3_char_level_mlp_nn_demo.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import matplotlib.pyplot as plt\n",
    "import torch.nn.functional as F"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['emma',\n",
       " 'olivia',\n",
       " 'ava',\n",
       " 'isabella',\n",
       " 'sophia',\n",
       " 'charlotte',\n",
       " 'mia',\n",
       " 'amelia',\n",
       " 'harper',\n",
       " 'evelyn']"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# read the text file\n",
    "words = open('names.txt', 'r').read().splitlines()\n",
    "words[:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "chars = sorted(list(set(''.join(words))))\n",
    "stoi = {s:i+1 for i, s in enumerate(chars)}\n",
    "stoi['.'] = 0\n",
    "itos = {i:s for s,i in stoi.items()}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# hyperparameters\n",
    "seed = 2147483647\n",
    "block_size = 3 # context length\n",
    "embedding_dim = 2\n",
    "hidden_layer_size = 100\n",
    "alphabet_size = len(itos)\n",
    "learning_rate = 0.1\n",
    "mini_batch_size = 32"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(torch.Size([228146, 3]), torch.Size([228146]))"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# build the dataset\n",
    "x, y = [], []\n",
    "\n",
    "for w in words:\n",
    "    context = [0] * block_size\n",
    "    for ch in w + '.':\n",
    "        ix = stoi[ch]\n",
    "        x.append(context)\n",
    "        y.append(ix)\n",
    "        # print(''.join([itos[i] for i in context]), '--->', itos[ix])\n",
    "        context = context[1:] + [ix] # slide the context\n",
    "\n",
    "x = torch.tensor(x)\n",
    "y = torch.tensor(y)\n",
    "x.shape, y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# parameters\n",
    "def reset_parameters():\n",
    "    g = torch.Generator().manual_seed(seed)\n",
    "    C = torch.randn((alphabet_size, embedding_dim), generator=g)\n",
    "    # first layer\n",
    "    W1 = torch.randn((block_size * embedding_dim, hidden_layer_size), generator=g)\n",
    "    b1 = torch.randn((hidden_layer_size), generator=g)\n",
    "    # second layer\n",
    "    W2 = torch.randn((hidden_layer_size, alphabet_size), generator=g)\n",
    "    b2 = torch.randn((alphabet_size), generator=g)\n",
    "    return C, W1, b1, W2, b2\n",
    "\n",
    "C, W1, b1, W2, b2 = reset_parameters()\n",
    "parameters = [C, W1, b1, W2, b2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3481"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# number of parameters\n",
    "sum(p.numel() for p in parameters)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "for p in parameters:\n",
    "    p.requires_grad_()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 406,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0th iteration loss: 19.505229949951172\n",
      "1th iteration loss: 17.084491729736328\n",
      "2th iteration loss: 15.776532173156738\n",
      "3th iteration loss: 14.833338737487793\n",
      "4th iteration loss: 14.002596855163574\n",
      "5th iteration loss: 13.253254890441895\n",
      "6th iteration loss: 12.579914093017578\n",
      "7th iteration loss: 11.983097076416016\n",
      "8th iteration loss: 11.470491409301758\n",
      "9th iteration loss: 11.05185604095459\n"
     ]
    }
   ],
   "source": [
    "for i in range(10):\n",
    "    # forward pass\n",
    "    emb = C[x] # (32, 3 , 2)\n",
    "    h = torch.tanh(emb.view(-1, block_size * embedding_dim) @ W1 + b1)\n",
    "    logits = h @ W2 + b2\n",
    "    # counts = logits.exp()\n",
    "    # prob = counts / counts.sum(dim=1, keepdim=True)\n",
    "    # loss = -prob[torch.arange(5), y].log().mean()\n",
    "    loss = F.cross_entropy(logits, y)\n",
    "    print(f'{i}th iteration loss: {loss.item()}')\n",
    "\n",
    "    # backward pass\n",
    "    for p in parameters:\n",
    "        p.grad = None\n",
    "    loss.backward()\n",
    "\n",
    "    # update the parameters\n",
    "    for p in parameters:\n",
    "        p.data -= learning_rate * p.grad\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 407,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0th iteration loss: 10.426352500915527, learning rate: 0.0010000000474974513\n",
      "100th iteration loss: 11.404221534729004, learning rate: 0.001996642677113414\n",
      "200th iteration loss: 7.499330997467041, learning rate: 0.003986582159996033\n",
      "300th iteration loss: 7.412033557891846, learning rate: 0.007959775626659393\n",
      "400th iteration loss: 7.594030380249023, learning rate: 0.01589282788336277\n",
      "500th iteration loss: 5.07920503616333, learning rate: 0.03173229098320007\n",
      "600th iteration loss: 2.9455490112304688, learning rate: 0.06335804611444473\n",
      "700th iteration loss: 2.717862129211426, learning rate: 0.1265033781528473\n",
      "800th iteration loss: 3.491755247116089, learning rate: 0.25258201360702515\n",
      "900th iteration loss: 6.334482669830322, learning rate: 0.504315972328186\n"
     ]
    }
   ],
   "source": [
    "training_iterations = 1000\n",
    "lri = []\n",
    "lossi = []\n",
    "\n",
    "lre = torch.linspace(-3, 0, training_iterations)\n",
    "lrs = 10**lre\n",
    "\n",
    "# use mini batches\n",
    "for i in range(training_iterations):\n",
    "    ix = torch.randint(0, x.shape[0], (mini_batch_size,))\n",
    "\n",
    "    # forward pass\n",
    "    emb = C[x[ix]]\n",
    "    h = torch.tanh(emb.view(-1, block_size * embedding_dim) @ W1 + b1)\n",
    "    logits = h @ W2 + b2\n",
    "    loss = F.cross_entropy(logits, y[ix])\n",
    "    if i % (training_iterations/10) == 0:\n",
    "        print(f'{i}th iteration loss: {loss.item()}, learning rate: {lrs[i]}')\n",
    "\n",
    "    # backward pass\n",
    "    for p in parameters:\n",
    "        p.grad = None\n",
    "    loss.backward()\n",
    "\n",
    "    # keep track of loss and lr\n",
    "    lri.append(lre[i])\n",
    "    lossi.append(loss.item())\n",
    "\n",
    "    # update the parameters\n",
    "    for p in parameters:\n",
    "        p.data -= lrs[i] * p.grad"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 408,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor(7.8875, grad_fn=<NllLossBackward0>)"
      ]
     },
     "execution_count": 408,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "emb = C[x] \n",
    "h = torch.tanh(emb.view(-1, block_size * embedding_dim) @ W1 + b1)\n",
    "logits = h @ W2 + b2\n",
    "loss = F.cross_entropy(logits, y)\n",
    "loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 409,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x13794bfd0>]"
      ]
     },
     "execution_count": 409,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot the learning rate / loss curve\n",
    "plt.plot(lri, lossi) # 0.1 looks like a good learning rate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 410,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0th iteration loss: 9.557682037353516\n",
      "2000th iteration loss: 2.508436441421509\n",
      "4000th iteration loss: 2.514707326889038\n",
      "6000th iteration loss: 2.353314161300659\n",
      "8000th iteration loss: 2.5112898349761963\n",
      "10000th iteration loss: 2.5420806407928467\n",
      "12000th iteration loss: 2.222245931625366\n",
      "14000th iteration loss: 2.4578216075897217\n",
      "16000th iteration loss: 2.408839702606201\n",
      "18000th iteration loss: 2.476323366165161\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "tensor(2.3926, grad_fn=<NllLossBackward0>)"
      ]
     },
     "execution_count": 410,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "training_iterations = 20000\n",
    "learning_rate = 0.5\n",
    "\n",
    "# use mini batches\n",
    "for i in range(training_iterations):\n",
    "    ix = torch.randint(0, x.shape[0], (mini_batch_size,))\n",
    "\n",
    "    # forward pass\n",
    "    emb = C[x[ix]]\n",
    "    h = torch.tanh(emb.view(-1, block_size * embedding_dim) @ W1 + b1)\n",
    "    logits = h @ W2 + b2\n",
    "    loss = F.cross_entropy(logits, y[ix])\n",
    "    if i % (training_iterations/10) == 0:\n",
    "        print(f'{i}th iteration loss: {loss.item()}')\n",
    "\n",
    "    # backward pass\n",
    "    for p in parameters:\n",
    "        p.grad = None\n",
    "    loss.backward()\n",
    "\n",
    "    # lr decay\n",
    "    if i % (training_iterations/4) == 0:\n",
    "        learning_rate /= 5\n",
    "\n",
    "    # update the parameters\n",
    "    for p in parameters:\n",
    "        p.data -= learning_rate * p.grad\n",
    "\n",
    "emb = C[x] \n",
    "h = torch.tanh(emb.view(-1, block_size * embedding_dim) @ W1 + b1)\n",
    "logits = h @ W2 + b2\n",
    "loss = F.cross_entropy(logits, y)\n",
    "loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([182625, 3]) torch.Size([182625])\n",
      "torch.Size([22655, 3]) torch.Size([22655])\n",
      "torch.Size([22866, 3]) torch.Size([22866])\n"
     ]
    }
   ],
   "source": [
    "# let's split the dataset into training, validation and test sets\n",
    "def build_dataset(words):\n",
    "    block_size = 3\n",
    "    x, y = [], []\n",
    "    for w in words:\n",
    "        context = [0] * block_size\n",
    "        for ch in w + '.':\n",
    "            ix = stoi[ch]\n",
    "            x.append(context)\n",
    "            y.append(ix)\n",
    "            context = context[1:] + [ix] # slide the context\n",
    "\n",
    "    x = torch.tensor(x)\n",
    "    y = torch.tensor(y)\n",
    "    print(x.shape, y.shape)\n",
    "    return x, y\n",
    "\n",
    "import random\n",
    "random.seed(42)\n",
    "random.shuffle(words)\n",
    "n1 = int(0.8 * len(words))\n",
    "n2 = int(0.9 * len(words))\n",
    "\n",
    "Xtr, Ytr = build_dataset(words[:n1])\n",
    "Xdev, Ydev = build_dataset(words[n1:n2])\n",
    "Xte, Yte = build_dataset(words[n2:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0th iteration loss: 32.0539665222168\n",
      "10000th iteration loss: 2.2922236919403076\n",
      "20000th iteration loss: 2.8701987266540527\n",
      "30000th iteration loss: 2.657052993774414\n",
      "40000th iteration loss: 2.6812124252319336\n",
      "50000th iteration loss: 2.3045685291290283\n",
      "60000th iteration loss: 2.0472702980041504\n",
      "70000th iteration loss: 2.200258255004883\n",
      "80000th iteration loss: 2.189608097076416\n",
      "90000th iteration loss: 2.2814347743988037\n"
     ]
    }
   ],
   "source": [
    "# let's train a bigger model\n",
    "embedding_dim = 10\n",
    "hidden_layer_size = 300\n",
    "\n",
    "C, W1, b1, W2, b2 = reset_parameters()\n",
    "parameters = [C, W1, b1, W2, b2]\n",
    "for p in parameters:\n",
    "    p.requires_grad_()\n",
    "\n",
    "stepi = []\n",
    "lossi = []\n",
    "\n",
    "training_iterations = 100000\n",
    "learning_rate = 0.5\n",
    "\n",
    "for i in range(training_iterations):\n",
    "\n",
    "    # mini batch contruct\n",
    "    ix = torch.randint(0, Xtr.shape[0], (mini_batch_size,))\n",
    "\n",
    "    # forward pass\n",
    "    emb = C[Xtr[ix]]\n",
    "    h = torch.tanh(emb.view(-1, block_size * embedding_dim) @ W1 + b1)\n",
    "    logits = h @ W2 + b2\n",
    "    loss = F.cross_entropy(logits, Ytr[ix])\n",
    "    if i % (training_iterations/10) == 0:\n",
    "        print(f'{i}th iteration loss: {loss.item()}')\n",
    "\n",
    "    # backward pass\n",
    "    for p in parameters:\n",
    "        p.grad = None\n",
    "    loss.backward()\n",
    "\n",
    "    # lr decay\n",
    "    if i % (training_iterations/4) == 0:\n",
    "        learning_rate /= 5\n",
    "\n",
    "    # track stats\n",
    "    stepi.append(i)\n",
    "    lossi.append(loss.log10().item())\n",
    "\n",
    "    # update the parameters\n",
    "    for p in parameters:\n",
    "        p.data -= learning_rate * p.grad"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor(2.2330, grad_fn=<NllLossBackward0>)"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# validation loss\n",
    "emb = C[Xdev] \n",
    "h = torch.tanh(emb.view(-1, block_size * embedding_dim) @ W1 + b1)\n",
    "logits = h @ W2 + b2\n",
    "loss = F.cross_entropy(logits, Ydev)\n",
    "loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor(2.1972, grad_fn=<NllLossBackward0>)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# training loss\n",
    "emb = C[Xtr] \n",
    "h = torch.tanh(emb.view(-1, block_size * embedding_dim) @ W1 + b1)\n",
    "logits = h @ W2 + b2\n",
    "loss = F.cross_entropy(logits, Ytr)\n",
    "loss"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "at this point we're starting to see some overfitting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 454,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x16e19a550>]"
      ]
     },
     "execution_count": 454,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAicAAAGdCAYAAADJ6dNTAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjkuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8ekN5oAAAACXBIWXMAAA9hAAAPYQGoP6dpAABOpElEQVR4nO3deVhU5eIH8O+wDaACIrKKgvsOiIq4lAuKS7aXmalZWZr+0uiWkts1M8zSa3U1yzK7lWuaLZpmKLlvKO47IggCIsIgO8z5/YEMM8zCDMxyYL6f5+F5nDNneTmDc77nPe8iEQRBABEREZFI2Fi6AERERETKGE6IiIhIVBhOiIiISFQYToiIiEhUGE6IiIhIVBhOiIiISFQYToiIiEhUGE6IiIhIVOwsXQB9yOVypKWloUmTJpBIJJYuDhEREelBEATk5eXB19cXNjb614fUi3CSlpYGf39/SxeDiIiIaiElJQUtWrTQe/16EU6aNGkCoOKXc3FxsXBpiIiISB8ymQz+/v6K67i+6kU4qXyU4+LiwnBCRERUzxjaJIMNYomIiEhUGE6IiIhIVBhOiIiISFQYToiIiEhUGE6IiIhIVBhOiIiISFQYToiIiEhUGE6IiIhIVBhOiIiISFQYToiIiEhUGE6IiIhIVBhOiIiISFTqxcR/pvLtwZtIyS7AC7390dGbEwoSERGJgVXXnOw4m4Z1h5OQfK/A0kUhIiKihwwOJ/v378fo0aPh6+sLiUSC7du3673toUOHYGdnh+DgYEMPS0RERFbC4HCSn5+PoKAgrFy50qDtcnJyMGHCBAwZMsTQQxIREZEVMbjNyYgRIzBixAiDDzRlyhS8+OKLsLW1Nai2xRwESxeAiIiIFMzS5uS7775DYmIiFixYoNf6xcXFkMlkKj+mIJFITLJfIiIiqj2Th5Nr165h9uzZ+PHHH2Fnp19FTUxMDFxdXRU//v7+Ji4lERERiYVJw0l5eTlefPFFLFy4EO3bt9d7u+joaOTm5ip+UlJSTFhKIiIiEhOTjnOSl5eHkydP4vTp05g+fToAQC6XQxAE2NnZ4a+//sLgwYPVtpNKpZBKpaYsmgqBjU6IiIhEw6ThxMXFBefOnVNZtmrVKuzduxc///wzAgMDTXn4GrHFCRERkfgYHE4ePHiA69evK17fvHkTCQkJcHd3R8uWLREdHY3U1FT873//g42NDbp27aqyvaenJxwdHdWWExEREQG1CCcnT57EoEGDFK+joqIAABMnTsS6detw584dJCcnG6+EREREZFUkgiD+FhcymQyurq7Izc2Fi4vx5sB59svDOHnrPla/1APDu/oYbb9ERERU++u3Vc+tw2FOiIiIxMeqwwkRERGJD8MJERERiQrDCREREYkKwwk4CBsREZGYWHU4kXAYNiIiItGx6nBCRERE4sNwQkRERKLCcAKATU6IiIjEw7rDCZucEBERiY51hxMiIiISHYYTIiIiEhWGE3CcEyIiIjGx6nDCJidERETiY9XhhIiIiMSH4YSIiIhEheEEgMCRToiIiETDqsOJhI1OiIiIRMeqwwkRERGJD8MJERERiQrDCTjOCRERkZhYdTiRcKQTIiIi0bHqcEJERETiw3BCREREosJwAnCUEyIiIhGx6nDCcU6IiIjEx6rDCREREYkPwwkRERGJCsMJERERiQrDCQCBo7ARERGJhlWHEzaIJSIiEh+rDidEREQkPgwnREREJCoMJ0RERCQqVh1OOPEfERGR+Fh1OCEiIiLxYTghIiIiUWE4AcBhToiIiMTDqsMJxzkhIiISH6sOJ0RERCQ+DCdEREQkKgwnAASw0QkREZFYGBxO9u/fj9GjR8PX1xcSiQTbt2/Xuf62bdswdOhQNG/eHC4uLggPD8fu3btrW16jOp2cAwC4fCfPsgUhIiIiBYPDSX5+PoKCgrBy5Uq91t+/fz+GDh2KnTt3Ij4+HoMGDcLo0aNx+vRpgwtrbA+KywAAX+1PtHBJiIiIqJKdoRuMGDECI0aM0Hv9FStWqLz+6KOP8Ouvv+L3339HSEiIoYcnIiKiBs7gcFJXcrkceXl5cHd317pOcXExiouLFa9lMpk5ikZEREQiYPYGsZ9++ikePHiA559/Xus6MTExcHV1Vfz4+/ubsYRERERkSWYNJ+vXr8fChQuxefNmeHp6al0vOjoaubm5ip+UlBQzlpKIiIgsyWyPdTZu3IjXXnsNW7ZsQUREhM51pVIppFKpmUpGREREYmKWmpMNGzZg0qRJ2LBhA0aNGmWOQxIREVE9ZXDNyYMHD3D9+nXF65s3byIhIQHu7u5o2bIloqOjkZqaiv/9738AKh7lTJw4EZ999hnCwsKQnp4OAHBycoKrq6uRfg0iIiJqKAyuOTl58iRCQkIU3YCjoqIQEhKC+fPnAwDu3LmD5ORkxfpff/01ysrKMG3aNPj4+Ch+ZsyYYaRfgYiIiBoSg2tOBg4cCEHQPtz7unXrVF7HxcUZeggiIiKyYpxbh4iIiESF4YSIiIhEheGEiIiIRIXhhIiIiESF4YSIiIhEheGEiIiIRIXhhIiIiESF4YSIiIhEheGEiIiIRIXhhIiIiESF4YSIiIhEheGEiIiIRIXhhIiIiESF4YSIiIhEheGEiIiIRIXhhIiIiESF4YSIiIhEheGEiIiIRIXhhIiIiESF4YSIiIhEheGEiIiIRIXhhIiIiESF4YSIiIhEheHkoasZeZYuAhEREYHhROFaxgNLF4GIiIjAcEJEREQiw3BCREREosJwQkRERKLCcEJERESiwnDykADB0kUgIiIiMJwQERGRyDCcEBERkagwnBAREZGoMJwQERGRqDCcEBERkagwnBAREZGoMJw8JLAnMRERkSgwnBAREZGoMJw8dDev2NJFICIiIjCcKHzwx0VLF4GIiIjAcEJEREQiY3A42b9/P0aPHg1fX19IJBJs3769xm3i4uLQo0cPSKVStG3bFuvWratFUYmIiMgaGBxO8vPzERQUhJUrV+q1/s2bNzFq1CgMGjQICQkJmDlzJl577TXs3r3b4MISERFRw2dn6AYjRozAiBEj9F5/9erVCAwMxLJlywAAnTp1wsGDB/Gf//wHkZGRhh6eiIiIGjiTtzk5cuQIIiIiVJZFRkbiyJEjWrcpLi6GTCZT+SEiIiLrYPJwkp6eDi8vL5VlXl5ekMlkKCws1LhNTEwMXF1dFT/+/v6mLiYRERGJhCh760RHRyM3N1fxk5KSYukiERERkZkY3ObEUN7e3sjIyFBZlpGRARcXFzg5OWncRiqVQiqVmrpoREREJEImrzkJDw9HbGysyrI9e/YgPDzc1IcmIiKiesjgcPLgwQMkJCQgISEBQEVX4YSEBCQnJwOoeCQzYcIExfpTpkxBYmIi3nvvPVy+fBmrVq3C5s2b8fbbbxvnNyAiIqIGxeBwcvLkSYSEhCAkJAQAEBUVhZCQEMyfPx8AcOfOHUVQAYDAwEDs2LEDe/bsQVBQEJYtW4ZvvvmG3YiJiIhII4kgCIKlC1ETmUwGV1dX5ObmwsXFxWj7DZi9Q+V10pJRRts3ERGRtavt9VuUvXWIiIjIejGcEBERkagwnBAREZGoMJwQERGRqDCcEBERkagwnBAREZGoMJwQERGRqDCcEBERkagwnBAREZGoMJwQERGRqDCcEBERkagwnBAREZGoMJwQERGRqDCcKCkrl1u6CERERFaP4URJdkGJpYtARERk9RhOlAmWLgARERExnBAREZGoMJwoYcUJERGR5TGcKBGYToiIiCyO4UTJzax8SxeBiIjI6jGcKJm19ayli0BERGT1GE6UFJSUW7oIREREVo/hRAUbnRAREVkaw4kSNoglIiKyPIYTJffyOUIsERGRpTGcEBERkagwnBAREZGoMJwQERGRqDCcEBERkagwnBAREZGoMJwQERGRqDCcEBERkagwnBAREZGoMJxUI5dzmFgiIiJLYjip5uf425YuAhERkVVjOKnmva1nLV0EIiIiq8ZwQkRERKLCcEJERESiwnBCREREosJwQkRERKLCcEJERESiUqtwsnLlSgQEBMDR0RFhYWE4fvy4zvVXrFiBDh06wMnJCf7+/nj77bdRVFRUqwITERFRw2ZwONm0aROioqKwYMECnDp1CkFBQYiMjERmZqbG9devX4/Zs2djwYIFuHTpEr799lts2rQJ77//fp0LT0RERA2PweFk+fLlmDx5MiZNmoTOnTtj9erVcHZ2xtq1azWuf/jwYfTr1w8vvvgiAgICMGzYMIwdO7bG2hYiIiKyTgaFk5KSEsTHxyMiIqJqBzY2iIiIwJEjRzRu07dvX8THxyvCSGJiInbu3ImRI0fWodhERETUUNkZsnJWVhbKy8vh5eWlstzLywuXL1/WuM2LL76IrKws9O/fH4IgoKysDFOmTNH5WKe4uBjFxcWK1zKZzJBiEhERUT1m8t46cXFx+Oijj7Bq1SqcOnUK27Ztw44dO7Bo0SKt28TExMDV1VXx4+/vb+piEhERkUgYVHPi4eEBW1tbZGRkqCzPyMiAt7e3xm3mzZuH8ePH47XXXgMAdOvWDfn5+Xj99dcxZ84c2Nio56Po6GhERUUpXstkMrMGlHK5AFsbidmOR0RERFUMqjlxcHBAaGgoYmNjFcvkcjliY2MRHh6ucZuCggK1AGJrawsAEARB4zZSqRQuLi4qP+b0/eEksx6PiIiIqhj8WCcqKgpr1qzB999/j0uXLmHq1KnIz8/HpEmTAAATJkxAdHS0Yv3Ro0fjyy+/xMaNG3Hz5k3s2bMH8+bNw+jRoxUhRWx2nLtj6SIQERFZLYMe6wDAmDFjcPfuXcyfPx/p6ekIDg7Grl27FI1kk5OTVWpK5s6dC4lEgrlz5yI1NRXNmzfH6NGjsXjxYuP9FkZ2v6DE0kUgIiKyWhJB27MVEZHJZHB1dUVubq5RH/EEzN6h9b2kJaOMdhwiIiJrVNvrN+fWISIiIlFhOCEiIiJRYTghIiIiUWE40eLG3QeWLgIREZFVYjjRYsiyfyxdBCIiIqvEcEJERESiwnBCREREosJwokM9GAKGiIiowWE40eHAtSxLF4GIiMjqMJzocDMr39JFICIisjoMJ0RERCQqDCc6sM0JERGR+TGc6FAmZzghIiIyN4YTHT7cccnSRSAiIrI6DCdEREQkKgwnREREJCoMJ0RERCQqDCdEREQkKgwnREREJCoMJ0RERCQqDCc1WPbXFUsXgYiIyKownNTgi73XkZlXZOliEBERWQ2GEz1kyootXQQiIiKrwXBCREREosJwogc5JwAkIiIyG4YTPYz/9rjK68S7D7DvSqaFSkNERNSw2Vm6APVBbmGpyuvBy/4BAGyd2hehrZpaokhEREQNFmtO6uBCWq6li0BERNTgWHU4sZFYugRERERUnVWHk8gu3pYuAhEREVVj1eFk/ujOli4CERERVWPV4cTH1UnvdePYO4eIiMgsrDqcGOLl707g0h2ZyjIOf0JERGR8DCcGGPHZAUsXgYiIqMFjOCEiIiJRYTghIiIiUWE4ISIiIlFhOKkDgS1iiYiIjI7hhIiIiESF4YSIiIhEheGkDv6777qli0BERNTgMJwY6FpGnuLfWQ9KTHIMQRBwMU2G4rJyk+yfiIhIzGoVTlauXImAgAA4OjoiLCwMx48f17l+Tk4Opk2bBh8fH0ilUrRv3x47d+6sVYEtbeh/9qu8zi0sNfoxNhxPwcjPD+C1708afd9ERERiZ3A42bRpE6KiorBgwQKcOnUKQUFBiIyMRGam5rlnSkpKMHToUCQlJeHnn3/GlStXsGbNGvj5+dW58GKw4Xiy0ff5/eEkAMCBa1lG3zcREZHY2Rm6wfLlyzF58mRMmjQJALB69Wrs2LEDa9euxezZs9XWX7t2LbKzs3H48GHY29sDAAICAupWaiIiImqwDKo5KSkpQXx8PCIiIqp2YGODiIgIHDlyROM2v/32G8LDwzFt2jR4eXmha9eu+Oijj1Berr09RXFxMWQymcqPWHGoEyIiIuMyKJxkZWWhvLwcXl5eKsu9vLyQnp6ucZvExET8/PPPKC8vx86dOzFv3jwsW7YMH374odbjxMTEwNXVVfHj7+9vSDHNavvpVEsXgYiIqEExeW8duVwOT09PfP311wgNDcWYMWMwZ84crF69Wus20dHRyM3NVfykpKSYupi1dkWp9w4RERHVnUFtTjw8PGBra4uMjAyV5RkZGfD29ta4jY+PD+zt7WFra6tY1qlTJ6Snp6OkpAQODg5q20ilUkilUkOKRkRERA2EQTUnDg4OCA0NRWxsrGKZXC5HbGwswsPDNW7Tr18/XL9+HXK5XLHs6tWr8PHx0RhMGqIMWRHyiozf5biSIAi496DYZPsnIiIyJ4Mf60RFRWHNmjX4/vvvcenSJUydOhX5+fmK3jsTJkxAdHS0Yv2pU6ciOzsbM2bMwNWrV7Fjxw589NFHmDZtmvF+CxE4k5KDgZ/sw18XVNveZD0oRthHsej277+w4+wdRG1KQFGp7sHVJBLdxxIEQWUfMzYmIPTDv3Hg2t1al5+IiEgsDO5KPGbMGNy9exfz589Heno6goODsWvXLkUj2eTkZNjYVGUef39/7N69G2+//Ta6d+8OPz8/zJgxA7NmzTLebyECr6w7gXv5JXj9h3gkLRmlWH4uNVfx72nrTwEAOvm4YPIjrWt9rEnrTiDuyl0cnzMEnk0c8duZNADA6n9uYEC75rXeLxERkRgYHE4AYPr06Zg+fbrG9+Li4tSWhYeH4+jRo7U5VL1RWENtiLKsfO2PYErK5LicrruRbdyVihqS3xLS8NqA2occQyzddRlO9rb4vyHtzHI8IiKyXrUKJ6ROrmXAkxqe0Kj54eitWpdBYvDR9HMntxCr4m4AAN54tA0c7DglExERmQ6vMkaQW1iKolK5xvckGhqQfPVPIkrLNa9/+36BUctmDMVafjciIiJTYDgxgqCFf6ktu56Zh+nrT6nMYqxs80n1sVvOpOTgp2Oqc/UMXhaHz2OvYd/lTHwRew2CBYak5SC4RERkTnysYwLbTt1G1OYzAIA/zt7RuE5aTqHasidWHlJblng3H8v3XFW87uLnovW4NfXyqS3lQFTXY1zLyMOr35/EjCHt8ExoC43ryOUCUnMK4e/uXLeDERFRvcSaExOoDCamcCe3yGT7Nod//XwWydkFeGeL9nP03tazGLB0HzaaYMZnIiISP4YTC9lx9g7K5QJmbjyNbw4k1mof5nrCo3yYulbOlJTV3H7l5/jbAIDPYq/V8WhERFQfMZxYSNK9Aqw/noztCWn4cMcl5BeXmfX4BSW1O56mBr4GbV+nrcUlLacQS3ddxp1c9Ud0RERUewwnFpSSXdUzZ/L/Tuq1TfXaEtX2IFWX/tv3C7BmfyLu55eo7eOHI0noPH83Np/Qb0JFU9XQjP/2GDJkdX9Mde9BMb765wbu5pl3CP+XvzuOVXE3MOm7E2Y9LhFRQ2f14WTj630sXQQAwOEb92q13W6l4fKVayUe/+8hLN55CSGL9ihqSRLvPsDNrHzM+/UCgIq2HZUEQUBOgXqQMTblipcD17Iw/9fzNW5z+36BzpqeKT/GI+bPy3hlnfFCQmm5HGVauntXuprxAABqHDSPiIgMY/XhpE/rZpYuQq1tPpmC+Q+DBgD8c/UuTiZlAwCylWpMZm09h6LScgxe9g8GfRqncV/v/3IewR/swb7LmQCAuCuZ6L34b+y/ehdxVzK1lkEuFxB3JVPviQerPxXK1lCzo1gXwI27D9D/430Ij9mrdb0TSfcBqE4VUBdl5XKEx8RiwNJ9kMvZkZqIyNzYldiCLt2RGbyN8ki01zIfqL3/7OojKnP7AMDvZ9Jw7naOzv1ueNgzZtmeKxjU0RMvP3xUMWHtcZ3bbTqZguht5wAAFz+IhLOD7j+pmkaxfVCt7c3+qxVD9ecWmm5W5+oy84qR9aAiNBWUlqOxlP9NiIjMyeprTiwp8W6+wdso15RoU31mZKCiAa4+BEF3j5rqj36Uj/XKuhMq7Wiq++NsWo21G8NX7K+2/wyd65sC60qIjK+wpNwig0iKWU2Pjq0Zw4kFpWoYiM0YXv8hXu91M/OK8PvDWY0B4EKaDKGL9mhdf8RnB1Relyt91xxNzMaApftwREP7maLSckxff1ptefWalNv3Vc/JkUTVfV1Mk+GxLw5g2V9X8Ogn+3Q+cqotlUbGRt+7ZtczHyBLz0djRPXN9cw8dJq/CzM3JVi6KKKxeMdFdJq/C4l31WvAieHE6j32+UH83wbV0JCno1tzZl4xrmbkYf6v55EpK8KZlBy1dbbEq/cCKtFyh3D8YRsZTfJL1Gd6fu37EzifKsMXe6/j1r0CxeMnY1K+uUsxw1xHKdkFiFj+D3p++LfJj0VkCd8cuAkA+DUhrYY1rceaAzdRWi7gi73XLV0UUWI4sXKZteh+O3zFfvzvyC3M2Jigsy2IIAgoKi2HXC4gKUv7I6zzWh71aNp3hhG7C2tq7HrrXj4GLN2neP2OCUf7rWSshrxE2pxJycGmE8l8rEL1Blv6kcEqr+nVH7lU98LXR3E8KRuPB/nqvGO6fb8QXf1ca3xEsyruOsqN1Hvm8PUsvPr9SXz4ZFeVOX4W/XFRZb3KcWJ+TUiFo70tBnXwxAd/XMAj7ZobpRxE5lA5b5ePqxMeac+/XTFhYNSMNSdkEp/uvoJjN7MhCDVX5abnFkIQhBof0SzddcVo5Xvl+xMoLC1Xm+OnTEP4uZtXjBkbE/DGD/H435Ek/Hg02aB2PbocvJalMrFjpW8OJOKHo7eMcozqyuUCbmblN7gvxeM3szHiswOK7vTGsOt8OvZpCc0lZXIkpOTUq+7m1zX08CMSI9ackNHtPp+usb2INv/+/aLevYmMRdt1ufrytNwiyIqqHi/pM/FizM5L8HZ1xKR+gTWu+9K3x9SWZeYV4cMdlwAAL/Tyh72tce8h3tmcgO0JaXi6hx/KygW8P7ITvF0djXoMS3j+qyMANHenr43s/BJM+bEihN74aCRsbVSbR7+9KQE7zt3BzIh2mBnRHkDFXfB//r6G1h6N8GSIX53LYKgVf19FIwc7TH6ktdmPrUsDy8FkBqw5IaMzJJhUWnc4yfgFMZK8Iv3nIbqQlouv9idi4e8Xa15Zg8y8IhQqnT+5Cb7Vtz+sydp2KhW/nUnDuz+bvl2NqV3PNP4ovcptnjR9DjvO3QEArNlfNXFn/K37+Dz2mkV6paTlFGLF39eweOcloz3+NIcr6Xn4PPZaref7qu/qzydlXgwnZJUM+UJQvkjV1LW4piBT+Sjlfn4JXvpGvdak9+JYldcPispMPhbCrVrUWtX24ldaLsecX85h13n1sXjqYtbWczrfLyotx5xfzilGQDYVXSMem1qBUqjV9ndalwthdn4JSmv5t6hrvtDIFfuxfM9V/EfD402yXgwnZFXK5ULFQHHaHutoWHbipn5tGM6n5uJqhvY7+O8PJ6HHoj24mCZDyKI9OHg9q8Z9hn74N4ZVG5jO0t7acBq9Fv+t8rhLl+Kychy8loWi0nJsPJGCn44lKx6XGEtNF83vDiXhp2PJmGTE+Zesya17+eixaA9GVhvnyJjO3m54vdZOJ9/HhuNVvaROJd9H9LZzKhOy8pGXZmxzQg1KYPQO3IzR3t5gxsbT+OPsHYP2+d99NY9DkFtYise+OKiybNup23i6R1VPoAW/VYzuO/Jz3V/w1QemS7ybj1fWncDLfQMUPS32Xc7E4RtZmDW8I+zq2CYlObsAQ5bF4akQP0wf3K7G9X97OGjfjrN3MLZ3yxrXX/DrBWw8kYIng33h7+5cp7LWVpqJBjw0pivpedh4IhnTBrWFR2NpLfZguqvcnw9rujRNmaEPfS7A1Vf5PPYatsSnYNvUfmjeRP18bDyejOIyOZKzCxDs74bRQb56lEPAmz+dgtTOBiteCNGz9LX31KrDAAA/t4peUk8/fF1Uavijb0PkFJTgTm4ROvm4mPQ4psSaEwABzSzzhUnGJwjqQ+wrMzSY6EvT6K5Rm8/U6vGHpvYNey9nqsxzNGndCaw5cBM/x99WWS89t6hWx7xxNx+f/qVerZ6QkoOluy5r/TIVBAHHb2brfJyx8UTFoHzb9RyA660NpzHmqyOKXjDnU3PxRew1k3+h63IhTYbH/3sQhzTUdhkrEkSu2I/vDiXhX1vq3gZI12MUTe7nl5h1/iqgYuqL7adTtb6/fM9VpGQXYqWGm4Oi0nLM3nYOC367gG8P3lQbSFKTqxl5uJAmw5/n07E9IQ35GgabLCot1zoFx+EbWXh13QncVhqYMT23CPuuZNbY8636KLCmHhW290exGPHZAY2DZCq7k1uIy+mGz/FmDgwnALZO7WvpIpBIlOqYVwgw/Eu/NiKW/6P1vcS7DzDum6OK18q9hw5dz0KfmFi8/J3uyRoN8eTKQ1gVd0PjBQIA/rqYgee/OoKBn+zT+H5t/HYmDcduZuPiw4kxH/viIJbtuYrV/9wAAPxw9BYCZu9QTDipD12f2/XMinN6XMfjuwnfHsPZ27kYp6GdkLFpG5TQGDRdRItKyxGyaA+CFv6ltVu0oV2Qd19Ix5Qf4pFboB54+i3Zi9ScQrz+Q7xqw2Et13dNYVtTl39dbtx9gGH/2a9WuwkAuQWl+ONsGopKyzH6i4MYsHQfTiXfV1vvxTXHEHs5UyU89omJxaTvTmDbKe0hSxNBw7/XHryJgNk7MPbro2qfQ6asCKvirus9xUXl/GiVE6dqEx6zF8NXHDDZVCp1wXACoFmtqlBJrBb+fhGpOYW4fb8A/ZbsxTcHEmveCBXPh2saWE7fLshVyw2/r9b1xTv1x1M4dF1zGb9/2OPpwDXtbVlqO7bJtYwHOJOSg61KNTX3C0oQe6liYkaZlobAx6qdz5oOf02pzU71dS+myXAntxDztp8HUDGTdl3vQO/nlyBi+T84dP2eoityJeU8o+33q05ijvRqZJmyqgte9WkmBEHAyn3X1WroavLGD/HYdSEd//m7ojZO+bSk5hSi35K9atvo6pkmCALO3s5R6cmmydHEe0jJLsDvZ9LQe/HfipARn6QeNgDgz3N3EPTBX5i+/jQ+3HFR8djqxyO3NNasAECGTD0gvLPlTJ3GDcovLsMHDweAPJJ4D1eqtV2btO4Elu66gp4f/o1kEwy7cDXd+L3d6orhhBqcX06n4rXvT2LJn5eRmlOoGDOkJpXPh3U5rGFSQ3O6k6t6h6P8pV990kRN3v/lfK2Om1tYiidWHlIZtG7pris1NmLUd2j+qE0JeOyLAxj6n6rGv1tPqV8Qc6rdiecX6/eoRzkyyOUCvj14E6eT72PpbsMH9tM063dNbmbl48A13XexlZSvcb8mpCoCoCHb/aL0uKT6RfNyugyvrjuBC2kVn42g48HUoev38ImGc3T4ehZGfHYAT686hItpMq09yu4+vNPX57r9oLgMK/6+qrGWZuupVDz+30NqAVLZmZQcvPD1UQxYug//t+E0MvOK8fr/Tmpd/6t/bmDqT6cUr389XfXYcdvpVHRZsBvFZeX45kCiynFvZuVrrBEypDan+vmoqUH3hbSqRy+PGFBLqalEcrmA65l5oh+EkQ1iHwpq4YozDbC1uLW6dEemd1uicrmg9xgLlY8a1Gn+jy6g4uJwuoZnv/qqfgdf2Xg2JbtArWxyuYBSuRxSO1vFsg3Hk/U+lvLjBW01SpfrcMeVnV8C90YOACouBtWtO5yEfz/eRfH67O1cTK1lL59ypS/i38+mKaYpGNbZy+B9vf5DvMogbwUl5bh9vwC2NhIc1XKeBn0aBwD4bXo/dG/hBqDicYqjvS2+O3RT5fFcZQ3CndxCzNiYAAA48N4ggxoTR20+g2aNpZi49jgaOVR9/hKJBGO/Por7BaU4fOMeLi0ajrtK81XtfDh2y1f/JOLrCaFIl2kedPBFpcdbIz8/AD83J3wzsSd+jr+NaYPaVq1owPXvcnoeLqfnYcXf19QG0dt8sqLdkq6we1rDo5jKwKApgH2ux4R7K/de17je+7+c01mW5HsFajcSypTLoykk1FQBV7lNSbnq/299fPDHRaw7nIQZQ2pu/G5JDCcPhbZyZzhpYPStYT+fmgtfN6c6HevUrRyNywWhoko27op+d82GqvwdNV0Un1l9GBdSZYifF4EmjvYG73tVnGGzpZbLBSSk5OB6Zh7CW3ugZTNn/H5GeyPYocv/wbpJvdFIqt+Xq6YL5R/n0tDFt+YeCT8erQplutpP/HYmDY/r0eujuv4fa76bFQRB5XgX0mTo3sINB69l4aVvj8HH1VHrqMP386vuzgcs3YfZIzrCx9URTwRrHnm2+iVu4sMG1MqDIgqCgPsP7/oLHzYwfnZ1Va1AlNJEl3N+Oa/3KLepOYUY8bCb8W0zzORtzCkKdNEUmoGqAfi00VS7odxj7Hyq6o2EpkqMpKx8/HYmDRP7Bqi9t/tCBr4/nISTt7Jxcu5Q3M0rQmGJHN1auOosF1A14OVnsdeqForwaSTDCTVYO8/pV/0ukdS9oet7W89qXF5aLjdZMAEqejRsiU/BxPAAleU/HbuF08k5ACqq5od39TZZGSq1eX+nyuukJaN0Bv57+SUY/d+D2PnWgFof86t/EtHOs4lB363KoxFXDwZvbTiNwpIyjOmlu4u0tvYI1f10LBlzt6s/Spv/63mNxweqQsbpFNWagCV/XgYAuDrZI0NWhOdC/SGRVDQwTcstwo+vhulVJn1pq03s/7F6exFlyo8gLqTl1nkQweq1Ht8duqlxBGZNI1OXlQvYdyXToFGelenzqLTSj0dvaZ2yYlXcDfxby6jRAoAxX1dv7yTByM8PoKCkXOPYScrjBM3YeFrxHRM/N0KlDaWpB3A0JYYTsgpb429rHfTsxt0HeGWd9mfTdWGOp7op2YVqo2vOUWpbMuXHeByePRhNHPX7755XVFqrmpbqvlC+M3tIU/V6mVz7F+jX+2/UeJxtp26r7XXh7xewYHQXjesrX6iS7uWrvT9r6zk839NfZ2DtsmB3jeUKmL1DbVnlHfItLd1VldeZo6V9UOUEmbO2nkMTRzvF76Npnqa60Pa3a8gFO+leAWZtPQcbA8O/cm+Vc7dzIbWvql3TNjWEprYxD4rLMKmGCUUr5ekZOLVZ+PtFdPF1Re9Ad7X3MvN097K5mqFam/eguEwx4u+JGmqJlG9+7uQWqYx7dF3PBuMirDhhOKlUDxvZkwGqzz6s7O1NpptbxlzzhdQ0n9GYr48gJbvmi8qvCamKdg6+dZwMcJmG4cg1Xdge/+8h7fvQMPZKdYdv3EP3atXZ3x1Kwt28Yjwb2qJW474ERu+seaVakBWVQhAEnWXKLSzFT8f0m5G6tjUClV74WnsD04LicqOMuaKpYXNN+sRUTeNw5nauxgu+GD3/lXEmnVx78Gattvs5/jZkWsaruZNbiGaNNPdMnbX1LI69H1GrY5oKwwmRCX1QywkAjU2fYAJAEUyAihmZje1XPQdiM1RZufrF/o+zd2ocdK+uF3dDLfnzsuLxjC7aak3q6mKaaluHo4na78qrd2fVlzE6gdRU09CgaDhfhUoDDmrquqxN9QlUKz+Lc7dzMfq/6mO81OYY5sJwQmRCphqR1loU1zAoXqVCC44eW59oa+BpTKYY0EvXAHli8+RK7TWB+tprpAkqU3MK8eq6E4g18YSXpsBxTh56ItjwFvpEJA6a2o4QWUKCgcMG3Ncx3UZdnb2dq3cw+WT3ZVGNfcJw8lD3Fm7474umnwiKiIxPRN+pRAax9MCOlVbuu4H4W5pH0rUEhhMlfnUc64KIiKi+qj4CsyUxnBAREZFZhj7QF8OJkg7eTSxdBCIiIqvHcKLE2YGdl4iIiCyN4YSIiIjYW4eIiIhIm1qFk5UrVyIgIACOjo4ICwvD8ePH9dpu48aNkEgkePLJJ2tzWCIiIrICBoeTTZs2ISoqCgsWLMCpU6cQFBSEyMhIZGbqHuglKSkJ//rXvzBgQO1nICUiIqKGz+Bwsnz5ckyePBmTJk1C586dsXr1ajg7O2Pt2rVatykvL8e4ceOwcOFCtG7duk4FJiIioobNoHBSUlKC+Ph4RERUzV5oY2ODiIgIHDmifXbLDz74AJ6ennj11Vf1Ok5xcTFkMpnKj7nYGTq3NxERUQMgnuawBoaTrKwslJeXw8vLS2W5l5cX0tPTNW5z8OBBfPvtt1izZo3ex4mJiYGrq6vix9/f35Bi1snmKeFmOxYRERGpM2lvnby8PIwfPx5r1qyBh4eH3ttFR0cjNzdX8ZOSkmLCUqrydnE027GIiIjEQkQ9iWHQqGMeHh6wtbVFRkaGyvKMjAx4e3urrX/jxg0kJSVh9OjRimVyecUU6HZ2drhy5QratGmjtp1UKoVUKjWkaERERNRAGFRz4uDggNDQUMTGxiqWyeVyxMbGIjxc/XFIx44dce7cOSQkJCh+Hn/8cQwaNAgJCQlmfVxDREREuoin6sTg8dqjoqIwceJE9OzZE71798aKFSuQn5+PSZMmAQAmTJgAPz8/xMTEwNHREV27dlXZ3s3NDQDUlouFs4OtpYtARERkdpfT8zC8q4+liwGgFuFkzJgxuHv3LubPn4/09HQEBwdj165dikayycnJsLGpvwPPujk7WLoIREREZicXT8UJJIKYBtPXQiaTwdXVFbm5uXBxcTH58QJm7zD5MYiIiMTkrcFtETWsg1H3Wdvrd/2t4iAiIiKjEVNNBcOJBl4u7ClERETWRUzPURhONNj/3iBLF4GIiMis5CJKJwwnGkjt2GOHiIjIUhhOtIh5upuli0BERGQ24qk3YTjRiu1OiIjImojoqQ7DCREREQGCiOpOGE6IiIgI5eUMJ0RERCQihaXlli6CAsOJFp5NHC1dBCIiIrMRT70Jw4lWXf1c0bwJG8USEZF1YIPYemJ8n1aWLgIREZFZiGmqPYYTIiIi4gix9YXE0gUgIiIyExFlE4YTXUT0OREREZmUXEQXPYYTIiIiYpsTIiIiEhe2OSEiIiJREU80YTghIiIiAD1aNrV0ERQYTnQY0slTbdnCx7uovG7WyMFcxSEiIjIZe1vxRALxlESEuvi6qrwe1d0HAR6NDNpHhIaAQ0REJDblbHPScEhqGAxlVHcf8xSEiIioLhhO6g+HatVc1bNITZ+lhEO5ERFRPSCeaMJwUqOEBUMxtrc/nOxt8c7Q9jXWlChbPznMdAUjIiIyonIRjcLGcFIDZwc7xDzdHecXRqJ188ZqNSG6woqbk4NBYYaIiMhS7GzEc8FiONGTrYg+NCIiImPr7Oti6SIoMJwYqHpNyPCu3jWsb3ioaefZ2OBtiIiI6kJE7WEZTupi98xHML5PgOK1n5tTjQ1o9TGgXfO6FYyIiMhAIsomsLN0Aeob5bDRwbsJAGDtyz3h7eKE9l6NUSYX0HHeLgCAk4NtrdqcsJ0KERGZW1ALN0sXQYE1JwaqDCTKBnf0QmdfF9jZ2sDR3hZzR3XC9EFtEejRyOCuxOcXRuLJYL9al89Uj4Q8GktNsl8iIhIHBzvxRALWnBioWWMpDrw3CI2k2k/dawNa13r/jaV26NbCFYdnD0bfJXtrXP/biT0R3qYZbCQS7L6QjgHtmqPHoj21Pr42J+dGoKxcjnHfHENKdgHScouMfgwiIiKANSe14u/uDHc959TR9IimX9tmNW7n6+ZU4zrRIzpicEdPODvYwdHeFk8E+8G9kQOe6dFCr7IZys7WBhtf74MDswabZP9EREQAw4nJKWeTOSM74c2BbfDdy70RPaJjnffdK9BdY2+g2bXY99sR7bW+t3VquOLfEomE3aqJiMikGE5MTDk7TH6kNd4b3hEOdjZ449E2NW77eJCv7n1rWd68iVRn2HgutAVe7hugsmxGRDut64e2cldb5uPqqLNstbFlSnjNKxERUYPHNici0slHdQCcpc92R2QXb5y9nYOxvVvicnoeWjR1wmNfHKxxX4722nPnJ88FAQDWHU6qdVlNUXfS2sAZn4mo/hjfpxV+OHrL0sWgeoI1JyamaxC29ZPD8G5kh6p1q73vaG+LUd19ED2yEwI8GmF4V2909XM1UUk1c3Wy17hcU394Q2tTpg2qqD1ysLPBvn8NRDP2CCJqsFo1c7Z0EageYTgxMV01DH3beGDaoLZV64qkKUdrj0ZYPzkMQf5u+PFV/Scv7NvGQ23Z/w1uq2HNCjOGtMdnLwTj0KzBCNRRaxLk78ZRc4nqueo1w0S6MJw0ULUJOi/1aQkAeH9kJ/Rt44Ffp/VDtxaaa2q0DXM8uKOn4t/rJvXCO8M6aF4RFTUmTwT7oXkT3TUm3f1c0d5LfXyZPq3V28KQdVn9Uqili0B66tdW/ebFFIzR2YAsj+HExGozt44phAVWXMgHtNP+BbHoia5ImD8UEZ29atyftsF6vpnQU23Z9cUjMFSPfeqk4TQ+2t5TfaEJjOrmg6dDVAfGmzwgsE77nD5Ie40S6a+mua0spabG7GQ6bzzaBkejhyCySx2/cwB09TN/bc9nLwSb/ZgA8Fr/un2nGRvDiYm1aFrzeCVRQyt61ix8vIvJyjGiqzfi/jUQa1/upVj2YlhFTclzoRXjokgkErg56zd+y6pxPeDn5oTPx4aoLLfR0M3YztYGjXUMWqePWZHGvxta/nxQjevMe6wzVo7rgbeHVvV+OhI9GLOGay/P/Mc6I35uhNb3343sgKih7c3yqOrJYF+cnje0xvUmhrcyyvEiOnkitFVTo+yrPlv8VNc67+Pk3Ahse7OvEUrTcIUFumOChr9db1dHg0fn1uTnKX0R+86jdd6PJV1YGKnXemKaVwdgODG5Tj4uWP58EDZM7qN1nbeGtMPVD0egZ4BpH1MEeDSCvdLEhAsf74JNr/fB4qe6Gbyvrn6uODR7MB4P8lVcjF7o7Q8AeKZHC3T2cTFaNa5EArRs5ozXH1EfeVf5MZKhNLWRASruXNa/FobEj0bi1Yd3E/7uztg6NRyx7zwKH1cn2NpItAbPV/oHam3cO2t4R0wb1BY2NoaNF6PpLlCfKQXmjOqMpo0c8Pv0/jrX01TesQ8/T0MENGuErVP74qmQ2k/BoIu+gx+awzAdtYFNHDU3JDeER2MperSsf0HvsxeCVRr6A0AHDY9ljeH7V3rj/ZGdNL5njEprR3tbtGle95sIPz0G1awNfW5oG0nt1M7FmQXDNNZyiwnDiRk83aMFwtvoHhXWkDkNegU0hZ+bE7r4au+5o89dg72tDcJaN6vzfAobX++DI9GD0ethuFr2fBB2zhigEoSMQdNv5OWifw+hUd19VPenYYfSh+1g+rb1UKsFCm3lrviikkgkiPvXQL2PXWnqwKrxbZY9H4RmjRww5dE2CPZ3UyyvrMlStuhJ9TvxZXrU/FTS1naokqZzG/N0d733X6ny8whp6aZ1nbUv98RMHePq6DJVaXygXgHmvXArD0YIADFP6w71i5/qqvY3VxsieTKsl+NzhuCJYD+1G4l1r/TSskXtDe/iDUd7Wzja22Lliz3g7GCLrVOrapqMed48GhsWiqs3/u3i64LvX+mNKXqMb2XITcvEvgH43yu9tb7v+7AH5dUPRyiWPdOjBVyd7BHR2Qvn/j1M72OZW62uHitXrkRAQAAcHR0RFhaG48ePa113zZo1GDBgAJo2bYqmTZsiIiJC5/pUs81vhGP/e4MMChWmrLKzt7WBj6vuO4NnNVxw66riy6d2v9nTIX4aL8jj++j/eMPO1gZtmqv2MlKuYv5yXA8sfLyLon1K9XYIXXxdcXJuBGaP6IgxvapqKV7V1J6l2q954L1BkOrx+Tdx1O9xmrYv8rG9W6q8dqghcE7oq/v8LX6qKwZ39MJMHYME6mJnW1XQ6o8U62pMT+01RWN6+qsMRvj1+FCttWPfTqy4Ix0X1gorX+xR53I52tnWeR9AVbuz6n+zplD9xqTy+yGohpCsr2uLR2D1+KrG0KO6++DiB8NVHilqu0HT1e5OG+XQo4+ITp4qg0rGPN0Nj7ZvjlnDtXcQqKRc6/Z31CM1rv9I++ZY/VIPjQNvPvmwBlP58xCUvkyMUcNnKgY3BNi0aROioqKwevVqhIWFYcWKFYiMjMSVK1fg6alexR4XF4exY8eib9++cHR0xMcff4xhw4bhwoUL8PMzTdVvQyeRSGBbQ7ju7CuubnvVH/Ecf3+I3ttW/qp1DViSai+UJ290sLVBSbnc4MdEv0zrhyM37sHORgI3Z3uVKcdHdKu4ay4rlyOyi7fG2gtNDaY1VYELAJ7u4Ydtp1IBVDxmSs0p1FimHi3dsOz5YAiCAEd7/S5s2hpuh7dphg3HkwEAS5/pjidCfLHuUBKy80uQU1CKM7dzcDk9T7G+VMeF9N3IDhgXZpy2LUDVBe+XN/viTEoOisrk2H/1Lg7fuFer/Y3q7oNNJ1MUr78eH4rXf4gHAIyuFix1PV4a0snwhpjDOnvhr4sZGt8Lb9MMey9nonkTKWYP74h3tpwxeP8AsOmNcJSUyWFrI0Gb93cqls8a3hEf77qstn5luypN7+njmR4tsPXUbZVlPQPcceZ2bq32BwAdvZtg/ujO+tXKaviTnjW8I/zdnXDgWpZBx23VrBF6B7rj+M1sAMCm1/tgzNdHdW7TK8Adf84YAAc7G0WQNbSDhI2e6w/v6gM7G9W/H/dGDpjUr+YbHbEyuOZk+fLlmDx5MiZNmoTOnTtj9erVcHZ2xtq1azWu/9NPP+HNN99EcHAwOnbsiG+++QZyuRyxsbF1Ljxp16+tB75QurNs5GD5wYBbulcNwuRpwOMYfb3WPxC2NhL8HaW5AVv1L4ZGUjtsfiMcW6aE4/icIfhtej/0NbCdjIujPSK7eGNIJy+EtnKHnYYvTTtbG/QMcNd54e778LGf1M4GEokE3yk1XA5t1RSeTaToXK2q2FNLF2wBQKBHI7Q24Fm5Pr1enu/lD6mdLd54tA2iR3bCx89211mlXOmbCT0xpqe/ov2OIT55tjs+eVb346WQlk3xcr9ATHm0DdbraNuliXLPiAHtPNBdKUAO66L9nBjzkUHr5o3w+dgQbJkSjvdHqje0XvZcEN6OaI9tU/vimdAWep1zbRzsbNQeG0x5VPMs6r0CmqqMNK1XTa3Shc/fXb02VXkgto2vq35WvfVoc7dr5iNa24pVp+kjEiDAqVpgH9DOQ7/2F0q/W1jrZpg1vCN6B6qW+bnQFnC0t8FLD2tgO/m4GNxmRdufVrMa2lsNqnZjdXJOhMZhGhwdjFMTZ2oGXbFKSkoQHx+P6OhoxTIbGxtERETgyJEjeu2joKAApaWlcHfX/odYXFyM4uJixWuZTGZIMemh0UG+yM4vwaHrWYrqvYYk0KMRkrLyFa/nPtYZs0Z01HpXJUHFxb+4TI5H2zcHAJUvF317KplCq2aNcOC9QWj68AtoYIfm+OP/+qNVM2c0ltpBIpGohavWzRvjk2e7Y96v51FUKlcsl9dwZ6TpLr1N88Y4/v4Q9P7IsJsGTxdHbH4jHM9/dUTrI7GIzl56dU//6Klu+PP8HZW72ud6+iMlu8CgMhniiWA/DO7oqaje/uCJrnhy5SG19QJN+CikR8umcLS3Ra8Ad5zVUKvQtJGDytxXyo/qLi8aDqmdDRb8dgEnk+7j4h3DvisHtPNQ+7t6MawlRnf3Rc8Ad5xPrSqPrRES2djeLZGeW4T+bT3Qp3UztHR3RnJ2Ad4f2RGvP9IGvyakYsbGBJWyrD+WXKtjuTlrfmQxsIMnhnb2Qjc/V0R28UagRyOcvZ1j8P6nDmyDqQPbIGD2DsWyT54LQszT3TTepOjLTik8Kj8uf6V/IJ4LbYF//XwW+6/eVduueuis3mZu8VNdsflEiqJ3aHXaxq6yFIPCSVZWFsrLy+HlpfpF4+XlhcuX9av6mzVrFnx9fRERob2rZUxMDBYuXGhI0UiLiX0DMLHaJH/1zdDO6new8x/rjGGdvbDvcqbK8spgsnvmI9hxNg1bT6WqPP44MGsQLqbJFOFETPyVapYkEonaVAWa2pg819MfT/dogTu5hej/8b6KhVq+ZbZODcfOc+mIGtoev5xORRNHOyTfK8DgThV3XJ4ujujT2h1HE7P16glUqXegOy4vGq7yCKk2X3QvhrXEcz1boN2cP1WWm7pBqPJzd6FaweP+NRA5haV69bZo59lYpe2QJv8a1h5/XcxQCSFuTtqPr0knHxc42NnA28VRcc4/eKIrfjp2C3N+OV/j9so0TTnh5+ZUYwN+ZY72NirhuJKbhqkv7G1t8J5SN/w/ZwzA5XSZolfSE8F+KuFk8ZNdcSU9D/G37utdnkpRQzvgSnoeTiRVbduiqTNsbSRYU5ueKnr+HdYUTN4c2Aar4m5oP4xEgoT5Q1EmF+CkVMvRt00zeLo4opN3E43hpCbjwloZ9bGqqZm1rn/JkiXYuHEj4uLi4OiovVo/OjoaUVFRitcymQz+/oZ3ayRx0eci4+PqiDu5RYrXv0/vr2irEdnFG1/vT4RnEyleqeERQQfvJujg3QG7LqSrLPds4gjPDsZ/pGQOz4a2wC+nU9Ua9FV0a3ZG6+aNkHg3X2sPkdBW7opGnS9pqeX4YmwPrDt8E2N6VjSE1ediCUDvti01sbe1QUQnT/x9KVPj+wFGnByyemNfQP1xvCHH26PlcaKy6YPb4c2BbVFSLsfAT+KQLisyeCA5R3tbnF0wTOUOu65WvtgD09afUlvuohQwNP3//XZiT5xJycHne68/XKnqvbFhLXEi6T4e7aD9RqCR1E7jrOcA0L+teq2OIdwbOeDH18LQYe4uxTJtY/B0b+EGPzcnlRuZEdU+F21zhz0R7ItfE9L0LlfjajVfl+7I8NSqw4plEolqLW783Aik5hSiu1J7NmtgUDjx8PCAra0tMjJUq4QzMjLg7a37P9inn36KJUuW4O+//0b37rqfIUulUkilnATOGu14awB6LNqjeK3ciDS0VVP8HfWISlVnZFdvbDyRorX9xav9AzFr6zkA9atLpiaO9rY6ew1sm9oXp5NzatUboVLzJlK8a4IB7+pC+QLVO8AdS5/tbpSxJ2rqCmwqNjYSONrYIvadR3EntxBtPQ0fA0RTGBza2UtRc/Jo++b4R4+768rsOaq7D6atV39/dJAv/r6Ugd4B7vh41xWN+3BVupA2V6pxk9rZYuW4uvdWMuZ/W3stgc7Bzgb73xuEudvPKxqAz32ss8o68x7rjOJSOcaGqYbaz14IwVMhfnqPZfJy3wAcup6l6Apdk2aNpSo9wwzpamyIxlJxtUUx6MGYg4MDQkNDVRqzVjZuDQ8P17rd0qVLsWjRIuzatQs9e4p74BeyLPdGDvjj//qjTfNGGqte23o2UellM7B9c/w2vR/+1jKK4/M6uoc2NG7ODhjU0bNOz7ur6yKyXl9AxWdqyCi0jTQ0ANQWVDt5V/y+Xi6mvzlqJLVTCyaV0zzUZtAuzyaOuLAwEjc+GonvXu6FDZP7KGrR9Jku4eW+AfB1dcRLSlX/9rY2WDUuFC9r6vXxkPKpNMZ0HZWTgI4Oqii7ckP6uhjW2UtnQ/zqF/3qn4FHYylWjw/V+Eh4YAdPtNNzoDlnBzv89FofjA8P0Gv96iYPaI3WzRtpbTtiqA+f7Io+rd0xWcMgl5Zk8GOdqKgoTJw4ET179kTv3r2xYsUK5OfnY9KkSQCACRMmwM/PDzExMQCAjz/+GPPnz8f69esREBCA9PSKavbGjRujcWPONGtN9P3a6urnith3Buq3T4lEZ3Wn8pdlPa84sYi2nk2wZUq41pqpuvh8bAje2nC6xvW8XRzR0t0Z9rYSONeip0FEZy/svZSJvOIyxTJtfwtODra49MFwlfFUzKlVs0Y4PmcIXDW019CHcnAPb9MMvQKa4s2BbRShS5d/P94FC0Z31howngzxxYbjKQj2d8PNrHzkFpYipGVTJBu5wfKv0/vhUppMMajjvMc6AxLgudC63WjMGaV5JFlLq36+a/rLa9rIAXv1/H7Ux0t9Wml9zGtJBoeTMWPG4O7du5g/fz7S09MRHByMXbt2KRrJJicnw8am6s7tyy+/RElJCZ599lmV/SxYsAD//ve/61Z6IgNYsjdOfdbLRNMqPB7ki1buznhi5SHFAGGa2NpIsO/haLy1vjM3YDMnPQJQQDPT9d7xbGK8NlF2tjY6R5KuTtf5nf9YF/Rp3QyPtm8OR3tbFJWWw83ZAcO7emPh7xfVurrXloujPcJaVzXIbdrIAcufDzbKvqn+qFWD2OnTp2P69Oka34uLi1N5nZSUVJtDUAMUFtgMSfcKVMZOMIfPXgjG1lOptR4ynUwnyN8NJ+dGoKlScNTUBteQ5+xrX+6JedsvKBo3dvNzxbnbucgrqqo5qW0D3sOzB6OgpEzr6LANmZODLZ4IrhqSoPIc+rg64ey/h4liLKXqlLs/N6rj5KOmom+jc2sjzk+LGqR5ozsjwKMRRnYz7zT3TwT7qXypkulFdvHGgt8uoJtfzXfthnRb1sfgjl4YPNsLVzPycOh6Fl7q0woDOzTH+9vOo1sLV/x1MR0rxgTXat++JprAzVL0qSXSh4tIh0G3s7XBqnE9UFRartff2VtD2uKvC+kae3KZizHa7bxcz4ePABhOyIwaS+1UJr6jhsvb1RHn/j0Mzha8m27v1QTtHzZSbOvZBJsfznUyr1ovDFN7OsQP2zR0AbekxU91xZaTtzFjSMOvTRzZTf/JF31cnXBiToTaAGb1TfXRYusjhhMiMgkxTypmTouf6oaIzl6iCif1bUAuc6rvwaShYDghIjIhJwdbg+7eybq08VTttWqM8ZgCTdhg21zM2zKRiIiIFFwc7XF8jv6ztOvy19uPYMPkPmjZzDhjw1gSwwkRUS1V9gCbPUJco+pS/WKs7uPtvZoYNC+SmPGxDhFRLc0Y0g4v9GoJby3zrhAZii1eKrDmhIioliQSCYMJkQkwnBAREYlEfZ+g1FgYTohIVHrpGMqeqKF6pkcL9ApoimB//Se1bMjY5oSIROXV/oFo4miHfm3EMy4Ikaktez7I0kUQFYYTIhIVe1sbDhBGZOX4WIeIiIhEheGEiIiIRIXhhIiIiESF4YSIiIhEheGEiIiIRIXhhIiIiESF4YSIiIhEheGEiIiIRIXhhIiIiESF4YSIiIhEheGEiIiIRIXhhIiIiESF4YSIiIhEpV7MSiwIAgBAJpNZuCRERESkr8rrduV1XF/1Ipzk5eUBAPz9/S1cEiIiIjJUXl4eXF1d9V5fIhgaZyxALpcjLS0NTZo0gUQiMdp+ZTIZ/P39kZKSAhcXF6Ptl1TxPJsPz7V58DybB8+zeZjyPAuCgLy8PPj6+sLGRv+WJPWi5sTGxgYtWrQw2f5dXFz4h28GPM/mw3NtHjzP5sHzbB6mOs+G1JhUYoNYIiIiEhWGEyIiIhIVqw4nUqkUCxYsgFQqtXRRGjSeZ/PhuTYPnmfz4Hk2DzGe53rRIJaIiIish1XXnBAREZH4MJwQERGRqDCcEBERkagwnBAREZGoWHU4WblyJQICAuDo6IiwsDAcP37c0kUSjZiYGPTq1QtNmjSBp6cnnnzySVy5ckVlnaKiIkybNg3NmjVD48aN8cwzzyAjI0NlneTkZIwaNQrOzs7w9PTEu+++i7KyMpV14uLi0KNHD0ilUrRt2xbr1q1TK481fFZLliyBRCLBzJkzFct4jo0nNTUVL730Epo1awYnJyd069YNJ0+eVLwvCALmz58PHx8fODk5ISIiAteuXVPZR3Z2NsaNGwcXFxe4ubnh1VdfxYMHD1TWOXv2LAYMGABHR0f4+/tj6dKlamXZsmULOnbsCEdHR3Tr1g07d+40zS9tZuXl5Zg3bx4CAwPh5OSENm3aYNGiRSrzqvA8G27//v0YPXo0fH19IZFIsH37dpX3xXRO9SmLXgQrtXHjRsHBwUFYu3atcOHCBWHy5MmCm5ubkJGRYemiiUJkZKTw3XffCefPnxcSEhKEkSNHCi1bthQePHigWGfKlCmCv7+/EBsbK5w8eVLo06eP0LdvX8X7ZWVlQteuXYWIiAjh9OnTws6dOwUPDw8hOjpasU5iYqLg7OwsREVFCRcvXhS++OILwdbWVti1a5diHWv4rI4fPy4EBAQI3bt3F2bMmKFYznNsHNnZ2UKrVq2El19+WTh27JiQmJgo7N69W7h+/bpinSVLlgiurq7C9u3bhTNnzgiPP/64EBgYKBQWFirWGT58uBAUFCQcPXpUOHDggNC2bVth7Nixivdzc3MFLy8vYdy4ccL58+eFDRs2CE5OTsJXX32lWOfQoUOCra2tsHTpUuHixYvC3LlzBXt7e+HcuXPmORkmtHjxYqFZs2bCH3/8Idy8eVPYsmWL0LhxY+Gzzz5TrMPzbLidO3cKc+bMEbZt2yYAEH755ReV98V0TvUpiz6sNpz07t1bmDZtmuJ1eXm54OvrK8TExFiwVOKVmZkpABD++ecfQRAEIScnR7C3txe2bNmiWOfSpUsCAOHIkSOCIFT8h7KxsRHS09MV63z55ZeCi4uLUFxcLAiCILz33ntCly5dVI41ZswYITIyUvG6oX9WeXl5Qrt27YQ9e/YIjz76qCKc8Bwbz6xZs4T+/ftrfV8ulwve3t7CJ598oliWk5MjSKVSYcOGDYIgCMLFixcFAMKJEycU6/z555+CRCIRUlNTBUEQhFWrVglNmzZVnPvKY3fo0EHx+vnnnxdGjRqlcvywsDDhjTfeqNsvKQKjRo0SXnnlFZVlTz/9tDBu3DhBEHiejaF6OBHTOdWnLPqyysc6JSUliI+PR0REhGKZjY0NIiIicOTIEQuWTLxyc3MBAO7u7gCA+Ph4lJaWqpzDjh07omXLlopzeOTIEXTr1g1eXl6KdSIjIyGTyXDhwgXFOsr7qFynch/W8FlNmzYNo0aNUjsPPMfG89tvv6Fnz5547rnn4OnpiZCQEKxZs0bx/s2bN5Genq5yDlxdXREWFqZyrt3c3NCzZ0/FOhEREbCxscGxY8cU6zzyyCNwcHBQrBMZGYkrV67g/v37inV0fR71Wd++fREbG4urV68CAM6cOYODBw9ixIgRAHieTUFM51SfsujLKsNJVlYWysvLVb7QAcDLywvp6ekWKpV4yeVyzJw5E/369UPXrl0BAOnp6XBwcICbm5vKusrnMD09XeM5rnxP1zoymQyFhYUN/rPauHEjTp06hZiYGLX3eI6NJzExEV9++SXatWuH3bt3Y+rUqXjrrbfw/fffA6g6V7rOQXp6Ojw9PVXet7Ozg7u7u1E+j4ZwrmfPno0XXngBHTt2hL29PUJCQjBz5kyMGzcOAM+zKYjpnOpTFn3Vi1mJybKmTZuG8+fP4+DBg5YuSoOSkpKCGTNmYM+ePXB0dLR0cRo0uVyOnj174qOPPgIAhISE4Pz581i9ejUmTpxo4dI1HJs3b8ZPP/2E9evXo0uXLkhISMDMmTPh6+vL80wGscqaEw8PD9ja2qr1esjIyIC3t7eFSiVO06dPxx9//IF9+/ahRYsWiuXe3t4oKSlBTk6OyvrK59Db21vjOa58T9c6Li4ucHJyatCfVXx8PDIzM9GjRw/Y2dnBzs4O//zzDz7//HPY2dnBy8uL59hIfHx80LlzZ5VlnTp1QnJyMoCqc6XrHHh7eyMzM1Pl/bKyMmRnZxvl82gI5/rdd99V1J5069YN48ePx9tvv62oGeR5Nj4xnVN9yqIvqwwnDg4OCA0NRWxsrGKZXC5HbGwswsPDLVgy8RAEAdOnT8cvv/yCvXv3IjAwUOX90NBQ2Nvbq5zDK1euIDk5WXEOw8PDce7cOZX/FHv27IGLi4viQhEeHq6yj8p1KvfRkD+rIUOG4Ny5c0hISFD89OzZE+PGjVP8m+fYOPr166fWFf7q1ato1aoVACAwMBDe3t4q50Amk+HYsWMq5zonJwfx8fGKdfbu3Qu5XI6wsDDFOvv370dpaalinT179qBDhw5o2rSpYh1dn0d9VlBQABsb1cuKra0t5HI5AJ5nUxDTOdWnLHozqPlsA7Jx40ZBKpUK69atEy5evCi8/vrrgpubm0qvB2s2depUwdXVVYiLixPu3Lmj+CkoKFCsM2XKFKFly5bC3r17hZMnTwrh4eFCeHi44v3Kbq7Dhg0TEhIShF27dgnNmzfX2M313XffFS5duiSsXLlSYzdXa/mslHvrCALPsbEcP35csLOzExYvXixcu3ZN+OmnnwRnZ2fhxx9/VKyzZMkSwc3NTfj111+Fs2fPCk888YTG7pghISHCsWPHhIMHDwrt2rVT6Y6Zk5MjeHl5CePHjxfOnz8vbNy4UXB2dlbrjmlnZyd8+umnwqVLl4QFCxbU2y6u1U2cOFHw8/NTdCXetm2b4OHhIbz33nuKdXieDZeXlyecPn1aOH36tABAWL58uXD69Gnh1q1bgiCI65zqUxZ9WG04EQRB+OKLL4SWLVsKDg4OQu/evYWjR49aukiiAUDjz3fffadYp7CwUHjzzTeFpk2bCs7OzsJTTz0l3LlzR2U/SUlJwogRIwQnJyfBw8NDeOedd4TS0lKVdfbt2ycEBwcLDg4OQuvWrVWOUclaPqvq4YTn2Hh+//13oWvXroJUKhU6duwofP311yrvy+VyYd68eYKXl5cglUqFIUOGCFeuXFFZ5969e8LYsWOFxo0bCy4uLsKkSZOEvLw8lXXOnDkj9O/fX5BKpYKfn5+wZMkStbJs3rxZaN++veDg4CB06dJF2LFjh/F/YQuQyWTCjBkzhJYtWwqOjo5C69athTlz5qh0T+V5Nty+ffs0fh9PnDhREARxnVN9yqIPiSAoDd1HREREZGFW2eaEiIiIxIvhhIiIiESF4YSIiIhEheGEiIiIRIXhhIiIiESF4YSIiIhEheGEiIiIRIXhhIiIiESF4YSIiIhEheGEiIiIRIXhhIiIiESF4YSIiIhE5f8BzEtEe96SIwkAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot the loss\n",
    "plt.plot(stepi, lossi)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 444,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x800 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# visualise the embeddings (works only if embedding_dim == 2)\n",
    "def visualise_embeddings():\n",
    "    plt.figure(figsize=(8, 8))\n",
    "    plt.scatter(C[:, 0].data, C[:, 1].data, s=200)\n",
    "    for i in range(C.shape[0]):\n",
    "        plt.text(C[i, 0].item(), C[i, 1].item(), itos[i], ha='center', va='center', color='white')\n",
    "    plt.grid('minor')\n",
    "# visualise_embeddings()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cerie\n",
      "zomalurailezity\n",
      "kell\n",
      "imitta\n",
      "nella\n",
      "kamanda\n",
      "samiyah\n",
      "javer\n",
      "gotot\n",
      "molie\n",
      "cavo\n",
      "kented\n",
      "jen\n",
      "emva\n",
      "sade\n",
      "enkamirry\n",
      "foba\n",
      "mhikingavtahlyn\n",
      "kasdro\n",
      "anell\n"
     ]
    }
   ],
   "source": [
    "# sample from the model\n",
    "seed = 2147483647\n",
    "g = torch.Generator().manual_seed(seed)\n",
    "\n",
    "# sample 20 names\n",
    "ix = 0\n",
    "for _ in range(20):\n",
    "    out = ''\n",
    "    context = [0] * block_size\n",
    "\n",
    "    while True:\n",
    "        emb = C[torch.tensor([context])]\n",
    "        h = torch.tanh(emb.view(1, -1) @ W1 + b1)\n",
    "        logits = h @ W2 + b2\n",
    "        probs = F.softmax(logits, dim=1)\n",
    "        ix = torch.multinomial(probs, num_samples=1, generator=g).item()\n",
    "        context = context[1:] + [ix]\n",
    "\n",
    "        if ix == 0:\n",
    "            break\n",
    "        out += itos[ix]\n",
    "        \n",
    "    print(out)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": ".venv",
   "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": "3.9.19"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}