZWPKk+3XSWXDNSxDX84SbuVwGn/+0n78u3ELmIffdv1Oig3lgRHeu7JOExayhJxGR1sTn4UYEcAebRY/DN8+5X59zDwx79MhtEY5jxfZDzPhiEz/uKQAgJiyQ317clTGD2hMY0DLuXC4iIt6lcCO+V2mH9Kmwqup6RSOfhCH3nnCTX3KKeOqLzSza7L57e2ighTvP78wd53Uk1KYfaxGR1kx/BcS3DmyGjydB9o/u15c9A4N/c9zVC8sd/PXLLby1chcuAwLMJsYObs9vL+5KbHjtU/5FRKT1UbgR33C5YPW/4KtpUFkOwW1g1N+g19V1rm4YBp/9uJ/HP/+ZnEL3ZOHLzkzgwZHd6RQb1pQtFxGRZk7hRppeSa67t2b7IvfrzpfA1S9CRGKdq3+fdZi/fPYz67LyAegYE8oT15zJ0C4xTdRgERFpSRRupGkV7IU3r4ZDWyEgCEY8DmffUefF+ArLHTyzYDNvr8wCINhq4a4LOvObCzp5rigsIiJyLIUbaTp5O+CNq6EgCyLawS0fHvc072+35fKH938gu7AcgF+flcxDl/YgPiKoKVssIiItkMKNNI2ibHePTUEWRHeCcf+FqJRaqzldBn/L2MrfF23FMCC1bQhP/ro3QztrCEpEROpH4UYaX3khzL3efUuF6E4wcQGEx9daLaewnPve+56VO/IAuOnsFKaNOoPgQA1BiYhI/SncSONylMG8WyD7JwiNhVs+qjPYLP3lIL+ft55DJXZCAy08+eveXN0v2QcNFhGRlk7hRhqPvRTevQl2fg3WUBj7vrvn5iiVThfPf/ULLy3ZjmFAz8QIXhzbX6d3i4jIKVO4kcbhKId3boTMbyAwDG7+EJLPqrFKXomdu95ey+qd7mGomwe355Ere+lMKBEROS0KN9I4FjxUFWzC4daPIWVQjbd3HSphwutr2JlbQpgtgBm/7s2ovkk+aqyIiPgThRvxvh/mwdo5gAlGv1kr2Py8r5Bxs1eRW2wnOSqYN247my5x4T5pqoiI+B+FG/GuA5vhs/vdzy94CDpfXOPtbQeKufW1VRwqsXNGUgSvTzibOF27RkREvEjhRrynKAfm3gCOUuh4AVzwxxpvZx0q5eZ/r+RQiZ0zkyOYe8c5RAZbfdRYERHxV2ZfN0D8REUxvHND1UX6OsP1r4P5yMTgXYdKuOlfK8gprKBrXBhv3jZYwUZERBqFem7EO/73O9j/A4TEuG+rENrW89b2g8WMfXUlOYUVdIoNZe4dg4kODfRhY0VExJ+p50ZOX+Yy2PARmMww5t0a17LZl1/mCTZd48J4785zNMdGREQalXpu5PS4nLBgivv5gIk1zowqLHcw8fU1NYJN2zCbjxoqIiKthXpu5PT88C5k/wi2SLjo/zyLHU4Xd7+9li05RcSF25hz2yAFGxERaRIKN3LqKoog4zH38wsehNAjd+5+cv4mvt12iJBAC7MnnE1yVLCPGikiIq2Nwo2cumXPQ3GOe47NoN94Fn+6fi+vf5sJwKzR/TgzOdJHDRQRkdZI4UZOTX4WLP+H+/nwv0CA++ynzdmF/OmjnwC496LOjDgjwVctFBGRVkrhRk5N+jRwVkDqedDjCsA9gfiut9ZS5nByXtcY0oZ393EjRUSkNVK4kQYz7V4FGz8GTDDySTCZcLkM0ub9QOahUpKjgvnbTf2xmE2+bqqIiLRCCjfSMIYLc/qf3c/PuhUS+wDwytLtfLUph8AAMy/fcpYu0iciIj6jcCMN0u7wCsz7v4fAcLj4EcA9z+b59F8AeOyqM+jTLsqHLRQRkdZO4Ubqz15Cr33vu5+flwZhcVQ6XTz4wY84nAYjesUz+uwU37ZRRERaPYUbqTfz6n8S7DiMEdkezrkHgH99s4Of9hYQERTA49ecicmkeTYiIuJbCjdSP45yzN+9CoDzwilgDWLP4VJmfbUVgKmjztA9o0REpFlQuJH6+el9TCUHKbVGY/S8BoDnvtyCvdLFOZ2iue6sZN+2T0REpIrCjZycYcCKFwHYETsCLFZ+2lPAJ+v3AfDw5b00HCUiIs1Gswg3L774IqmpqQQFBTF48GBWr1593HUvvPBCTCZTra8rrriiCVvcymzPgIObMQJD2dX2AgzD4Mn5mwC4pl8Svdvp9goiItJ8+DzczJs3j7S0NKZNm8a6devo27cvI0eO5MCBA3Wu//HHH7N//37P14YNG7BYLNxwww1N3PJWZMVLALj63UJlQChLfsllxY5DBAaYeWCkrkIsIiLNi8/DzcyZM5k0aRITJ06kV69evPLKK4SEhDB79uw614+OjiYhIcHzlZ6eTkhIiMJNYzm03d1zgwnXwEk4DXjmS/c1bSYOTaVdmxDftk9EROQYPg03drudtWvXMmzYMM8ys9nMsGHDWLFiRb328dprr3HTTTcRGhraWM1s3dbOcT92GQZtUll1wMS2gyVEhVi556IuPm2aiIhIXQJ8+eG5ubk4nU7i4+NrLI+Pj2fz5s0n3X716tVs2LCB11577bjrVFRUUFFR4XldWFgIgMPhwOFwnGLL61a9P2/v12cqKwj4/m1MQGX/ceSXlPHFbncevueCToQE+FGtVfzuGNbB32v09/pANfoDf68PvF9jQ/bj03Bzul577TV69+7NoEGDjrvOjBkzePTRR2stX7hwISEhjTOkkp6e3ij7bWrJecsZWJZHmTWa9K2VLFiyhEKHhbY2g7Z5G5k/f6Ovm9ho/OUYnoi/1+jv9YFq9Af+Xh94r8bS0tJ6r+vTcBMTE4PFYiEnJ6fG8pycHBISEk64bUlJCe+99x6PPfbYCdebMmUKaWlpnteFhYWkpKQwYsQIIiIiTr3xdXA4HKSnpzN8+HCsVqtX9+0LljfdE4kDh0zi/EGX8chflwKVPHT5GVx1VjvfNq6R+NsxrIu/1+jv9YFq9Af+Xh94v8bqkZf68Gm4CQwMZMCAAWRkZHDNNdcA4HK5yMjIYPLkySfc9oMPPqCiooJbbrnlhOvZbDZsNlut5VartdF+oBpz303mwGbYvRJMFiwDJ/DBuv0UlFUSE2RwVb/kll/fSfjFMTwJf6/R3+sD1egP/L0+8F6NDdmHz8+WSktL49VXX+WNN95g06ZN3H333ZSUlDBx4kQAxo0bx5QpU2pt99prr3HNNdfQtm3bpm5y67DuTfdjt0spD47nX9/sAGBYkguLWRfsExGR5svnc25Gjx7NwYMHmTp1KtnZ2fTr148FCxZ4JhlnZWVhNtfMYFu2bGHZsmUsXLjQF032f5UV8MO77ucDxvPB2j0cLKogMTKIs2OLfds2ERGRk/B5uAGYPHnycYehlixZUmtZ9+7dMQyjkVvVim36H5TlQUQyrk6X8Oon3wBwx69SCcjb4OPGiYiInJjPh6WkGVr3hvux/y18vT2PrLxSwoMCuP6sJN+2S0REpB4UbqSmvB2wcylggv638PaKXQDcMCCFkMBm0dEnIiJyQgo3UlP1ROIul7DbFcOiLe57fN1yTnsfNkpERKT+FG7kCKcDvp/rfn7WeOauysIw4LyuMXSKDfNt20REROpJ4UaO+OVLKDkAobGUdxrO+9/tBuDWczr4uGEiIiL1p3AjR1RPJO43loVbDpNXYicxMoiLe8T5tl0iIiINoHAjbgV7YNtX7udnjef9Ne5emxsGtCPAoh8TERFpOfRXS9y+nwuGC1LPY7cpkWXbcjGZ4IaBKb5umYiISIMo3Ijbxv+4H/vdzAdVc23O7RxDSnTj3DldRESksSjcCBzaDgc3gTkAZ9dL+WDtHgBuPFu9NiIi0vIo3Ij7dgsAqefxzR4H+wvKiQqxMqJXvG/bJSIicgoUbgQ2f+Z+7Hkl/12/D4Cr+iYRZLX4sFEiIiKnRuGmtSvcD3vWAFDe+VIW/pwDuMONiIhIS6Rw09pt+dz9mDyQxXstFFdUkhwVzFnt2/i2XSIiIqdI4aa121wVbnpeyf9+dA9JXdk3EbPZ5MNGiYiInDqFm9asvBB2fgNAcceRZGxy3yRzVB8NSYmISMulcNOa7VgMLge07UL6gQgqKl10ig3ljKQIX7dMRETklCnctGa/fOl+7HYpX/yUDbh7bUwmDUmJiEjLpXDTWrlcnnBj7zycb7flAjBc17YREZEWTuGmtdq3DkpzwRbJd67ulNidxIbb6JWoISkREWnZFG5aq18WuB+7XMLirYcBuKBbrM6SEhGRFk/hprXaUhVuul3Kki0HAbioe5wPGyQiIuIdCjetUeE+yPkJTGb2xQ5l64FiLGYTv+oa4+uWiYiInDaFm9ao6to2JPZjUZYLgLPaRxEZbPVho0RERLxD4aY1yqwKNx3P8wxJXaghKRER8RMKN61R5jIAHClDWb7dfQr4hd1jfdkiERERr1G4aW0K9sDhnWCy8J2rO6V2J3E6BVxERPyIwk1rk/mt+zGpH1/tKAPcvTa6KrGIiPgLhZvWpnq+TeqvWLLFfaNMzbcRERF/onDT2lTNtznY9my2HyzRKeAiIuJ3FG5ak6Pm22SUdARgQIc2RATpFHAREfEfCjetyVHzbdK3H5lvIyIi4k8UblqTqvk2le3P5duqU8B1ywUREfE3CjetSdV8my1BfSl3uIiPsNEjIdzHjRIREfEun4ebF198kdTUVIKCghg8eDCrV68+4fr5+fnce++9JCYmYrPZ6NatG/Pnz2+i1rZgR823+ao4FYBzO8foFHAREfE7Ab788Hnz5pGWlsYrr7zC4MGDmTVrFiNHjmTLli3ExdUeLrHb7QwfPpy4uDg+/PBDkpOT2bVrF1FRUU3f+JbmqPk2izMrABjaRWdJiYiI//FpuJk5cyaTJk1i4sSJALzyyit8/vnnzJ49mz/96U+11p89ezZ5eXksX74cq9V9hk9qampTNrnlqppvU9FuKD8uzQdgaOe2PmyQiIhI4/BZuLHb7axdu5YpU6Z4lpnNZoYNG8aKFSvq3Oa///0vQ4YM4d577+XTTz8lNjaWsWPH8tBDD2GxWOrcpqKigoqKCs/rwsJCABwOBw6Hw4sV4dmft/frDQGZyzABP1l64TIgtW0IsaEBDWprc67PW1Rjy+fv9YFq9Af+Xh94v8aG7Mdn4SY3Nxen00l8fHyN5fHx8WzevLnObXbs2MGiRYu4+eabmT9/Ptu2beOee+7B4XAwbdq0OreZMWMGjz76aK3lCxcuJCQk5PQLqUN6enqj7PdUBdkPMfLwTgxMvLrRHQKTA4pPea5Sc6uvMajGls/f6wPV6A/8vT7wXo2lpaX1Xtenw1IN5XK5iIuL41//+hcWi4UBAwawd+9enn322eOGmylTppCWluZ5XVhYSEpKCiNGjCAiwrs3i3Q4HKSnpzN8+HDPsFlzYPrpfdgIRmI/dpYkAMWMvrAfl52Z0KD9NNf6vEk1tnz+Xh+oRn/g7/WB92usHnmpD5+Fm5iYGCwWCzk5OTWW5+TkkJBQ9x/dxMRErFZrjSGonj17kp2djd1uJzAwsNY2NpsNm81Wa7nVam20H6jG3Pcp2e0e5itPHsovy4oB+FW3+FNuY7OrrxGoxpbP3+sD1egP/L0+8F6NDdmHz04FDwwMZMCAAWRkZHiWuVwuMjIyGDJkSJ3bnHvuuWzbtg2Xy+VZ9ssvv5CYmFhnsJEqe9cCsMHSE4BeiRFEh+r7JSIi/smn17lJS0vj1Vdf5Y033mDTpk3cfffdlJSUeM6eGjduXI0Jx3fffTd5eXncd999/PLLL3z++ec8+eST3Hvvvb4qofmzl8JB9xymJcXtADink86SEhER/+XTOTejR4/m4MGDTJ06lezsbPr168eCBQs8k4yzsrIwm4/kr5SUFL788kt+//vf06dPH5KTk7nvvvt46KGHfFVC85ezAQwXhMWzZJ/7cA9MbePjRomIiDQen08onjx5MpMnT67zvSVLltRaNmTIEFauXNnIrfIj+9YDUBnfl80/uydjDeigcCMiIv7L57dfkEa273v3Q0h3XAa0axNMfESQjxslIiLSeBRu/N3+9QB8X5kKqNdGRET8n8KNPztqMvFXBUkADFS4ERERP6dw48+qJhMbYfEs3uu+NtCADtE+bpSIiEjjUrjxZ1XzbYqjz6S4opIwWwDdE8J93CgREZHGpXDjz6rOlNph7QJA//ZRWMwmHzZIRESk8Snc+LOqycSryzsAcFZ7zbcRERH/p3Djr+wlnsnEnx9yXxRRF+8TEZHWQOHGX+1bD4YLZ1gi6/ODMZugX0qUr1slIiLS6BRu/FXVzTIPRpwJQPeECMKD/PvOsyIiIqBw47+qws1GU1cABnSI8mFjREREmo7Cjb/auw6AxcUpAAzU9W1ERKSVULjxR8UHoCALAxOf58YBuu2CiIi0Hgo3/qhqSKossguHncHEhdto1ybYx40SERFpGgo3/qgq3OwK7gm4TwE3mXTxPhERaR0UbvzRnu8AWGPvCOjifSIi0roo3Pgblwv2uScTf3G46k7gqZpMLCIirYfCjb/J2wHlBbgsQawpSyTQYqZXYoSvWyUiItJkFG78TdWdwPMjulNJAD0TwwkM0GEWEZHWQ3/1/E3VzTKr7wTep12U79oiIiLiAw0ON6+//joffPBBreUffPABb7zxhlcaJadh33oA1lS47wTep12kDxsjIiLS9BocbmbMmEFMTEyt5XFxcTz55JNeaZScIpcL9v8AwMLDCYB6bkREpPVpcLjJysqiY8eOtZZ36NCBrKwsrzRKTlHeDrAX4bLY+NGeSEighS5xYb5ulYiISJNqcLiJi4vjxx9/rLX8hx9+oG3btl5plJyiqvk2+eHdcGLhzKRILGZdvE9ERFqXBoebMWPG8Lvf/Y7FixfjdDpxOp0sWrSI++67j5tuuqkx2ij1VRVutgW4JxP31nwbERFphQIausFf/vIXMjMzueSSSwgIcG/ucrkYN26c5tz4WtVk4tXl7QFNJhYRkdapweEmMDCQefPm8fjjj7N+/XqCg4Pp3bs3HTp0aIz2SX0ZBux3Dxem5ycC0FeTiUVEpBVqcLip1rVrV7p27erNtsjpyNsBFQW4LDY2licRERRAh7Yhvm6ViIhIk2vwnJvrrruOp59+utbyZ555hhtuuMErjZJTUDXf5nBYVyoJoE+7KN0JXEREWqUGh5ulS5dy+eWX11p+2WWXsXTpUq80Sk5B1W0Xtlo6A5pvIyIirVeDw01xcTGBgYG1llutVgoLC73SKDkFe913Al9Rngro4n0iItJ6NTjc9O7dm3nz5tVa/t5779GrVy+vNEoayFnp6bn5siAZUM+NiIi0Xg2eUPzII4/w61//mu3bt3PxxRcDkJGRwTvvvMOHH37o9QZKPRzcDI5SnNYwfilPIibMRmJkkK9bJSIi4hMNDjejRo3ik08+4cknn+TDDz8kODiYvn37smjRIqKjoxujjXIye78DICe8F64iM33bRWoysYiItFoNHpYCuOKKK/j2228pKSlhx44d3HjjjTzwwAP07dv3lBrx4osvkpqaSlBQEIMHD2b16tXHXXfOnDmYTKYaX0FBrbyXYo873Pxscp+arysTi4hIa3ZK4QbcZ02NHz+epKQk/vrXv3LxxRezcuXKBu9n3rx5pKWlMW3aNNatW0ffvn0ZOXIkBw4cOO42ERER7N+/3/O1a9euUy3DP1RNJl5a4r4ysS7eJyIirVmDhqWys7OZM2cOr732GoWFhdx4441UVFTwySefnPJk4pkzZzJp0iQmTpwIwCuvvMLnn3/O7Nmz+dOf/lTnNiaTiYSEhFP6PL9TUQwHNwGwoKAdoMnEIiLSutU73IwaNYqlS5dyxRVXMGvWLC699FIsFguvvPLKKX+43W5n7dq1TJkyxbPMbDYzbNgwVqxYcdztiouL6dChAy6Xi7POOosnn3ySM844o851KyoqqKio8LyuPl3d4XDgcDhOue11qd6ft/d7Iqbd3xFguCgPSeRAeRuSo4KIsJkbpQ2+qK+pqcaWz9/rA9XoD/y9PvB+jQ3Zj8kwDKM+KwYEBPC73/2Ou+++u8ZtF6xWKz/88MMp9dzs27eP5ORkli9fzpAhQzzL//jHP/L111+zatWqWtusWLGCrVu30qdPHwoKCnjuuedYunQpGzdupF27drXWnz59Oo8++mit5e+88w4hIS3/9gRdcj7njH3z+N52NtcW/J6+0S5u6+7ydbNERES8qrS0lLFjx1JQUEBERMQJ1613z82yZct47bXXGDBgAD179uTWW2/lpptuOu3GNtSQIUNqBKGhQ4fSs2dP/vnPf/KXv/yl1vpTpkwhLS3N87qwsJCUlBRGjBhx0m9OQzkcDtLT0xk+fDhWq9Wr+z4ey0cfwD7YGdofCmD4gO5cfn7HRvksX9TX1FRjy+fv9YFq9Af+Xh94v8aGXCi43uHmnHPO4ZxzzmHWrFnMmzeP2bNnk5aWhsvlIj09nZSUFMLDwxvU0JiYGCwWCzk5OTWW5+Tk1HtOjdVqpX///mzbtq3O9202Gzabrc7tGusHqjH3XUvOTwAsKXH3Wp3VIbrRP7tJ6/MR1djy+Xt9oBr9gb/XB96rsSH7aPDZUqGhodx2220sW7aMn376iT/84Q889dRTxMXFcdVVVzVoX4GBgQwYMICMjAzPMpfLRUZGRo3emRNxOp389NNPJCYmNuiz/YK9BA5nAvBNQRwAZyRrMrGIiLRup3wqOED37t155pln2LNnD+++++4p7SMtLY1XX32VN954g02bNnH33XdTUlLiOXtq3LhxNSYcP/bYYyxcuJAdO3awbt06brnlFnbt2sUdd9xxOqW0TAc2A1ARFMthIugUE0pksH//D0BERORkGnyF4rpYLBauueYarrnmmgZvO3r0aA4ePMjUqVPJzs6mX79+LFiwgPj4eACysrIwm49ksMOHDzNp0iSys7Np06YNAwYMYPny5a3zvlYHNgKQE+SeY6NTwEVERLwUbk7X5MmTmTx5cp3vLVmypMbr559/nueff74JWtUC5PwMwGZXCgC9dfE+ERGR0xuWEh874A43K0vck6/7qudGRERE4aZFqwo3a0oTsZhNnJGkcCMiIqJw01IVH4SSgxiY2Gok0zUujOBAi69bJSIi4nMKNy1V1WTi/KB2lGPTZGIREZEqCjct1QH3zTK3mdx3Au+jycQiIiKAwk3LlePuuVlb5p5M3C8lyoeNERERaT4UblqqqsnEP9qTCbKa6Z7QsFtfiIiI+CuFm5bI5fJcnXiLkUKf5CisFh1KERERULhpmXJ/AUcJdnMQmUYC/dtH+bpFIiIizYbCTUu0Zw0Am81dcGJRuBERETmKwk1LVBVulpe77ynVv30bX7ZGRESkWVG4aYn2fAfAOlcXkiKDiI8I8nGDREREmg+Fm5amoshzptT3ri7qtRERETmGwk1Ls+97wOBQQBwHaaP5NiIiIsdQuGlpqubbrHN2AVC4EREROYbCTUtTNd9mpb0TVovuBC4iInIshZuWxDA8PTffu7rQKzGCIKvuBC4iInI0hZuWJH8XlBzEaQpgo5GqycQiIiJ1ULhpSXavBmCHpRMVBGq+jYiISB0UblqSrBUALKuomkycop4bERGRYynctCRZKwFY6exG29BAUqKDfdwgERGR5kfhpqUoO+y5eN9aV3f6t4/CZDL5uFEiIiLNj8JNS1E13ybHmkIukZpMLCIichwKNy3FruUArHF1A6B/SpQPGyMiItJ8Kdy0FFXzbb4u74zJBL3b6eJ9IiIidVG4aQkc5bBvHQBrXN3pFhdOeJDVx40SERFpnhRuWoJ934PTTrE1mkwjQde3EREROQGFm5ag6vo2Gy29AJPCjYiIyAko3LQEVTfL/Lo0FUBnSomIiJyAwk1LUD3fxtGJcFsAXWLDfNwgERGR5kvhprkr3A9F+3FhZoORSt+UKMxmXbxPRETkeBRumruqXpscWwfKCNJ8GxERkZNQuGnu9rrDzXpnJwCFGxERkZNoFuHmxRdfJDU1laCgIAYPHszq1avrtd17772HyWTimmuuadwG+lJVz82ysg4A9NOdwEVERE7I5+Fm3rx5pKWlMW3aNNatW0ffvn0ZOXIkBw4cOOF2mZmZPPDAA5x33nlN1FIfMAxPz80Prk6ktg0hOjTQx40SERFp3nwebmbOnMmkSZOYOHEivXr14pVXXiEkJITZs2cfdxun08nNN9/Mo48+SqdOnZqwtU0sbweU51NpsrLFaK9TwEVEROohwJcfbrfbWbt2LVOmTPEsM5vNDBs2jBUrVhx3u8cee4y4uDhuv/12vvnmmxN+RkVFBRUVFZ7XhYWFADgcDhwOx2lWUFP1/ry1X9PuNQQAu6ydcJQF0Cc53Ottbghv19ccqcaWz9/rA9XoD/y9PvB+jQ3Zj0/DTW5uLk6nk/j4+BrL4+Pj2bx5c53bLFu2jNdee43169fX6zNmzJjBo48+Wmv5woULCQkJaXCb6yM9Pd0r+zljz8d0AVaWtQegZNcG5h/a4JV9nw5v1decqcaWz9/rA9XoD/y9PvBejaWlpfVe16fhpqGKioq49dZbefXVV4mJianXNlOmTCEtLc3zurCwkJSUFEaMGEFERIRX2+dwOEhPT2f48OFYrad/Y0vLmy8BsM7ZGVuAmduuuxSrxXcjid6urzlSjS2fv9cHqtEf+Ht94P0aq0de6sOn4SYmJgaLxUJOTk6N5Tk5OSQkJNRaf/v27WRmZjJq1CjPMpfLBUBAQABbtmyhc+fONbax2WzYbLZa+7JarY32A+WVfbtckP0TAD8anejTLpKQoNp1+EJjfu+aC9XY8vl7faAa/YG/1wfeq7Eh+/DphOLAwEAGDBhARkaGZ5nL5SIjI4MhQ4bUWr9Hjx789NNPrF+/3vN11VVXcdFFF7F+/XpSUlKasvmNK287OEqwm2xsN5I0mVhERKSefD4slZaWxvjx4xk4cCCDBg1i1qxZlJSUMHHiRADGjRtHcnIyM2bMICgoiDPPPLPG9lFRUQC1lrd4+38AYLs5FRdm+qdE+bY9IiIiLYTPw83o0aM5ePAgU6dOJTs7m379+rFgwQLPJOOsrCzMZp+fsd709q8H4Du7ezJxP12ZWEREpF58Hm4AJk+ezOTJk+t8b8mSJSfcds6cOd5vUHNQ1XPzkyuVhIggEiODfdwgERGRlqEVdom0AIYB+38EYKOro+4nJSIi0gAKN81Rfpb7ysQE8IvRTuFGRESkARRumqOqIaltpvY4CNDNMkVERBpA4aY5qgo36x3tMZngjCTvXmxQRETEnyncNEdV4WaD0ZHOsWGE2prFvG8REZEWQeGmuTEMz2ngG12p9E6O9G17REREWhiFm+amaD+UHMSFmU1Gew1JiYiINJDCTXOzdx0AO0wplGNTz42IiEgDKdw0N/vc4Wato6N7MrHCjYiISIMo3DQ3VT03Pxid6RgTSpgmE4uIiDSIwk1zYhienpsfXJ00JCUiInIKFG6ak7wdUF6Aw2Rli5GicCMiInIKFG6ak6ohqS10pJIAzlS4ERERaTCFm+Zk71oA1jg6AroysYiIyKlQuGlOqubb/OjqRPvoEMKDrD5ukIiISMujcNNcOB2w/0fAfaZU94RwHzdIRESkZVK4aS4ObILKMsrNoew0EuihcCMiInJKFG6ai33fA7DF0gUDMz0SNN9GRETkVCjcNBdVdwL/zt4eQMNSIiIip0jhprmoCjfrHR0IDDCT2jbExw0SERFpmRRumgNnJeRsBGCD0ZGucWEEWHRoREREToX+gjYHh7ZCZRl2SwiZRryGpERERE6Dwk1zUDUktcvauWoyscKNiIjIqVK4aQ6qws2PzlQAnSklIiJyGhRumoOqcLO8NBlAPTciIiKnQeHG11wuz5WJf3J1pE2Ildhwm48bJSIi0nIp3Pja4Z1gL8JptrHdSKJ7Qjgmk8nXrRIREWmxFG58bf96ALKDO+PEovk2IiIip0nhxteqhqQ20wnQlYlFREROl8KNr2X/BMCqMk0mFhER8QaFG1+rujLxmrIkALrFK9yIiIicDoUbXyrJheJsDExsMVJoHx1CqC3A160SERFp0RRufKmq16YouB2lBGm+jYiIiBco3PhSzgYAsqzuycSabyMiInL6mkW4efHFF0lNTSUoKIjBgwezevXq46778ccfM3DgQKKioggNDaVfv3689dZbTdhaL6rqufmxsh2gM6VERES8wefhZt68eaSlpTFt2jTWrVtH3759GTlyJAcOHKhz/ejoaB5++GFWrFjBjz/+yMSJE5k4cSJffvllE7fcC6p6blYUJwC6p5SIiIg3+DzczJw5k0mTJjFx4kR69erFK6+8QkhICLNnz65z/QsvvJBrr72Wnj170rlzZ+677z769OnDsmXLmrjlp8lZCQc2A/CDox2BAWZS24b4uFEiIiItn09PzbHb7axdu5YpU6Z4lpnNZoYNG8aKFStOur1hGCxatIgtW7bw9NNP17lORUUFFRUVnteFhYUAOBwOHA7HaVZQU/X+6rXf3F+wOiuoDAhhd3ksPWNDMVxOHC6nV9vkTQ2qr4VSjS2fv9cHqtEf+Ht94P0aG7Ifk2EYhlc+9RTs27eP5ORkli9fzpAhQzzL//jHP/L111+zatWqOrcrKCggOTmZiooKLBYLL730Erfddlud606fPp1HH3201vJ33nmHkBDf9ZQkHV7J2ZkvsSOgCxcXP8bZsS5u6eLyWXtERESas9LSUsaOHUtBQQERESeextEiL6oSHh7O+vXrKS4uJiMjg7S0NDp16sSFF15Ya90pU6aQlpbmeV1YWEhKSgojRow46TenoRwOB+np6QwfPhyr1XrCdc2Lv4dM2BPSA4rh4rN6cPmvUr3aHm9rSH0tlWps+fy9PlCN/sDf6wPv11g98lIfPg03MTExWCwWcnJyaizPyckhISHhuNuZzWa6dOkCQL9+/di0aRMzZsyoM9zYbDZsNlut5VartdF+oOq174M/A7Cuwn3bhV7JUS3mB7wxv3fNhWps+fy9PlCN/sDf6wPv1diQffh0QnFgYCADBgwgIyPDs8zlcpGRkVFjmOpkXC5XjXk1LULVaeDLi+IBXeNGRETEW3w+LJWWlsb48eMZOHAggwYNYtasWZSUlDBx4kQAxo0bR3JyMjNmzABgxowZDBw4kM6dO1NRUcH8+fN56623ePnll31ZRsOUHYbCPQBscrWnTYiVuPDavUsiIiLScD4PN6NHj+bgwYNMnTqV7Oxs+vXrx4IFC4iPd/doZGVlYTYf6WAqKSnhnnvuYc+ePQQHB9OjRw/efvttRo8e7asSGi7HPSRVEpxEUXkI5ySEYzKZfNwoERER/+DzcAMwefJkJk+eXOd7S5YsqfH68ccf5/HHH2+CVjWiqiGpvbbOgC7eJyIi4k0+v4hfq1R1ZeKfXSmAbrsgIiLiTQo3vlDVc7OqJBFQuBEREfEmhZum5nLCAfecm1Wl7nDTLV7hRkRExFsUbpra4UxwlOK02Mg0EmgfHUKYrVlMfRIREfELCjdNrWq+TV5oZ1yYNSQlIiLiZQo3Ta1qvk2mpSOgi/eJiIh4m8JNU6sKN+vt7QBNJhYREfE2hZumVjUstUy3XRAREWkUCjdNqaLIPaEY+MGeTGCAmdS2ob5tk4iIiJ9RuGlKVbddKA+OJ59wusaFEWDRIRAREfEm/WVtStk/ArA/uCug+TYiIiKNQeGmKWX/BMBmIxXQfBsREZHGoHDTlKrCzeqyJAC664aZIiIiXqdw01SclZ7bLiwpSADUcyMiItIYFG6ayqFtUFmO0xpKphFHVIiVuHCbr1slIiLidxRumkrVkNThsG4YmOmREI7JZPJxo0RERPyPwk1TqTpTKtPaCYAemm8jIiLSKBRumkpVz816R3tAp4GLiIg0FoWbpmAYnnDzTaF7MrHCjYiISONQuGkKRdlQmothMrOyxH1PqW7xCjciIiKNQeGmKVT12pRFdKaCQFKigwmzBfi4USIiIv5J4aYpeG670AWA7vGaTCwiItJYFG6aQs4GALaQCkDPRA1JiYiINBaFm6ZQNSy1qiwZ0GRiERGRxqRw09gqiuHQdgAyDscBuu2CiIhIY1K4aWwHfgYMKkMT2GMPIzDATGrbUF+3SkRExG8p3DS2qsnEh8O7A9AlNowAi77tIiIijUV/ZRtb1XybI7dd0JCUiIhIY1K4aWxV4eZHRwoAPXSmlIiISKNSuGlMzkrI2QjA0qLq2y7oGjciIiKNSeGmMeVth8pyDGsoyw9HAhqWEhERaWwKN42p+rYLbXrgcJmICrESF27zcaNERET8m8JNY6p124VwTCaTL1skIiLi95pFuHnxxRdJTU0lKCiIwYMHs3r16uOu++qrr3LeeefRpk0b2rRpw7Bhw064vk9V9dz8bHQAoFeS5tuIiIg0Np+Hm3nz5pGWlsa0adNYt24dffv2ZeTIkRw4cKDO9ZcsWcKYMWNYvHgxK1asICUlhREjRrB3794mbnk9VIWb5cVJAPROjvRla0RERFoFn4ebmTNnMmnSJCZOnEivXr145ZVXCAkJYfbs2XWuP3fuXO655x769etHjx49+Pe//43L5SIjI6OJW34SRTlQchDDZObL3GhA4UZERKQp+DTc2O121q5dy7BhwzzLzGYzw4YNY8WKFfXaR2lpKQ6Hg+jo6MZq5qmp6rVxRHUmzx5ASKCFTrFhPm6UiIiI/wvw5Yfn5ubidDqJj4+vsTw+Pp7NmzfXax8PPfQQSUlJNQLS0SoqKqioqPC8LiwsBMDhcOBwOE6x5XWr3p/D4cC8bz0WILtqMnHPhHBczkpcTq9+ZJM6uj5/pRpbPn+vD1SjP/D3+sD7NTZkPz4NN6frqaee4r333mPJkiUEBQXVuc6MGTN49NFHay1fuHAhISEhjdKu9PR0BuxcSDtgRdX1bULtecyfP79RPq+ppaen+7oJjU41tnz+Xh+oRn/g7/WB92osLS2t97o+DTcxMTFYLBZycnJqLM/JySEhIeGE2z733HM89dRTfPXVV/Tp0+e4602ZMoW0tDTP68LCQs8k5IgI75695HA4SE9PZ/jw4QS/9hgAv9jOAOCKob25vH+yVz+vqR1dn9Vq9XVzGoVqbPn8vT5Qjf7A3+sD79dYPfJSHz4NN4GBgQwYMICMjAyuueYaAM/k4MmTJx93u2eeeYYnnniCL7/8koEDB57wM2w2GzZb7QvnWa3WRvuBshp2TIe2A5CeFwdAv/Zt/eYHuDG/d82Famz5/L0+UI3+wN/rA+/V2JB9+HxYKi0tjfHjxzNw4EAGDRrErFmzKCkpYeLEiQCMGzeO5ORkZsyYAcDTTz/N1KlTeeedd0hNTSU7OxuAsLAwwsKax4Rd08FNgEFlSBxZeeEEWc10jg31dbNERERaBZ+Hm9GjR3Pw4EGmTp1KdnY2/fr1Y8GCBZ5JxllZWZjNR07qevnll7Hb7Vx//fU19jNt2jSmT5/elE0/LlPVmVKHwrpBHvRKjCDA4vOz7kVERFoFn4cbgMmTJx93GGrJkiU1XmdmZjZ+g05XzgYAtpk7Arq+jYiISFNSd0IjMFWFm9Vl7gnEvdtF+bA1IiIirYvCjZeZDCemAz8DsPBQLAD920f5sEUiIiKti8KNl0WU7cZUWYYzMILNlfFEBlvp2FaTiUVERJqKwo2XtSlxnwKeE34GBmb6pURhNpt83CoREZHWQ+HGy9qUbgPgJ1M3QENSIiIiTU3hxsuiq3puFhWlANC/fRtfNkdERKTVUbjxprLDhFW4Lyr4ZYE73PTTmVIiIiJNSuHGi0z71gFQEtaBfMLpHBtKZIh/X1ZbRESkuVG48SLT3u8A2BnUC9CQlIiIiC8o3HiRae9aAFY7OgGaTCwiIuILCjfe4nJh2ucON/Pz2wHQLyXKhw0SERFpnRRuvOXQNkzlBThMgayvSCbcFkCPhAhft0pERKTVaRY3zvQLhXsxQtqS5YihkgAGprbBoov3iYiINDn13HhL54uovH8zD1vSABjcqa2PGyQiItI6Kdx4kcuAH4vDABjUMdrHrREREWmdFG68aNvBYkorTQRbzfROjvR1c0RERFolhRsvWp15GHCfAm616FsrIiLiC/oL7EVrqsLNoFQNSYmIiPiKwo2XGIbhCTdnp0b5tjEiIiKtmMKNl+zMLeFgsZ0Ak0FfzbcRERHxGV3nxkv25ZcTHWqljdmOzWrxdXNERERaLfXceMmvusaw8qELub2709dNERERadUUbrzIZDIRavV1K0RERFo3hRsRERHxKwo3IiIi4lcUbkRERMSvKNyIiIiIX1G4EREREb+icCMiIiJ+ReFGRERE/IrCjYiIiPgVhRsRERHxKwo3IiIi4lcUbkRERMSvKNyIiIiIX1G4EREREb8S4OsGNDXDMAAoLCz0+r4dDgelpaUUFhZitfrf7cH9vT5Qjf7A3+sD1egP/L0+8H6N1X+3q/+On0irCzdFRUUApKSk+LglIiIi0lBFRUVERkaecB2TUZ8I5EdcLhf79u0jPDwck8nk1X0XFhaSkpLC7t27iYiI8Oq+mwN/rw9Uoz/w9/pANfoDf68PvF+jYRgUFRWRlJSE2XziWTWtrufGbDbTrl27Rv2MiIgIv/1hBf+vD1SjP/D3+kA1+gN/rw+8W+PJemyqaUKxiIiI+BWFGxEREfErCjdeZLPZmDZtGjabzddNaRT+Xh+oRn/g7/WBavQH/l4f+LbGVjehWERERPybem5ERETEryjciIiIiF9RuBERERG/onAjIiIifkXhxktefPFFUlNTCQoKYvDgwaxevdrXTTplM2bM4OyzzyY8PJy4uDiuueYatmzZUmOdCy+8EJPJVOPrrrvu8lGLG2b69Om12t6jRw/P++Xl5dx77720bduWsLAwrrvuOnJycnzY4oZLTU2tVaPJZOLee+8FWubxW7p0KaNGjSIpKQmTycQnn3xS433DMJg6dSqJiYkEBwczbNgwtm7dWmOdvLw8br75ZiIiIoiKiuL222+nuLi4Cas4vhPV53A4eOihh+jduzehoaEkJSUxbtw49u3bV2MfdR33p556qokrOb6THcMJEybUav+ll15aY53mfAzh5DXW9e/SZDLx7LPPetZpzsexPn8f6vM7NCsriyuuuIKQkBDi4uJ48MEHqays9Fo7FW68YN68eaSlpTFt2jTWrVtH3759GTlyJAcOHPB1007J119/zb333svKlStJT0/H4XAwYsQISkpKaqw3adIk9u/f7/l65plnfNTihjvjjDNqtH3ZsmWe937/+9/zv//9jw8++ICvv/6affv28etf/9qHrW24NWvW1KgvPT0dgBtuuMGzTks7fiUlJfTt25cXX3yxzvefeeYZXnjhBV555RVWrVpFaGgoI0eOpLy83LPOzTffzMaNG0lPT+ezzz5j6dKl3HnnnU1VwgmdqL7S0lLWrVvHI488wrp16/j444/ZsmULV111Va11H3vssRrH9be//W1TNL9eTnYMAS699NIa7X/33XdrvN+cjyGcvMaja9u/fz+zZ8/GZDJx3XXX1VivuR7H+vx9ONnvUKfTyRVXXIHdbmf58uW88cYbzJkzh6lTp3qvoYactkGDBhn33nuv57XT6TSSkpKMGTNm+LBV3nPgwAEDML7++mvPsgsuuMC47777fNeo0zBt2jSjb9++db6Xn59vWK1W44MPPvAs27RpkwEYK1asaKIWet99991ndO7c2XC5XIZhtOzjZxiGARj/+c9/PK9dLpeRkJBgPPvss55l+fn5hs1mM959913DMAzj559/NgBjzZo1nnW++OILw2QyGXv37m2yttfHsfXVZfXq1QZg7Nq1y7OsQ4cOxvPPP9+4jfOSumocP368cfXVVx93m5Z0DA2jfsfx6quvNi6++OIay1rScTz270N9fofOnz/fMJvNRnZ2tmedl19+2YiIiDAqKiq80i713Jwmu93O2rVrGTZsmGeZ2Wxm2LBhrFixwoct856CggIAoqOjayyfO3cuMTExnHnmmUyZMoXS0lJfNO+UbN26laSkJDp16sTNN99MVlYWAGvXrsXhcNQ4nj169KB9+/Yt9nja7Xbefvttbrvttho3i23Jx+9YO3fuJDs7u8Zxi4yMZPDgwZ7jtmLFCqKiohg4cKBnnWHDhmE2m1m1alWTt/l0FRQUYDKZiIqKqrH8qaeeom3btvTv359nn33Wq139TWHJkiXExcXRvXt37r77bg4dOuR5z9+OYU5ODp9//jm33357rfdaynE89u9DfX6Hrlixgt69exMfH+9ZZ+TIkRQWFrJx40avtKvV3TjT23Jzc3E6nTUOEkB8fDybN2/2Uau8x+Vycf/993Puuedy5plnepaPHTuWDh06kJSUxI8//shDDz3Eli1b+Pjjj33Y2voZPHgwc+bMoXv37uzfv59HH32U8847jw0bNpCdnU1gYGCtPxjx8fFkZ2f7psGn6ZNPPiE/P58JEyZ4lrXk41eX6mNT17/D6veys7OJi4ur8X5AQADR0dEt7tiWl5fz0EMPMWbMmBo3JPzd737HWWedRXR0NMuXL2fKlCns37+fmTNn+rC19XfppZfy61//mo4dO7J9+3b+7//+j8suu4wVK1ZgsVj86hgCvPHGG4SHh9ca9m4px7Guvw/1+R2anZ1d57/V6ve8QeFGTujee+9lw4YNNeakADXGuHv37k1iYiKXXHIJ27dvp3Pnzk3dzAa57LLLPM/79OnD4MGD6dChA++//z7BwcE+bFnjeO2117jssstISkryLGvJx6+1czgc3HjjjRiGwcsvv1zjvbS0NM/zPn36EBgYyG9+8xtmzJjRIi7zf9NNN3me9+7dmz59+tC5c2eWLFnCJZdc4sOWNY7Zs2dz8803ExQUVGN5SzmOx/v70BxoWOo0xcTEYLFYas0Ez8nJISEhwUet8o7Jkyfz2WefsXjxYtq1a3fCdQcPHgzAtm3bmqJpXhUVFUW3bt3Ytm0bCQkJ2O128vPza6zTUo/nrl27+Oqrr7jjjjtOuF5LPn6A59ic6N9hQkJCrUn+lZWV5OXltZhjWx1sdu3aRXp6eo1em7oMHjyYyspKMjMzm6aBXtapUydiYmI8P5f+cAyrffPNN2zZsuWk/zaheR7H4/19qM/v0ISEhDr/rVa/5w0KN6cpMDCQAQMGkJGR4VnmcrnIyMhgyJAhPmzZqTMMg8mTJ/Of//yHRYsW0bFjx5Nus379egASExMbuXXeV1xczPbt20lMTGTAgAFYrdYax3PLli1kZWW1yOP5+uuvExcXxxVXXHHC9Vry8QPo2LEjCQkJNY5bYWEhq1at8hy3IUOGkJ+fz9q1az3rLFq0CJfL5Ql3zVl1sNm6dStfffUVbdu2Pek269evx2w21xrKaSn27NnDoUOHPD+XLf0YHu21115jwIAB9O3b96TrNqfjeLK/D/X5HTpkyBB++umnGkG1Oqz36tXLaw2V0/Tee+8ZNpvNmDNnjvHzzz8bd955pxEVFVVjJnhLcvfddxuRkZHGkiVLjP3793u+SktLDcMwjG3bthmPPfaY8d133xk7d+40Pv30U6NTp07G+eef7+OW188f/vAHY8mSJcbOnTuNb7/91hg2bJgRExNjHDhwwDAMw7jrrruM9u3bG4sWLTK+++47Y8iQIcaQIUN83OqGczqdRvv27Y2HHnqoxvKWevyKioqM77//3vj+++8NwJg5c6bx/fffe84Weuqpp4yoqCjj008/NX788Ufj6quvNjp27GiUlZV59nHppZca/fv3N1atWmUsW7bM6Nq1qzFmzBhflVTDieqz2+3GVVddZbRr185Yv359jX+X1WeXLF++3Hj++eeN9evXG9u3bzfefvttIzY21hg3bpyPKzviRDUWFRUZDzzwgLFixQpj586dxldffWWcddZZRteuXY3y8nLPPprzMTSMk/+cGoZhFBQUGCEhIcbLL79ca/vmfhxP9vfBME7+O7SystI488wzjREjRhjr1683FixYYMTGxhpTpkzxWjsVbrzk73//u9G+fXsjMDDQGDRokLFy5UpfN+mUAXV+vf7664ZhGEZWVpZx/vnnG9HR0YbNZjO6dOliPPjgg0ZBQYFvG15Po0ePNhITE43AwEAjOTnZGD16tLFt2zbP+2VlZcY999xjtGnTxggJCTGuvfZaY//+/T5s8an58ssvDcDYsmVLjeUt9fgtXry4zp/L8ePHG4bhPh38kUceMeLj4w2bzWZccskltWo/dOiQMWbMGCMsLMyIiIgwJk6caBQVFfmgmtpOVN/OnTuP++9y8eLFhmEYxtq1a43BgwcbkZGRRlBQkNGzZ0/jySefrBEMfO1ENZaWlhojRowwYmNjDavVanTo0MGYNGlSrf8kNudjaBgn/zk1DMP45z//aQQHBxv5+fm1tm/ux/Fkfx8Mo36/QzMzM43LLrvMCA4ONmJiYow//OEPhsPh8Fo7TVWNFREREfELmnMjIiIifkXhRkRERPyKwo2IiIj4FYUbERER8SsKNyIiIuJXFG5ERETEryjciIiIiF9RuBGRVic1NZVZs2b5uhki0kgUbkSkUU2YMIFrrrkGgAsvvJD777+/yT57zpw5REVF1Vq+Zs2aGndGFxH/EuDrBoiINJTdbicwMPCUt4+NjfVia0SkuVHPjYg0iQkTJvD111/zt7/9DZPJhMlkIjMzE4ANGzZw2WWXERYWRnx8PLfeeiu5ubmebS+88EImT57M/fffT0xMDCNHjgRg5syZ9O7dm9DQUFJSUrjnnnsoLi4GYMmSJUycOJGCggLP502fPh2oPSyVlZXF1VdfTVhYGBEREdx4443k5OR43p8+fTr9+vXjrbfeIjU1lcjISG666SaKiooa95smIqdE4UZEmsTf/vY3hgwZwqRJk9i/fz/79+8nJSWF/Px8Lr74Yvr37893333HggULyMnJ4cYbb6yx/RtvvEFgYCDffvstr7zyCgBms5kXXniBjRs38sYbb7Bo0SL++Mc/AjB06FBmzZpFRESE5/MeeOCBWu1yuVxcffXV5OXl8fXXX5Oens6OHTsYPXp0jfW2b9/OJ598wmeffcZnn33G119/zVNPPdVI3y0ROR0alhKRJhEZGUlgYCAhISEkJCR4lv/jH/+gf//+PPnkk55ls2fPJiUlhV9++YVu3boB0LVrV5555pka+zx6/k5qaiqPP/44d911Fy+99BKBgYFERkZiMplqfN6xMjIy+Omnn9i5cycpKSkAvPnmm5xxxhmsWbOGs88+G3CHoDlz5hAeHg7ArbfeSkZGBk888cTpfWNExOvUcyMiPvXDDz+wePFiwsLCPF89evQA3L0l1QYMGFBr26+++opLLrmE5ORkwsPDufXWWzl06BClpaX1/vxNmzaRkpLiCTYAvXr1Iioqik2bNnmWpaameoINQGJiIgcOHGhQrSLSNNRzIyI+VVxczKhRo3j66adrvZeYmOh5HhoaWuO9zMxMrrzySu6++26eeOIJoqOjWbZsGbfffjt2u52QkBCvttNqtdZ4bTKZcLlcXv0MEfEOhRsRaTKBgYE4nc4ay8466yw++ugjUlNTCQio/6+ktWvX4nK5+Otf/4rZ7O6Efv/990/6ecfq2bMnu3fvZvfu3Z7em59//pn8/Hx69epV7/aISPOhYSkRaTKpqamsWrWKzMxMcnNzcblc3HvvveTl5TFmzBjWrFnD9u3b+fLLL5k4ceIJg0mXLl1wOBz8/e9/Z8eOHbz11lueicZHf15xcTEZGRnk5ubWOVw1bNgwevfuzc0338y6detYvXo148aN44ILLmDgwIFe/x6ISONTuBGRJvPAAw9gsVjo1asXsbGxZGVlkZSUxLfffovT6WTEiBH07t2b+++/n6ioKE+PTF369u3LzJkzefrppznzzDOZO3cuM2bMqLHO0KFDueuuuxg9ejSxsbG1JiSDe3jp008/pU2bNpx//vkMGzaMTp06MW/ePK/XLyJNw2QYhuHrRoiIiIh4i3puRERExK8o3IiIiIhfUbgRERERv6JwIyIiIn5F4UZERET8isKNiIiI+BWFGxEREfErCjciIiLiVxRuRERExK8o3IiIiIhfUbgRERERv6JwIyIiIn7l/wGrIaKhzEjbOgAAAABJRU5ErkJggg==",
|
|
"text/plain": [
|
|
"<Figure size 640x480 with 1 Axes>"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
}
|
|
],
|
|
"source": [
|
|
"\n",
|
|
"plt.plot(history.history['loss'],label=\"train\")\n",
|
|
"plt.plot(history.history['val_loss'],label=\"validation\")\n",
|
|
"plt.title('Model Loss')\n",
|
|
"plt.yscale('log')\n",
|
|
"plt.ylabel('Cross Entropy')\n",
|
|
"plt.xlabel('Iteration')\n",
|
|
"plt.grid()\n",
|
|
"plt.legend()\n",
|
|
"plt.show()\n",
|
|
"\n",
|
|
"#print(\"history\",history.history)\n",
|
|
"plt.plot(history.history['sparse_categorical_accuracy'],label=\"train\")\n",
|
|
"plt.plot(history.history['val_sparse_categorical_accuracy'],label=\"validation\")\n",
|
|
"plt.title('Model Accuracy')\n",
|
|
"#plt.yscale('log')\n",
|
|
"plt.ylabel('Acc')\n",
|
|
"plt.xlabel('Iteration')\n",
|
|
"plt.grid()\n",
|
|
"plt.legend()\n",
|
|
"plt.show()\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"## Inspect how well the system is working on a sample of 25 new images (from the test set)...\n",
|
|
"- The test set has a lot of images in it, but we can only view 25 at a time.\n",
|
|
"- Hence rerun this code block several times, to get a different random set of samples from the test set"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 14,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"image/png": "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",
|
|
"text/plain": [
|
|
"<Figure size 1000x1000 with 25 Axes>"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
}
|
|
],
|
|
"source": [
|
|
"import matplotlib.pyplot as plt\n",
|
|
"import numpy as np\n",
|
|
"plt.figure(figsize=(10,10))\n",
|
|
"# plot 25 random images from the test set.\n",
|
|
"first_index=np.random.randint(len(test_images)-25)\n",
|
|
"for i in range(first_index,first_index+25):\n",
|
|
" # define subplot\n",
|
|
" plt.subplot(5,5,i+1-first_index)\n",
|
|
" # plot raw pixel data\n",
|
|
" plt.imshow(test_images0[i], cmap=plt.get_cmap('gray'))\n",
|
|
" plt.xticks([])\n",
|
|
" plt.yticks([])\n",
|
|
" plt.grid(False)\n",
|
|
" if class_names!=None:\n",
|
|
" prediction=keras_model(test_images[i:i+1])[0,:] # This will be a vector of length 10\n",
|
|
" prediction_class=np.argmax(prediction) # Pick the index of the largest element of the length-10 vector\n",
|
|
" # Add a label underneath...\n",
|
|
" true_label=test_labels0[i]\n",
|
|
" class_name=class_names[prediction_class]\n",
|
|
" plt.xlabel(class_name+\" \"+(\"CORRECT\" if prediction_class==true_label else \"WRONG\"))\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": "3.11.0"
|
|
}
|
|
},
|
|
"nbformat": 4,
|
|
"nbformat_minor": 4
|
|
}
|