kernels/baselines/python-api.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Evaluations of PyTorch Extension API of Kernels"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Assistant function: l2 cache flusher"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "device L2 cache size: 72 MB\n"
     ]
    }
   ],
   "source": [
    "import pycuda.driver as cuda\n",
    "\n",
    "class L2Flush:\n",
    "    def __init__(self):\n",
    "        cuda.init()\n",
    "        self.device = cuda.Device(1)\n",
    "        self.context = self.device.make_context()\n",
    "        self.m_l2_size = self.device.get_attribute(cuda.device_attribute.L2_CACHE_SIZE)\n",
    "        if self.m_l2_size > 0:\n",
    "            print(\"device L2 cache size: %d MB\" % (self.m_l2_size / 1024 / 1024))\n",
    "            self.m_l2_buffer = cuda.mem_alloc(self.m_l2_size)\n",
    "\n",
    "    def __del__(self):\n",
    "        if self.m_l2_buffer:\n",
    "            self.m_l2_buffer.free()\n",
    "\n",
    "        # Detach the CUDA context when the object is deleted\n",
    "        self.context.pop()\n",
    "\n",
    "    def flush(self, stream):\n",
    "        if self.m_l2_size > 0:\n",
    "            self.context.push()\n",
    "            cuda.memset_d8_async(self.m_l2_buffer, 0, self.m_l2_size, stream)\n",
    "            self.context.pop()\n",
    "l2Flusher = L2Flush()\n",
    "l2Flusher.flush(cuda.Stream(0))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## W8A8 Kernel Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "\n",
    "# set torch device as 1. Since RTX4090 is 1 on our local machine.\n",
    "torch.cuda.set_device(1)\n",
    "\n",
    "test_M_pool = [8, 16, 32, 64, 128, 256, 512, 1024, 2048]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[2.928233438775072, 6.058944027846304, 12.147133542431966, 24.49705443214467, 49.57134767891193, 99.02931334169796, 192.23851561133415, 369.4828550922088, 433.98997666107266]\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "from torch_int._CUDA import linear_a8_w8_b8_o8\n",
    "\n",
    "@torch.no_grad()\n",
    "def test_quant_linear_a8_w8_b8_o8(M,N,K):\n",
    "    # Init inputs with corresponding shapes\n",
    "    # Ref: https://github.com/Guangxuan-Xiao/torch-int/blob/main/torch_int/kernels/linear.cu#L369\n",
    "\n",
    "    weight = torch.randint(-128, 127, (N, K), dtype=torch.int8).cuda()\n",
    "    bias = torch.randint(torch.iinfo(torch.int8).min, torch.iinfo(\n",
    "        torch.int8).max, (N,), dtype=torch.int8).cuda()\n",
    "    x = torch.randint(-128, 127, (M, K), dtype=torch.int8).cuda()\n",
    "\n",
    "    torch.cuda.synchronize()\n",
    "\n",
    "    elapsed_time_ms = 0\n",
    "    iterations = 30\n",
    "\n",
    "    for _ in range(iterations):\n",
    "        start_event = torch.cuda.Event(enable_timing=True)\n",
    "        end_event = torch.cuda.Event(enable_timing=True)\n",
    "        torch.cuda.synchronize()\n",
    "        start_event.record()\n",
    "        y = linear_a8_w8_b8_o8(x, weight, bias, 1, 1)\n",
    "        end_event.record()\n",
    "        torch.cuda.synchronize()\n",
    "        elapsed_time_ms += start_event.elapsed_time(end_event)\n",
    "\n",
    "        l2Flusher.flush(cuda.Stream(0))\n",
    "\n",
    "    total_ops = M * N * K * 2 * iterations\n",
    "    gflops = total_ops / elapsed_time_ms / 10**9\n",
    "    return gflops, elapsed_time_ms\n",
    "\n",
    "w8a8_flops = []\n",
    "for m in test_M_pool:\n",
    "    gflops, elapsed_time_ms = test_quant_linear_a8_w8_b8_o8(m, 4096, 4096)\n",
    "    w8a8_flops.append(gflops)\n",
    "\n",
    "print(w8a8_flops)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## W4A16 Kernel Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[5.093059212625437, 6.536323152461491, 7.543259966351492, 7.947865778756361, 8.344728762237368, 8.54029590437908, 8.499885651479563, 8.183972536588632, 8.167991369592459]\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "from awq_inference_engine import gemv_forward_cuda\n",
    "\n",
    "@torch.no_grad()\n",
    "def test_quant_linear_a16_w4_gemv(M, N, K):\n",
    "    groupSize = 128\n",
    "\n",
    "    # Init inputs with corresponding shapes\n",
    "    # Ref: https://github.com/mit-han-lab/llm-awq/blob/main/awq/kernels/csrc/quantization/gemv_cuda.cu#L189\n",
    "\n",
    "    x = torch.rand(M, K, dtype=torch.float16, device='cuda')\n",
    "    weight = torch.randint(0, 128, (N, K // 8), dtype=torch.int32, device='cuda')\n",
    "    scailing = torch.rand(N, K // groupSize, dtype=torch.float16, device='cuda')\n",
    "    zeros = torch.randint(0,128,(N, K // groupSize // 8), dtype=torch.int32, device='cuda')\n",
    "    \n",
    "    elapsed_time_ms = 0\n",
    "    iterations = 30\n",
    "\n",
    "    for _ in range(iterations):\n",
    "        start_event = torch.cuda.Event(enable_timing=True)\n",
    "        end_event = torch.cuda.Event(enable_timing=True)\n",
    "        torch.cuda.synchronize()\n",
    "        start_event.record()\n",
    "        y = gemv_forward_cuda(x, weight, scailing, zeros, groupSize)\n",
    "        end_event.record()\n",
    "        torch.cuda.synchronize()\n",
    "        elapsed_time_ms += start_event.elapsed_time(end_event)\n",
    "\n",
    "        l2Flusher.flush(cuda.Stream(0))\n",
    "\n",
    "    total_ops = M * N * K * 2 * iterations\n",
    "    gflops = total_ops / elapsed_time_ms / 10**9\n",
    "    return gflops, elapsed_time_ms\n",
    "\n",
    "a16w4_flops = []\n",
    "for m in test_M_pool:\n",
    "    gflops, elapsed_time_ms = test_quant_linear_a16_w4_gemv(m, 4096, 4096)\n",
    "    a16w4_flops.append(gflops)\n",
    "\n",
    "print(a16w4_flops)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## FP16 Baseline Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[3.5913723582696826, 7.396274389981544, 14.601368378822112, 29.153260582063147, 55.5314758714023, 95.79681813261212, 136.6930123791791, 152.77694542825068, 161.1156499269412]\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "from torch.nn import functional as F\n",
    "\n",
    "@torch.no_grad()\n",
    "def test_quant_linear_a16_w16(M, N ,K) -> float:\n",
    "    weight = torch.rand(K, N, dtype=torch.float16).cuda()\n",
    "    x = torch.rand(M, K, dtype=torch.float16).cuda()\n",
    "\n",
    "    elapsed_time_ms = 0\n",
    "    iterations = 30\n",
    "    \n",
    "    for _ in range(iterations):\n",
    "        start_event = torch.cuda.Event(enable_timing=True)\n",
    "        end_event = torch.cuda.Event(enable_timing=True)\n",
    "        torch.cuda.synchronize()\n",
    "        start_event.record()\n",
    "        y = F.linear(x, weight)\n",
    "        end_event.record()\n",
    "        torch.cuda.synchronize()\n",
    "        elapsed_time_ms += start_event.elapsed_time(end_event)\n",
    "\n",
    "        l2Flusher.flush(cuda.Stream(0))\n",
    "    total_ops = M * N * K * 2 * iterations\n",
    "    gflops = total_ops / elapsed_time_ms / 10**9\n",
    "    return gflops, elapsed_time_ms\n",
    "\n",
    "a16w16_flops = []\n",
    "for m in test_M_pool:\n",
    "    gflops, elapsed_time_ms = test_quant_linear_a16_w16(m, 4096, 4096)\n",
    "    a16w16_flops.append(gflops)\n",
    "\n",
    "print(a16w16_flops)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Atom (Our W4A4) GEMM Kernel Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[6.644442034767834, 13.6618572262898, 27.373496811519512, 54.63260018922175, 108.40621174030697, 211.1060843567243, 375.1194162714706, 464.8434121405477, 539.3925668781076]\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "from punica.ops import dense_layer_gemm_i4_fp16\n",
    "from punica.ops import scale_size\n",
    "\n",
    "@torch.no_grad()\n",
    "def test_quant_our_w4a4_gemm(M):\n",
    "    bs = M\n",
    "    hidden_dim = 4096\n",
    "    group_size = 128\n",
    "\n",
    "    a = torch.randint(\n",
    "        16,\n",
    "        128, (bs, (hidden_dim - group_size) // 2),\n",
    "        dtype=torch.uint8).cuda()\n",
    "    b = torch.randint(\n",
    "        16,\n",
    "        128, (hidden_dim, (hidden_dim - group_size) // 2),\n",
    "        dtype=torch.uint8).cuda()\n",
    "    a_scale = torch.randn((hidden_dim // group_size - 1, scale_size(bs)),\n",
    "                            dtype=torch.float16).cuda()\n",
    "    b_scale = torch.randn((hidden_dim // group_size - 1, scale_size(bs)),\n",
    "                            dtype=torch.float16).cuda()\n",
    "    a_keeper = torch.randint(\n",
    "        0, 255, (bs, group_size), dtype=torch.uint8).cuda()\n",
    "    b_keeper = torch.randint(\n",
    "        0, 255, (bs, group_size), dtype=torch.uint8).cuda()\n",
    "    a_keeper_scale = torch.randn((scale_size(bs),),\n",
    "                                dtype=torch.float16).cuda()\n",
    "    b_keeper_scale = torch.randn((scale_size(bs),),\n",
    "                                dtype=torch.float16).cuda()\n",
    "\n",
    "    iterations = 30\n",
    "    elapsed_time_ms = 0\n",
    "    \n",
    "    for _ in range(iterations):\n",
    "        torch.cuda.synchronize()\n",
    "        start_event = torch.cuda.Event(enable_timing=True)\n",
    "        end_event = torch.cuda.Event(enable_timing=True)\n",
    "        start_event.record()\n",
    "        y = dense_layer_gemm_i4_fp16(a, b, a_scale, b_scale, a_keeper,\n",
    "                                    b_keeper, a_keeper_scale, b_keeper_scale)\n",
    "        end_event.record()\n",
    "        torch.cuda.synchronize()\n",
    "        elapsed_time_ms += start_event.elapsed_time(end_event)\n",
    "\n",
    "        l2Flusher.flush(cuda.Stream(0))\n",
    "\n",
    "    total_ops = M * hidden_dim * hidden_dim * 2 * iterations\n",
    "    gflops = total_ops / elapsed_time_ms / 10**9\n",
    "    return gflops, elapsed_time_ms\n",
    "\n",
    "w4a4_flops = []\n",
    "for m in test_M_pool:\n",
    "    gflops, elapsed_time_ms = test_quant_our_w4a4_gemm(m)\n",
    "    w4a4_flops.append(gflops)\n",
    "\n",
    "print(w4a4_flops)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjwAAAI1CAYAAADSACBSAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy81sbWrAAAACXBIWXMAAA9hAAAPYQGoP6dpAADayUlEQVR4nOzdeVxU1fvA8c8M+46AoLgirrlroibuW+a+pqlpZpm2W33T+rVofbOszL6tmhpprqnlkma5K2rgLmYuCG6obLLDADP39wfNDWRngGF53q8XOcw9955nCJiHc895jkZRFAUhhBBCiCpMa+4AhBBCCCHKmiQ8QgghhKjyJOERQgghRJUnCY8QQgghqjxJeIQQQghR5UnCI4QQQogqTxIeIYQQQlR5kvAIIYQQosqThEcIIYQQVZ4kPEIIIYSo8iptwpOQkMC6det45ZVX6NmzJ40bN8bFxQVra2s8PT3p1asXCxcuJCYmxqR+AgIC0Gg0RfoICAgonRcnhBBCiFJlae4ASiooKIgJEybkeSwqKooDBw5w4MABPv74Y3788UcGDhxYzhEKIYQQoqKotAkPQL169ejduzcdO3akXr161K5dG4PBwM2bN9m4cSObN28mOjqaYcOGERQURNu2bU3qb9euXXh7e+d7vG7duiZdXwghhBBlQ1NZd0vX6/VYWFgU2OaXX35h5MiRAIwcOZLNmzcXu5+AgACeeOIJAMLCwmjYsGGxryGEEEII86q0c3gKS3YARowYQbNmzQA4dOhQWYckhBBCiAqq0iY8ReXk5ARAWlqamSMRQgghhLlU6YTn4sWLnD59GoDmzZubNxghhBBCmE2VS3hSUlK4fPkyixYtomfPnmRmZgLw0ksvmXztJ554Am9vb6ytrfHw8KBLly783//9H7du3TL52kIIIYQoO5V20nJ22ScW52XOnDl88MEHaDSaUr82gK2tLYsXL2bGjBlFuqZOp0On06mfGwwGYmNjcXd3L1GMQgghRHWlKAqJiYl4e3uj1eY/jlOpl6UXpl27dixdupROnTqZdJ1GjRoxatQounbtSr169QC4evUqmzZtYuPGjaSlpfHMM8+g0Wh4+umnC73eggULmDdvnkkxCSGEEOJfN27cKLA8TJUY4YmLi+PmzZsApKamEhoayoYNG/j555/x9fVl8eLFDBkypETXjo+Px9nZOd+Rl+3btzNq1CgyMjKwt7cnNDSUWrVqFXjN+0d44uPjqV+/Pjdu3MDZ2blEcQohhBDVUUJCAvXq1SMuLg4XF5d821WJhCc/q1atYsqUKWg0GpYvX87UqVPLpJ/333+ft956S3385ptvFuv8hIQEXFxc1ORKCCGEEEVT1PfQKjdpObvJkyczduxYDAYDzz33HLGxsWXSz9NPP62OAB04cKBM+hBCCCFEyVXphAdg+PDhACQnJ/Pbb7+VSR+enp64u7sDyIotIYQQogKq8glPzZo11cfXrl0rs35kdZUQQghRcVX5hCf7iIujo2OZ9BEVFUV0dDRAgZuLCiGEEMI8qnzC89NPP6mPW7duXSZ9LF26FOPc7549e5ZJH0IIIYQouUqb8AQEBBS6P9Znn33Gjh07APDx8aF79+45ju/fvx+NRoNGo8lzBVd4eDinTp0qsI/t27czf/58AOzs7AotUiiEEEKI8ldpCw++++67vPLKK4wePRp/f398fX1xdHQkMTGRc+fOsXr1agIDAwGwtrZm6dKlRdphPbvw8HB69+5N165dGTp0KG3btsXT0xPIKjy4ceNGNm7cqI7ufPLJJ9SpU6d0X6gQQgghTFZpEx6A2NhYvvvuO7777rt829StW5cVK1bQr1+/Evdz9OhRjh49mu9xe3t7PvvssyJVWRZCCCFE+au0Cc+uXbv49ddfCQwM5MqVK9y9e5eYmBjs7Ozw9PSkXbt2DBkyhHHjxmFvb1+iPjp27MiPP/7I0aNHOX78OLdv3yY6OprMzExq1KhBy5Yt6du3L9OnT1dHfoQQQghR8VTpSsuVhVRaFkIIIUqmqO+hlXaER+SmKAoZGRkYDAZzhyKEqEa0Wi1WVlZSj0xUaJLwVAF6vZ7o6GgSExPJyMgwdzhCiGrIysoKJycnPDw8ir1ARIjyIAlPJafX67lx4wY6nQ4XFxccHR2xsLCQv7SEEOVCURT0ej1JSUnExcWRmppKvXr1JOkRFY4kPJVcdHQ0Op2O+vXrY2dnZ+5whBDVlKOjIy4uLly/fp3o6Gi8vLzMHZKoYAx6A6EhN0iITcbZzQHfVvXQWpRfOUBJeCoxRVFITEzExcVFkh0hhNnZ2dnh7OxMYmIinp6eMtIsVGcOX2Tz17uJi05Un3P1cGLUrH609W9WLjFU2krLAjIyMsjIyCizPcKEEKK4nJyc1N9NQkBWsrNi/s85kh2AuOhEVsz/mTOHL5ZLHJLwVGLG1Vhyr1wIUVEYfx/JalEBWbexNn+9u8A2m7/ZjUFf9t8vkvBUATJsLISoKOT3kcguNORGrpGd+8VFJRIacqPMY5GERwghhBBlIiE2uVTbmUISHiGEEEKUutTkNI7tOlukts5uDmUcjazSEkIIIUQpu3L2Oj8u3M69yIRC27rWdMK3Vb0yj0lGeIQQQghRKjLTM9m6bB9fvraGe5EJuNd2ZcgTPQs8Z9TMfuVSj0cSHiGEKAcNGzZEo9EwderUXMfCw8PRaDRoNBoCAgLKPTYhSsPt8CgWvbCSPRv+RFGgy8Nt+M83T9B/QlemvT0SVw+nHO1dazox7e2R5VaHR25pCVEE+/fvp3fv3kVu//333+d4YyvofDs7O2rWrEn79u0ZN24c48aNw9Iy7x/NK1euEBwcTFBQEEFBQZw6dYrU1NQ8+yyKjIwMVq9ezU8//cS5c+e4e/cuTk5O1K5dm86dOzNw4EDGjh1brGsaBQQE8MQTT+R5TKPR4OjoSP369enWrRtPP/00HTt2LFE/QgjzMhgUDm45zrZl+8nM0OPgYsf4lwbRpltTtU1b/2a07tpEKi0LUZ2lpqZy/fp1rl+/zpYtW1i8eDFbt26lVq1aOdodOHCAXr16lVq/Z8+eZeLEiYSEhOR4PiYmhpiYGEJCQti4cWOJE56CGKuEnz9/nvPnz/Pdd98xZ84cPvjgg1LvSwhRduKiE1n98a9cOhUOQItOjXjslUdwdstdEFdroaVJ2wblHOG/JOERophmzpzJrFmzCmxTt27dIp+flJTE8ePH+fTTTwkPDyc4OJjhw4dz7NixHDVNFEVRH2u1Wlq0aIGDgwNBQUHFfg1nz56ld+/exMbGYmtry/Tp0xkwYAB16tQhPT2dy5cv89tvv3Ho0KFiXzsv77//PsOHD1c/NxgMREVFsX//fhYvXkxSUhILFiygUaNGTJ8+vVT6rEwaNmyY4/+vEJXBqYN/s+Hz30hJTMPKxpLhT/XBf2j7CluLSRIeIYrJ09OTVq1aler5Xbp0YeLEifj5+XHlyhWCgoLYvn07Q4cOVdvUqVOHjz/+mE6dOtGxY0ccHR0JCAgodsKTlpbG2LFjiY2NpX79+uzevZsmTZrkimfy5Mmkp6eX+HVmV6dOnTy/Zn379qVXr170798fRVH4+OOPq2XCI0Rlkpqcxqav/iB493kA6jWpxeQ5Q/Gq527myAomk5aFqCBq1KjB3Llz1c9/++23HMebNGnCq6++Ss+ePU3aP+2TTz7h0qVLaLVaNmzYkCvZyc7a2rrE/RRV37596dChAwCXLl0iIaHwZaxCCPMIPXeDhc+sIHj3eTRaDQMee4iXP59c4ZMdkIRHiArFz89PfXzt2rVSv75er+fbb78FoF+/fnTu3LnU+ygJHx8f9bFOp8uzTUhICO+//z4DBw6kbt262NjY4OjoSJMmTZgyZQrHjh0rtJ+IiAjmzJlDhw4dcHFxwcrKCi8vL1q3bs2ECRMICAgoMOGKj49nwYIFdOvWjZo1a2JtbU3t2rUZOnQoGzduLPFtqcJWab377rvqccgapfv444/p0KEDTk5OODk54efnx5dffklmZmah/aWlpfHll1/St29fatWqhbW1NZ6envTr14/ly5cX6RqiesnM0LNt+X6+eHU1sXcTcK/lwoufTmTw1B5YWFaO/RzllpYQFYiVlZX6WK/Xl/r1jxw5wq1btwAYNmyY+nxaWhq3bt3C1taWWrVqlfuGtMbkzrhi7X75rXJLT0/nypUrXLlyhZUrVzJnzhwWLFiQZx+HDh1iyJAhuRKayMhIIiMjCQkJYd26dXh4eDBkyJBc5+/Zs4dHH32UmJiYHM/fuXOH7du3s337dh555BHWr19v0ghcYe7evcvDDz/M6dOnczwfHBxMcHAwv//+O7/88gtabd5/z545c4bhw4fnSqijoqLYs2cPe/bsYcmSJWzbtg0vL6+yehmiErlzLZpVH23j5pW7AHQe2JpRM/tha29j5siKRxIeISqQc+fOqY+9vb1L/frZR0Fat27N5cuXmTNnDtu2bSMjIwMAFxcXhg0bxjvvvIOvr2+px3C//fv3c+LECSBnEpZdZmYmDg4ODB48mD59+tC8eXOcnZ2JjIzk/Pnz/O9//+PatWt8+OGHNG3aNNdyeJ1Ox/jx40lISMDJyYmZM2fSu3dvPD09SU9PJywsjCNHjvDzzz/n2X9gYCCDBg0iIyMDLy8vnn/+edq2bYu3tzcRERGsX7+eH3/8kR07djBlyhQ2bdpUul+kbEaNGsVff/3FCy+8wNChQ3Fzc+PixYu89957XLhwgW3btvHdd98xY8aMXOdeuXKFnj17Eh8fj7OzM88++yx+fn7Uq1ePmJgYtm7dypIlS9SJ84cOHcqRhIvqxWBQOLz1BFuX7ScjPRMHZzsefenhcqubU+oUYXbx8fEKoMTHxxfrvNTUVOWvv/5SUlNTyygyYbRv3z4FUABl5syZyrlz5/L9uHv3boHnv/POO3n2kZGRoXTp0kVtt3LlykLj+v7779X233//faHtp06dqrZfsmSJ4uDgoH5+/4eTk5Pyxx9/FHrNosT2/vvv5/ganTlzRtmzZ4/y9ttvK87Ozgqg1KlTR7l8+XKe14qKilLu3buXb186nU7p37+/AigNGjRQMjMzcxzfs2ePGsu2bdvyvU5GRkaun8P09HSlYcOGCqA8/PDDSnJycp7nLl26VO3j999/z3W8QYMGCqBMmTIl17GwsLAC/z++88476nErKytl3759udrExMQoXl5eCqC0adMmzxgfeughBVDat2+vREVF5dlm586dilarVQBl6dKlebYpiPxeqhriohOUr+esU17ov0B5of8C5Zu565W46ERzh5Wnor6HSsJTAZRXwpOaoc/3Q5epL5e2aZn5t00zoW1Zy56wFPaRV0JTUMKTlJSk7N+/X+nVq5fapkGDBkpaWlqhcRU34Rk2bJja3tbWVgGUl19+Wbl8+bKi0+mU0NBQ5bXXXlM0Go0CKK6ursq1a9eK+FXKP7aCPmxsbJTXX39duXXrVon6MTp9+rR6zePHj+c4tnr1avVYcX/OVq5cqX69IiMjC2zr5+enAMpjjz2W61hpJTyzZ8/Ot/85c+YogKLRaJS4uLgcxw4ePKhe4+zZswW+jnHjximA8tBDDxXYLi+S8FR+pw5cUOaM+kx5of8C5ZXBHysHtxxXDAaDucPKV1HfQ+WWVjUyfc+lfI+19XDgtY7/bt727P7L6PR5T8BsXsOO//P7t3jUywdDSczIe76Jj7Mt73VtqH7++uGrRKflPSGyjoM1H/k3Uj9/+2g4t5LzXhbtYWvJ4p6N8309Fdm8efOYN29evsc9PT355ZdfsLEp/fvjycnJ6uO0tDTmz5/PW2+9pT7XqFEjFi5cSI0aNXjjjTeIi4tjwYIFfPPNN6Uei5FOp+OHH37A0dGRN954I9+5J/efc/fuXZKSkjAYDEDOOkVnzpzJUbm5du3a6uPvv/+eF198scjxbd26FYCePXvmOb8oux49ehAUFMTRo0eLfP3imjhxYr7HjK9ZURTCwsJo166desz4Opo1a0br1q0L7KNHjx5s2LCB4OBgMjMz8638LaqWtGQdm77+g6A/soqR1m3sxeQ5Q6lV38PMkZUOWaUlRDG98847KFmjo3l+vPvuuyW6ro+PD6+99hrnzp3L8UZVmmxtbdXHHh4evP7663m2e+2119RKzxs2bDC5KN7333+f6+uUmJhIUFAQ06dP586dO7z11ls8+uijagJzv+TkZBYsWEDbtm1xcHCgQYMGtGzZktatW9O6dWvat2+vto2Ojs5xrr+/P40aZSXTL730En5+fixYsIDAwMBCaw0dP34cgF27dqkrpfL7+OSTT4CsicxlpXnz5vkec3NzUx8nJibmOGZ8HRcvXiz0dTz33HNA1tYjsbGxZfAqREUTeu4GHz2zgqA/QtBoNfSf0JWXP3+8yiQ7IJOWq5VlfZvme0x7X2HMr3rlX5vl/raf9ch/Yuv9bT/yb0R+7533F+ec37VhkdtWJtkrLWs0GmxtbfHw8MDFxaXM+3Zy+nfzvl69euVbZ8fS0pK+ffuyevVqYmNjuXr1aqlPYHZ0dKRTp0506tQJT09PPvjgAzZu3MiKFStyFR8MDw+nT58+hIWFFenaxv3FjKysrNi2bRtjxozhwoUL6oomyFoZ1qNHDx5//HEeffTRXCvUIiMji/3a7u+/NNnb2+d7LPvo2P2r/EryOgBSUlJKdJ6oHDIz9OxcdZg964+iKODm5cKk14fg26pe4SdXMpLwVCO2lkUf0CurtjbF2CiuOG0rE1MrNZuiXr16eT4urG1UVFSZrth65ZVX+PDDDzEYDHkmPJMnTyYsLAyNRsMTTzzB+PHjadGihVoLR6PRYDAY1GQlrxGpBx54gHPnzrFt2za2bdvGwYMHuXLlCqmpqezatYtdu3axaNEiduzYgaenp3qeMXEYNGgQCxcuLLOvQVkzvo62bdvy448/Fvm8OnXqlFVIwszuXI9m1Yf/Ljf3G9Ca0TP7YetQuZabF5UkPEJUIy1btlQfF1bnJ/vxsp7D4ebmRs2aNbl7926OpfkAf//9N4cPHwbgjTfe4P3338/zGkW59WJhYcGIESMYMWIEALdv3+a3337jq6++4sSJE5w4cYIZM2bkWJ7u7u5OREQE6enpZktUS4O7e1Yl3KSkpEr9OoTpFEXh0NaTbP1uHxnpmdg72fLoSw/Trnv+t0urgqr5J7QQIk89evRQH1+9erXAtqGhoerj8vgr31jd9/4qv+fPn1cfP/roo/meb5yjUhy1a9fmiSee4OjRo+r2Ftu3b89xS8o4N+j48eOltreYORhfx9WrV8t0jpGo2OJjkljy5k9s+uoPMtIzaf6gD3OWPlnlkx2QhEeIasXHx0d949u3bx/x8fF5tktMTGT37t0A+Pr65ljlVBbCw8PVCsb332rLngBlX2V2P+OWGSVhZWVFz5491f7i4uLUY8ZiiPHx8Xz//fcl7sPcjK9DURQ+//xzM0cjzOHM4Yt8NGM5F45fxcraktHP9ueZ/47Dxd2p8JOrAEl4hKhm5syZA2QlDy+99FKebWbPnq1uwfDMM8+UaTwGg0GNCWDw4ME5jmff3DSvfaYAvvnmG7Zs2ZJvH4cOHeLKlSv5Hk9PT+fAgQNA1mTq7MvPp0yZoiZhr776KgcPHsz/xQCHDx9Wr1WRDBgwQN2r7eOPP2bDhg0FtjfOdxKVX1qKjjWf/sqK+T+TnJBKHV9PXv1qKj2Gd1T3Z6sOZA6PEJXIxo0bSUpKUj83zm25/zFArVq1ePjhh3NdY9y4cfzwww/s2LGDgIAAbt++zcyZM6lfvz43btxgyZIl7NixA8i6DWJcomyKW7duERISkuO55ORk/vrrL7777ju1bo27u3uupfLt27enVatWhISEsGTJEu7du8fkyZOpXbs2N2/e5Mcff2Tjxo1069aNwMDAPPvfs2cP7733Ht27d2fw4MG0adOGmjVrkpqayqVLl/j22285efIkAE8++WSOOUs2NjZs2LCBXr16kZSURJ8+fRg/fjwjRozAx8cHg8HA7du3OXHiBD///DPnzp3jiy++UEeMKpI1a9bg5+dHbGwsjz76KD/++COPPvooTZo0wcLCgsjISE6dOsW2bds4duwYr7zyCkOHDjV32MIEV8/f5MePthFzJx6NBvqO68Kgx7tjaVU5NvwsTZLwCFGJvPrqq/nuor58+XKWL1+uft6zZ888Ex6A9evXM3r0aH7//Xd1hdL9OnXqxNatW3PU7imp//u//+P//u//Cmzj4+PDxo0b1fo/RhqNhlWrVtGnTx/u3bvHhg0bco1OtG7dmp9++qnA/ccMBgMHDhwocPRl+PDheW4+2qVLF/bv38+4ceO4ceMGq1evZvXq1flex9nZOd9j5uTr68vRo0cZPXo0ISEh6oq1/FTU1yEKp8/U89uqQP5YfxTFoODm5cyk/wzFt3XVW25eVJLwCFENOTo68ttvv7F+/Xp++OEHTp8+TUxMDK6urrRr144JEybw+OOPl+mu6cb6Q23btmXYsGFMmjQp3xoz7dq14/Tp0yxYsICdO3cSERGBk5MTjRs3Zty4cTz77LMFJmavvvoqbdq0Yffu3Zw6dYqIiAi1Lk2tWrXw8/Pj8ccfz3U7LbsuXbpw+fJlAgIC2LZtG6dOnSI6OhqtVkvNmjVp0aIFPXv2ZPTo0TRrVnE3V2zatCmnT59mw4YNbNq0ieDgYKKiotDr9bi7u9OsWTP8/f0ZOXKkOpFbVC53r8ew6qNt3LicNTm9U79WjH62H3YOpv/xUplpFFNLqAqTJSQk4OLiou5gXFRpaWmEhYXh4+NTKn+FCyGEqeT3kvkoisLhbafY8t1eMnRZy83Hvfgw7XtU7RVYRX0PlREeIYQQopJLiE1izac7uBCcVW6iWYeGPPbqYFw9qscKrKKQhEcIIYSoxM4GXmLd4p0kx6diaWXB0Om96DH8QbT37+1TzUnCI4QQQlRCaSk6fv52D8d+OwtAHV9PJr8+lNoNaxZyZvUkCY8QQghRyYT9dYtVH20j5nYcGg30GdeFRyb7Y2ktb+v5ka+MEEIIUUnoM/XsWh3I72uzlpvX8HRm0n+G0LhNfXOHVuFJwiOEEEJUMAa9gdCQGyTEJuPs5oBvq3pE345j1UfbuH7xNgCd+rVk9LP9q/1y86KShEcIIYSoQM4cvsjmr3cTF52oPmfnaEN6Wgb6TAN2jjaMe+FhOvRqYcYoK59Ku5dWQkIC69at45VXXqFnz540btwYFxcXrK2t8fT0pFevXixcuFDdkLA07Ny5k5EjR1K3bl1sbGyoW7cuI0eOZOfOnaXWhxBCiOrrzOGLrJj/c45kByA1SYc+00Btn5rMWfKkJDslUGlHeIKCgpgwYUKex6KiotQS8h9//DE//vgjAwcOLHFfBoOBp59+OkfZfsjaH+jWrVv88ssvTJ8+nSVLlqDVVtocUgghhBkZ9AY2f727wDapSWk4uzmWU0RVS6VNeADq1atH79696dixI/Xq1aN27doYDAZu3rzJxo0b2bx5M9HR0QwbNoygoCDatm1bon7efPNNNdlp3749//nPf/D19SU0NJSFCxdy6tQpli1bRs2aNfnggw9K8yUKIYSoJkJDbuQa2blfXFQioSE3aNK2QTlFVXVU2q0l9Hp9ofv8/PLLL4wcORKAkSNHsnnz5mL3c+nSJVq2bElmZiYPPvggBw8exM7OTj2ekpJCz549OX78OJaWlly4cIHGjRsXqw/ZWkIIUVXI76WSO7HvL1Yu2Fpou8fnDqNj7wfKIaLKoajvoZX2/ktRNjUcMWKEuonfoUOHStTP4sWLyczMBOCLL77IkewA2Nvb88UXXwCQmZnJZ599VqJ+hBBCVG+J95KL1M7ZzaGMI6maKm3CU1ROTln7iKSlpRX7XEVR2LJlCwDNmzenS5cuebbr0qWLmlht2bKFSjpoJoQQwgwURWHvxj/5ZcmeQtu61nTCt1W9coiq6qnSCc/Fixc5ffo0kJWwFFdYWBgREREA9OzZs8C2xuO3bt0iPDy82H0JIYSoftJSdAS8/wtblu5DUcC3dcHJzKiZ/dBaVOm37jJTqSct5yUlJYVbt26xbds2Fi5cqN6Oeumll4p9rb/++kt9XFjClP34hQsX8PHxKXZ/Qgghqo8716NZMe9n7t6IwcJSy8hn+uI/tANnAy/lqsPjWtOJUTP70da/mRkjrtyqRMITEBDAE088ke/xOXPm8NhjjxX7ujdv3lQf161bt8C29er9m5XfuHGjwLY6nQ6dTqd+npCQUOzYhBBCVF6nDlxgzac7SE/LwMXDiWlvjaBhizoAtPVvRuuuTXJVWpaRHdNUiYQnP+3atWPp0qV06tSpROcnJv6bXTs6Flz3wMHh30lkSUlJBbZdsGAB8+bNK1FMQgghKi99pp6ty/azf3MwAE3a1mfKG8NxqpFzIrLWQitLz0tZlUgXR4wYwblz5zh37hxBQUGsXbuWkSNHcvr0aSZMmMD27dtLdN3sE52tra0LbGtjY6M+Tk1NLbDt3LlziY+PVz8KGxESQghR+cXHJPHVf9aqyU7fR7sw88PxuZIdUTaqxAiPq6srrq6u6uedOnVi/PjxrFq1iilTpjB8+HCWL1/O1KlTi3Xd7DUk0tPTC2yb/RbV/UvX72djY5MjQRJCCFG1hZ67QcB/fyEhNhlbexsmvjaYNt2amjusaqVKjPDkZ/LkyYwdOxaDwcBzzz1HbGxssc43LmmHwm9TJSf/Wz+hsNtfQgghqgdFUdi3KYgvX1tDQmwytRvW5JUvp0iyYwZVOuEBGD58OJCVkPz222/FOjf7ROXsE5jzkv22VPYJzKLy0+v1ODs7o9Fo6NChQ4FtFUXB3d0djUaDRqNhxYoVBbb/4Ycf1LbffPNNobHcvn2bGjVqqOf06tWrOC+F8PBwtFqtev6aNWuKfG5cXBx//PEH//3vfxk+fDje3t4ljsMoKCiIWbNm0aJFC5ydnXF0dMTX15fBgwezaNEioqKiSnRdISqCtBQdP3ywhV+W7MVgUOjY+wFe/nwynnXdzB1atVQlbmkVpGbNmurja9euFevcBx74t3T333//XWDb7MdbtJBdbKsSCwsLHnroIXbt2sWZM2dISEjIt3z5+fPnc4wkHjp0iGnTpuV77ewVwHv06FFoLM8//zxxcXFFD/4+K1euzFEYc+XKlUVewdi+fftSqzGl0+l47rnnWL58ea5CnVevXuXq1avs2LGDRo0aMWLEiFLpU4jydPd6DMvnb+bu9Ri0FllLzrsP64BGozF3aNVWlR/huXXrlvq4uLeafHx88Pb2BuDAgQMFtj148CAAderUoWHDhsULUlR4xmTEYDBw5MiRfNsZExjj1ieFbWliPO7h4ZEjwc7Ltm3b2LRpE56enkWO+36rVq0C/v1Z2L17N7dv3y7SudkTEy8vL4YMGVKiGNLT0xk5ciTLli1DURT69OnD8uXLCQwM5MSJE2zbto25c+fSpEmTEl1fCHM7ffBvPn3+B+5ej8HF3ZEXPnmMHsM7SrJjZlU+4fnpp5/Ux61bty7WuRqNRr0l9vfff3Ps2LE82x07dkwd4Rk+fLh8U1dB2UdfjMltXozHxo4dC0BoaKharft+kZGRXLp0CQB/f/8Cv2+SkpJ49tlnAfjkk0+KF/w/jhw5wpUrVwD47LPPsLCwQK/Xs3r16iKd/9xzz7Fx40auX7/OnTt32LZtW4nieP/999m5c6d6G2/Pnj1MmzaNhx56iA4dOjBkyBA++OADLl26xODBg0vUhxDmoNcb+GXpXr5//xd0qek0blufV79+Ap+WBddxE+VEqaS+//57JTU1tcA2ixYtUgAFUHx8fJTMzMwcx/ft26cenzJlSp7XuHjxomJhYaEAyoMPPqikpKTkOJ6SkqI8+OCDCqBYWloqly5dKvZriY+PVwAlPj6+WOelpqYqf/31V6FfB2G6tLQ0xdbWVgEUf3//fNvVqVNHAZSNGzcqvr6+CqCsW7cuz7YbN25Uv/8WLVpUYP/PP/+8Aii9e/dWFEVRz+vZs2eRX8OMGTMUQPHw8FDS09OVAQMGKIDSpk2bIl/jfsWNIzQ0VLGyslIA5dlnny1xv6Liqq6/l+JjEpXPZ/+ovNB/gfJC/wXKL0v3KpmZenOHVS0U9T200o7wvPvuu9SpU4enn36alStXEhgYyJkzZzh8+DDffPMN/v7+zJ49G8iqobN06dIi7bB+v6ZNm/Laa68BcPz4cbp168b69es5fvw469evp1u3bhw/fhyA1157TYbhqygbGxv8/PwACA4OzlGGwCgsLEy9herv74+/vz+Q/22tos7fCQoK4quvvsLa2rpIE5vzotPp2LBhAwDjxo3DysqKyZMnA3D27Fl1z7mytnTpUjIyMtBqtcydO7dc+hSirIWG3ODjWQGEnruBjb01094eyfCnemMhlZErlEo9aTk2NpbvvvuO7777Lt82devWZcWKFfTr16/E/fz3v/8lMjKSFStWcOrUKcaPH5+rzZNPPsn7779f4j6qCoPBwIXoK8SlJuBq50wLj8ZotVXjh75Hjx4cPHgQnU7Hn3/+mStJMd7OatKkCV5eXvj7+/PDDz/kewvMmPA4OzvTrl27PNtkZmby1FNPYTAYeP3112nWrGT76Gzbto179+4BMGnSJABGjhyJg4MDycnJrFy5Mt8YSpPxFnOHDh2oUyerjL6iKNy5c4fU1FS8vLxyVC0XoiJTFIUDvxxny9J9GPQGajXwYNrbI/Gq527u0EQeKm3Cs2vXLn799VcCAwO5cuUKd+/eJSYmBjs7Ozw9PWnXrh1Dhgxh3Lhx2Nvbm9SXVqtl+fLljB49mqVLlxIcHEx0dDQeHh506tSJGTNmMGjQoFJ6ZZVX0M3TBJzcSGxqnPqcm50rUzuMwa9uO7PFVVqyJziHDh3KlfAYExjjyI7x35CQEO7du0eNGjXUtomJiZw5cwaAhx56KN/Rx08++YSzZ8/SuHFj3njjjRLHvnLlSgB8fX3p2rUrkLUdyogRI1i9ejVr1qzh448/LtEoaFFFRUVx9epVIGs+XXp6OgsWLODbb7/lzp07QNbPWufOnXn11VcZNWpUmcUihKl0qemsXbSTUwcuANCh9wOMf+lhbOwKrsovzKfSJjzNmjWjWbNm6m2rkujVq1euJbEFeeSRR3jkkUdK3F9VFnTzNIsCl+V6PjY1jkWBy5jdbXqlT3q6du2KpaUlmZmZed6mMj7XvXt3AJo3b46HhwfR0dEEBgbmWNV05MgR9Ho9kP/trNDQUObPnw/AV199laPyd3FERUWpNagmTpyY49ikSZNYvXo1d+/eZdeuXWX6/f3XX3+pj+3t7enZs2euhQAGg4GjR48yevRoZs6cyddff11m8QhRUndvxLBi/s/cuRaN1kLLiKf70GOErMKq6KrGvQZRIEVRSMvUldlHSkYq35/8qcAYAk5uJCUjtUzjKE7yWhKOjo60b98eyJmwQO4VV0bdunUDcs/jKcr8nWeeeYbU1FQeffRRBgwYUOK4165dS0ZGBvDv7Syj/v37U6tWLeDfUaCykr0+0fLlyzl27Bh+fn7s3buXlJQU7t27x5o1a9RSEN988w3/+9//yjQmIYrr9KGsJed3rkXj7ObI8588Rs+RD0qyUwlU2hEeUXQ6fTpTN71i1hhiU+OYtvm1Mu0jYPSn2FqW7R5lPXr0IDg4mMTERE6fPk3Hjh2Bf+fveHl55Zi47u/vz5YtW3LN4zEmPLa2tnTq1ClXPytXrmT37t04Ozvz2WefmRSzMZHx8/PLNanewsKC8ePHs3jxYrZu3VpgUUVTZd9+JS0tjVatWrFv3z71lrOdnR0TJkzgwQcfpH379iQnJzNv3jymT59u8m1pIUyl1xvYvnw/ezcGAdC4TT2mvDEcZzfZSqiykBEeIYrBeLsKco7S3D9/5/72J06cIDU1FcgqvBcUlPVLs3Pnzlhb57znHx0dzSuvZCWo//3vf6ldu3aJ4z1//jwnTpwAco/uGBmfT01NzVG3qrTdf0tu/vz5eSYyTZo0YebMmUDWqNDu3bvLLCYhiiLhXjJfz1mnJjt9xvgx66MJkuxUMjLCUw3YWFgTMPrTMrv+hagrfHSw8OXSr/eYSYuajcssDhuLsp8s2L17dzQaDYqicOjQIV566SUg/4SnQ4cO2Nvbk5KSwrFjx+jduzfBwcGkpaUBed/Omj17NtHR0Tz44IPMmjXLpHiNozuWlpZ5ri4E6NixIy1atODChQusXLmSJ5980qQ+85N9M16NRlPgysmBAweqBRaDg4MZNmxYmcQkRGHCzt/k+/d/IT4mCRs7ax575RHa9Whu7rBECUjCUw1oNJoyvdXT1qsFbnauOVZn3c/dzpW2Xi0q/RJ1Nzc3WrZsSUhIiJrkJCQkqCuu7k94rKys8PPzY//+/Rw8eJDevXsXOH8nIiJC3f6hT58+au2c/ERGRrJu3TogayuUzp07q8cMBoNaRTkzM7NIW1IcOnSI8PDwMtkeJfumuq6urjkSoILaygaiwhwUReHQlhP8vGQvBr0Br/ruPPn2KLzqy5LzykoSHmEyrVbL1A5j8lylZTSlw5hKn+wY9ejRg5CQEKKiovj7778JCwvDYDDkmNScnb+/P/v371cTHeN8HisrK3WJuFF6err6eOHChYXGcuHCBSZMmADAlClTciQ8e/bsybGXXFEoisKqVat46623inVeUTRp0gQrKysyMjJyTPjOS/bjlpbya0qUL11qOusW/8bJfVkrC9v3bMGE2YNkyXklJ79JRKnwq9uO2d2m56rD427nypQqUofHqHv37upy6UOHDhEWFgZAly5d8qxjYxz1OXbsGDqdTt18tEOHDmVaZM94O8vGxoYVK1YUmnAuXLiQU6dOlVnCY0zwDh48SEJCglrLKi+hoaHqY2OBQiHKQ+TNWJbP26wuOR/+dG96jpBVWFWBJDyi1PjVbceD3m2qbKVlo/s3EjUmPPffzjLq2rUrFhYWJCcnExAQQHx8fK7rGDVs2LBIy+uNv3x79uzJ/v37cx1PSkri559/BrKWnj/22GOFXvPOnTucOnWKy5cvc/To0VyjT6Vh9OjR6gjXL7/8wvTp0/Nst3nzZvVx9oniQpSlM4cvsvqTX9GlpOPs5sDU/xuBb6t6hZ8oKoWq9U4kzE6r1dLSsyndGjxIS8+mVS7ZAfD29sbX1xeAffv2qXup5ffG7OzsTOvWrYGct6kK2j/LVJs2bVKXgY8ZM6ZI54wePVpNpMqqJs+0adPUuUTvvPNOnjvJHzhwQJ3H1KpVK7WWkRBlRa83sHXZPlbM/xldSjq+rery6ldPSLJTxcgIjxAl0L17d0JDQ9U5MpaWlnTp0iXf9v7+/pw+fVrdWkGr1eY7IlQajAmLlZVVkVc41atXj06dOhEUFMT69ev5/PPPcyyZP336dL6bjN65c4eAgIAcz40ZMwZHx5zLdh0dHfnf//7HhAkTiIiIoFOnTsydO5fOnTuj0+n47bffWLRoEXq9HktLS7799lu5lSDKVOK9ZH74YAuXz1wHoNfoTgx7shcWlmW3zYowD0l4hCiBHj165HiDb9++fYHF8fz9/fnyyy/Vz1u3bo2rq2uZxHbjxg31NlefPn1y7OFVmDFjxhAUFMS9e/fYtm0bo0ePVo/98ssvzJs3L8/zLl68yBNPPJHjuV69euVKeAAeffRRoqOjmT17NhERETz//PO52jg6OvLjjz/K6I4oU2F/3cpach6diI2dNRNeeYT2suS8yqp69xuEKAf3344qbLTm/ttdZXk768cff8RgMADkSFiKInv7stxq4tlnn+XkyZPMnDmTxo0bY2dnh6OjI23atOE///kPly5dYvjw4WXWv6jeFEXh4JYTfPHqauKjE/Gq787sLx6XZKeK0yhlvQGRKFRCQgIuLi7Ex8cXq6x/WloaYWFh+Pj4lHhjSSGEKE0V/feSLjWd9Z/v4sTe8wC069GcCbMHYWtfttvSiLJT1PdQuaUlhBCiWoi8FcuKeT9zOzwKrVbDsKd602tUJ5knVk1IwiOEEKLKOxt4idUf/0paig5nNwemvDGcxm3qmzssUY4k4RFCCFFl6fUGdgQcZPf6YwA0alWXqW+OwMVdNv6sbiThEUIIUSUl3kvmhwVbuXz6GgC9RnVi2HRZcl5dScIjhBCiUjPoDYSG3CAhNhlnNwd8W9Xj+qU7fP/ez8RFJ2Jta8VjrzxC+54tzB2qMCNJeIQQQlRaZw5fZPPXu4mLTlSfs3O0QZeSjsGg4FXPnWlvj6RWg7z3bRPVhyQ8QgghKqUzhy+yYv7PuZ5PTdIB0LCFNzMXPCpLzgUghQeFEEJUQga9gc1f7y6wTVx0ItY2VuUUkajoJOERQghR6YSG3MhxGysvcVGJhIbcKKeIRGEMBgORkee5di2QyMjzakX48iK3tIQQQlQ6CbHJpdpOlK2bN4M4eTKA1NRY9Tk7Ozc6dJhK3bp+5RKDjPAIIYSodJzdHEq1nSg7N28GERi4KEeyA5CaGktg4CJu3gwqlzgk4RFCCFHp+Laqh429dYFtXGs64duqXjlFJPJiMBg4eTKgwDYnT/5QLre3JOERQghR6Zz/8wq6lPQC24ya2Q+thbzNmVN09IVcIzv3S02NITr6QpnHIt8JQgghKpW712NYtXA7AA/4+eLq4ZTjuGtNJ6a9PZK2/s3MEZ7IJjU1rlTbmUImLQshhKg00pJ1LHt3E7qUdBq3qcf0d0eh0WhyVVqWkZ2KIT29aJPG7excyzYQTEx4pk2bhkaj4f3336d27dpFOicqKorXX38djUbD8uXLTeleCCFENWIwKPy4cDuRN2Nx9XBi6psj1H2xmrRtYOboRHZ6fTohIT/x99/bCm1rZ+eOh0fZb/thUgocEBBAQEAA9+7dK/I5CQkJ6nlCCCFEUf2x5gjnjl7G0sqCae+MwqmGrMCqiKKjL7Nr1xw12alZs3mB7Tt0mIJWW/YjcnJLSwghRIUXcuwKO1cdAmDciwNp0KxodxVE+cnMTCckZAOXLv2KoijY2rry4IPTqVPnwXzq8LjTocOUcqvDU+4JT1paGgA2NrK3iRBCiMJF3oxl1YfbUBToPqwDnQe0MXdI4j7R0ZcICvqWxMQIABo08Kd9+6nY2DgCULeuH97eD/6zaisOOztXPDxalMvIjlG5JzyBgYEAeHl5lXfXQgghKpm0FB3L520mLUVHo1Z1GTGjr7lDEtlkjeqs5+LFHUDOUZ37abVaPD1bln+Q/yhWwjN//vw8n//666/x9PQs8FydTkdoaChbt25Fo9HQrVu34nQthNno9Xpq1KhBYmIi7du35+TJk/m2VRQFDw8PYmOzhm2XL1/OtGnT8m3/ww8/MHXqVCDr52jmzJk5joeEhPDtt99y4MABrl27RlpaGi4uLrRs2ZJhw4bx1FNP4eTklMeV8xYeHk6jRo1QFAWA1atX89hjjxX5fIPBwE8//cTatWs5ceIEUVFRaLVavLy88PPzY+rUqQwaNKjI1xOiIIqisPqTX7lzLRoXd0ee+L8RWFpZmDss8Y+oqIsEB39LYuJtABo27EG7do+rozoVjlIMGo1G0Wq16odGo8n1XGEfGo1GsbOzU06fPl2crqu0+Ph4BVDi4+OLdV5qaqry119/KampqWUUmTAaOHCgAiharbbA/0/nzp1TAPVj6tSpBV73ySefVNuGhITkOPbhhx8qFhYWOa53/0e9evWUU6dOFfl1zJs3L8f5AwcOLPK5sbGxSvfu3QuMB1BGjx6tpKWlFfm6omopzd9Lv685orzQf4Hy8iMLlbC/bpZCdKI0ZGSkKSdP/qCsWzdeWbfuUeWXX55Rbt06brZ4ivoeWuybZ4qiqB8ajQaNRpPjufw+bGxsaNiwIRMnTuTo0aO0bdu25FmaEOWsR48eQNYIx5EjR/Jtd+hQ1qRKCwuLHJ8X1t7Dw4MHHnhAfX7t2rXMmTMHvV6PtbU1L7/8Mr/++it//vkna9aswd/fH4AbN27w8MMPExcXV6TXsWrVKgAcHbP+Atu9eze3b98u0rnjx49X4/Xx8eHrr7/m0KFD7N27l48//hgPDw8ANm3axIsvvlikawqRnwvBV/k14AAAY58bQMMWdcwckYCsUZ1du17n0qWsW1gNG/Zg0KCP8fbuaO7QCmdKVmUc3Tl//rwpl6n2ZISn4jt06JA6gjF37tx8240fP14B1H8B5datW3m2vXv3rtpmxIgROY61bNlSPbZ9+/Y8zx81apTa5uOPPy70NQQGBqrtv/vuO3X0qCjnBgcHq+c2atRISUhIyNXm2rVriqurqzoSdvfu3UKvK6qe0vi9FHUrVnl95CLlhf4LlHWLd5ZidKKk7h/V2bLlGeXWrRPmDktRlDIc4cmufv361K9fH2vrgjdwE6Ky69SpE7a2tkDBozbGY2PGjMHX17fA9tmfN44gQVatqvPnzwPQoUMHBg8enOf577zzjvr46NGjhb6GlStXAlmjSVOmTKFv36zJn8ZRn4JkH9V66aWX8pw3VL9+fZ544gkgayTszz//LPS6QtxPl5rOsnc3k5qko+EDdRg9s5+5Q6r2oqL+vm9UpycPP/wJ3t4dzB1asZiU8ISHhxMWFkbjxo1LKx4hKiQbGxv8/LJqRQQHB6PT6XK1CQsL49atWwD4+/urt52Km/Ckp/+7IWKjRo3yjcmYUN1/Tl50Oh0bNmwAYNy4cVhZWTF58mQAzp49y+nTpws8vyxiEuJ+iqKwdtEObodH4ezmwLS3RmBpLeXizCUzU8fJkz+wd+88kpLuYGfnRvfur9O580ysrStf0UfZbESUKoPBQGTkea5dCyQy8jwGg8HcIZUaY1Ki0+nyHL04ePAgAE2aNMHLy0tNeIzP38+Y8Dg7O9OuXTv1eQ8PD9zc3AC4evVqvvGEhoaqj5s1K3iTxG3btqkV0SdNmgTAyJEjcXDI+qVlHP3JT/brl1ZMQtxv709BnDrwNxaWWp54ayQu7kVfgShKV1TUBXbt+g+XL+8EFHx8evHwwx/j7d3e3KGVmCQ8otTcvBnE9u3PsW/fexw79gX79r3H9u3PcfNmkLlDKxXZR2HyGrUxPmdMdIz/hoSE5Np+JTExkTNnzgDw0EMPqZOcjZ555hkATp48yW+//ZZnPO+99x4AlpaWTJ8+vcDYjQmNr68vXbt2BcDBwYERI0YAsGbNGvR6fb7nDxw4EB8fHwA+//xzkpNzbwh48+ZNdcsYf39/WrVqVWBMQmT394kwtq3YD8CoWf1p1LKueQOqpjIz0zh5MoC9e+eTlHQXOzs3evSYg5/fM5VyVCe7Ukl4Lly4wMsvv8yDDz6Im5sbVlZWWFhYFPhhaSnDlFXJzZtBBAYuylE2HCA1NZbAwEVVIunp2rWr+n1bUMLTvXt3AJo3b46HhweKoqgFN42OHDmiJhjZEymjN954g4EDBwIwYsQIXn31VXbu3ElwcDDr16+nV69ebNy4EQsLC7788kuaN89/r5qoqCg1aZo4cWKOY8bRnrt377Jr1658r2Ftbc2aNWvw8PAgNDSUtm3bsmTJEgIDA9m/fz+ffvopHTt25N69ezRq1IgVK1bkey0h7hdzO44fPtiCYlDo8nAbug1uZ+6QqqXIyAvs2vU6ly//RtaoTm8efvgTatduZ+7QSoXJWceiRYuYO3cumZmZajGz8nL8+HF27NjB4cOH+euvv4iKisLKygpvb2+6devGk08+qf6VbYp3332XefPmFantvn376NWrl8l9liZFUdDrc885Kb3rGzh58vsC25w8GYCXVys0mrIbVLSwsEGj0ZTZ9R0dHWnfvj3BwcFqwmIcmYmMjOTSpUsAOb7nunXrxpYtWzh06BBDhgxRn89v/o6Rg4MD27dvZ+XKlXzwwQd8+umnfPrppznajBo1ijlz5tCpU6cC4167di0ZGRnAvwmOUf/+/alVqxZ37txh5cqVPPLII/lep0uXLpw6dYrPP/+czz//XB2Fyv71ee+995g1a5Z6S06IwuhS01k2bxMpiWk0aFabMc8NKNOfY5FbZmYaZ8+u+yfRATs7Nzp1mkHt2lWrfIxJCc9vv/3Gq6++CoBGo6FLly507NgRNze3Mt8fo0ePHnn+lZ2ens7ly5e5fPkyAQEBPP7443z33XfVeiWZXq9j06apZo0hNTWWzZvzrzhcGkaPDsDS0rZM++jRowfBwcEkJiZy+vRpOnbMqj1hnKfj5eVFkyZN1Pb+/v5s2bIl1zwe4/eura1tvglLUFAQP/74Y75zZv744w9q1qxJ06ZNcXFxyTdm4+0sPz+/HLFBVr2g8ePHs3jxYrZu3UpCQgLOzs55XkdRFNatW8eGDRvUBCq7pKQkVq9ejbe3d4HVpYUwUhSFdYt/I+JqFI6u9kx7eyRWMkm5XEVGnicoaAnJyZEANGrUh7ZtJ2FtbW/myEqfSd9ZixcvBqBGjRps3bq1XLeLiIjI2qDM29ubsWPH0r17d+rXr49er+fo0aN8+umn3Lp1i5UrV5KRkcGaNWtKpd9z584VeNw4z0FUTd27d1dHWg4dOqQmPPfP38neHuDEiROkpqZiZ2dHeno6QUFZt/g6d+6cZzK+ceNGJk2ahE6no02bNsybN48ePXrg5OTEjRs3WL9+Pe+99x5Llizh4MGD7N69G29v71zXOX/+PCdOnAByj+4YTZo0icWLF5OamspPP/3Ek08+mauNwWBg/Pjx/PTTTwA8+eSTPPvss7Ro0QK9Xs/p06dZuHAhW7du5cknn+Ts2bPq7wch8rN/UzAn9/2F1kLLE/83AteaeSfbovRlZKRx9uwarlz5HQB7e3c6dXqaWrWq1qhODqYU+3F3d1e0Wq2yePFiUy5TIoMHD1bWr1+vZGZm5nk8KipKadq0qVos7cCBAyXu65133lGvUxbKuvCgwWBQMjJSy+wjIuKksm7do4V+REScLNM4DAaDKf8biiQmJkbRaDQKoIwaNUp9vn379gqgfPbZZznap6enK/b29gqg7N27V1EURTl8+LD6/fTWW2/l6uPOnTuKg4ODAigtW7ZUkpKS8oxl9+7decaS3X/+8x8FUCwtLZXIyMh8X1eLFi0UQOnRo0eex7/88ks15nfffTff60yePFltt3Xr1nzbiaqrqL+XLp4MU14c+KHyQv8FyoFfzLctQXV0506Ism3b8+rv5qCgpUp6erK5wyqxcik8mJKSAuT+q7Y8bN++nXHjxuVa3WLk4eGRY87Dxo0byyu0Ckej0WBpaVtmH15ebbGzK3jOhp2dO15ebcs0jvK47+/m5kbLllm7/RpHdRISEtQVV/f/LFhZWan1e4y3tQqbv7Nu3Tp1FdQbb7yhLh2/X9++fdXigb/88kuulWAGg4HVq1cDkJmZiaenp7odzP0fFy5cUGMLDw/P1deyZcsAcHJyYs6cOXnGA/DBBx+oj2XisshP7N14Av6bNUnZr38rug+rXAXsKquMjDROnFjB/v3vkZwcib29Bz17vkGnTk9hZVX1bmHdz6SEp06drL1NKmqBsd69e6uPs9cHEaVLq9XSocPUAtt06DClzOd1lRdjkhIVFcXff/9NYGAgBoNBndR8v/sLEBoTHysrK3WJeHbG5AOyKi0XxHhLzWAwqJOmjfbs2aMWQiwqRVHyrLxsjOmBBx7AxsYm3/Pr1q2Ll5cXAH///Xex+hbVQ7oug+XzNpOckEq9JrUY+8JAmaRcDu7eDWHXrtfUW1i+vv14+OGF1KrVxsyRlR+T5vAMHTqUzz//nMDAwDx/cZtb9mq4+Y0EidJRt64f3brN5uTJgBxL0+3s3OnQYQp16/qZMbrS1b17d77++msgK4kJCwsDslYx5fV9Zkx4jh07hk6nU7dp6NChQ56jN9lLNmRmZhYYS/bJw/eXejBOVraxsWHFihWFJpwLFy7k1KlTrFq1irfeeitXTDqdrtB4ssckpSfE/RRFYf3i37h55S4OLnZMe2ck1jZW5g6rSsvISOXMmTWEhv4BgL29B506zaBWrdZmjqz8mfQb6dVXX2XVqlV8+umnTJo0iVq1apVWXKXiwIED6uMWLVqUyjUHDBjA6dOniYuLw9XVlQceeICHH36YGTNmUKNGjVLpo7KqW9cPb+8HiY6+QGpqHHZ2rnh4tKgyIztG2W9DHTx4UE148ru127VrVywsLEhOTiYgIID4+Phc18ku+8T3Q4cOFVjAzzhapNFoaNiwofp8UlISP//8M5C19Pyxxx4r9HXduXOHU6dOcfnyZY4ePZrjjxgfHx9CQkIICQlRv/fzEhISQmxsbK7XIQTAwV9OcHzPebRaDU+8OQI3z/xXFwrT3b17jqCgJaSkRANZozpt207EysrOzJGZh0nvRN7e3mzZsgW9Xs9DDz3Ejh07SisukxkMBj788EP183HjxpXKdf/44w+ioqLIyMggKiqKAwcOMHfuXBo1asSWLVuKdA2dTkdCQkKOj6pCq9Xi6dmSBg264enZssolO5D1fW/cM2rfvn0cP34c+HdF1v2cnZ1p3Trrr6mFCxeqz+eX8AwePFgd4v/vf/+b722ppUuXqn136dIFd3d39dimTZvUeUBjxowp0usaPXq02u/9W00MHToUyPrenT17dp41t9LS0njhhRfUz7PXHRLiytnr/LJkDwDDn+5Dk3YNzBxR1ZWRkcrx48vYv/+/pKREY2/vQa9eb/Lgg9OrbbIDoFHy+s1VRH369AGylohfunQJjUaDq6srTZo0wd6+4AlQGo2GPXv2lLTrQn366adqjaBRo0axadOmEl/r3XffZfPmzYwYMQI/Pz+8vb3JyMjg4sWLrF69mt9/z7onamFhwbZt2xg0aFCh18urkGF8fHy+NVDykpaWRlhYGD4+PupO3qJ8PPHEE+o2CpB1+yY+Pj7f7/vnn3+eL7/8Uv1cq9USExOT70jJk08+qU76rVmzJi+99BLdu3dXl6WvW7dOLbVgYWHB7t27cxS87Nu3L3v37sXKyoq7d+8WefSxc+fOBAUFUaNGDe7cuaMumY+OjqZly5ZERmbV6vD39+eZZ56hefPm6PV6Tp06xf/+9z/++usvIGtE9fTp09W6/lV1ldfvpXuRCXzybABJ8Sl07NOSya8PkXk7ZeTOnXMEB/87qtO48QDatJlQpROdhIQEXFxcCn8PNWUpmEajUbRaraLVahWNRlOkD2NbrVZrStcF2r9/v2JpaakAiqenp3L37l2Trnfv3r0Cj3/77bfqUlxvb+9Cl2OmpaUp8fHx6seNGzfKdFm6KH0rVqxQ/58DSqdOnQpsv27duhzt27ZtW2D7tLQ05dFHH81xTl4fDg4OyurVq3Oce/36dUWr1SqAMnDgwGK9roULF6rX3rhxY45jp06dUnx8fAqNqV27dkp4eHix+hVVx/2/l9J1GcrHz36vvNB/gfLRM8sVXWq6mSOsmtLTk5WgoKXqUvNt255T7twJMXdY5aKoy9JNmsPTo0ePCpelnz9/npEjR5KZmYmtrS0//fQTnp6eJl0zv7/CjWbMmEFwcDDLly8nIiKCTZs25dqzKDsbG5sCV7qIiu/+21GFlWa4/3ZXfrezjGxsbFi3bh0zZswgICCAY8eOcevWLXQ6Hc7OzjRr1ox+/frx9NNPU7duzk0Wf/zxR3WX+tGjRxf1Jant//Of/wBZt7Wyn9+uXTvOnTvHDz/8wJYtWzh79iyxsbFoNBo8PT1p3749Y8eO5dFHH8XKSiaiiqxJyhv+t4sbl+5g72TLk++MwtpWvjdK2507ZwkOXnrfqM5jWFnJyH92Jt3SqmjCwsLw9/cnIiICCwsLNm3axPDhw8ul7+DgYLXeylNPPcXSpUuLfG6Rh+PuI7e0hBAVTfbfS8G//8XGL39Ho9Uw84NHadahobnDq1IyMlI4ffpHrl7dC4CDgyd+fjPw9Gxp5sjKV1HfQ6vMutGIiAj69etHREQEGo2GFStWlFuyA1n1SYyKW/tECCGqmuuXbrP5m90ADHuylyQ7pez27TMcP76UlJQYABo3HvjPXB354zc/VSLhiY6Opn///uomi1988QWPP/54ucZQ0W7tCSGEuRj0BrZ8sweD3kD7ni3oPabq1OEyt/T0FE6fXkVY2D7AOKrzDJ6eDxRypij1hOfmzZvcuXOHlJQUOnXqhJ1d2c4Mj4+PZ+DAgerqkA8//JBnn322TPvMi7F/IM9NHIUQojpQDApJ8SkkJ6RQ26cmE2YPkj8IS8BgMOSqaXb3btZcHWNx1yZNHqZNm/FYWsqoTlGUSsKTmJjIwoULCQgIUHcxh6ydxbPf6lm3bh2bN2/GxcWF7777zuR+U1JSGDx4MCdPngTgzTff5PXXXzf5uiWxZMkS9XHPnj3NEoMQQphbfGwSmRl67BxsmP7OKGzspDRBcd28GZSrar2FhQ16fdbuAY6OXnTq9AyenqVTULe6MLkq3OXLl+nQoQMffPABt27dQlGUPIuSQVZxtM2bN7NixQoOHz5sUr/p6emMHDmSwMBAAF588UXef//9Yl8nICBA3UDx3XffzXX83LlzXLlypcBrLF26VN1csVatWowcObLYcQghRGWXFJdCSmIqkFVc0MO7elefL4mbN4MIDFyUI9kB1GSndu32DBz4kSQ7JWDSCE9aWhqDBw8mNDQUBwcHnn32WXr06JFvhdWGDRvSu3dv9u7dy9atW03aZX3ChAlqwb8+ffrw5JNPEhISkm97a2trmjZtWux+Tpw4wfTp0+nduzeDBg2idevWuLu7k5mZyd9//52r8ODSpUvz3d1aCCGqKl1qOveisqrG2zna4tuqnpkjqnwMBgMnTwYU2CYu7jparYyalYRJCc8333zDlStXcHBw4NChQ7Rr167QcwYNGsSePXs4evSoKV2zefNm9fHevXtp06bgHV8bNGhAeHh4ifrS6/Xs3r2b3bt359vG3d2d5cuXqyX4hRCiusjM1BMdEYeiKNg52KBD6oyVRNacndgC26SmxhAdfaHaLT0vDSYlPJs3b0aj0fDiiy8WKdkBaNu2LZB1K6wyeOSRR1i+fDlHjx7l1KlT3L17l5iYGBRFwc3NjbZt2/Lwww8zderUYtXQEUKIqkAxKMRExKHX67GytsTFw4G45ILftEXeUlPjSrWdyMmkhOfChQtA1g7iRWXc4DAuLs6UrvOdJ1RcU6dOZerUqfke9/T0ZNq0aUybNq1U+hNCiKrkXlQCurR0tFotHt410BsyzR1SpWVn51qq7UROJk1aTkpKAsDR0bHI5+h0WROvpPS8EEJUbknxKSTFpwDgXssFK+sqUdrNbDw8WmBn51ZgGzs7dzw8ZMJySZiU8BhHa4ozN+b8+fNA1momIYQQlZMuNZ17kVmTlF3cnbBzlFowptJqtbRsWfD+dx06TEGrNXmBdbVk0letQ4cOABw8eLDI56xcuRKNRkPXrl1N6VoIIYSZ6DP1RN/+Z5Kyoy3ObrIytbSkpcUBoNXmHC2zs3OnW7fZ1K0rVatLyqTxxzFjxvDrr7+ydOlSZs+eTf369Qtsv3jxYg4ePIhGo2HChAmmdC2EEMIMFEUh+nYc+sysScruXi5SSbmUGAx6QkP3APDgg0/j4OCeo9KyjOyYxqSv3uTJk2nTpg1paWn06tWLnTt35phMrNFoUBSF4OBgJk6cyCuvvIJGo6F79+4MGjTI5OCFEEKUr7ioRHSp6Wi1Gjy8a6C1kDfh0nL79ilSU2OxsXGifv2ueHq2pEGDbnh6tpRkpxSYNMKj1WrVAoLh4eEMGTIEe3t7Ndvv1asXiYmJ6kRlRVHw9fVlw4YNpkcuhBCiXCXHp5AYlwyAWy1XmaRcyq5c+QMAH59eWFjIwp7SZnLKWL9+fU6fPs2ECRPQarUkJyer20tERUWRlpamjvqMGzeOoKAgPD09TQ5cCCFE+UlPyyDWOEnZzRF7maRcqpKS7nLnzhkAfH37mjmaqqlU0nM3NzdWr17NBx98wK+//srx48eJjIxEr9fj7u5O+/btGTp0aIm2dhBCCGFe+kw9URH31ErKzu5FL0UiiiY0NKuSf61abXF0lFXMZaFUxyMbNGjArFmzSvOSQgghzCjXJOVarjJJuZTp9RmEhe0HwNe3n3mDqcJkFpQQQoh85ZikXNtVJimXgZs3/0SnS8TOzg1v7w7mDqfKku9cIYQQeUpOSP13krKXK1Y2MpG2LBgnK/v69kWrtTBzNFVXqd3SMhgM/PXXX1y9epXExET0en2h5zz++OOl1b0QopK5ePEirVu3RqvVEhoaSp06dcwdUqUzePBgduzYwdtvv828efNK9drpaRnE3o0HwNnNEXsnmaRcFuLirhMdfRGNRouPT29zh1O1KSZKTk5W5s6dq9SsWVPRarVF/rCwsDC16yojPj5eAZT4+PhinZeamqr89ddfSmpqahlFJhRFUTIzMxUnJycFUNq3b19gW4PBoLi5uSmAAijLly8vsH1AQIDa9uuvvy40loiICMXV1VU9p2fPnsV5KUpYWJii0WjU81evXl2s8/Py9ddfq9cDlO+//75I5z3yyCMKoMyaNavIffXq1Uvtp3///iWMuOo4evSoAij29vbKjRs3Su26mRmZyq3Qu8q1ixHK3ZsxisFgKPK58nupeI4fX66sW/eocvjwp+YOpdIq6nuoyZuH9uzZk48++ojo6Gh1OXpRP4SoDCwsLHjooYcAOHPmDAkJCfm2PX/+PLGxsernhw4dKvDa2Y/36NGj0Fief/554uLiCm2Xn5UrV+b42Vu5cmWJrwUQERHBnDlzin3ekSNH2LFjB9bW1kU+/9q1axw4cED9fM+ePURERBR6XkBAABqNBo1GU6x9/yqDLl260L9/f1JSUvjggw9K5ZqKohBzJ47MTD2WVhYySbkMZWSkER6e9TvA17e/maOp+kxKeN5//31OnDiBoih06dKFFStWcOLECUJDQwkLCyvw4+rVq6X1GoQoc8ZkxGAwcOTIkXzbGRMYCwuLHJ8X1t7Dw4MHHnigwLbbtm1j06ZNJtWxWrVqFQCOjlnLinfv3s3t27dLfL3nnnuOhISEYsf0/vvvAzB27Fjq1atXpHNWrVqFoijY2NhgaWmJwWDgxx9/LHbMVc0rr7wCwPLly036f2kUH51IWko6Gk1WJWULmaRcZq5fP0xmZiqOjrXw8mpp7nCqPJO+kzdu3IhGo2Hw4MEcPnyYqVOn0r59e3x8fGjQoEGhH0JUFtlHXwraLNd4bOzYsQCEhobmOwoRGRnJpUuXAPD39y/wr+ikpCSeffZZAD755JPiBf+PI0eOcOXKFQA+++wzLCws0Ov1rF69ukTX27JlCz///DM1a9bk9ddfL/J5Fy9e5LfffgNg0qRJRT7PmKwNGTKEAQMG5HiuOuvXrx+enp6kp6ezZMkSk66VnJBKwr2sScrutVywlknKZUZRFK5cyaq94+vbD41GEsuyZtJX+NatWwC88MILss+HqNI6deqErW3WpM2CRm2Mx8aMGYOvr2+B7YtzO+uNN97gxo0b9O7dm8mTJxcrdiPj7SsPDw+mTJlC375Z1VxLkjQkJiby3HPPAVkJmJubW5HP/f7771EUBU9PT/r1K1rNkWPHjqnJ4cSJE9VEKSQkhJMnTxYz+qrFwsKCRx99FPj3a1sSOSYp13DE3smu1GIUucXGXiEuLhyt1gofn57mDqdaMClLMQ5je3h4lEowovIz6A1cPnONE/v+4vKZaxj0BnOHVCpsbGzw8/MDIDg4WN0fLruwsDD1jwB/f3/8/f0B0xOeoKAgvvrqK6ytrfnmm29KFL9Op1P3sBs3bhxWVlZq4nT27FlOnz5drOvNnTuXmzdv0qtXr2KvtjTGMXz4cCwti7ZQ1Jis1ahRg8GDBzNixAicnJxyHLvf/v370Wg0PPHEE+pzPj4+6nwe48f+/ftznZuUlMSHH35I165dcXNzw8bGhrp16zJmzBi2b99eYKy9evVCo9HQq1cvAK5cucIzzzxDo0aNsLOzo2HDhjz55JNcu3Ytx3khISE88cQTNGrUCFtbW+rVq8fMmTOJjIws9OszevRoAK5fv05gYGCh7e+n1xuIvp1VSdnW3gYXD6mkXNaMS9Hr1++KjY2TmaOpHkxKeIxvABcvXiyVYETldubwReZN/oYvX1vLygVb+fK1tcyb/A1nDleN7w9jUqLT6fjzzz9zHTfezmrSpAleXl5qwpPfLTBjwuPs7Ey7du3ybJOZmclTTz2FwWDg9ddfp1mzZiWKfdu2bdy7dw/49zbSyJEjcXBwAIo3efnYsWN88803JUrArl27RlhYGJA14bYo0tPTWb9+PZB1q9Da2ho7OztGjRoFwNq1a8nMzCxWHAU5deoUzZo1Y+7cuRw7dox79+6Rnp7OrVu32LRpE0OHDmX06NGkpaUVeq3du3fToUMHlixZQlhYGGlpaVy7do0VK1bg5+fH33//rb6Gjh07EhAQQFhYGDqdjps3b/Ltt9/i5+dX6OTsTp06qfPGdu7cWazXqygKMbfjyMz4Z5JybZmkXNZ0uiRu3DgKSGXl8mRSwvPyyy8D8OWXX8qqq2ruzOGLrJj/M3HRiTmej4tOZMX8n6tE0pN9FCavURvjc8ZEx/hvSEiImmwYJSYmcuZM1kaBDz30kPpmdb9PPvmEs2fP0rhxY954440Sx25MaHx9fenatSsADg4OjBgxAoA1a9YUqXZWRkYGTz/9NAaDgddee43mzZsXK47sX7dOnToV6Zzt27erK9+yz/kxPo6MjFTnBGXXqVMnzp07p06QBti1axfnzp3L8ZE9jlu3btG3b18iIiLU0aFdu3Zx/PhxVq5cSdu2bQHYvHkzU6dOLTDuiIgIxo0bh6urK1988QV//vknhw4d4qWXXkKj0RAZGcn06dMJDg7m8ccfx9fXl2XLlhEUFMS+ffvUEbhr164xe/bsAvuyt7enZcusSa/ZV7LlRVEU0lJ0JCekkpai+2eSsi5rknJtmaRcHsLDD6DXZ+Dq2gB39ybmDqfaMKnw4EMPPcRHH33Ef/7zH8aPH8+SJUtwdXUtpdBEaVEUhfS0jDK7vkFvYNNXfxTYZtPXu2narkGZlqW3trUq079Mu3btiqWlJZmZmQUmPN27dwegefPmeHh4EB0dTWBgIEOGDFHbHjlyRE0w8rudFRoayvz58wH46quv1DlExRUVFaUmBBMnTsxxbNKkSaxevZq7d++ya9cuHnnkkQKv9fHHH3Pu3DkaNWrE//3f/xU7FuMKN2tr60JXpRkZk7WGDRuqSSRAnz598Pb2JiIigpUrV+b4+kJWQteqVSuOHz+uPte0aVMaNmyYb18vvfSSmpx+9913PPnkk+qxjh07Mm7cOAYNGsS+fftYv349U6ZMYdCgQXle6/LlyzRp0oTAwEBq1qypPu/v74+lpSWffPIJgYGBDB48GD8/P/744w/s7e3Vdr169SItLY2ffvqJTZs2ERUVleM69+vYsSNnz57l1KlTKIqS589CSmIa96IS0GfmTm7dvFywtpVJymVNURR1o1Bf3/4ymlaOTK60/Oqrr+Lr68tTTz1FvXr16N+/P02bNs3xg5uft99+29TuRRGkp2Xwn+GLzBpDfHQic0YtLtM+Fm6ZjY2ddZld39HRkfbt2xMcHKwmLMaRmftXXBl169aNLVu2cOjQoRxvyEWZv/PMM8+QmprKo48+qq5KKom1a9eSkZGV8N6/Kqp///7UqlWLO3fusHLlygITnitXrvDee+8BJU/Abt68CYC7u3u+o1rZxcTEsGPHDgAee+yxHG8OWq2Wxx57jE8++YRt27YRFxdn0h9cERER/PzzzwA8/PDDOZIdIxsbG1asWEGTJk3IzMzkyy+/zDfhAfjf//6XZ5Iya9YsdbVddHQ0Bw4cyPN35syZM/npp5/IzMzk6NGjDBs2LN++jHMqU1JSuHPnDrVr185xPCUxjejb9/I6FUDeeMtJZOR5EhNvY2lpS4MG3cwdTrVicsITGRnJzz//THx8PAaDgS1bthT5XEl4RGXTo0cPgoODSUxM5PTp03Ts2BH4d56Ol5cXTZr8O0Tt7+/Pli1bcs3jMSY8tra2ed7aWblyJbt378bZ2ZnPPvvMpJiNIyR+fn45YoOsFT7jx49n8eLFbN26lYSEBJydnfO8zowZM0hLS2Ps2LE8/PDDJYolKioKyJp8XBQFJWvG5z755BN1JOSpp54qUVyQNcnZOOqWV7Jj1LBhQ/r378/OnTvVc/JK3lxdXRk4cGCe1/Dx8cHJyYnExETatGlDixYt8mxnvIUGFFq7LPtKufsTHkVRuBeVf8FMgHtRCdg52kjiU8aMk5UbNOiOlZWshCtPJiU8MTEx9OjRg8uXL8scngrM2taKhVsKngNgitBzN1jyfz8V2m7G+2PxbV20InMlUR7D8d27d+fTTz8FspIWY8Jz//yd7O0BTpw4QWpqKnZ2dqSnpxMUFARA586dsbbOOSoVHR2tFpP773//m+sv9eI4f/48J06cAPKveTNp0iQWL15MamoqP/30U55v9gEBAezduxdnZ2cWL15c4niMc3GKmvD88MMPAHTo0CHPpKBt27a0atWKkJAQVq5caVLCExISoj7u3LlzgW07d+7Mzp07SUlJ4erVq7kSSciavF5Q8uDq6kpiYiJNmzYtsI1RYmJivu0g59c0OTk5xzFdanqet7Gy02fq0aWmY2tvU2A7UXKpqfe4dSvrFmvjxjJZubyZNKHigw8+4NKlSyiKwpgxY9i7dy8xMTHo9XoMBkOhH6J8aDQabOysy+yjeUcfXD0KXlbpWtOJ5h19yjSO8vjLtHv37mo/2W9L5ZfwdOjQAXt7ezIyMjh27BiQtazduMInr9tZs2fPJjo6mgcffJBZs2aZFK9xdMfS0pLx48fn2aZjx45qMpHXaq2oqCheffVVAN577z28vb1LHI/xNlhqamqhbS9cuKDOvymoQKFxcm9gYKC6Aqwksm8JUljl6Fq1auV5XnaF3dY31i4rqF32+maFTSrP/jW1ssqZ/Oszi/b7tqjtRMlcvboXRdHj7t4UV1cpvlveTBrh2bp1KxqNhkmTJql/iYnqR2uhZdSsfqyY/3O+bUbN7FemE5bLi5ubGy1btiQkJERNchISEtQVV/cnPFZWVvj5+bF//34OHjxI7969C5y/ExERoRYC7NOnj1qzJj+RkZGsW7cOyLpNkn1kwmAwqFWUMzMzi7T9w6FDhwgPD88xsXfZsmXExMTg6uqKu7u72l922Zfp//nnn2pi06dPnxz9Guez5JckZJc9+Zo9e3ahK5UURWHlypW88847hV67MJXxtk72r2muuUxFfDkWlpX/Z7SiMhj0hIbuAaBxY9k3yxxMSniMRdamTZtWKsGIyqutfzOmvT2SzV/vzrE03bWmE6Nm9qOtf8nqx1REPXr0ICQkhKioKP7++2/CwsIwGAzqpOb7+fv7s3//fjXRMc7nsbKyUpeIG6Wnp6uPFy5cWGgsFy5cYMKECQBMmTIlR8KzZ88e9We0qBRFYdWqVbz11lvqc8Yii3FxcUXaCuLbb7/l22+/BWDfvn15Jjz3L9O/X/ZkrThWrVpV4oQn+xyYu3fvFrjH1507d/I8z5yyf02NsSuKQlJcSq5yEXmxsLQo00n/1d3t26dITY3F2tqJevUKvmUqyoZJCY+Hhwe3bt1SK56K6q2tfzNad21CaMgNEmKTcXZzwLdVvSoxspNd9+7d+frrr4GsEZHshfTymrxqHPU5duwYOp1OXZrdoUMHtfBfWTCOkBhXFhW2/cvChQs5depUroSnNLVu3Zp169YRHx9PZGRkvqNO+/bt48aNG0DWDvHG3erz8+eff7J48WJCQ0MJDAykW7d/V78UdbSmVatWOa5XUMJjnINlb29Po0aNinT9smZcJejj44O9vT3puqytIowlKaysLclIz79AY42azpVyZKuyME5W9vHpiYWFJJbmYFLC0717d9atW0dISAgdOnQorZhEJaa10NKkbdW+N33/RqLGhOf+21lGXbt2xcLCguTkZAICAoiPj891HaOGDRsWaQGA8Y2pZ8+e+W6NYFxi3b9/fx577LFCr3nnzh1OnTrF5cuXOXr0qDr69O677/Luu+8WeG5AQIC6hcP333+fb1E+4yRuyJrLNHjw4DzbGZM1CwsL/u///q/Q23H9+vXjyy+/JDMzk5UrV+ZIeLIvn89rSxCjXr16qRuqrlixgjFjxuTZ7vr16/zxxx85zqkIjPOd/Pw6ExedSGJsMgoKWq0WVw8nHFzsSE3S5arDY2FpQY2aztg7lazOkyhcUtJd7tw5C0hlZXMy6U/vV155BSsrK3VZqBDVgbe3t7ox6L59+9Q3muxv5tk5OzvTunVrIOdtqsI2DDXFpk2b1JU6+b1x32/06NFqIlWcrSaKw8/PDxubrFVAxlGS+yUnJ7N582Yg62talLlHHh4e9OyZtQHjhg0bciQ22Ve5hYaG5nsNb29vRo4cCWRtz5DXvMT09HSmTZumLpU3bqBqblevXiU6OhqAzh27khCbhIKCnYMttRp44Ohqj0ajwd7JFm+fmnjWdcO9liuedd3w9qkpyU4Zy5q7o1CrVhucnGoV2l6UDZMSng4dOrBs2TIuXbrEgAED1CFVIao6Y3Jz69YtdDodlpaWBe4NZRz9MdZS0Wq1+Y4IlQZjwmJlZVVgsbrs6tWrp9YEWr9+fY75RKXFxsZGrU2zZ8+ePNts3ryZpKQk4N9NMYvC2DYuLo6tW7eqz7dv314d5Xnrrbf4448/uHTpEleuXOHKlSs5Vjd99tln6vLuadOm8dRTT7F7925OnDjB6tWr6dy5sxq3sepyRWAccbK0tKRX9z5YWGjxqO2Kh7crllY5R6A0Gg229jY4ONthay91d8qaXp9BWNg+IKuysjAfk25pGScrP/DAAxw+fJgHHniANm3aFKnSskajYfny5aZ0L4TZ9OjRg4CAAPXz9u3bF/g97+/vz5dffql+3rp16zLbhuXGjRvqba4+ffoUueYNZI0GBQUFce/ePbZt21ashKOonnrqKbZu3cqRI0e4du0aDRrkvAVqTNY0Go26QWhRjBo1iueeew6DwcDKlSsZO3YsAE5OTrzwwgssXLiQkydP5qpavW/fPnVn87p167Jnzx6GDBlCREQEy5YtY9myZXn2VVFWpqYkpfHD91lfsx7+vWjgUw8XDyfZE6uCuHkzCJ0uETs7N7y9ZeqHOZmU8AQEBKh/HWg0GgwGA2fOnFGX6ObHuM+LJDyisrr/dlRhozX33+4qy9tZP/74o1rnqrgJy+jRo/nPf/4DZCUeZZHwDBo0iLp163Lz5k3Wrl3LnDlz1GO3bt1i7969QNbcp+LU/PHy8qJbt24cOnSI3377LcfeUx9++CFNmjRh5cqVnD9/nvj4+Hzr2rRv356LFy/y5Zdf8ssvv3Dx4kVSUlLw8PCgS5cuTJ06laFDh5rwFSgdmZl64iITuBoaxp/BWTWenn/+Ody8XMwcmcguNDRr9K1Roz5otRVjvld1pVFMKJHcsGFDk4ZDTSkSVpUkJCTg4uJCfHx8vmX985KWlkZYWBg+Pj4l3lhSCHNYuHAhr7/+Ok2bNuXChQuFriAT/1IUheSEVOKiEjEYDHzx9WI++XwhLVq04Pz582a/RSW/l/4VF3eDXbteQ6PRMmTIl9jbV4wSBlVNUd9DTRrhCQ8PN+V0IUQ19fzzz/O///2PS5cusWHDhnyrQIucMtIzib0bjy41a35VRqaO73/MGil/5513zJ7siJyMu6J7e3eUZKcCkD+rhBDlzs7Ojnnz5gHw/vvvy158hVAUhYTYJO5ci0aXmo5Go8HVw5mftq4jJiYGPz8/xo0bZ+4wRTYZGWlcu5ZVZFQqK1cMJu+WLoQQJTF16lTu3r1Leno6t2/fNmmPrqpMl5bOvbsJpOuylsLb2ttQw9MZK2tLnJ2deeeddxg1apSM7lQw168HkpGRiqNjLby8WhV+gihzpZ7wGAwGYmNjSUlJoU6dOhWmKJcQomKxsLDgjTfeMHcYFZbBYCA+JonEe1n1lLRaLTU8nbB3slOTG1M3lxVlQ1EUtbKyr29fNBq5mVIRlErCo9frCQgIICAggODgYDIyMtBoNJw9e5YHHnhAbbd9+3YOHjyIi4sLb775Zml0LYQQVU5qso57kfFkZmStJLN3sqNGTScsLOUPyMogNjaUuLhwtForfHx6mTsc8Q+TE57IyEhGjBjBn3/+Weh9+IYNGzJs2DA0Gg2DBw+mXbt2pnYvhBBVhl5vIC4qgeSErGKIFpYWuHk6Y+dYvVc7VTbG0Z169bpgYyN7TVYUJo2z6fV6hg4dyrFjx9BoNIwbNy5HcbX7tWrVSt3N2bjPjxBCVHfGpeZ3wqPUZMfJ1YHaDTwk2alk0tOTuHEja4Pgxo1l36yKxKSE54cffiA4OBgrKyt+/fVX1q1bV+g95WHDhqEoCocPHzalayGEqBIyMzKJjrhHzJ049HoDVtaWeNVzp4anM1qpllzphIUdRK/PwMWlPu7uTc0djsjGpJ+mtWvXotFomDFjhro/TmHat28PwMWLF03pGsjaHXj+/PkMGDCAunXrYmNjg6OjI02bNuWJJ54ok6Rq7dq1DBgwgFq1amFra0uDBg2YNGkSR48eLfW+hBBVl6IoJMYlc+daNKnJOjQaDS7ujtSq74GNnbW5wxMloCiKWnunceP+snKugjFpDs/Zs1nb3Rd1c0JA3fk4JibGlK7p0aMHhw4dyvV8eno6ly9f5vLlywQEBPD444/z3XffYW1t2i+Q1NRUxowZw44dO3I8f/36dVavXs3atWt5++23eeedd0zqpySkhokQlUu6LoN7dxPQpWUVELSxtcbNyxkrGyszR2a66vz7KDLyLxITI7C0tKVBg7LbHFiUjEkJT1xcHADu7u5FPse4f42py9UjIiIA8Pb2ZuzYsXTv3p369euj1+s5evQon376Kbdu3WLlypVkZGSwZs0ak/qbNm2amuz07t2bF198EW9vb86dO8cHH3xAaGgo7777LrVr1+bpp582qa+iMpbjN+6bJISo2BRDVgHBhHvJKIqCVqvBxcMJRxf7KjMaYPx9VB23CzHum9WggT9WVnZmjkbcz6SEx83NjcjISG7cuKHeqirM5cuXAdRN/UqqefPmfPDBB4wePTpX8tSlSxcmT55Mt27duHTpEmvXruWZZ54p8YaNe/fuZd26dQAMHTqUn3/+We2zU6dODBs2jI4dO3L9+nVef/11xo4dW6wdqkvK0tISrVZLWloaDg4OZd6fEKLkdKnpxN6NJyM9EwA7BxtqeLpgaVW1lpqnpaWh1WqxtKxedW1TU+9x82YwIJWVKyqTUvCWLVsCEBwcXORz1q9fj0ajoVOnTqZ0zfbt2xk3bly+I0UeHh58+umn6ucbN24scV+ffPIJkJVgfP3117n69PDw4KOPPgKyRr2WLVtW4r6KQ6vVYm9vT1JSUrn0J4QoPoPeQGxkPHdvxJCRnomFhRaP2q54eNeocskOQFJSEvb29tVuhOfq1X0oih5396a4ujYwdzgiDyZ9R44YMQJFUfjyyy+5d+9eoe03btzItm3bABg9erQpXRdJ79691cehoaElukZiYiJ79uwBoF+/ftStWzfPdqNGjVJ3aS3PJffOzs6kpKQU6esvhChfqUlp3L4WTVJcCgAOzvbUalgzR7XkquTevXukpKQUuGN1VWQwGLh6Net9QpaiV1wmjTk+9dRTfPLJJ9y4cYMBAwbwww8/5KisbBQZGcnnn3/Oxx9/jEajoVWrVuWy0Z1Op1Mfl3TOUHBwMOnpWRMLe/bsmW87a2trunTpwu+//65Wm7ayKvsJiC4uLqSmpnLnzh2Sk5NxcXHB0tKySv4yFaKy0GfqSYhNIjU563eQpZUFLu5O2NhZk5GRTkaGmQMsRYqikJmZSXx8PImJidSoUQMXFxdzh1Wubt8+RUpKDNbWTtSr18Xc4Yh8mJTw2NjYsGXLFnr16sWJEydo3bo1zZo1U49PmjSJpKQkrl69iqIoKIqCu7s7mzZtKpc35AMHDqiPW7RoUaJr/PXXX+rj5s2bF9i2efPm/P7772RmZnL58uU8k7+y4OXlhbW1NXFxcdy8ebNc+hRC5E2Xmk5KUhqKIWu1kq29DXaONiTfSTBzZGXLxsYGLy+vcpm/WNEYJyv7+PTEwkJKClRUJs8qa9u2LcHBwUyZMoWjR4/y999/q8fOnDmTY4min58fa9asoVGjRqZ2WyiDwcCHH36ofl7SEaXsCUR+t7OM6tWrpz6+ceNGvgmPTqfLMfqUkGDaL0KNRoObmxs1atQgMzNTXQknhCg/9yLj2bHyEOEXslaQ1qrvwSNTulO7oWkLNCoDCwuLajuynJQUye3bZ4CsjUJFxVUq0+gbN25MYGAghw8fZuvWrRw/fpzIyEj0ej3u7u60b9+eYcOG0b9/+c1c/+yzzwgKCgKy5td07NixRNdJTExUHzs6OhbYNvtKqYImEi9YsIB58+aVKJ6CaDQarKysyuVWmhAii15vYP+mYHauOkSGLhMra0sGPd6dXqM7YSGVkqu8rLk7Cl5erXFyqm3ucEQBSnXdoL+/P/7+5i+2dODAAebMmQNkFTr85ptvSnyttLQ09XFhxQttbGzUx6mpqfm2mzt3LrNnz1Y/T0hIyDE6JISoHG5eucPaRTu5eeUuAE3aNeDRFx+mZp3qd1unOtLrM7h6dR8gS9ErgypXKOH8+fOMHDmSzMxMbG1t+emnn9TqziVha/vvxn3Gycv5yX6bys4u/6JTNjY2OZIjIUTFZNAbCA25QUJsMs5uDvi2qofWQkt6Wga//XiYfRuDMBgU7J1sGf50HzoPaF0tb+tUV7duBaPTJWBnVwNv75LdRRDlx6SEZ/z48UyaNImHH364QhSZCgsLY8CAAdy7dw8LCwvWrVtX4mKDRk5OTurjwurdJCcnq48Lu/0lhKjYzhy+yOavdxMX/e9tbVcPJ7oMasPxPeeJjogDoH3P5oya2Q9nN/mZr26uXMmarNyoUR+02qpXU6mqMekG84YNGxg+fDi1atVi1qxZZt0BPSIign79+hEREYFGo2HFihUMHz7c5Otmn6hc2AqoGzduqI/lFpUQldeZwxdZMf/nHMkOQFx0Ir+tCiQ6Ig4XDyemzxvN1DdHSLJTDcXH3yAq6gIajZZGjWSycmVgUsLj5OSEoijExsayZMkSevbsiY+PD2+++Sbnz58vrRgLFR0dTf/+/bl69SoAX3zxBY8//nipXDv7SqvsK9DyYjxuaWlJkyZNSqV/IUT5MugNbP56d4FtrG2tmPPtNFp3lZ/z6io0NKvQoLd3R+zt3cwcjSgKkxKeyMhIfvrpJ0aOHIm1tTWKonDt2jU+/PBD2rRpQ7t27fjkk0/KtDZMfHw8AwcOVOvlfPjhhzz77LOldv1OnTqpk5Wz1/W5X3p6OseOHVPPkZVSQlROIceu5BrZuV96Wga3wiLLKSJR0WRmphEenvV+IJWVKw+TCw+OHj2a0aNHk5CQwMaNG1mzZg379+/HYDBw9uxZXn/9debMmUP37t2ZNGkSY8aMKbUqnCkpKQwePJiTJ08C8Oabb/L666+XyrWNnJyc6Nu3Lzt37mT37t3cvHkzz3o8mzdvVuvpjBw5slRjEEKUPn2mnsibsdy6GknE1Uj134TY5MJPhiK3E1XPtWtHyMhIxdHRCy+v1uYORxSRRsleGbCU3L59m3Xr1rFmzRpOnDiR1dE/Kxesra0ZNGgQkyZNYtSoUSXuIz09naFDh/L7778D8OKLL7J48eJiXycgIIAnnngCgHfeeYd33303V5u9e/fSt2/WPdphw4axefPmHFtVREdHq7ulu7q6cvXq1WJVG01ISMDFxYX4+PhqtweNEOUhOSFVTWqMic3ta9HoM0pepPO5jyfQpK1sElkd/f77G9y7d5W2bSfSvPlQc4dT7RX1PbRMllbVrl2bl19+mZdffpnLly/z448/sm7dOi5fvoxOp+OXX35h69atZGZmlriPCRMmqMlOnz59ePLJJwkJCcm3vbW1NU2bNi1RX3369GH8+PGsW7eOrVu30r9/f1566SW8vb05d+4c//3vf7l+/ToAH330UbUsrS5ERWDQG4i8FUtEaCS3wqLUJCc+n1tUNnbWePvUxLuRJ3UaeVLH1xOveu58+PTyAm9rudZ0wreVLEyojmJiQrl37yparRU+PvnvrygqnjJfS96kSRPmzZvHvHnzWLduHbNmzSIuLg5TB5Y2b96sPt67dy9t2rQpsH2DBg0IDw8vcX8rVqwgISGBHTt2sG/fPvbt25fjuFar5a233uLpp58ucR9CiKJLSUrLSmhCI4kIy0ps7oRHk5Ge9x9S7rVd8fapSZ1GnlkJjq8Xbl4uaLW56+aMmtWPFfN/zrfvUTP7oZUqytWScd+sevU6Y2MjI/KVSZknPFFRUaxfv57Vq1erWz1URnZ2dvz666+sWbOGgIAAzpw5Q1xcHF5eXnTv3p3nnnuOrl27mjtMIaocg95A9O24rFtR2ZKbe5F570FnbWuVNWrjY0xsPPFuWBNbh6IX+2zr34xpb4/MXYenphOjZvajrX+zAs4WVVV6ehLXrx8BpLJyZVQmc3iSk5PZvHkza9asYc+ePej1enVER6PR8NBDDzFx4kSeeeaZ0u66UpI5PNVPfhV8K7PSeE1pyTpuhUVmG7mJ4nZYFOm6jDzbu3k5493IE2+frFtS3r6eeNSukeeoTUlUxf9PouQuXdrBqVMrcXGpz8CBH0lV7Qqi3OfwZGZmsnPnTtasWcO2bdvUvaSMic4DDzzAxIkTeeyxx2jQQCb6ieorvwq+o2ZV3pGD4r4mg0Eh9k7cv5OI/xm5ibkTn+f1rWwsqd3wn9tR/8y58W7kib2jbZ7tS4vWQisTkwWQ9V525UpWfabGjftJslMJmZzwHDp0iNWrV7Nx40bu3bsH/Jvk1KlThwkTJjBx4kTatm1raleiiOSv0orLWMH3fnHRiayY/zPT3h5Z6ZKewl7T5NeH4F67Ro7l37fDotCl5r03nauHE96+/4zY/DNyU7NODfkeFmYVFfUXiYkRWFra0qBBd3OHI0rApISnQYMGalFBY5Lj4uLCmDFjmDhxIj179pQsuJxVxdGDqqIoFXw3ffUHPg/UQaPRoAAoClk/Wlk/X4ry73MKCij//Owp5Gif+9i/n/97E1v593rZ+/qnffa+7m9rvK5er2fD57sKfE2rPtqe5/OWVhbUauCRNcemkSd1/plz4+Cc/8a7QpiLcd+sBg38sbKS79HKyKQ5PFpt1l9cNjY2DB48mIkTJzJ48GC1MrEomtKaw5PfX9pGlXH0wKiij1opikJ6WgZJ8SkkJ6SSHJ+a83FCCneuRXM1pOyqjldk9k62NGjmnW3kpiae9dyxqED/D4XIT2pqHNu2PYui6Bkw4CNq1JDbnBVJuczh6d27NxMnTmT06NGlVj1ZlExRRg82f7Ob1l2bVKhEoSjMMWqVmZ5JUkIqyfEp//ybSnJCCkn//KsmNPGpartME4rY5UejMf4HNP88YXwu6x8NaEDDP/9m+zyr3b9FP43Hsq71b/uswzmva+wn17Fsn2s0GtJSdEWqODzmuQF07P1Aoe2EqIjCwvahKHrc3ZtIslOJmZTw7Nmzp7TiECYKDblR6P4/cVGJLJ+3GY86NbC2tsLKxhIra8usf20ssfrnOWub+45ZW2JlY4X1P59bWFoU2E9pKo05L3q9gZSEVJITco+65ExkUkmKyzqe3/ySwlhaWeDoYo+Dix0OzvY4utipnycnpHHwl+OFXuPZhRNo2q5y/FK9fOYaX762ttB2zm4O5RCNEKXPYDCoG4X6+spS9MqszOvwiPJR1H19Qo5dMbkvrVaDlY3VfQnRP0lRtgQpv4TJKp821rZWOZ63sNSy6es/Coxl/eLfSE5MIzUxLdcIjDHBSUlMK/HrdHCxx8HZDsd/EhgHFzscne1xcP3nX5d//3VwtsPa1irfeWsGvYGzhy8WWsG3cevKU8HXt1U9XD2cpCqxqLLu3DlFSko01taO1K/fxdzhCBOUasJz4sQJdu/eTUhICLGxsQC4ubnRqlUr+vXrR8eOHUuzO5FNUf+C7tSvFU41HMhIzyBDl5n1kW78N+u59PTsz2eoj40MBgVdanqJR0FKU3JCKus/21mktvZOtlmjLc52OUZg/k1k7HB0tVc/t3OwKdVJ91oLbZWr4FsVX5MQ2RmXovv49MTCQuanVmalUnjw3LlzPP3004VWUu7cuTNLliyhdWvZXTa70pi0bNAbmDf5m0L/0n5n5cwSvfkoipItMcqZIBk/T9dl/Hvsn8fpeSROOZKpbJ+np2XkuHZRvzOz9kGqmXME5p8RF0cXexxd7LFzsq0wE2TznJNUySv4VsXXJERyciTbt78IKDzyyGc4OdU2d0giD0V9DzU54dm9ezdDhw4lPT1dXZpuZWWFu7s7ADExMWRk/Fsl1cbGhu3bt6u7jwtZpZUXRVG4eCKMb97YUGjbyrhrdUVfdVYSVfE1iert7Nm1XLiwBS+v1vTq9aa5wxH5KOp7qEm/jaKjoxk7diw6nQ6NRsP06dP5888/SU5OJiIigoiICFJSUggKCuKpp57CwsICnU7H2LFjiYmJMaVrkQfj/j+uHk45nnet6VSpkh3IWgHUtH3DXK/lfpV1foixgm/H3g/QpG2DKpEYVMXXJKovvT6Tq1f3A7JvVlVh0hyezz//nPj4eKytrdmyZQsDBw7M1cbCwoIHH3yQBx98kNGjRzN06FDi4+P5/PPPmT9/vindizy09W9G665NqsRf2jI/RAhhLrduBaHTxWNnVwNv7w7mDkeUApPeKX799Vc0Gg3PPfdcnsnO/QYMGMDzzz+Poij8+uuvpnQtClCV/tKuSqNWQojKw1hZuVGjPmi1sqC5KjDp/2JYWBgAw4YNK/I5w4YNY9GiRVy9etWUrkU1UpVGrYQQFV98/E2ioi6g0Whp1KiPucMRpcSkhCctLau+iYND0YuKGdvqdDpTuhbVjOxaLYQoL6GhWUvRvb07Ym/vbuZoRGkx6U/kWrVqAXDq1Kkin2Ns6+XlZUrXQgghRKnLzEwjPPwgAL6+/cwcjShNJiU83bt3R1EUPvzwQxISEgptn5iYyEcffYRGo6F79+6mdC2EEEKUuuvXj5CRkYKjoxe1aknNuKrEpIRnxowZQNZcnh49enD8eP77BB0/fpyePXsSGhqa41whhBCiojBWVvb17YdGI/MEqxKT5vB069aNWbNm8fXXX3Pu3Dk6d+5My5Yt6dy5M56enmg0Gu7evcuff/7J+fPn1fNmzZpFt27dTA5eCCGEKC2xsaHcu3cVrdaShg17mjscUcpMXmv3xRdfYG9vz6JFizAYDISEhORIbgC1ArNWq+XVV1/lww8/NLVbIYQQolQZl6LXq9cFW9uSV70XFZPJ43UajYaFCxdy+vRpZs6cSZMmTVAUJcdHkyZNmDlzJqdPn1bn8AghhBAVRXp6MtevHwHA11cqK1dFpVZNqVWrVnz11VcApKenc+/ePQBq1KiBtbXsMCuEEKLiCg8/iF6fjotLfTw8mpo7HFEGTBrhadSoEY0aNeLLL7/M8by1tTVeXl54eXlJsiOEEKJCUxRFrb2TNVlZ7kJURSaN8Ny8eRO9Xk+7du1KKRwhhBCifEVFXSAh4RaWljY0bOhv7nBEGSmVwoN2dnalEowQQghR3oyTlRs06I6Vlb2ZoxFlxaSEp3PnzgC5VmUJIYQQlUFaWhy3bgUBUlm5qjMp4Zk5cyaKovDZZ5+RkZFRWjEJIYQQ5eLq1X0YDHrc3ZtQo0ZDc4cjypBJCU+fPn2YO3cuZ86cYciQIdy4caO04hJCCCHKlMFgIDR0DyBL0asDkyYtz58/HxsbG1q3bs0ff/xBo0aN6NatG23atKFGjRpYWFgUeP7bb79tSvdCCCFEid25c5qUlGisrR2oV6+LucMRZUyjGMsgl4BWq82xfE9RlGIt59Pr9SXtukpJSEjAxcWF+Ph4nJ2luqcQQpSHgwc/4vbtUzRtOpj27SebOxxRQkV9DzW58OD9+ZIJ+ZMQQghRLpKTo7h9+zQAjRv3NW8wolyYlPAYDIbSikMIIYQoN1lzdxS8vFrh5ORt7nBEOTB5Ly0hhBCiMtHrM7l6dR8gk5WrE0l4hBBCVCu3bgWj08Vja1uDOnU6mjscUU4k4RFCCFGthIZmVVb29e2DVltqe2iLCq7U/k/HxMTwww8/sHv3bs6dO0dsbCwAbm5utGrVin79+jFlyhQ8PDxKq0shhBCiWBISbhEZ+RcajYZGjXqbOxxRjkol4fniiy948803SU5OBnKu1Lp16xYRERH8/vvvvPvuu7z//vu8+OKLpdGtEEIIUSxXrmTtiu7t3RF7e/kDvDoxOeF57bXXWLRokZrkuLq60r59e7y8vAC4e/cup0+f5t69eyQnJzN79myuXbvGokWLTO1aCCGEKLLMTB3h4QcBmaxcHZmU8OzatYtPP/0UgLp16/Lpp58ycuRILC1zXlav17N582Zee+01rl+/zueff87DDz/MgAEDTOleCCGEKLLr14+QkZGMg4MntWq1Nnc4opyZNGn5iy++AMDb25s///yTsWPH5kp2ACwsLBg7dizHjh3D2zur3sHnn39uStdCCCFEsYSGZt3O8vXth0Yja3aqG5P+jwcFBaHRaJg7dy61a9cutH2tWrWYO3cuiqIQHBxsStdCCCFEkcXGXiU2NhSt1hIfn17mDkeYgUkJj3GScufOnYt8jrGt8VxTREZGsn37dt5++20GDRqEh4cHGo0GjUbD1KlTTb6+0bvvvqtet7CP/fv3l1q/QgghSseVK1lL0evW7YytrexZWB2ZNIenbt26XLlyBZ1OV+RzjG3r1KljStcA6sRoIYQQIj/p6clcv34EgMaNZbJydWVSwjN48GA+//xzdu7cSbdu3Yp0zo4dO9RzS1P9+vVp3rw5v//+e6le937nzp0r8LiPj0+Z9i+EEKJ4wsMPodfrcHGph4dHM3OHI8zEpITn1VdfZfXq1SxatIhBgwYVmvQcOXKEzz77jJo1a/Lqq6+a0jUAb7/9Np06daJTp054eXkRHh5e5glHq1atyvT6QgghSo+iKNkqK/dDo9GYOSJhLibN4fH29mbHjh3UqlWLvn378tJLL3H69OkchQcVReH06dO8/PLL9OnTh1q1arFz585SuaU1b948hgwZIre2hBBC5Ckq6m8SEm5haWlDw4bdzR2OMCOTRngaNWoEQEpKCunp6XzxxRd88cUXWFtb4+bmhkajISYmhvT0dCAr+UlJSWHMmDH5XlOj0RAaGmpKWEIIIQTw775Z9ev7Y2Vlb+ZohDmZlPCEh4fn+Nw4sqPT6bh9+3ae50RGRhZ4TRluFEIIURrS0uK4efNPABo37mfmaIS5mZTwTJkypbTiqDQGDBjA6dOniYuLw9XVlQceeICHH36YGTNmUKNGDXOHJ4QQ4h9Xr+7HYNDj5taYGjVkQUl1Z1LC8/3335dWHJXGH3/8oT6OioriwIEDHDhwgI8++oiAgACGDx9uxuiEEEIAGAwGrl7dA8hSdJGlVHZLrw5at27NiBEj8PPzw9vbm4yMDC5evMjq1av5/fffiYuLY/To0Wzbto1BgwYVeC2dTpejdlFCQkJZhy+EENXKnTtnSE6OwtragXr1upo7HFEBSMJTBC+99BLvvvturuc7d+7M448/zpIlS3jmmWfQ6/VMnz6d0NBQbG1t873eggULmDdvXhlGLIQQ1ZtxsnLDhj2xtLQ2czSiIpDd04rA1dW1wOMzZszgySefBCAiIoJNmzYV2H7u3LnEx8erHzdu3CitUIUQotpLTo4iIuIUIJOVxb9KbYQnJiaGo0ePcvXqVRITE9Hr9YWe8/bbb5dW92Y3Y8YMli9fDsCBAweYOHFivm1tbGywsbEpr9CEEKJauXp1L6Dg5dUKJydvc4cjKgiTE547d+4we/ZsNm3aRGZmZrHOrUoJzwMPPKA+vnXrlhkjEUKI6kuvz/wn4cmqrCyEkUkJT1RUFA899BDXrl3LUV25OpL6QUIIYX4REcdJS4vH1rYGdeo8aO5wRAVi0hyed955h/DwcBRFYezYsezdu5eYmBj0ej0Gg6HQj6rkr7/+Uh97e5t3CNVgMHA+8hKB145zPvJSlftaCyFEfq5cyZqs3KhRb7RaWZcj/mXSd8P27dvRaDRMnjyZgICAUgqpclqyZIn6uGfPnmaLI+jmaQJObiQ2NU59zs3OlakdxuBXt53Z4hJCiLKWkHCLyMjzaDQafH37mDscUcGYNMITFRUFwLRp00olGHMICAhAo9Gg0WjyXHp+7tw5rly5UuA1li5dyrJlywCoVasWI0eOLItQCxV08zSLApflSHYAYlPjWBS4jKCbp80SlxBClCWDwUBk5HlOnVoJQK1a7bG39zBzVKKiMWmEx9vbm/DwcBwcHEornmI5fPhwjmQkOjpafXzlypVco05Tp04tdh8nTpxg+vTp9O7dm0GDBtG6dWvc3d3JzMzk77//VgsPAlhYWLB06VKzfD0MBgMBJzcW2OaHkxt50LsNWq1UIxBCVA03bwZx8mQAqamx6nMxMZe5eTOIunX9zBiZqGhMSnh69OhBeHg4586do2PHjqUVU5EtW7aMH374Ic9jgYGBBAYG5niuJAkPgF6vZ/fu3ezevTvfNu7u7ixfvpyhQ4eWqA9TXYi+kmtk534xqXFciL5CS8+m5ROUEEKUoZs3gwgMXJTr+fT0RAIDF9Gt22xJeoTKpD/1X331VaysrPj0009JS0srrZgqlEceeYTly5czffp0OnbsSN26dbGzs8PW1hZvb28GDRrE559/ztWrV826j1ZcatG2pyhqOyGEqMgMBgMnTwYU2ObkyR9k0YZQaRQT15OvXbuWKVOm0KVLF5YtW0bTpjJ6UFwJCQm4uLgQHx+Ps7Nzia5xPvIS7+37X6Ht3ur9gozwCCEqvcjI8+zb916h7Xr3fgtPz5blEJEwl6K+h5q8Zm/ChAk0adKEwYMH88ADD9CmTRuaNm2Kvb19gedpNBq1MrEwXQuPxrjZuRZ4W8vdzpUWHo3LLyghhCgjqYXcwi9uO1H1mZzwXLp0idmzZ6sThs+cOcOZM2cKPEdRFEl4SplWq2VqhzEsClyWb5vH24+WCctCiCrB2tqpSO3s7FzLNhBRaZiU8Fy/fp0ePXoQFRWlVlp2cnLC1dVV3ljNwK9uO2Z3m56rDo9RvC6p/IMSQohSlpGRwoULvxTazs7OHQ+PFmUfkKgUTEp45s+fT2RkJFqtlldeeYVZs2bRsGHDUgpNlIRf3XY86N2GC9FXiEtNwNXOmSsx11h7dgs/nNpIoxr1aOze0NxhCiFEiaSm3uPgwY+IiwtHq7XCYMjIt22HDlPkj2+hMinh2bNnDxqNhhdffJGFCxeWVkzCRFqtNsfE5AdqNiE0Npygm2f47MhyFgx4HWcbRzNGKIQQxZeYGMGBAwtITo7C1taF7t3nkJISlasOj52dOx06TJEl6SIHkxKeu3fvAjB69OhSCUaUDY1GwzOdJnE9LoI7SVF8cTSAuT1myV8+QohKIyYmlEOHPkSnS8TRsRY9e87F0dELNzcfvL0fJDr6AqmpcdjZueLh0UJ+v4lcTPqOqF27NgDW1talEowoO/bWdrzS7SlsLKw5d/dvNp7fYe6QhBCiSG7fPs2+ffPR6RKpUaMRffvOw9HRSz2u1Wrx9GxJgwbd8PRsKcmOyJNJ3xX9+/cHIDg4uFSCEWWrnqs3Tz04AYDNf/3GyYgQM0ckhBAFCws7yKFDH6PX66hVqw29e7+Nra2LucMSlZDJlZYdHBz46KOPiI2NLfwEYXb+DTsxoHF3AL76cyWRSdGFnCGEEOVPURQuXNhKUNDXKIqeBg388ff/D1ZWtuYOTVRSJiU8jRs35ueffyYxMZFu3brxxx9/lFZcogxNbjcKX7cGJKen8NmR5aTr81/lIIQQ5U1RDJw+vZKzZ9cA0KzZEDp3noWFhcml40Q1ZtLWEn369AHg1q1bXL58GY1Gg6urK02aNClSpeU9e/aUtOsqpTS2liiu6JR7zN31IYnpyfRp9BBPd3qsXPoVQoiC6PUZ/Pnn19y4cRSAdu0m06zZYDNHJSqyor6HmpTwaLVaNBoNAEW9jEajUSst6/X6knZdpZgj4QE4e+cCCw58jYLCjE4T6d2oa7n1LYQQ98vISCEwcBF374ag1Vrg5zeLBg26mTssUcGVy15aPXr0UBMeUfm0qdWCsa0GsyFkOytObqBhjbr41Khn7rCEENVQamocBw9+SFxcOJaWtnTrNptatdqYOyxRhZi8W7ownblGeAAMioGPDy3h1O3zeDq488GA13G0Lvh2pBBClKbExNv/FBSMxMbGhR495uDm5mPusEQlUdT3UClWUM1pNVqe7TKFmg7uRCbH8PWfKzEoBnOHJYSoJmJiQtmz522SkyNxdPSiX795kuyIMiEJj8DR2p7ZD03HSmvJyYgQtlyQ1XZCiLKXu6DgfBwda5k7LFFFScIjAPBxq8cTHccBsCFkO+fu/G3miIQQVVl4+P0FBd+SgoKiTBVp0vL169fVx/Xr18/z+ZLIfi1hfn0aPcSl6DD2hx3lf8cCWDDgdTzsa5g7LCFEFaIoChcvbufMmdUANGjgT6dOz0iNHVHmijRp2cLCIquxRkNmZmau50vU8X3Xqs7MOWn5fumZ6by9ZxHhcTdp7NaQd/u8hKX8IhJClIKsgoKruHRpJwDNmg2mbduJaDRys0GUXKlOWlYURf3I7/mSfIiKx9rSmpe7TcfByo4rseGsOr3Z3CEJIaoAvT6DY8e+VJOdtm0n0a7dZEl2RLkp0p/u33//fbGeF5Wbl6MHs7o8zseHlrDrykGauPvg37CTucMSQlRSuQsKzqRBA39zhyWqGanDUwFUpFta2a0/t42f/9qFjYU17/d7lXqu3uYOSQhRyUhBQVHWpA6PMNnYloNp7dUMnT6dRUeWkZKRau6QhBCVSGLibfbseZu4uHBsbFzo3fttSXaE2UjCI/Kl1Wp5vstU3OxcuZ0YybdBq2XulRCiSLIKCr5zX0HBRuYOS1RjkvCIAjnbOvHyQ09iobUg6OZpfr2419whCSEquNu3z7B//3x0ugRq1PChb995UlBQmJ0kPKJQTTx8eLzdKADWnN3ChcgrZo5ICFFRhYcf4tChhWRm6vDyak3v3m9ja+tq7rCEkIRHFM2Axj3wb9AJg2Lg86MruJcab+6QhBAViKIo/P33Nv788ysURU/9+t3o3v11rKzszB2aEIAkPKKINBoN0x8cT13n2sSlJfD50RVkGvTmDksIUQEYCwoaqyc3bTqYLl2elerJokKRhEcUma2lDbO7TcfO0pa/o0JZe3aLuUMSQpiZXp/5T0HBHUBWQcH27aWgoKh45DtSFIu3sxfP+E0E4NeLezl245SZIxJCmEtGRgqHDn3E9etH0Ggs6Nz5OZo3H2LusITIkyQ8otg612vPkGZ9AVgStJqIhLtmjkgIUd7S0uLYu3c+d++ew9LShh49/kPDhlI9WVRckvCIEpnQZhgtajYmNTONRYHLSMvUmTskIUQ5SUy8w+7dxoKCzv8UFGxr7rCEKJAkPKJELLQWvND1CVxtnbmZcJvvgtdKUUIhqoHY2FD27Hmb5ORIHBw86dt3Pm5uvuYOS4hClXnCs23bNiZPnsygQYOYNWsWJ0+eLOsuRTmpYefCiw9NQ6vREnj9OL9fOWjukIQQZejOnTPs2/dvQcF+/ebj5CQFBUXlYFLCs2/fPjw9Palfvz5xcXG5jr/11luMGDGCNWvW8Pvvv7NkyRK6dOnCqlWrTOlWVCAtajZmYtvhAKw8vZnL0WFmjkgIURbCww9z8KAUFBSVl0kJz44dO4iOjqZTp064urrmOHb27Fk++OADFEVBURRcXV1RFIXMzExmzJhBeHi4KV2LCuSRpn3wq9sOvUHPZ0eWk5CWaO6QhBCl6O+/t/Pnn1/+U1DwISkoKColkxKew4cPo9Fo6NevX65j33zzDYqiUKNGDU6cOEFMTAxBQUG4ubmh0+n49ttvTelaVCAajYZn/Cbi7eRFbGocXxwLwGAwmDssIYSJFMXAqVOrOHPmRwCaNn2ELl2ek4KColIyKeG5ffs2AC1btsx1bPv27Wg0Gp577jnat28P/9/encdFVe4PHP/MMCwDyC4Kgrgg7iluabhc91vqzTJbvOaSZpptZnmre0tbvd3Kmy1W5trPm1q2mNpiuaZiLmBuuG8gKKDsDDDMnN8fyAjCDMM6zPB9v17z4nDOc87zPRyY+fKc53kO0KNHD5544gkUReG3336rTtWinnF31jIraiquTi4cuXqSr49tsnVIQohquDmhYNHfctGEghNkQkFht6r1m5uSkgJQ5nbW2bNnuXz5MgD33HNPqW39+vUzlRGOJdQ7iEd7PgTAd8d/4WDiERtHJISoCr1ed8uEgjNlQkFh96qV8BQPQ87IKP0gyd9//x0Ab29vunbtWmqbv78/ALm5udWpWtRTfcN6Mjy8PwCL9n7B1exUG0ckhKiMvLx0tm27OaFgv35zaNGin63DEqLaqpXwNG1aNBwxLi6u1PpffvkFgKioqDL75OTkAODr61udqkU99nDXe2nj34IcvY7/7l5CQWGBrUMSQliheELBtLTzpgkFg4JkQkHhGKqV8PTu3RtFUfjkk09MLTbnzp1j/fr1qFQqhg4dWmafU6dOATeTJeF4NE4anrljCo1cPbmQnsCymK9tHZIQogIyoaBwdNVKeKZOnQoUDUHv1KkT9913H7179yYvLw+tVsu4cePK7LNzZ9HkdBEREdWpGoDk5GQ2btzIK6+8wp133klAQAAqlQqVSsWkSZOqffzyrF69mmHDhtG0aVPc3NwICwtj/PjxREdH10p99srf3Zen+kxGpVKx/Xw0W8/tsXVIQgjAaDSSnHyMixd3k5x8DKPRWGpCQR+fFgweLBMKCsdTrbGFgwYN4umnn2bhwoVcuHCBixcvmvr1vPPOOwQEBJQqn5eXZ2r96d+/f3WqBqBJkybVPoa1dDod9913Hz/++GOp9ZcuXeJ///sfq1ev5pVXXmHu3Ll1FlN917lJW+7vNJK1Rzaw/OBXtPQJpaVfqK3DEqLBSkjYR0zMCnS666Z1Li6eFBTkAkaaNOlEVNSzODu72y5IIWpJtccX/ve//+WHH37g4YcfZsiQIUyYMIHffvuNGTNmlCn7ww8/4OXlRfPmzRk1alR1qy6lefPmDBs2rEaPWdIjjzxiSnYGDhzI999/z759+1i6dCmtW7fGaDQyb948Fi9eXGsx2KO72w+le3An9MZCFuxZQnZ+jq1DEqJBSkjYx+7dC0olOwAFBdmAkYCAtvTr94IkO8JhqRQ7fuLj3Llz6dmzJz179qRJkyZcuHCBli1bAjBx4kRWrFhRI/Vs3bqVwYMHAzBq1Ci+++47nJycTNtTU1Pp3r07ly5dwsfHh3PnzlWqU3ZmZibe3t5kZGTg5eVVIzHXJ9kFuby0+T8k56QSGdSR5/s9hlrm8hCizhiNRjZufKJMslOSVuvPyJEfolbL36awL9Z+htr1b/arr77KyJEja/3W1rvvvguARqNh0aJFpZIdgICAAN5++20A0tPTWbJkSa3GY288Xdx5Nmoqzk7OxCYd4/vjm20dkhANSmpqnMVkB0Cnu0ZqapzFMkLYs1pJeAoLC0lJSSElJYXCwsLaqKLOZGVlsWXLFgCGDBlCSEhIueXuvfdeU2b53Xff1Vl89qKFbwhTut0PwNdHN3H4iryxClEXijolWzcJqE6XXrvBCGFDNZbwHD9+nKeeeooOHTrg5uZG06ZNTSOZ2rdvz5NPPsnRo0drqro6s3//fgoKiuaRGTBggNlyLi4u9O7d27SPXq+vk/jsyV9a9WFQqztQUPgwegWpOZb/4xRCVI2iGElJiePgwWVs2DCDuLjvrdpPq/Wp1biEsKVqJzxGo5HZs2fTpUsXPv74Y06cOIHRaDQ9Jd1oNHLy5EkWLVpEZGQks2bNsqsHSx4/fty03K5dO4tli7cXFhZy+vTpWo3LXk3qNpaWvqFkFeTw3z1L0RskMRSiJiiKwrVrp4mN/YING55g69ZXOXNmM3l5GTg7e+Dk5Gpxf63Wn4CA9nUUrRB1r9qPvB03bhxff/21aTh6x44d6dWrl6lfzdWrV9m/fz9Hjx7FYDDwwQcfkJiYyNq1a6tbdZ1ISEgwLZu7nVUsNPTmkOv4+Hg6dOhQqbryCo24FJZNBtUqcHFSlypnTnXK5huMmOvCrlKBaxXLFhiMGE1lnXj89keYt+Udzl6/yPKYb5jW80EzZcty05g7bvXKujoVzd8EoDcaMVjIyStT1sVJhfpG2UKjQqGFIOpDWWe1Cid15csajAp6C2U1ahWaKpQ1KgoFhvpd1kkNzjc6+iqKQn4NlbXmb1lRFDIyLpIYH03i5b3k5KTcPAdnd4KCe9AspA+NAztyJSmG/XvfN1tfpy4PU2AESvxDapv3iLJq6+9e3iMqX7Y+vkdY+pwrtZ9VpcxYs2YNX331FSqVii5durB48WJ69uxZbtn9+/czffp0YmNjWbduHWvWrOHBBx8st2x9kpWVZVr29PS0WNbDw8O0nJ2dbbZcfn4++fn5pu8zMzMBeHL7GZw9ytbRJcCD57vfTKZmbj9t9o2yna+Wf/UKM30/a+dZsvSGcsu29HLj9T4tTN//Y9c5UvPK73PVzMOFt/u2Mn3/SvQFLueU/8iIADcN7w8IN33/+r5LnM/MK1XG6DII9BvZem4X7Ru3ol+LXgD852A8J9J05R7X1UnF0iFtTd8vPHSZP1PND3NfNfxmi9ynR5LYdzXLbNklgyNw0xT9AS07doXfEzPNll00MBwvl6I/nf+dSOa3+HSzZf/bvxWNtS4AfHU6hR8vmL+N9++oloR4Fv0Xvv5cKt+dvWa27Ku9w2jtrQXg54vXWXMqxWzZl3qG0sGv6HdzW0I6K+Oumi07u1sIkY2Lfgf3JGWw+OgVs2Wf7BLM7U2L+q0dSM7iwz8TzZad1qkp/Zv5AHD4Wg7vxSSYLTuxfROGNi8a5XgiLZe39sebLftgRGNGtix6Pt/5zDzm7r1otuw9rf0ZE94YgMScAl7Yfd5s2bta+DGubSAA1/L0zNp5zmzZIaE+TOpQNElflt7A49vOmC3bL9iLxzoHA5BvUJi65ZTZsr2aNOKprs1M35cs62lIplnBUYL1R/E03vyd0mhcCQ7uwfq05iSqW2PM0EAGcOwc4ENT9/vpmv8Lzoabzz/MU3tzxG04G074wInS8djyPaJYI2cnPhnUxvS9vEfIe8St7xGvbjf/N1dStRKe4jlnIiIi2LVrV6kP/Fv17NmTnTt30qNHD06ePMlnn31mFwlPXt7NP0IXFxeLZV1dbzYZ63Tl/0ECzJ8/n1dffbX6wdkxtXMLFNeeKPn7+fzAasJ8mtHcp1nFOwrRQHkYUgnWHyO44ChexpsfXgY05Ht0YEiXQQQFRaLRuLLkt5MYy/mn6IpLBy4ERjKtZS46XTparQ9zj2jIKrTb2UmEsFq15uHx9/cnPT2dpUuXWv0ohxUrVvDII4/g4+PD9es122m1NubhmTlzJosWLQKKHpJqqR/PJ598wuOPPw7AunXrGDNmTLnlymvhCQ0N5eq1tHLnEHCsW1o3GRUjC3Z9ytHkEzT1bMxbQ+egcXKV5moblq2PzdUN+ZZWgS6V+PhoLl2KJj39ws1tag2BTbrQLKQ3TYK64eqidcj3iGJyS6v+lK2P7xGpaRk08fetcB6earXwFI9euu2226zep7isvYxiatSokWnZ0m0quPkkeLB8+8vV1bVUa1AxN4261B+gOdaUqUrZkm9ANVnWxWxZNU/dMYkXN7/NlewUPtm3imejppreKKp+3OqVdVarcbayeGXKlvwjdbSyTiXe2GqyrFqlMt1GsIeyqhoom5t7jfj4vVy6tIfr18+WKO9EkyadaN78Dpo164GLi/kWdcd5j6ifZeU9ovJla/c9wrofcLUSnrCwMOLi4sjIyKi48A3F/VXCwsIqKFk/lOyonJCQQI8ePcyWjY+/2degZAdmYZ6Xqyez7pjCvK3vs//yn2w8uYVR7YbYOiwh6pROl05Cwl4uXYomNfWkab1KpSIwsCOhoX0ICemFq2sjC0cRQlhSrYRnzJgxvP7663zzzTcMHDjQqn3WrVuHSqXinnvuqU7VdabkSKsTJ05YLFu8XaPR0KZNG4tlxU3h/i2YGDmGpQfX8uXh9bTya07HwAhbhyVErcrPzyQhYR+XLkWTknKckr0LGjduR2hoH0JDb8fNzcd2QQrhQKqV8Dz77LP873//47PPPqN///7cf//9FsuvW7eOzz77jJYtW/Lcc89Vp+o607NnT1xcXCgoKGDHjh288MIL5ZYrKChg7969pn2cnZ3rMky7N6R1X06lnuf3i/v4IHo584f9Az+ZBE04mIKCHC5f3s+lS9FcvXoERbnZucPPL5zmzfsQGtobd3d/G0YphGOqVsLj7e3Nb7/9xgMPPMBDDz3El19+yaRJk+jZsyeBgYGoVCrTPDwrV67khx9+oEePHnz11Vd4e3vX1DnUqkaNGjF48GB++uknfvvtNxISEsqdj+fbb7813a6zl9ar+kSlUjG1x4NcTE/gUkYiC/cs5+WBT6FRO1W8sxD1mF6vIzHxIJcuRXPlyp8YjTeHdfv4tKB58zto3rw3Hh6BNoxSCMdXrVFaJR+iqShKhZ1NrSmjUqmq/PytqozSWrFiBZMnTwaKnr4+b968MmVKPi39b3/7G99++608Lb2WJGUl89Kv/0Gnz+OuiIFMiCx/pJsQ9VlhYT5JSbFcurSHpKRYDCVmFPfyCjElOY0aBdswSiEcg7WfodVq4bk1V7Imd6pGflXGrl27OHPm5oRDqamppuUzZ86USXisHTp/q0GDBvHggw+yZs0afvjhB4YOHcozzzxDcHAwR44c4c033+TSpUsAvP3225VKdkRpQY0CebzXw7y3+3N+PLWNiICW9A7tZuuwRANlNBpvPGm8aM6agID2qNXljwgxGPRcufInly7tITHxIIWFN6eeaNQoiNDQPjRv3gdvbxnQIIQtVCvhmTt3bk3FUSVLlixh5cqV5W7bvXs3u3fvLrWuqgkPwLJly8jMzOTHH39k27ZtbNu2rdR2tVrNyy+/zLRp06pchyjSM6QLo9oNYcOJ3/h03/8I9Q4myDOQuNQzpOsy8dF60T4g3OwHjxA1ISFhHzExK9Dpbs4XptX60a3bJEJCimYGNxoLuXr1CJcuRXP58gH0+lxTWQ+PxqYkx8enhdXTLQghaoddJzx1SavVsmnTJr788ktWrFjBn3/+SXp6Ok2aNKFfv3488cQT9OnTx9ZhOowHO4/izLULxKWc4a3tH2FEIU13c/oDP60Pk7rdR6+QrrYLUjishIR97N69oMx6ne46u3cvoGPHMeh0aSQk7KOg4Ob8XFqtH6GhvWne/A78/FpLkiNEPVKtPjyiZkgfnvKl6zKZ/dPr5OjNP6bj2aipkvSIGmU0Gtm48YlSLTuWuLl5ExLSm+bN+xAQEIFKJS2PQtSlOunDI0Rt8nL1xKmCUVorY9bRI/g2ub0laoTBUEh8fLRVyU5QUDfatr2Lxo07yO+fEHagRhMevV5PTEwMR48eNT0ny8/Pj06dOtGtWzeZm0ZUSlzqGTLzLT/O45ounbjUMzJRobCK0WhAp7tGdnYKOTkp5OQkk5OTQm5uCtnZyeh0aYB1jd5hYVE0adKpdgMWQtSYGkl4cnNzef311/n8889JS0srt4yvry/Tpk3jX//6F+7u7jVRrXBw6brMGi0nHJ/RaCQv77opoSlOZG4mNNdLTfZXHrVaU2quHHO0MjGmEHal2gnPpUuXGDJkCGfPnrU45Pz69eu8/fbbfPPNN2zZsqXcyfuEKMlHa11/JmvLidpVmSHcVaUoRvLy0snJSSE7u3RCU5TgpGI0GiweQ612xsMjAA+PQDw8Gt/yCsTZ2ZNNm560eFtLq/UnIKB9jZ6bEKJ2VSvh0ev13Hnnnaa5cNq1a8fkyZO5/fbbadq0KQBXrlxh3759rFixguPHj3P69GnuvPNOYmNj0WikC5Ewr31AOH5aH67r0s2W8df60D4gvO6CqiF1kRzUJWuGcFtDURTy8zNu3G4q+Uq+8TUVo1Fv8RhqtRPu7gGmBObWpMbNzafCjsXduk0qd5TWze0T7fp6CdEQVWuU1ieffMLMmTNRqVS89NJLzJs3r9QMxCUZjUbmzZvHG2+8gUql4uOPP2b69OlVDtyRyCgt8/YlHGLB7iVmt3dq0pYX+z9eYefm+qSmkoP6wtwQ7mJRUc+azktRFAoKsiwkNCkYDAUW61Op1Li7+5dKaNzdG+PpWZzQ+NVIMlL+dfKnW7eJdnmdhHBU1n6GVivhGTRoEDt27GD06NF88803Vu0zZswYvvvuOwYOHMiWLVuqWrVDqemEx9FaD/YlHGLFwa9R56XghkIeKnQaL3IK8wCIDOrI030ewc3Z1caRVqwyyYE9sGYIt0bjRuPGHUy3nQpvXDfzVGi1vnh6Bt5IZIq+engUJTVarT/qOkpwHe1vSQhHVCfD0o8ePQrAI488YvU+U6ZM4bvvvuPIkSPVqVqY4WitBwDBFPBXVRo6bnZO1jpr8Ar/C8tORxObdIzXti3kH/1n4O3WyIaRWmY0GomJWWGxTEzMCho3boeiKBiNBozGQhSl8MZy0avo+8IS3xtKfF9Yosyt282vu/UYRd8bKtxeWFiAwWA5gSkszCMpKabUOjc331v60dy89eTuHoCTU/243a1WqwkM7GjrMIQQNaBaLTyurq4UFhZy4MABIiMjrdonNjaW7t274+LiQl5eRf/pNQw11cLjaK0HUPE5tez8dz4/uYusghwCPQJ4ccDjBDWq3FOnjUYjRmMBBoMeg0GP0agvsXzr+gIMhkIz60vvf/M4Rdvy87PIybla3R+JXWrZ8i+Ehva5kdQE4OTkYuuQhBAOok5aePz8/EhOTub8+fNWJzznz5837StqjnWtBysJDu5hdZO8oig3XkbAaFo2t67o+1vXKRiNVdvfaDSwf/9iizEmnPiWSaFR/HHpIPk5F1j980t0CgzHQ+NSJnkxGArKSUb0KIrlUT22olI5oVY7oVZrUKudUKk0N753urFNY9pW3rqS+9/cdus6c8fX4ORUskzJ7TfLp6WdZ9++RRWeS4sW/aSlRAhhU9VKeLp168ZPP/3Exx9/zL333mvVPosWLUKlUlmdIAnrFPUzsDw7rE53je+/fxS1WmNVwmEP9HodCed+o1nxCiOkXTlE+bNBVUylcsLJyRknJ2fUamecnFxQqzU4Obncsr54263rzK/Pykrkzz+/rDCGAQNeokmTznbxHCYvr2YcObJGhnALIeo9qxMetVqNWq3m8OHDdOjQAYCHHnqIn376ie3bt/PII4/w4Ycf4uHhUe7+ubm5PPXUU2zduhWVSsW4ceNq5gwEADoLQ7dL0utzarxulUp1Y5iv2rRc9GGtNi3fXFe8rC6xX/nrCgqyyclJrrD+pk1vw8cnDAU10Zf/JCErBSMqosJ60TmoY5nEQ612KZGcOJdKTmqzM6zRaOTUqZ8rTA4CAzvZRbIDRe8LMoRbCGEPrO7Do1YXfRgdOXLElPAoikK/fv3Ys2cPKpWKxo0bc//993P77bcTGBiISqXi6tWr/PHHH3z11VekpKSgKAp9+/Zl586dtXpi9qQm+vAkJx9j27bXKyzXo8c0/P3b3EgsVLckGdYkLCXXFe9XO6w9p4EDXzbdLik0GlhyYA3bz0cDcE+Hv3J/pxH1JoFwxH5WIEO4hRC2U+PD0stLeADS0tIYMWIEe/fuLTqgmQ+W4mr69OnDxo0b8fX1tfpkHF1NJDzWDA/Wav0ZOfJDu/lvu6rnpCgK6479yDfHfgKgf4vbmdZzHJp6MlePoyYHMoRbCGELdfa0dF9fX3bt2sUnn3zCokWLiIuLK7dc+/btmTlzJtOnT5c3wVrgiLcWqnpOKpWKsZ1G4K/1ZcnBNey88AcZeZk8c8cUtM5utR12hUJCehEc3MPhkgMZwi2EqM+q3cJzq6SkpHKflh4UFFQzETugmpx40BFbD6pzTjGJR1m4Zxn5hgJa+obyj34z5NlbQgjhQOrslpaoPplpuWLVOaez1y7y9u+fkJmfTWMPf17s/zjBXk1qOWIhhBB1QRIeOyLP0qp9V7JT+PeORVzJTsHTxZ3n+02nbUArW4clhBCimmot4enRo4fZoeeVoVKp5FlaN0jCUzcy87J4+/dPOXv9Is5OzjzZexK9QrrYOiwhhBDVUGsJT01QFAWVSoXBUD9nuK1rkvDUnbzCfD6IXk5M4lFUqJjU7T6Gtxlg67CEEEJUkbWfoZXu2HHzcQNVfwlhK24aV2ZHPcrgVlEoKCyP+Zov/1yP0U5mlhZCCFE1lU54jh49euNhi9V7SeuOsBUntRNTezzI/Z1GAvDDiV9Z9Mf/UWgotHFkQgghaot9D90RoopUKhX3dvwr03uNx0mlZtfF/bz9+yfk6nW2Dk0IIUQtkIRHNGh/admbOf2m46px4cjVk7y69X2uW/lcMiGEEPZDEh7R4HUJ6sC8gbPwdmvExfTLvPLbeyRkJNk6LCGEEDVIEh4hgJZ+obw+eDZBjQJJzU1j7pYFxKWcsXVYQgghaogkPELcEOgZwKuDn6WNf0ty9Dre2v4Re+NjbR2WEEKIGiAJjxAleLl68vJfnqRns9vQGwtZuGcZP57aZuuwhBBCVJPVCc/58+c5d+4cERERtRmPEDbnonFh1h1TGRbeDwWFL2K/YdWh72SuHiGEsGMaawuGhYXVZhxC1CtqtZrJ3e7H392P1YfXs/HkFq7r0pnRazzOTs62Dk8IIUQlyS0tIcxQqVTc3X4oM2+fgJNKzZ5LB5m/cxE5Bbm2Dk0IIUQlScIjRAX6tejFC/0fR6tx43jyaeZt/S/XctNsHZYQQohKkIRHCCt0btqOuYOewdfNi/iMJF7+7T3i0xNtHZYQQggrScIjhJVa+Ibw2pDnaObVlOu6dOZu/S/Hkk/ZOiwhhBBWkIRHiEpo7OHHq4Nm0a5xa3L1OubvWMSeSwdsHZYQQogKSMIjRCV5unrw0oAnuD0kkkJjIR9Er2DTyS22DksIIYQFkvAIUQUuTs483Wcyf23zFwD+79B3rIz9RubqEUKIekoSHiGqSK1WMzFyDOO73APAT6e2sXDPcgoMehtHJoQQ4laS8AhRDSqVipHtBvNUn0k4qZ34IyGWt3Z8RLbM1SOEEPWKJDxC1IA7mvfgpf4zcXfWciLlLHO3LCA157qtwxJCCHGDJDxC1JCOTSKYN+gZ/LQ+XM68wstb3uNiWoKtwxJCCIEkPELUqOY+zXhtyGxCvIJI02Uwb+v7HLl60tZhCSFEg+cwCc/FixeZPXs27dq1w8PDAz8/P3r27Mk777xDbm71+lOsWLEClUpl1WvFihU1c0LCbgW4+/Lq4Fm0bxyOrjCPf+9cxK4L+20dlhBCNGgOkfBs2LCB2267jQULFnDy5Elyc3NJS0vjwIEDzJkzh8jISM6cOWPrMEUD4uHizksDZtIntBsGo4GP/ljJ+rhfURTF1qEJIUSDpLF1ANUVGxvLAw88gE6nw9PTkxdffJGBAwei0+lYs2YNn3/+OadOnWLEiBEcOHCARo0aVau+X375heDgYLPbQ0JCqnV84TicnZx5ss8k/Nx92HRyK6sPr+dabhqTIu8DIC71DOm6THy0XrQPCEetdoj/P4QQol6y+4Tn6aefRqfTodFo2Lx5M3369DFtGzRoEG3atGHOnDmcOnWK9957j3nz5lWrvoiICFq0aFG9oEWDoVapebjrvfi7+/J/sd+y+cxOzl6/SJouneu6DFM5P60Pk7rdR6+QrrYLVgghHJhd/0u5b98+fv/9dwCmTJlSKtkpNnv2bNq3bw/AwoUL0etlUjhR9+6KGMjTd0zGSaXm7PWLpZIdgOu6dBbsXsK+hEO2CVAIIRycXSc833//vWl58uTJ5ZZRq9VMmDABgPT0dLZt21YXoQlRRq9mXfFwcbdYZmXMOoxGeTyFEELUNLtOeHbt2gWAh4cH3bt3N1tuwIABpuXdu3fXelxClCcu9QyZ+dkWy1zTpROXKh3shRCiptl1whMXFwdAeHg4Go357kjt2rUrs09VTZ48meDgYFxcXAgICKB3797861//4vLly9U6rnB86brMGi0nhBDCenab8OTl5ZGamgpUPDLK19cXDw8PAOLj46tV7/bt20lKSkKv13Pt2jX++OMP3nzzTcLDw/nss8+sOkZ+fj6ZmZmlXsLx+Wi9rCrnWcFtLyGEEJVntwlPVlaWadnT07PC8sUJT3a25VsK5rRq1YrnnnuOb775hn379rFv3z7WrFnD2LFjUalU5OXlMX36dBYvXlzhsebPn4+3t7fpFRoaWqWYhH1pHxCOn9anwnKL969mX8IhmbNHCCFqkEqx03fV+Ph4mjdvDsDDDz/MF198YbF88+bNiY+Pp3Xr1pWehDAjIwMvLy9UKlW52zdu3Mi9996LXq/H3d2ds2fP0rRpU7PHy8/PJz8/3/R9ZmYmoaGhpnqE49qXcIgFu5eY3e7p4m560nrnJm2Z1G0szbzM/y4JIURDl5mZibe3d4WfoXbbwuPm5mZaLigoqLB8cYKh1WorXZe3t7fZZAdg5MiRvPLKKwDk5uaydOlSi8dzdXXFy8ur1Es0DL1CuvJs1NQyLT3+Wh+ejZrKR6Ne554Of8VZreHI1ZPM+fkt/i/2W3L1OtsELIQQDsJuW3jy8vJMycuIESPYuHGjxfKenp7k5OTQu3dvoqOjazye5ORkmjZtiqIoDB06lM2bN1u9r7XZqXAcRqPR4kzLV7NT+SL2Gw4mHgHA260R4267m34teqFW2e3/KUIIUeOs/Qy125mW3dzc8Pf359q1ayQkJFgsm5aWRk5ODkCt9ZcJDAzE39+f1NRUGbElKqRWq+kYGGF2exPPAJ7v9xiHko6zMnYdSVnJfLJvFb+d3cXkbvfTyq95HUYrhBD2z67/VezQoQMAZ86cobCw0Gy5EydOmJaLZ12uDZZuewlRFV2DOvDO8JcYd9vduGlcOX3tAv/89R0W7/+SzLysig8ghBACsPOEp2/fvgDk5ORw8OBBs+V27NhhWo6KiqqVWFJSUkzD5C09XFSIytI4afhb+6EsuOtl+ob1REFh67k9zPrxNX4+tR2D0WDrEIUQot6z64Rn9OjRpuXly5eXW8ZoNJpGcPn4+DBw4MBaiWXx4sWmYcQlZ3YWoqb4aX14ovdE5g2aRQufEHL0OlbEruPFzW9zPPm0rcMTQoh6za4Tnl69etGvXz8Ali5dWm5n5Pfee880u/LTTz+Ns7Nzqe3bt29HpVKhUqmYNGlSmf0vXLhAbGysxTg2btzIa6+9BhSNAjP3XC8hakK7xq15a+gcpnR/AE8Xdy5lJPLatoUs3LOMa7lptg5PCCHqJbvttFxs4cKFREVFodPpGDZsGC+99BIDBw5Ep9OxZs0a00SAERERzJ49u9LHv3DhAgMHDqRPnz6MGjWKLl26EBgYCMC5c+dYt24d69atM7XuvPvuuzRr1qzmTlCIcqjVaoaG96N3aCRfHdnEb+d2ER0fQ0ziUUZ3GM6ItoNwcXKu+EBCCNFA2H3CExkZydq1axk/fjyZmZm89NJLZcpERESwadMmGjVqVOV6oqOjLQ5nd3d357///S/Tpk2rch1CVFYjV0+m9HiAQa3vYEXM15xMPcfaIxvYfj6aCZFj6B7c2dYhCiFEvWD3CQ/AqFGjOHz4MAsXLmTTpk0kJCTg4uJCeHg4Y8eO5YknnsDdvWrPJ+revTurVq0iOjqaAwcOkJSURGpqKoWFhfj6+tKxY0cGDx7M1KlTTS0/QtS1lr6hzBs0i90XD/C/P7/janYq7/z+GZFBHZkQOYagRvK7KYRo2Ox24kFHIhMPipqk0+fx7fGf+fHUNgxGA05qJ0ZEDOLeDn/FzdnV1uEJIUSNsvYzVBKeekASHlEbEjOvsjL2G/68chwAX603f+8ymqjmPWTOKCGEw5CEx45IwiNqi6IoHEw8whex35KcUzRPVLvGrZkcOZYw3xAbRyeEENUnCY8dkYRH1LYCg56NJ7bwfdwvFBj0qFQqhrbuy/2dRuLp6mHr8IQQosok4bEjkvCIupKac51Vf37P3vgYADxd3Hmg8ygGt4oq9fBSIYSwF5Lw2BFJeERdO3b1FCtivyY+IwmAFj4hTO42lraNW9s4MiGEqBxJeOyIJDzCFgxGA5vP/M7XRzeRq9cB0DesJ3/vMhpfrbeNoxNCCOtIwmNHJOERtpSRl8Wawz+w/fxeFBTcNK6M6Xgnd7b5Cxonh5iqSwjhwCThsSOS8Ij64Oy1iyyL+Yqz1y8CENyoCRMjx9AlqIONIxNCCPMk4bEjkvCI+sKoGNl5/g9WH/6BjPwsAHo0u42Hu95LE88AG0cnhBBlScJjRyThEfVNboGOr4/9yC+nd2BUjDirNYxqN4S72w/DVeNi6/CEEMJEEh47IgmPqK/iM5JYEfM1x5JPAeDv7svDXe/h9pBIma1ZCFEvSMJjRyThEfWZoij8kXCIVYe+JTU3DYCOgRFM6jaWUO8gUzmj0Uhc6hnSdZn4aL1oHxAuc/sIIWqdJDx2RBIeYQ/yCwv44cSv/BD3K3pjIWqVmuHh/bmv010cSz7Fiph1XNelm8r7aX2Y1O0+eoV0tVnMQgjHJwmPHZGER9iT5OxU/u/Qt+y/fBgArcYNXWGe2fLPRk2VpEcIUWus/QyV9mYhRKUEegYwu+80XhwwkyDPQIvJDsDKmHUYjcY6ik4IIconCY8Qokq6NG3P5G5jKyx3TZdOXOqZOohICCHMk4RHCFFlWQU5VpVLykqu5UiEEMIymTdeCFFlPlrr+pwtObCGHef/oFtwJ7oFd6K5d7AMaxdC1ClJeIQQVdY+IBw/rU+p0Vm3clI5YVAMnL52ntPXzrP2yAb8tD6m5KdjYIRMZiiEqHUySqsekFFawp7tSzjEgt1LzG5/NmoqrfzCOJR4jJikoxy5ehK9QW/a7uzkTKfACLoFdyIyuBMB7r51EbYQwkHIsHQ7IgmPsHf7Eg6VmYfHX+vDxHLm4SkoLOBY8mliEo8Sk3SUazcmMyzW3LsZ3YI70i24E+F+LWTyQiGERZLw2BFJeIQjqMpMy4qiEJ+RWJT8JB7j9LXzKNx8S2rk6knXph3oFtyJ25q2w8PFvbZPQwhhZyThsSOS8AhRJDM/mz+TjhOTeJQ/r8SRq9eZtqlVato1bk23oE5EBnckuFET6fgshJCEx55IwiNEWYVGA6dSz91o/TlKYtbVUtubeAaYkp/2jcNxdnK2UaRCCFuShMeOSMIjRMWuZKcQm3iMmMSjxKWcodBYaNrmpnHltqbtiAzqRGRQR6uHywsh7J8kPHZEEh4hKkenz+PI1ZPEJB7lUNIx0vMyS21v7RdGZFBRx+cWviGoVdLxWQhHJQmPHZGER4iqMypGzqfFE5N4lNjEY5xLu1Rqu6+bF5HBRS0/nZu0w83Z1UaRCiFqgyQ8dkQSHiFqTpoug9ikY8QmHuPw1TjyCwtM2zRqDR0D25hafwI9Ayweqyojz4QQdUsSHjsiCY8QtUNv0HM85cyNvj9HSM65Vmp7iFdTIoM70S2oExEBLXFSO5m2lTe3kJ/Wh0nlzC0khLAdSXjsiCQ8QtQ+RVFIzLpqGvV1MvUcRsVo2u7hrKVLUAe6BXXCoBj4ZN8qs8d6NmqqJD1C1BOS8NgRSXiEqHvZBbkcvhJn6vicXZBr9b7+Wh8+HPma3N4Soh6w9jNUHh4qhGiQPF3cuaN5d+5o3h2j0cjp6xeISTxK9KWDZW593eqaLp3fL+4jqnkPNE7yNiqEPZAWnnpAWniEqD92XzzAh3tXWFVWpVIR6OFPU89AghoFEtSo8Y2vTfB395Hh8ELUAWnhEUKIKrB20kJntQa9sZCr2alczU7lzyvHy2xv4lmcABW9mt5IiLxdG8ljMYSoY5LwCCFECe0DwvHT+pQanXUrf60PH4x4lcyCbJKykku9rmSncCU7Bb2xkITMJBIyk8rsr3V2I8izdBIU1CiQIM9A3F20tXh2QjRckvAIIUQJarWaSd3uY8HuJWbLTOx2H05OTvhqvfHVetMhsE2p7QajgdTctJtJUFYySdnJJGWlkJpzHZ0+j3Npl8pMkgjg7daoxC2ym68mngG41MDzwmRuIdFQSR+eekD68AhR/5Q3D4+/1oeJ1ZyHp8Cg52p2alESVPzKLvqakZdldj8VKvzdfcu9RdbY3a/UHEKVOSeZW0jYOxmWbkck4RGifqrr1pBcvY4rWSllbpElZl1Fp88zu5+T2okmHgFlb5E1CsTXzRuVSsW+hEMWW61kbiFhryThsSOS8AghLFEUhcz87BJJUImEKKuov5A5rhoXmno0Jik7mQKD3mw5e55bSG7TNWyS8NgRSXiEEFVlVIxcy00v9xZZSs71UrNJV6SZVxMae/jj4eyOh4sWDxd3PJzdcXfR4lli2cPFHU9nd9ycXW0+9N5Rb9NJEmc9SXjsiCQ8QojaUGgoJDnnGlvP7WHjyS01fnyVSoW7sxYPZ3c8XbS430iKbiZLN766aG+sdzetd3dxR2NFvyNLHPU2nSRxlSPz8AghRAOncdIQ7NWEyOCOViU8YzuNwF/rQ44+l+wCHbkFueTodeQU5JJToCNHn3tjORe9sRBFUUzfJ+dUPj43jSvuzjdaj1zci5KnG8uepb4vmTAVLWtUTqyIWWfx+Ctj1tEj+Da7ahkxl8Rd16WzYPcSSeKqwSESnosXL/LBBx+wadMm4uPjcXV1pXXr1tx///3MnDkTd3f3Gqnnp59+YvHixezfv5+UlBQaN25Mz549mTZtGnfeeWeN1CGEEDXN2rmF7mk/3OrkoMCgNyU7N5Oi0su5N5azbyRMufqir7rCog7YeYX55BXmW4zLHLVKXeHtumu6dP67ZymBHv5onJxwUjnhpHZCo9agUatxUhUtO6mL1xe9itY7lVivQaNywkmtNn3vpHZCU6acU7Vu8RmNRkniapHd39LasGED48ePJzMzs9ztERERbNq0ifDw8CrXYTQamTZtGkuXLjVbZurUqXz22WdV+iWUW1pCiNpWn27/GIyGG8nQjeToRiJkKXnK0eeSW6AjR6+rVL+kuqZSqW4kQprSCZKqREJVnFQ5OZVan1Og40Tq2Qrr6B0aSWMPfwDUKhUqVKhUalQ36ler1DfWqW6sU5coByrURdtUKtSo4MZXS+uKj1defUX7qEFFiX3UqFFhxMiH0SvIzM82ez7V7TDfIPrwxMbGEhUVhU6nw9PTkxdffJGBAwei0+lYs2YNn3/+OVCU9Bw4cIBGjRpVqZ4XX3yRf//73wBERkYyZ84cWrduzdmzZ/nPf/5DbGysqdxbb71V6eNLwiOEqAu1NbdQXVIUBV1hHocSj/PB3uUVlu8b1hNfrTeFRgOGG69CYyEGxXhjXWHRV8VgKlPqa4n1xfsWKgYMRiOFFkbHicp5eeBTdAyMqNK+DSLh6d+/P7///jsajYadO3fSp0+fUtvfeecd5syZA8DcuXOZN29epes4deoUHTt2pLCwkB49erBz50602ptTv+fm5jJgwAAOHDiARqMhLi6u0q1JkvAIIeqKo4z+MRqNPLHxlQpv09XmUHtFUTAqxqJEqExidDNZKjQW3lhvLJEw3ZJcKQbiM5L46dS2CuvtE9oNf3dfFEVBQTF9NSoK3PhqeV1RC9nNdUYUhRvHMhatR7mxznjjPJUb+xhNdZZ37JJfjYpCjj6XNF1Ghef0ZO9JRIX1qNJ1cPiEZ9++fdx+++0APPbYY3z66adlyhiNRjp16kRcXBw+Pj4kJyfj7Fy5qdkff/xxPvnkEwCio6Pp3bt3mTJ79+41JVuPP/44H3/8caXqkIRHCCEqrz7dpqsJ9SGJq2nHkk/x+rYPKixXFy089vETK8f3339vWp48eXK5ZdRqNRMmTAAgPT2dbdsqzpxLUhSF9evXA9CuXbtykx2A3r1707ZtWwDWr1+PneaQQghhV3qFdOXZqKn4aX1KrffX+thdsgM3n+NmycRu99lNsgM3O8xb4q/1oX1A1fvZWstuR2nt2rULAA8PD7p372623IABA0zLu3fvZtiwYVbXcf78eRITE8scx1w9J0+e5PLly1y4cIGWLVtaXY8QQoiq6RXSlR7BtznEbTq4mcTZe1+rYtY+jLcurpfdJjxxcXEAhIeHo9GYP4127dqV2cdax48fL/c41tQjCY8QQtQNtVpd5dsh9ZEkcbXDLhOevLw8UlNTAQgJCbFY1tfXFw8PD3JycoiPj69UPQkJCabliuoJDQ01LVdUT35+Pvn5+abvzQ2pF0II0TBJElfz7DLhycrKMi17enpWWL444cnONj8PQHXr8fDwMC1XVM/8+fN59dVXKxWLEEIIYc9sncTZZftYXl6eadnFxaXC8q6urgDodLpaq6e4DmvqefHFF8nIyDC9KtvyJIQQQojKscsWHjc3N9NyQUFBheWLbx+VnD+npuspeYuqonpcXV1LJUhCCCGEqF122cJTcsZka25T5eQUPdXOmttfVa2nuI6q1COEEEKI2mWXCY+bmxv+/kXPESnZsbg8aWlppmSkZMdia5TsqFxRPSVvS1W2HiGEEELULrtMeAA6dOgAwJkzZygsNP88kxMnTpiW27dvX6U6bj1OTdcjhBBCiNpltwlP3759gaJbSQcPHjRbbseOHablqKioStXRsmVLgoODyxynPDt37gSgWbNmtGjRolL1CCGEEKJ22W3CM3r0aNPy8uXlPzHXaDTyxRdfAODj48PAgQMrVYdKpeLuu+8Gilpw9u7dW265vXv3mlp47r77blQqVaXqEUIIIUTtstuEp1evXvTr1w+ApUuXEh0dXabMe++9Z5pd+emnny7z4NDt27ejUqlQqVRMmjSp3HqeeeYZnJycAHjyySfLDDnX6XQ8+eSTAGg0Gp555pnqnJYQQgghaoHdJjwACxcuRKvVUlhYyLBhw5g/fz579+5l27ZtPPbYY8yZMweAiIgIZs+eXaU6IiIieP755wE4cOAAUVFRrF27lgMHDrB27VqioqI4cOAAAM8//zxt2rSpmZMTQgghRI2xy3l4ikVGRrJ27VrGjx9PZmYmL730UpkyERERbNq0qdQQ88p68803SU5OZtmyZcTGxvLggw+WKTNlyhTeeOONKtchhBBCiNpj1y08AKNGjeLw4cPMmjWLiIgI3N3d8fHxoUePHrz99tvExsYSHl69x86r1WqWLl3Kpk2buPvuuwkODsbFxYXg4GDuvvtufvzxR5YsWWK3D3YTQgghHJ1KURTF1kE0dJmZmXh7e5ORkYGXl5etwxFCCCHshrWfodIkIYQQQgiHZ9d9eBxFcSNbZmamjSMRQggh7EvxZ2dFN6wk4akHsrKyAHkkhRBCCFFVWVlZeHt7m90ufXjqAaPRSGJiIo0aNZJJCy3IzMwkNDSU+Ph46etUT8k1sg9yneyDXCfrKIpCVlYWwcHBFgcPSQtPPaBWq0s9qFRY5uXlJX/89ZxcI/sg18k+yHWqmKWWnWLSaVkIIYQQDk8SHiGEEEI4PEl4hN1wdXVl7ty5uLq62joUYYZcI/sg18k+yHWqWdJpWQghhBAOT1p4hBBCCOHwJOERDuXtt99GpVKhUqnYu3evrcMRt/juu+8YOnQo/v7+uLm50bJlSx566CHi4+NtHVqDpygK3377LQMHDiQoKAh3d3fatm3LY489xrlz52wdXoOzatUqHnvsMXr06IGrqysqlYoVK1ZUuN/+/fu566678PHxwcPDg969e/PVV1/VfsB2QIalC4dx9OhR5s6di4eHBzk5ObYOR5SgKArTp09n8eLFtG7dmgcffJBGjRqRmJjIjh07uHjxoky8aWPPPfccCxYsICgoiNGjR+Pl5cWff/7J559/zurVq9mzZw+dOnWydZgNxr/+9S8uXrxIQEAAQUFBXLx4scJ9tm3bxvDhw3FzczP9jX3zzTc88MADxMfHM3v27DqIvB5ThHAABQUFSrdu3ZTbb79dGT9+vAIo0dHRtg5L3PD+++8rgPL4448rhYWFZbbr9XobRCWKJSUlKWq1WgkLC1PS09NLbVuwYIECKJMnT7ZRdA3Tr7/+qly4cEFRFEWZP3++AijLly83W16v1yutW7dWXF1dldjYWNP69PR0JSIiQnFxcTEdr6GSW1rCIbz55pscO3aMZcuW4eTkZOtwRAk6nY5XX32VVq1asXDhwnKvj0Yjjc22dOHCBYxGI1FRUWUmcBs5ciQAKSkptgitwRoyZAhhYWFWl9+6dStnz55l3LhxdO3a1bTe29ubl156iYKCAlauXFkLkdoPeZcRNnXgwAF+/PFHdu3axfHjx0lJScHZ2Zng4GCioqKYMmUKffv2tXiMmJgY3nzzTV577TU6dOhQR5E3LNW5Tps3byYtLY3JkydjMBj44YcfOHXqFD4+PgwZMoTw8PA6PhvHVJ1r1KZNG1xcXNi9ezeZmZmlZvXduHEjAIMHD66T83AUNfHeVhnbt28HYNiwYWW2DR8+HIAdO3bUWH12ydZNTKLh6tevnwJU+JowYYKSn59f7jHy8vKUjh07Kj169DDdKpk4caLc0qpB1b1OL7/8sgIozz//vBIREVFqH7VarcyePdsGZ+VYauJvacGCBYpKpVKCgoKU6dOnK3PmzFGGDx+uODs7K48//rjcdqyEmrgeJVlzS+u+++5TAOXAgQPlbvf09FRCQ0OrekoOQVp4hM0kJiYCEBwczNixY+nXrx/NmzfHYDAQHR3Ne++9x+XLl/niiy/Q6/V8+eWXZY7xyiuvcPr0aQ4ePCi3smpJda9TcnIyAAsWLKBbt27s27eP9u3bExsby7Rp03jvvfdo3bo1M2bMqPNzcxQ18bc0a9YsmjVrxtSpU/n0009N6/v27cu4cePktmMl1MT1qKyMjAzA/DOlvLy8TGUaLFtnXKLhGjFihLJ27dpyO7EqiqKkpKSUahHYsWNHqe179uxR1Gq18tprr5VaLy08Nau61+nRRx9VAEWr1SqXL18ute3IkSOKWq1WWrduXWvxNwTVvUaKoiivvvqq4uzsrMyfP1+Jj49XsrKylN9//13p0aOHotFolPXr19f2aTiMmrgeJVnTwjN06FAFUE6fPl3u9uDgYMXLy8vqc3BEkvCIem3Dhg2mN4Unn3zStF6v1ytt2rRRunbtqhQUFJTaRxKeumfuOimKojz33HMKoPTr16/cfcPDwxVASUtLq4NIGy5L1+jXX39VAGXWrFll9ktKSlK0Wq0SHh5eV6E2CJaux63kllbNkFFaol4bOHCgafns2bOm5ezsbE6fPs2hQ4dwcXExTTaoUqlMIxH69OmDSqXi+++/r+uwGxxz1wmgbdu2APj4+JS7b/F6nU5XK7GJIpau0U8//VSmTLGmTZvSrl07zpw5Q3Z2du0G2YBYuh5V0aZNGwBOnz5dZtuVK1fIzs42lWmo5KasqNfy8/NNyyX76Li6ujJlypRy99m5cyenT5/mb3/7G40bN6ZFixa1HWaDZ+46wc039ri4uDL76fV6zpw5g4eHB40bN67dIBs4S9eooKAAMD/0PCUlBbVajbOzc+0F2MBYuh5VMWDAAObPn8/mzZt58MEHS2375ZdfTGUaMkl4RL1Wchhl+/btTctarZYlS5aUu8+kSZM4ffo0L774Ir179671GIX56wTQunVrhg0bxubNm1myZAlTp041bfv3v/9Neno648ePl06xtczSNYqKiuKjjz5iwYIFjBkzplTH108//ZSEhASioqLkqd01yNL1qIrBgwfTqlUrvvzyS5566inTXDwZGRm89dZbuLi4MGHChGrXY8/kaemi3jIajfTp04d9+/YBRfNadO/evcL9Jk2axMqVK4mOjpaEpw5Yc53Onj3LHXfcQXJyMiNGjKBdu3bExsaydetWwsLC2Lt3L02bNrVF+A1CRdfIYDAwaNAgdu7cSWBgIH/729/w8fEhJiaGrVu3otVq2b59O7169bLVKTgUa/5mlixZwq5duwA4cuQIMTExREVFmeat6tu3b6l/HsD8oyUuXrzIu+++K4+WsHUnIiHMeffdd02d+u69916r95NOy3XL2ut06dIlZdKkSUrTpk0VZ2dnJTQ0VJk5c6Zy9erVOoy2YbLmGuXl5Snz589XIiMjFXd3d0Wj0SjNmjVTxo8frxw/fryOI3Zs1lyP4vcxc6+JEyeWu98ff/yh/PWvf1W8vLwUrVar9OrVS1mzZk0tno39kBYeUS/t2LGDIUOGUFhYSGBgIEeOHCEwMNDWYYlbyHWq/+Qa1S9yPWxHRmmJeufYsWPcc889FBYW4ubmxtdffy1vCPWQXKf6T65R/SLXw7Yk4RH1yvnz5xk2bBhpaWk4OTmxZs0a+vfvb+uwxC3kOtV/co3qF7keticJj6g3EhMTGTJkCImJiahUKpYtW8bdd99t67DELeQ61X9yjeoXuR71gyQ8ol5ITU1l6NChnDt3DoAPP/ywwQ+hrI/kOtV/co3qF7ke9YckPMLmMjIyGD58OMePHweK5maZOXOmjaMSt5LrVP/JNapf5HrUL5LwCJvKzc1lxIgRxMTEAPDPf/6Tf/zjHzaOStxKrlP9J9eofpHrUQ/Zely8aLjy8/OVYcOGmeaVePrpp20dkiiHXKf6T65R/SLXo36SeXiEzYwZM4Zvv/0WgEGDBvH++++jUqnMlndxcSEiIqKuwhM3yHWq/+Qa1S9yPeonSXiEzVh6AyhPWFgYFy5cqJ1ghFlyneo/uUb1i1yP+kn68AghhBDC4UkLjxBCCCEcnrTwCCGEEMLhScIjhBBCCIcnCY8QQgghHJ4kPEIIIYRweJLwCCGEEMLhScIjhBBCCIcnCY8QQgghHJ4kPEIIIYRweJLwCCGEEMLhScIjhBBCCIcnCY8QQgghHJ4kPEIIIYRweJLwCCHsxvbt21GpVKhUKrZv327rcMp14cIFU4wrVqywdTgWnTx5EhcXF9zc3Lh8+XKd1asoCp07d0alUrF8+fI6q1c0bJLwCGGHSn7wl/fy9PQkIiKChx9+mK1bt9o6XFFPPfvss+j1eqZMmUKzZs0qte/w4cNRqVS89dZbZY6pUql45JFHzO6rUqn45z//CcA///lPcnJyKh+8EJUkCY8QDignJ4fTp0+zatUqBg8ezMSJEzEYDLVWnz20vIjS9uzZw48//oiLiwsvvPBCpfY1Go3s3bsXgL59+5ba9vvvv5e7/lb3338/bdu2JSkpiY8//rhS9QtRFZLwCGHnZsyYwZEjR0yvw4cPs337dubPn09gYCAAX3zxBXPnzrVxpA1DixYtUBQFRVGYNGmSrcMx64033gBg7NixhIaGVmrfI0eOkJmZiYuLC7169TKtz87O5tChQ0DFCY9arWbWrFkAvPvuu+Tl5VUqBiEqSxIeIexcYGAgnTp1Mr06d+7MgAEDeOGFF9i2bRtubm4AfPDBBxQUFNg4WlEfnDx5kp9//hmA8ePHV3r/PXv2ANC9e3fT7xfA3r17KSwsJDAwkIiIiAqPM3bsWJydnUlJSWHNmjWVjkOIypCERwgH1qFDB0aMGAFAVlYWJ06csHFEoj5Yvnw5iqIQGBjIkCFDKr3/7t27AfO3s6Kioqw6jp+fH3/9618BWLp0aaXjEKIyJOERwsG1bNnStJyfn19umXPnzvHee+8xatQoWrRogVarRavVEhYWxgMPPGBqDbhV8YikgQMHmtYNHDiwTCdqc6OVdu/ezdSpU2nbti1eXl64uLgQEhLCyJEj+fjjj0lPT6/w/L766isGDx5M48aN0Wq1tG3bljlz5nD9+vUK963IwYMHmTJlChEREXh4eODm5kZoaCjdu3dn5syZ/PDDDyiKUmofS6O0VqxYYbGz+a2vefPmlRuXwWBg5cqVjBw5kuDgYFxdXfH396dv374sWLAAnU5n8by++uorAO6++240Gk2lfy4VJTwV3c4qacyYMaZjxsfHVzoWIaymCCHszrZt2xRAAZS5c+daLDt27FhT2StXrpTZfu7cOdN2S6/x48crer2+1L7nz5+3at/ly5eX2i83N1d56KGHKtzv1nMred5btmxRxo8fb3bf8PBwJSkpqSo/XkVRFGXBggWKWq2uMMasrCyzP5Nbz3v58uVW/bwsXduLFy8qXbp0sbhfeHi4cvLkyXLP68KFC6ZyS5curfDnEBYWVqmYy3tNnDjR7PFPnDhhKrd48eIK4xGiqiqf2gsh7MaJEyfYuHEjAL1796ZJkyZlyhgMBlxcXBg+fDhDhw6lQ4cO+Pn5cf36dU6dOsXHH3/MsWPHWLVqFa1ateLVV1817dusWTOOHDnC/v37TcOQly1bRs+ePUvVERISYlo2Go3cfffd/PrrrwC0adOGxx9/nB49euDu7k5SUhJ79uwxtUKY8/LLL7Nnzx5Gjx7NhAkTCAsL4+rVq3z88cds2rSJM2fOMGvWLFavXl3pn9vhw4d57rnnMBqNtGzZkieeeIKuXbvi5+dHVlYWJ0+eZNu2baxfv75Sxx09ejQ9evSwWOb55583taiFhYWV2nbt2jX69u1LfHw8rq6uPProowwYMIAWLVqQnZ3N5s2bWbhwIWfOnOHOO+8kJiYGb2/vUscoboUBylwnW4iIiMDHx4f09HR27NjBo48+auuQhKOydcYlhKi8ki0dM2bMUI4cOWJ6HT58WNm5c6fy9ttvK02bNlUAxdvbW4mOji73WNnZ2UpiYqLZuoxGozJp0iQFUDw8PJT09HSL8Wzbts1i7AsXLjSVveeee5S8vLxyyxkMBiUhIcFsPYDyxhtvlBvvsGHDFEDRaDRKcnKyxXjK8/LLL5vOt7xWsWLp6emKwWAotc5SC09FPvroI9O+f//738tsHzdunAIoYWFhyrlz58o9RkxMjOLh4aEAyksvvVRm+4wZMxRAcXFxUQoLCyuM6cyZM0pcXJzpNWbMGAVQHnrooVLrZ82apQDKX/7yl1Lr4+LiLP5+KYqiDBw4UAGUdu3aVRiPEFUlCY8QdujWD35zL7VarUyfPt3s7Q1rXbt2TXFyclIAZd26dRbjsZTwGAwGJSQkRAGUkJCQMreDKlKynu7duytGo7Hccj///LOp3Pr16ytVh6IoyqOPPqoASmRkZKX3rWrC89tvvykajUYBlF69eik6na7McYuvwYYNGywea86cOQqgBAcHl9k2atQoBVCCgoKsjq2kDh06KICycuXKUutHjx6tAMrrr79e6WM+8MADCqC4urqavaZCVJd0WhbCgRmNRtasWcMnn3xitsPyrfR6PQkJCcTFxXH06FGOHj1KYmIi/v7+APz5559VjufQoUMkJCQA8Oijj+Lp6VnlY40bNw6VSlXutu7du5uWz507V+ljBwUFAXD8+HH27dtXtQAr4fTp04wdO5bCwkKaNWvG999/X2q4N8CmTZswGAy4u7tz5513Wjxe//79AUhMTOTSpUultqWkpADg6+tb6TivXbtGXFwcULZjcvFQ9TvuuKPSx/Xz8wOKOtVb01FdiKqQhEcIOzd37lzTRHfFr9zcXA4fPszzzz9PdnY277//PkOGDCE3N7fcY+j1ej7++GN69+6Np6cnoaGhdOjQgc6dO5teycnJAKSmplY51tjYWNNyv379qnwcgHbt2pndVvwBCkXD8SvroYcewtnZmfz8fKKiohg1ahSffvopR48eLTMqq7rS09MZNWoUaWlpaLVa1q9fb0q4Sjpw4AAAubm5aDQai6O7Ro4cadrvypUrpY5TPHqtKgnP7t27URSF4OBgWrVqZVp/5swZkpOT0Wg03H777ZU+bslY5DETorZIwiOEA9JqtXTu3Jn//Oc/LFq0CIBdu3aVee4RFH0A9unThyeeeII//vijwskJKxrybEnJZKm8D/XKcHd3N7tNrb751laVR2q0a9eO1atX4+vrS2FhIRs3bmTGjBl07tyZwMBAHn744VKdf6vKYDDwwAMPcPLkSdMw9pKtUyUVJ5yVdWuSW9xyVJXruGvXLqDsPDvFw9Rvu+02PDw8Kn3ckrE4OztXen8hrCGjtIRwcFOmTOGFF17g+vXrLFu2zPRIgWJPP/00Bw8eBIpGET3yyCPcdtttBAYG4ubmZrpt1Lx5c+Lj42u8haO+GjNmDEOGDGHt2rX88ssv/P7776SkpJCamsqqVatYtWoVEydOZNmyZaUSrMqYNWsWmzdvBopGnd1///1myxYnbgEBAWzbts3qOkrOwwTQuHFjgArnKVqxYgWTJ08ud9vXX39d7u3EmJiYMuut+X0pGcuto8qEqCmS8Ajh4NRqNW3atOGPP/4gKSmJa9eumfrjZGZmsnbtWgD+/ve/s2rVKrPHSUtLq3YsAQEBpuWkpCSLt6XqA29vb6ZNm8a0adMAiIuLY/369Xz44YckJiaycuVKIiMjefrppyt97MWLF/Phhx8CcN9995mdZLBY8TXLysqiffv2ODk5VbpOuJnw1MT1rCnFsRQn2ULUBrmlJUQDUFhYWO7y6dOn0ev1ADzwwANm9z9x4gTZ2dlmt5vrPHyrbt26mZZ37txp1T71Sfv27XnhhRfYu3ev6dZNRfMFlWf79u088cQTAERGRrJy5coKf4aRkZFAUcfe4v48VdG5c2cAMjIyLN4mGz16dKmH0hY/+sHX17fU+uL5lKDollfJbUeOHLEqplOnTgHQsWPHqp6WEBWShEcIB5ebm8vx48eBor49JVtZSiY/ljqLfvrppxbrKPlfuaXRYF26dDE9mXvJkiUWk6j6LDQ01PRwzMp24j579iz33Xcfer2eJk2asH79eov9kYqNGjXKlBS9//77lY65WMnO4vv37zdbzsfHp9RDaYs7P/ft27fU+uLWmbZt2xIVFVVqW6dOnSqMJzMzk5MnTwJUqcOzENaShEcIBzdv3jxTp9Dhw4eXuhUSHh5u+hBduXJluf0tNmzYwEcffWSxjpIdkM+ePWu2nFqt5vnnnwcgISGBCRMmmO0kbTQaSUxMtFhvbfn+++8tDo+Oj483PYj11j4ylmRkZDBq1CiuXbuGq6sr33//vSkBrEjbtm0ZO3YsAGvWrGHBggUWy58/f77cWaZ79eqFq6srQKWG3Be3yN06uq6483bxUPjKOnDggOn3btiwYVU6hhDWkD48Qti55ORkjh49WmpdXl4ep0+f5osvvjA9psDNzY3XXnutVDl/f3/uuusuNm3axM8//8ywYcOYMWMGYWFhJCcn880337BixQpatWpFenq6aQ6XWzVv3pyQkBASEhJ49913CQkJoW3btqbkqkmTJjRq1AiAmTNnsmHDBn799Ve+++47OnfuXOrREleuXGHv3r2sXr2acePGVdi3pTa8//77/P3vf2fEiBEMGjSI9u3b4+3tTVpaGgcOHODDDz80JZHTp0+3+rhPPvmkaR6bZ555Bk9PzzLXrqTAwEACAwNN33/yySccOHCAc+fOMXv2bNavX8+ECRPo2LEjrq6uXLt2jT///JOff/6ZrVu3cs899/DQQw+VOqarqyvDhw/nhx9+YMuWLaUeFWKOwWAwzbNjLuEZMGCAdT+EW2zZsgUo6t9VmYeOClFptprxUAhRddbOtFz8aty4sfLLL7+Ue6xLly4pzZs3N7tv8+bNlWPHjpkeImnuQZCLFi0ye4xbZxzOyclR7rvvvko/PLMyj7AwdwxrDBgwwKpZrMubVdjSTMvWHNfS+SuKoiQlJSn9+vWzav/JkyeXe34bNmxQAEWlUikXLlyo8Oexf/9+BVC0Wq1SUFBgWp+enm56wOqlS5cqPE55WrZsqQDKzJkzq7S/ENaSFh4hHJCLiwt+fn507NiRu+66i8mTJ5udaC40NJSYmBjefvtt1q9fz8WLF3Fzc6NFixaMHj2ap59+2qpJ6mbMmEGTJk347LPPOHToENevXy/VR6gkd3d3vv76a7Zt28by5cvZtWsXV65cwWAw0KRJE7p27crIkSPLtE7UldWrV7Nx40a2b9/O8ePHuXLlCqmpqbi5uREWFkb//v2ZPn06t912W53H1rRpU3bu3MmmTZtYvXo10dHRXLlyBb1ej4+PD23atKFPnz787W9/M3ub6c477zS1yK1evZoXXnjBYp3FrTi9e/cuNU/Onj17MBqNtGjRwupbcyVFR0dz/vx5oOj3R4japFKUBjKphhBCCJP//Oc//OMf/yAiIoK4uLgqzyVUHVOnTmXp0qUMHz7cdOtViNoiCY8QQjRAOp2ONm3acPnyZVavXs2DDz5Yp/VfunSJ8PBw9Ho9e/fulRFaotbJKC0hhGiAtFqtqcPyG2+8UeczaM+fPx+9Xs/YsWMl2RF1QvrwCCFEAzVp0iSuXr1KQUEBSUlJBAcH10m9iqIQFhbG3LlzeeSRR+qkTiHklpYQQgghHJ7c0hJCCCGEw5OERwghhBAOTxIeIYQQQjg8SXiEEEII4fAk4RFCCCGEw5OERwghhBAOTxIeIYQQQjg8SXiEEEII4fAk4RFCCCGEw5OERwghhBAO7/8B9fKwyy6kOPYAAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 600x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ai = [8, 16, 32, 64, 128, 256, 512, 1024, 2048]\n",
    "fp16 = [3.5913723582696826, 7.396274389981544, 14.601368378822112, 29.153260582063147, 55.5314758714023, 95.79681813261212, 136.6930123791791, 152.77694542825068, 161.1156499269412]\n",
    "w4a4 = [6.639533644869575, 13.565385101531655, 27.254175615471844, 54.4611666014685, 107.93834080480016, 211.59420293722238, 372.91334478435743, 466.0135395676664, 538.1759856711493]\n",
    "w4a16 =[5.093059212625437, 6.536323152461491, 7.543259966351492, 7.947865778756361, 8.344728762237368, 8.54029590437908, 8.499885651479563, 8.183972536588632, 8.167991369592459]\n",
    "w8a8 = [2.928233438775072, 6.058944027846304, 12.147133542431966, 24.49705443214467, 49.57134767891193, 99.02931334169796, 192.23851561133415, 369.4828550922088, 433.98997666107266]\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "bar_colors_default = [\"#5BB8D7\", \"#57A86B\", \"#A8A857\", \"#6E4587\", \"#ADEBCC\", \"#EBDCAD\"]\n",
    "# set font size\n",
    "plt.rcParams.update({'font.size': 20})\n",
    "plt.figure(figsize=(6, 6))\n",
    "for i in range(len(ai)):\n",
    "    w4a4[i] = w4a4[i] / fp16[i]\n",
    "    w4a16[i] = w4a16[i] / fp16[i]\n",
    "    w8a8[i] = w8a8[i] / fp16[i]\n",
    "plt.axhline(y=1, color=bar_colors_default[0], linestyle='--', label='FP16 Baseline')\n",
    "plt.plot(ai, w4a16, label='W4A16', marker='o', color=bar_colors_default[1])\n",
    "plt.plot(ai, w8a8, label='W8A8', marker='o', color=bar_colors_default[2])\n",
    "plt.plot(ai, w4a4, label='W4A4(Atom)', marker='o', color=bar_colors_default[3])\n",
    "plt.xscale('log', base=2)\n",
    "plt.legend()\n",
    "plt.xlabel('Batch size(#)')\n",
    "plt.ylabel('Tops improvement')\n",
    "plt.show()\n",
    "plt.close()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "int",
   "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": "3.10.13"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}