Skip to content

Instantly share code, notes, and snippets.

@mdouze
Created July 14, 2021 09:51
Show Gist options
  • Save mdouze/4d8cd80b97eee9db2225133d4e931cc6 to your computer and use it in GitHub Desktop.
Save mdouze/4d8cd80b97eee9db2225133d4e931cc6 to your computer and use it in GitHub Desktop.
Display the source blob
Display the rendered blob
Raw
{
"cells": [
{
"cell_type": "markdown",
"id": "9f85f826",
"metadata": {},
"source": [
"### It is useful to cluster >1B datasets to around 262k - 1M clusters for IVF indexing with Faiss.\n",
"However, it is not feasible to do the clustering within the allocated time for indexing. \n",
"\n",
"Therefore, here we evaluate other options to break down the clustering cost, while getting the same number of clusters.\n",
"The model that we use is: Deep1M (1M database vectors), 16384 clusters (which conveniently breaks down to 2^7 * 2^7)\n"
]
},
{
"cell_type": "code",
"execution_count": 15,
"id": "ongoing-first",
"metadata": {},
"outputs": [],
"source": [
"import numpy as np\n",
"import faiss\n",
"from faiss.contrib import datasets\n",
"import time\n",
"from matplotlib import pyplot"
]
},
{
"cell_type": "code",
"execution_count": 16,
"id": "c9e0fce4",
"metadata": {},
"outputs": [],
"source": [
"\n",
"def two_level_clustering(xt, nc1, nc2, clustering_niter=25, spherical=False):\n",
" d = xt.shape[1]\n",
"\n",
" print(f\"2-level clustering of {xt.shape} nb clusters = {nc1}*{nc2} = {nc1*nc2}\")\n",
" print(\"perform coarse training\")\n",
"\n",
" km = faiss.Kmeans(\n",
" d, nc1, verbose=True, niter=clustering_niter,\n",
" max_points_per_centroid=2000,\n",
" spherical=spherical\n",
" )\n",
" km.train(xt)\n",
"\n",
" print()\n",
"\n",
" # coarse centroids\n",
" centroids1 = km.centroids\n",
"\n",
" print(\"assigning the training set\")\n",
" t0 = time.time()\n",
" _, assign1 = km.assign(xt)\n",
" bc = np.bincount(assign1, minlength=nc1)\n",
" print(f\"done in {time.time() - t0:.2f} s. Sizes of clusters {min(bc)}-{max(bc)}\")\n",
" o = assign1.argsort()\n",
" del km\n",
"\n",
" # train sub-clusters\n",
" i0 = 0\n",
" c2 = []\n",
" t0 = time.time()\n",
" for c1 in range(nc1):\n",
" print(f\"[{time.time() - t0:.2f} s] training sub-cluster {c1}/{nc1}\\r\", end=\"\", flush=True)\n",
" i1 = i0 + bc[c1]\n",
" subset = o[i0:i1]\n",
" assert np.all(assign1[subset] == c1)\n",
" km = faiss.Kmeans(d, nc2, spherical=spherical)\n",
" xtsub = xt[subset]\n",
" km.train(xtsub)\n",
" c2.append(km.centroids)\n",
" i0 = i1\n",
" print(f\"done in {time.time() - t0:.2f} s\")\n",
" return np.vstack(c2)"
]
},
{
"cell_type": "code",
"execution_count": 24,
"id": "d3441c8b",
"metadata": {},
"outputs": [],
"source": [
"def pp_time(t): \n",
" if t < 600: \n",
" return f\"{t:.2f}s\"\n",
" if t < 3600: \n",
" return f\"{int(t/60)}m {int(t)%60}s\"\n",
" m = int(t/60)\n",
" return f\"{m//60}h {m%60}m\"\n",
"\n",
"\n",
"\n",
"def eval_quantizers(ds, sqrt_nlist, maxtrain=10**5, nprobes=[1, 4, 16, 64, 256], include_hsnw=False):\n",
" xt = ds.get_train(maxtrain)\n",
" d = ds.d\n",
" xb = ds.get_database()\n",
" xq = ds.get_queries()\n",
" gt = ds.get_groundtruth()\n",
" nlist = sqrt_nlist**2\n",
"\n",
" print(\"flat\")\n",
" \n",
" quantizer = faiss.IndexFlatL2(d)\n",
" index = faiss.IndexIVFFlat(quantizer, d, nlist)\n",
" t0 = time.time()\n",
" index.train(xt)\n",
" t1 = time.time()\n",
" index.add(xb)\n",
" index.invlists.imbalance_factor()\n",
"\n",
" xy = [] \n",
" stats = faiss.cvar.indexIVF_stats \n",
" for nprobe in nprobes: \n",
" index.nprobe = nprobe \n",
" stats.reset()\n",
" D, I = index.search(xq, 100)\n",
" rank = 1\n",
" recall = (I[:, :rank] == gt[:, :1]).sum() / len(xq)\n",
" print(f\"nprobe={nprobe} 1-recall @ {rank}: {recall} dis/q={stats.ndis/len(xq):.2f}\")\n",
" xy.append((recall, stats.ndis))\n",
" xy = np.array(xy) \n",
" pyplot.semilogy(xy[:, 0], xy[:, 1] / len(xq), label=f'Flat train {pp_time(t1-t0)}')\n",
" \n",
" if include_hsnw:\n",
" print(\"flat + HNSW\")\n",
"\n",
" quantizer = faiss.IndexHNSWFlat(d, 32)\n",
" quantizer.hnsw.efConstruction = 200 \n",
" quantizer.hnsw.efSearch = 80 \n",
" index = faiss.IndexIVFFlat(quantizer, d, nlist)\n",
"\n",
" t0 = time.time()\n",
" index.train(xt)\n",
" t1 = time.time()\n",
" index.add(xb)\n",
" index.invlists.imbalance_factor()\n",
"\n",
" xy = [] \n",
" stats = faiss.cvar.indexIVF_stats \n",
" for nprobe in nprobes: \n",
" index.nprobe = nprobe \n",
" stats.reset()\n",
" D, I = index.search(xq, 100)\n",
" rank = 1\n",
" recall = (I[:, :rank] == gt[:, :1]).sum() / len(xq)\n",
" print(f\"nprobe={nprobe} 1-recall @ {rank}: {recall} dis/q={stats.ndis/len(xq):.2f}\")\n",
" xy.append((recall, stats.ndis))\n",
" xy = np.array(xy) \n",
" pyplot.semilogy(xy[:, 0], xy[:, 1] / len(xq), label=f'Flat + HNSW train {pp_time(t1-t0)}')\n",
" \n",
" print(\"IMI\")\n",
" \n",
" quantizer = faiss.MultiIndexQuantizer(d, 2, int(np.log2(sqrt_nlist)))\n",
"\n",
" index = faiss.IndexIVFFlat(quantizer, d, nlist)\n",
" index.quantizer_trains_alone = 1\n",
"\n",
" t0 = time.time()\n",
" index.train(xt)\n",
" t1 = time.time()\n",
" index.add(xb)\n",
"\n",
" stats = faiss.cvar.indexIVF_stats\n",
" xy = []\n",
" for nprobe in nprobes: \n",
" index.nprobe = nprobe \n",
" stats.reset()\n",
"\n",
" D, I = index.search(xq, 100)\n",
" rank = 1\n",
" recall = (I[:, :rank] == gt[:, :1]).sum() / len(xq)\n",
" print(f\"nprobe={nprobe} 1-recall @ {rank}: {recall} dis/q={stats.ndis/len(xq):.2f}\")\n",
" xy.append((recall, stats.ndis))\n",
" xy = np.array(xy) \n",
" pyplot.semilogy(xy[:, 0], xy[:, 1] / len(xq), label=f'inverted multi-index train {pp_time(t1-t0)}')\n",
" \n",
" print(\"RCQ\")\n",
" \n",
" quantizer = faiss.ResidualCoarseQuantizer(d, 2, int(np.log2(sqrt_nlist)))\n",
"\n",
" index = faiss.IndexIVFFlat(quantizer, d, nlist)\n",
" index.quantizer_trains_alone = 1\n",
"\n",
" t0 = time.time()\n",
" index.train(xt)\n",
" t1 = time.time()\n",
" quantizer.set_beam_factor(-1)\n",
"\n",
" index.add(xb)\n",
" \n",
" xy = []\n",
" stats = faiss.cvar.indexIVF_stats\n",
" for nprobe in nprobes: \n",
" index.nprobe = nprobe \n",
" stats.reset()\n",
"\n",
" D, I = index.search(xq, 100)\n",
" rank = 1\n",
" recall = (I[:, :rank] == gt[:, :1]).sum() / len(xq)\n",
" print(f\"nprobe={nprobe} 1-recall @ {rank}: {recall} dis/q={stats.ndis/len(xq):.2f}\")\n",
" xy.append((recall, stats.ndis))\n",
" xy = np.array(xy) \n",
" pyplot.semilogy(xy[:, 0], xy[:, 1] / len(xq), label=f'residual coarse quantizer train {pp_time(t1-t0)}')\n",
" \n",
" print(\"2-level clustering\")\n",
"\n",
" t0 = time.time()\n",
" centroids12 = two_level_clustering(xt, sqrt_nlist, sqrt_nlist)\n",
" t1 = time.time()\n",
" quantizer = faiss.IndexFlatL2(d)\n",
" quantizer.add(centroids12)\n",
" index = faiss.IndexIVFFlat(quantizer, d, nlist)\n",
"\n",
" index.add(xb)\n",
" index.invlists.imbalance_factor()\n",
" xy = []\n",
" stats = faiss.cvar.indexIVF_stats\n",
" for nprobe in nprobes: \n",
" index.nprobe = nprobe \n",
" stats.reset()\n",
" D, I = index.search(xq, 100)\n",
" rank = 1\n",
" recall = (I[:, :rank] == gt[:, :1]).sum() / len(xq)\n",
" print(f\"nprobe={nprobe} 1-recall @ {rank}: {recall} dis/q={stats.ndis/len(xq):.2f}\")\n",
" xy.append((recall, stats.ndis))\n",
"\n",
" xy = np.array(xy) \n",
" pyplot.semilogy(xy[:, 0], xy[:, 1] / len(xq), label=f'2-level clustering {pp_time(t1-t0)}')\n",
" \n",
" \n",
" if include_hsnw:\n",
" print(\"2-level clustering + HNSW\")\n",
"\n",
" t0 = time.time()\n",
" centroids12 = two_level_clustering(xt, sqrt_nlist, sqrt_nlist) \n",
" quantizer = faiss.IndexHNSWFlat(d, 32)\n",
" quantizer.hnsw.efConstruction = 200 \n",
" quantizer.hnsw.efSearch = 80 \n",
" quantizer.add(centroids12)\n",
" t1 = time.time()\n",
" index = faiss.IndexIVFFlat(quantizer, d, nlist)\n",
"\n",
" index.add(xb)\n",
" index.invlists.imbalance_factor()\n",
" xy = []\n",
" stats = faiss.cvar.indexIVF_stats\n",
" for nprobe in nprobes: \n",
" index.nprobe = nprobe \n",
" stats.reset()\n",
" D, I = index.search(xq, 100)\n",
" rank = 1\n",
" recall = (I[:, :rank] == gt[:, :1]).sum() / len(xq)\n",
" print(f\"nprobe={nprobe} 1-recall @ {rank}: {recall} dis/q={stats.ndis/len(xq):.2f}\")\n",
" xy.append((recall, stats.ndis))\n",
"\n",
" xy = np.array(xy) \n",
" pyplot.semilogy(xy[:, 0], xy[:, 1] / len(xq), label=f'2-level clustering + HNSW {pp_time(t1-t0)}')\n",
" \n",
" pyplot.gcf().set_size_inches(11, 7)\n",
" pyplot.xlabel(\"1-recall @ 1\")\n",
" pyplot.ylabel(\"nb distances per query\")\n",
" \n",
" pyplot.legend()\n",
" pyplot.grid()"
]
},
{
"cell_type": "code",
"execution_count": 25,
"id": "finnish-giant",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"flat\n",
"nprobe=1 1-recall @ 1: 0.2786 dis/q=125.44\n",
"nprobe=2 1-recall @ 1: 0.4059 dis/q=246.83\n",
"nprobe=4 1-recall @ 1: 0.5571 dis/q=482.08\n",
"nprobe=8 1-recall @ 1: 0.6957 dis/q=936.89\n",
"nprobe=16 1-recall @ 1: 0.8111 dis/q=1808.96\n",
"nprobe=32 1-recall @ 1: 0.8973 dis/q=3452.74\n",
"nprobe=64 1-recall @ 1: 0.9507 dis/q=6549.73\n",
"nprobe=128 1-recall @ 1: 0.9787 dis/q=12361.58\n",
"nprobe=256 1-recall @ 1: 0.994 dis/q=23190.52\n",
"IMI\n",
"nprobe=1 1-recall @ 1: 0.3733 dis/q=1600.51\n",
"nprobe=2 1-recall @ 1: 0.5038 dis/q=2594.21\n",
"nprobe=4 1-recall @ 1: 0.6238 dis/q=4024.67\n",
"nprobe=8 1-recall @ 1: 0.7274 dis/q=6136.54\n",
"nprobe=16 1-recall @ 1: 0.8131 dis/q=9053.74\n",
"nprobe=32 1-recall @ 1: 0.8798 dis/q=13071.86\n",
"nprobe=64 1-recall @ 1: 0.9279 dis/q=18776.15\n",
"nprobe=128 1-recall @ 1: 0.9594 dis/q=27262.57\n",
"nprobe=256 1-recall @ 1: 0.9807 dis/q=40308.77\n",
"RCQ\n",
"nprobe=1 1-recall @ 1: 0.2443 dis/q=229.39\n",
"nprobe=2 1-recall @ 1: 0.36 dis/q=425.64\n",
"nprobe=4 1-recall @ 1: 0.4967 dis/q=776.16\n",
"nprobe=8 1-recall @ 1: 0.6355 dis/q=1402.61\n",
"nprobe=16 1-recall @ 1: 0.7589 dis/q=2528.22\n",
"nprobe=32 1-recall @ 1: 0.8579 dis/q=4514.27\n",
"nprobe=64 1-recall @ 1: 0.9218 dis/q=8036.60\n",
"nprobe=128 1-recall @ 1: 0.9659 dis/q=14269.52\n",
"nprobe=256 1-recall @ 1: 0.9871 dis/q=25584.58\n",
"2-level clustering\n",
"2-level clustering of (100000, 96) nb clusters = 128*128 = 16384\n",
"perform coarse training\n",
"\n",
"assigning the training set\n",
"done in 0.04 s. Sizes of clusters 203-1683\n",
"done in 4.68 sing sub-cluster 127/128\n",
"nprobe=1 1-recall @ 1: 0.283 dis/q=132.08\n",
"nprobe=2 1-recall @ 1: 0.4238 dis/q=261.18\n",
"nprobe=4 1-recall @ 1: 0.5719 dis/q=512.17\n",
"nprobe=8 1-recall @ 1: 0.7066 dis/q=998.44\n",
"nprobe=16 1-recall @ 1: 0.8197 dis/q=1929.67\n",
"nprobe=32 1-recall @ 1: 0.8973 dis/q=3698.68\n",
"nprobe=64 1-recall @ 1: 0.9498 dis/q=7048.08\n",
"nprobe=128 1-recall @ 1: 0.9787 dis/q=13304.22\n",
"nprobe=256 1-recall @ 1: 0.9912 dis/q=24936.67\n"
]
},
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAApwAAAG5CAYAAAAjwuS9AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAACbAklEQVR4nOzde1yO9x/H8dfV+YxEJFRyTkKIHHJKzoeZwzA7YJvNfjswdjI7z84HO9gYhmG2YVZyzvmY8zkURUlROh/u+/v746axhRvVXfk8H489Vtd139f17pJ8+h41pRRCCCGEEEIUFzNTBxBCCCGEEOWbFJxCCCGEEKJYScEphBBCCCGKlRScQgghhBCiWEnBKYQQQgghipUUnEIIIYQQolhJwSmEEEIIIYqVFJxCCHEHmqbFaJqWpWlamqZpKZqmbdM07WlN04r1Z6imaVaapv1+7f5K07Sgf52fc+14338d//La8ceKM58QQhhLCk4hhDBOH6WUI1Ab+AiYBMwqgftuAUYACbc4fxIYdf0TTdMsgIeB08UfTQghjCMFpxBC3AWlVKpS6i9gCDBK0zQfTdOsNU37VNO0c5qmXdQ07QdN02yvv0fTtN6apu2/oXXU94ZzMZqmvapp2lFN065omjZb0zSba/fKVUp9qZTaAuhuEWkFEKhpWqVrn4cAB7l1gSqEECVOCk4hhLgHSqldQBzQHpgG1AP8AG+gBjAFQNO05sDPwFNAZWAG8JemadY3XG440B2oc+06b9xFlGzgL2Dotc8fBX65l69JCCGKixScQghx7y4AzsAY4EWl1GWlVBrwAf8UgGOAGUqpnUopnVJqLpADBNxwnelKqVil1GXgfWDYXeb4BXhU07QKQEdg2T1/RUIIUQwsTB1ACCHKsBoYfo7aAZGapl0/rgHm1z6ujaHrffwN77MC3G74PPaGj8/+69wdKaW2aJpWBUPL6N9KqawbsgghhMlJC6cQQtwDTdNaYig4lwFZQGOlVMVr/1VQSjlce2ks8P4N5yoqpeyUUgtvuFzNGz6uhaHl9G7NB15GutOFEKWQFJxCCHEXNE1z0jStN7AImK+UOgD8BHyhaVrVa6+poWla92tv+Ql4WtO01pqBvaZpvTRNc7zhss9qmuauaZoz8Bqw+Ib7WV+fRARYaZpmoxXefPk10A3YVKRfsBBCFAEpOIUQwjgrNE1Lw9Bi+TrwOfD4tXOTgFPADk3TrgJrgfoASqk9GMZxTgeuXHvdY/+69q/AauDMtf/eu+HcCQwtqDWAVdc+rv3vcNfGj65TSqn7/UKFEKKoafKzSQghTEfTtBhgtFJqramzCCFEcZEWTiGEEEIIUayk4BRCCCGEEMVKutSFEEIIIUSxkhZOIYQQQghRrMr0wu8uLi7Kw8PD1DFKlYyMDOzt7U0do0yQZ2UceU7Gk2dlHHlOxpNnZRx5TsYp7ucUGRmZpJSqUti5Ml1wenh4sGfPHlPHKFUiIiIICgoydYwyQZ6VceQ5GU+elXHkORlPnpVx5DkZp7ifk6ZpZ291TrrUhRBCCCFEsZKCUwghhBBCFCspOIUQQgghRLEqk2M4NU3rA/Tx9vb+z7m8vDzi4uLIzs4u+WClQIUKFTh27JipY5QJ8qyMc7fPycbGBnd3dywtLYsxlRBCiLKkTBacSqkVwAp/f/8x/z4XFxeHo6MjHh4eaJpmgnSmlZaWhqOjo6ljlAnyrIxzN89JKUVycjJxcXF4enoWczIhhBBlRbnrUs/OzqZy5coPZLEphKlpmkblypUf2B4GIYQQhSt3BScgxaYQJiR//4QQQvxbuSw4hRBCCCFE6SEFZzEwNzfHz8+v4L+YmBgiIiLo3bv3bd+3f/9+wsLC7vrc7Vy4cIFBgwbd1XumT5+Ot7c3mqaRlJRUcHzBggX4+vri6+tL27ZtOXDgwG2vM378eBwcHP5zfPfu3Zibm/P777/fVS4hhBBClE1ScBYDW1tb9u/fX/Cfsdtv3mvBmZ+ff8trurm53XVhFxgYyNq1a6ldu/ZNxz09Pdm4cSMHDx7kzTffZOzYsbe8xp49e0hJSfnPcZ1Ox6RJk+jevftdZRJCCCFE2SUFpwns2rWLtm3b0qxZM9q2bcuJEyfIzc1lypQpLF68GD8/PxYvXlzw+sLOTZ06lbFjxxIcHMyjjz5KTEwM7du3p3379jRv3pxt27YBEBMTg4+PDwBz5sxh4MCBhISEULduXV555ZVC8zVr1qzQIrlt27ZUqlQJgICAAOLi4gp9v06nY+LEiXz88cf/OffNN9/w0EMPUbVq1YJjGRkZ9OrVi6ZNm+Lj43PT1y6EEEKIsq9MLotkrLdXHOHohatFes1Gbk681afxbV+TlZWFn58fYGgVXLp06U3nGzRowKZNm7CwsGDt2rW89tpr/PHHH7zzzjvs2bOH6dOn3/R6Kyur/5ybOnUqkZGRbNmyBVtbWzIzM1mzZg15eXkkJCQwbNiwQveZ379/P/v27cPa2pr69eszfvx4atasedfPYdasWfTo0aPQc9OnT6dv375Ur179puPnz59n6dKlrF+/nt27dxccDw8Px83NjdDQUABSU1PvOo8QQgghSq9yXXCayvUu9VtJTU1l1KhRREVFoWkaeXl593Sfvn37YmtrCxgWvH/uuefYu3cvlpaWnDx5stD3dOnShQoVKgDQqFEjzp49e9cF54YNG5g1axZbtmz5z7kLFy6wZMkSIiIi/nPuhRdeYNq0aZibm990vEmTJkyYMIFJkybRu3dv2rdvf1d5hBBCCFG6leuC804tkaby5ptv0qlTJ5YuXUpMTAxBQUH3dB17e/uCj7/44gtcXV3Ztm0b9vb22NjYFPoea2vrgo/Nzc1vO/6zMAcPHmT06NGsXLmSypUr/+f8vn37OHXqFNd3gcrMzMTb25tTp06xZ88ehg4dCkBSUhJhYWFYWFjQv39/IiMjCQsL49VXXyU4OJgpU6bcVS4hhBBClF7luuAsrVJTU6lRowZgGFd5naOjI2lpaYW+53bnrl/T3d0dMzMz5s2bh06nK9LMAOfOnWPgwIHMmzePevXqFfqaXr16kZCQUPC5g4MDp06dAiA6Orrg+GOPPUbv3r3p378/Fy5cwNnZmREjRuDg4HDTMxFCCCHEDZJPg5U9OFYzdZK7IpOGTOCVV17h1VdfJTAw8KbCsFOnThw9evQ/k4budA5g3LhxzJ07l86dO3Py5MmbWj/v1tdff427uztxcXH4+voyevRoAN555x2Sk5MZN24cfn5++Pv7F7ynZ8+eXLhw4Z7ud+jQIVq1aoWfnx/vv/8+b7zxxj1nF0IIIcotXR78/gTM6Q36om9YKk6aUsrUGe6apml9gD7e3t5joqKibjp37NgxGjZsaJpgpYDsD248eVbGuZfn9KD+PYyIiLjnITIPEnlOxpNnZZwH5jlt/Bg2vA+Df4FG/e767cX9nDRNi1RK+Rd2rky2cCqlViilxl6f/CKEEEIIUa7FH4SN08Bn0D0Vm6ZWJgtOIYQQQogHRn4OLHsG7CpDz09MneaeyKQhIYQQQojSbOM0uHgYhi0GO2dTp7kn0sIphBBCCFFaxUXCli/AbwTUDzF1mnsmBacQQgghRGmUlwXLngZHNwj5wNRp7ot0qQshhBBClEbr34OkkzByKdiU7YnS0sJZDNq2bVvs95gzZ85dr3sZExODj49PkWe58br79+8nLCys4Nxff/3FRx99VOj7evbsSUpKyj3d5378O6OxLly4wKBBg+7qPdHR0bRu3Zq6desyZMgQcnNz//OaDRs24OfnV/CfjY0Ny5YtAwwL5Ht6ehacu92WqUIIIcqRs9th+7fg/wTU6WzqNPdNCs5isG3btmK9vk6nu6eCsyT8u5jr27cvkydPLvS1YWFhVKxYsYSS/eN2Befttvp0c3Pj999/v6t7TZo0iRdffJGoqCgqVarErFmz/vOaTp06sX//fvbv38/69euxs7MjODi44Pwnn3xScN7Pz++u7i+EEKIMys0wzEqvWAu6vWvqNEVCCs5i4ODgAPyzwOqgQYNo0KABw4cPRynFypUrGTx4cMHrIyIi6NOnDwCrV6+mTZs2NG/enIcffpj09HQAPDw8eOedd2jXrh0LFy5kz549DB8+HD8/P7KysoiMjKRjx4506NCB7t27Ex8fD0BkZCRNmzalTZs2fPvtt4XmjYiIoGPHjgwePJh69eoxefJkFixYQKtWrWjSpAmnT58GDK1tNxZc17/O63Jzc5kyZQqLFy8u2BFpzpw5PPfcc4Xe18PDg6SkJGJiYmjYsCFjxoyhcePGBAcHk5WVddv8Op2OiRMn0rJlS3x9fZkxYwYAn3/+OU888QRg2MHIx8eHzMzMQjMGBgayePFipk6dytixYwkODubRRx8lJiaG9u3b07x5c5o3b17wC8SNLaxz5sxh4MCBhISEULduXV555ZX/fH1KKdavX1/QKjpq1KiClstb+f333+nRowd2dna3fd3GjRsLWj2bNWt2221PhRBClDFr3oIrMdD/O7B2uOPLy4LyPYZz5WRIOFS016zWBHoU3kVcmH379nHkyBHc3NwIDAxk69atdOvWjaeeeoqMjAzs7e1ZvHgxQ4YMISkpiffee4+1a9dib2/PtGnT+Pzzz5kyZQoANjY2bNmyBYCZM2fy6aef4u/vT15eHuPHj2f58uXY2NgQFhbG66+/zs8//8zjjz/ON998Q8eOHZk4ceItcx44cIBjx47h7OyMl5cXo0ePZteuXXz11Vd88803fPnll3f8Wq2srHjnnXfYs2cP06dPBzB6X/SoqCgWLlzITz/9xODBg/njjz8YMWLELfPPmjWLChUqsHv3bnJycggMDCQ4OJgXXniBoKAgli5dyvvvv8+MGTNuKt5uzPjhhx/i6OjI1KlTiYyMZMuWLdja2pKZmcmaNWuwsbEhKiqKYcOGsWfPnv9k3r9/P/v27cPa2pr69eszfvx4atasWXA+OTmZihUrYmFh+Gvm7u7O+fPnb/scFi1axEsvvXTTsddff5133nmHLl268NFHH2Ftbc2nn37Kt99+S2BgIOnp6djY2Bj1nIUQQpRyZyJg908QMA482pk6TZGRFs5i1qpVK9zd3TEzM8PPz4+YmBgsLCwICQlhxYoV5OfnExoaSr9+/dixYwdHjx4lMDAQPz8/5s6dy9mzZwuuNWTIkELvceLECQ4fPky3bt0IDAzkvffeIy4ujtTUVFJSUujYsSMAI0eOvGXOli1bUr16daytralTp05Bl26TJk2IiYkpugdyC9fHKQK0aNGCmJiY2+ZfvXo1v/zyC35+frRu3Zrk5GSioqIwMzNjzpw5jBw5ko4dOxIYGGjU/fv27YutrS0AeXl5jBkzhiZNmvDwww9z9OjRQt/TpUsXKlSogI2NDY0aNbrpzwoMLZz/pmnaLTPEx8dz6NAhunfvXnDsww8/5Pjx4+zevZvLly8zbdo0AAIDA3nppZf4+uuvSUlJKShqhRBClGHZV2H5c1DZG7pMMXWaIlW+/5W6i5bI4mJtbV3wsbm5ecEYwSFDhvDtt9/i7OxMy5YtcXR0RClFt27dWLhwYaHXsre3L/S4UorGjRuzffv2m/a9TklJuW2Bc6ucZmZmBZ+bmZkVZLawsECv1xfcs7AJMLei0+lo0aIFYCju3nnnnVve39zcnKysLJRSt8yvlOKbb765qTi7LioqCgcHh7sa43rjs/3iiy9wdXXlwIED6PX6W7Ye3urP9joXFxdSUlLIz8/HwsKCuLg43Nzcbpnht99+Y8CAAVhaWhYcq169esG9Hn/8cT799FMAJk+eTK9evQgLCyMgIIC1a9fSoEEDo79eIYQQpdDq1+HqeXhiNVjamjpNkZIWThMJCgpi7969/PTTTwUtlwEBAWzdupVTp04BkJmZycmTJwt9v6OjY8G4vfr163Pp0iW2b98OGFrojhw5QsWKFalQoUJBN/yCBQvuK7OHhweRkZEALF++nLy8vNvmupG5uXnBxJd/F5u3crv83bt35/vvvy/IcPLkSTIyMkhNTeV///sfmzZtIjk5udBJPrfKeF1qairVq1fHzMyMefPmodPpjMr7b5qm0alTp4IMc+fOpV+/W+9/u3DhQoYNG3bTsetjcZVSLFu2rGAM6enTp2nSpAmTJk3C39+f48eP31NGIYQQpUTUGtj7C7R9Hmq2NHWaIicFp4mYm5vTu3dvVq5cSe/evQGoUqUKc+bMYdiwYfj6+hIQEHDLQuKxxx7j6aefxs/PD51Ox++//86kSZNo27Ytfn5+BRNdZs+ezbPPPkubNm0Kuozv1ZgxY9i4cSOtWrVi586dhba4durUiaNHjxZMGrpft8o/evRoGjVqRPPmzfHx8eGpp54iPz+fF198kXHjxlGvXj1mzZrF5MmTSUxMLDTj9UlD/zZu3Djmzp1LQEAAJ0+evGXLsjGuj8P19vYmOTmZJ598EoA9e/YwevTogtfFxMQQGxtbMHzguuHDh9OkSROaNGlCUlISb7zxBgBffvklPj4+NG3aFFtbW3r06HHPGYUQQphY1hX4azxUaQidXjN1mmKhFTbOrKzw9/dX/57McezYMRo2bGiiRKZ3Y5e6uD15Vsa5l+f0oP49vL4yhbg9eU7Gk2dlnDL/nP4cC4f/gNHrwM2v2G5T3M9J07RIpZR/YeekhVMIIYQQwlSO/Q0HF0P7CcVabJpamSw4NU3ro2naj6mpqaaOIoQQQghxbzKS4e8XoJovdJhg6jTFqkwWnEqpFUqpsRUqlO19RYUQQgjxAAt9CbJSYMAPYG55x5eXZWWy4BRCCCGEKNMO/wFHlxkmCbk2NnWaYicFpxBCCCFESUq7CKEvQw1/wzJIDwApOIUQQgghSopSsOJ/kJd1rSu9fO/Bc50UnKXM6NGjC91Kcc6cOTz33HP3fF0HB4f7iVWuLFu27KZnPGXKFNauXVuk99i/fz9hYWF3/b4LFy4waNCgu3rP9OnT8fb2RtM0kpKSCo6npqbSp08fmjZtSuPGjZk9e3ah7x8+fDj169fHx8eHJ554omAx/U8++QQ/Pz8CAwPx8fHB3Nycy5cv3/XXJIQQ4gZ7ZsHJlYatK13qmjpNiZGCsxgppQq2gjTWzJkzadSoUTElKn7/3t6xNPp3wfnOO+/QtWvX+7rmv7/u2xWct3tGbm5uhe6OdDuBgYGsXbuW2rVr33T822+/pVGjRhw4cICIiAhefvnlQrcjHT58OMePH+fQoUNkZWUxc+ZMACZOnMj+/fvZunUrH374IR07dsTZ2fmusgkhhLjBrp8MXene3aD1M6ZOU6Kk4CxiMTExNGzYkHHjxtG8eXNiY2P55JNPaNmyJb6+vrz11lsAZGRk0KtXL5o2bYqPj0/BjjdBQUFcX8x+9uzZ1KtXj44dO7J169aCezz22GM3FSXXWy/T09Pp06cPzZs3p0mTJixfvvyOeX/55Rd8fX1p2rQpI0eOBODs2bN06dIFX19funTpwrlz5wBYsWIFrVu3plmzZnTt2pWLFy8CMHXqVMaOHUtwcDCPPvooR44coVWrVvj5+eHr60tUVBQA8+fPLzj+1FNPFbplZHh4OA0aNKBdu3Y8//zzBbswTZ06tWAfcQAfHx9iYmIA6N+/Py1atKBx48b8+OOPNz2X119/naZNmxIQEMDFixfZtm0bf/31FxMnTiQwMJDTp08XPM89e/bg5+eHn58fTZo0KdjH/fTp04SEhNCiRQvat29fsPvTY489xksvvUSnTp2YNGlSwX1zc3OZMmUKixcvLthx6d/PKCYmhvbt29O8eXOaN29esDNUTExMwfaVc+bMYeDAgYSEhFC3bl1eeeWVQv8MmzVrhoeHx3+Oa5pGWloaSinS09NxdnbGwuK/XTc9e/ZE0zQ0TaNVq1bExcX95zU3brt5q+9dIYQQt7HlCwibAPV7wpD5YPZglWDleuDAtF3TOH65aPeYbuDcgEmtJt32NSdOnGD27Nl89913rF69mqioKHbt2oVSir59+7Jp0yYuXbqEm5sboaGhgKH780bx8fG89dZbREZGUqFCBTp16kSzZs1ue18bGxsWLFhAjRo1SEpKIiAggL59+xYUTv925MgR3n//fbZu3YqLi0tBd+lzzz3Ho48+yqhRo/j55595/vnnWbZsGe3atWPHjh1omsbMmTP5+OOP+eyzzwCIjIxky5Yt2NraMn78eP73v/8xfPhwcnNz0el0HDt2jMWLF7N161YsLS0ZN24cCxYs4NFHHy3Ik52dzZgxY1i/fj3e3t4Fe8zfyc8//4yzszNZWVm0bNmShx56iMqVK5ORkUFAQADvv/8+r7zyCj/99BNvvPEGffv2pXfv3nTv3v2mHXT8/f3Zv38/YGjdCwkJAWDs2LH88MMP1K1bl507dzJu3DjWr18PGPZwX7t2Lebm5gXXsbKy4p133mHPnj1Mnz4dMBTMNz6jzMxM1qxZg42NDVFRUQwbNox/75oFhpbSffv2YW1tTf369Rk/fjw1a9Y06rk899xz9O3bFzc3N9LS0li8eDFmt/kBl5eXx7x58/jqq69uOp6ZmUl4eHjB1xIeHn7b710hhBA3UAo2vA+bPgGfh2DAjCJfAikzLxM7S7sivWZRK9cFp6nUrl2bgIAAAFavXs3q1asLisX09HSioqJo3749EyZMYNKkSfTu3Zv27dvfdI2dO3cSFBRElSpVABgyZAgnT5687X2VUrz99tvs2LEDMzMzzp8/z8WLF6lWrVqhr1+/fj2DBg3CxcUFoKC7dPv27fz5558AjBw5sqBlLS4ujiFDhhAfH09ubi6enp4F1+rbt2/BXudt2rTh/fffJy4ujoEDB1K3bl3WrVtHZGQkLVu2BCArK4uqVavelOf48eN4enpSt65hTMuIESNuarG8la+//pqlS5cCEBsbS1RUFJUrV8bKyqqghbRFixasWbPmjtcC+O2339i7dy+rV68mPT2dbdu28fDDDxecz8nJKfj44YcfvqnYvJ0bn1FeXh7PPfcc+/fvx9zc/JZ/tl26dOH6erONGjXi7NmzRhecq1atws/Pj/Xr13P69Gm6detG+/btcXJyKvT148aNo0OHDv/5Xly5ciWBgYEF3x9NmjS57feuEEKIa5SCVa/Djm+h2Ujo8xWYGfdvhrF0eh2DVgyiu0d3/tf8f0V67aJUrgvOO7VEFhd7e/uCj5VSvPrqqzz11FP/eV1kZCRhYWG8+uqrBAcHM2XKlJvO36pl0sLComBsqFKqYFzeggULSE5OJjIyEktLSzw8PMjOzr5lTqXULe9RWI7x48fz0ksv0bdvXyIiIpg6dWqhX/MjjzxC69atCQ0NpXv37sycOROlFKNGjeLDDz806l63+5qBgq8rIiKCtWvXsn37duzs7AgKCio4Z2lpWXA9c3Nzo8aXHjlyhLfeeotNmzZhbm6OXq+nYsWKBS2f/3bj130nN772iy++wNXVlQMHDqDX67GxsSn0PdbW1gUfG/s1XDd79mwmT56Mpml4e3vj6enJ8ePHadWq1X9e+/bbb3Pp0iVmzJjxn3N//PFHQXc6QL169e74vSuEEA88vQ7+fhH2zoXWT0P3D4ulG3197Hpi02JpVNkw/yN1+XKsatfG1s+vyO91Px6sAQQm0L17d37++WfS09MBOH/+PImJiVy4cAE7OztGjBjBhAkT2Lt3703va926NRERESQnJ5OXl8eSJUsKznl4eBAZGQnA8uXLC2YVp6am4uLigqWlJRs2bODs2bO3zdalSxd+++03kpOTAQq61Nu2bcuiRYsAQxHbrl27guvXqFEDgLlz597yumfOnMHLy4vnn3+evn37cvDgQbp06cLvv/9OYmJiwb3+na9BgwZER0dz+vRpwDBu8Mav+foz2rt3L9HR0QWZKlWqhJ2dHcePH2fHjh23/ZoBHB0dSUtL+8/x1NRUhg4dyi+//FLQsuzk5ISnp2fB81dKceDAgXu+x433ql69OmZmZsybN6/Q8az3q1atWqxbtw6AixcvcuLECby8vP7zupkzZ7Jq1SoWLlz4ny731NRUtmzZQr9+/QqO3el7VwghHni6fFj6tKHYbP8yhHxUbGM25xyZg7uDO51rdkbpdCS8+x4pfy4tlnvdDyk4i1lwcDCPPPIIbdq0oUmTJgwaNIi0tDQOHTpUMIHm/fff54033rjpfdWrV2fq1Km0adOGrl270rx584JzY8aMYePGjbRq1YqdO3cWtJwNHz6cffv24e/vz4IFC2jQoMFtszVu3JjXX3+djh070rRpU1566SXA0EU9e/ZsfH19bxrTN3XqVB5++GHat29f0A1fmMWLF+Pj44Ofnx/Hjx/n0UcfpVGjRrz33nsEBwfj6+tLt27diI+Pv+l9NjY2/Pjjj/Tq1Yt27drdNOv6oYce4vLly/j5+fH9999Tr149AEJCQsjPz8fX15c333yzYCjD7QwdOpRPPvmEdu3aFRS3YJi9fvbsWcaMGVMweQgMRfesWbMKlhcyZjJWp06dOHr0aMGkoX8bN24cc+fOJSAggJMnT95VS+m/ff3117i7uxMXF4evry+jR48G4M0332Tbtm00adKELl26MG3atII/t549e3LhwgUAnn76aS5evEibNm3w8/PjnXfeKbj20qVL6dy580357vS9K4QQD7T8HFgyCg79Zlj6qMsUMKI38V7sT9zPwUsHGdloJOZm5mQfO44+PR27a8PXShNNKWXqDPfM399f/XuixbFjx2jYsKGJEpleWlraTRNhyrKIiAg+/fRT/v7772K5fnl6VsXpXp7Tg/r3MCIigqCgIFPHKPXkORlPnpVxSs1zys2ExSPg9DoImQYBTxfr7V7Y8AK7E3azZtAa7CztSJ49h8Rp0/DeGIGlq+t/Xl/cz0nTtEillH9h56SFUwghhBDifuWkwYJBcHo99J1e7MXmuavnWH9uPUPqDymYoZ65ezeWtWth6epKYlo2en3paVSUglOUWkFBQcXWuimEEEIUmczL8Es/OLcDHpoJzUcW+y3nHZ2HhZkFwxoYJnUqnY7MPXsKutOfmhfJ43N2F3sOY0nBKYQQQghxr9Ivwdw+kHAIhsyDJne3PfG9SMlOYdmpZfTy6kUVO8Mk15yTJ9FfvYp9y5ZcSsthf2wKLWpXKvYsxirXyyIJIYQQQhSb1POGls3UOHhkMdTpXCK3/e3kb2TrshnVaFTBsczdhtZMu5YtWXU8EaWga8P/juM0FWnhFEIIIYS4W5ejYXYIpCXAyD9LrNjM0eXw67FfCawRiHcl74Ljmbt3Y+nujqWbG2uPXcStgg0Nq5eeibFScAohhBBC3I1LJ2F2D8NEoVF/Qe22JXbrsDNhJGcn39S6qfR6Mnftxq5lS7LzdGyOSqJLQ1ejNncpKVJwFrHY2Fg6depEw4YNady48X/2pb6Rg4NDkd+/evXqRr926tSpfPrpp3d9j5SUFL777ru7fh8Y1n9MSUm5p/f+m7m5ecF6mX379i30NefOnSvYh97X15ewsLCCc1OmTMHHxwcfH59C18oUQggh/iP+oKHY1OvgsVCo0fzO7ykiSinmHplL/Ur1Caj+z7rTOVGn0KWmYteyJdtPJ5OVp6Nro9LTnQ4yhrPIWVhY8Nlnn9G8eXPS0tJo0aIF3bp1o1GjRqaOVmSuF5zjxo0z+j1KKZRSNxV898vW1vaWW05e99577zF48GCeeeYZjh49Ss+ePYmJiSE0NJQDBw6wf/9+cnJy6NixIz169LjlPuNCCCEEcXtg/kCwcoRHl4OL953fU4S2nN/C6dTTfNDug5taLwvGb7ZqxdrdF7G3MifAy7lEs92JtHAWserVqxfsCuTo6EjDhg05f/78Hd/3ySef0LJlS3x9fXnrrbcAmDRp0k0tiVOnTuWzzz675etv55dffsHX15emTZsycuR/l2sICgri+iL6SUlJeHh4AIa9xa/vKuPr60tUVBSTJ0/m9OnT+Pn5MXHixFvmiYmJoWHDhowbN47mzZsTGxuLh4cHSUlJBefGjBlD48aNCQ4OJisrC4Ddu3fj6+tLmzZtmDhxIj4+Pnf8+m5F0zSuXr0KGLZpdHNzA+Do0aMEBgZiYWGBvb09TZs2JTw8HIDJkyfTqFEjfH19mTBhwj3fWwghRDkSvdkwQcjWGZ5YWeLFJsDcI3OpaleVEI+Qm45n7t6NhVt1LGu4se5YIu3rVsHawrzE891OuW7hTPjgA3KOHS/Sa1o3bEC1114z6rUxMTHs27eP1q1b3/Z1q1evJioqil27dqGUom/fvmzatImhQ4fywgsvFLQk/vbbb4SHh9/y9R06dCj0+keOHOH9999n69atuLi4FOyZbowffviB//3vfwwfPpzc3Fx0Oh0fffQRhw8fLmhdvFWeWrVqceLECWbPnl1oF3xUVBQLFy7kp59+YvDgwfzxxx+MGDGCxx9/nB9//JG2bdsyefLkW2bLzs7G398fCwsLJk+eTP/+/f/zmqlTpxIcHMw333xDRkYGa9euBaBp06ZMmTKFV199lczMTDZs2ECjRo24fPkyS5cu5fjx42iaVmTd/0IIIcqwqLWweDhU8oCRy8DJ+OFrReX45ePsTNjJiy1exNLcsuC4UorM3btxaN+OIxeuknA1my4Nq5Z4vjuRFs5ikp6ezkMPPcSXX355x27a1atXs3r1apo1a0bz5s05fvw4UVFRNGvWjMTERC5cuMCBAweoVKkStWrVuuXrb2X9+vUMGjSoYB9tZ2fjm9nbtGnDBx98wLRp0zh79iy2trZG5weoXbv2Lfc39/T0LNivvEWLFsTExJCSkkJaWhpt2xoGYD/yyCO3zHbu3Dn27NnDr7/+ygsvvHDTvujXLVy4kMcee4y4uDjCwsIYOXIker2e4OBggoODadu2LcOGDaNNmzZYWFjg5OSEjY0No0eP5s8//8TOzs7oZyWEEKIcOvoXLBwKLvUMYzZNUGyCoXXTzsKOQfVuXucz9/RpdJcvY9eyJWuPXUTToHOD0ldwlskWTk3T+gB9vL1v35xtbEtkUcvLy+Ohhx5i+PDhDBw4EDBMJurTpw8ATz/9NE8//c+WV0opXn31VZ566qn/XGvQoEH8/vvvJCQkMHTo0Du+vjBKqTvOVLOwsECv1wOGlsPrHnnkEVq3bk1oaCjdu3dn5syZeHl5/ef6heWJiYnB3t7+lve0trYu+Njc3JysrCyUMn4bruvd415eXgQFBbFv3z7q1Klz02tmzZpV0FXepk0bsrOzSUpKomrVqkycOJF33nmn4OusW7cuFhYW7Nq1i3Xr1rFo0SKmT5/O+vXrjc4khBCiHDmwGJY9AzVawPAlYFvRJDESMhIIjw5naIOhOFnd3Ih10/jN5WdpXqsSlR2sC7uMSZXJFk6l1Aql1NgKFSqYOsp/KKV48sknadiwIS+99FLB8Zo1a7J//372799/U7EJ0L17d37++WfS09MBOH/+PImJiQAMHTqURYsW8fvvvzNo0KA7vr4wXbp04bfffiM5ORmg0C51Dw8PIiMjAfj9998Ljp85cwYvLy+ef/55+vbty8GDB3F0dCQtLc2o/HerUqVKODo6smPHDgAWLVpU6OuuXLlCTk4OYBhzunXr1kInZtWqVYt169YBcOzYMbKzs6lSpQo6na7geRw8eJCDBw8SHBxMeno6qamp9OzZky+//PKOk5KEEEKUU3tmw9KnwCMQRi41WbEJ8OuxX9GjZ0SjEf85l7l7NxauriQ7VeHw+aulsjsdymgLZ2m2detW5s2bR5MmTQq6iz/44AN69ux5y/cEBwdz7Ngx2rRpAxiWS5o/fz5Vq1alcePGpKWlUaNGjYIlj273+sI0btyY119/nY4dO2Jubk6zZs2YM2fOTa+ZMGECgwcPZt68eXTu/M/itYsXL2b+/PlYWlpSrVo1pkyZgrOzM4GBgfj4+NCjRw8++eSTQvOYm9/bgOVZs2YxZswY7O3tCQoKorBfLI4dO8ZTTz2FmZkZer2+YKIPGJY78vf3p2/fvnz22WeMGTOGL774Ak3TmDNnDpqmkZeXR0hICGZmZjg5OTF//nwsLCxIS0ujX79+ZGdno5Tiiy++uKevQQghRBmlFGz7GtZMgbrdYfBcsPzvcLKSkp6bzpKTSwiuHUwNhxo3nVNKkbFrN/YBAaw/YWjo6VaKdhe6kXY3XZiljb+/v7o+s/q6Y8eO0bBhQxMlMr20tDQcHUvPzgL3Ij09vWCN0o8++oj4+Pjbrmd6r8rDsyoJ9/KcHtS/hxEREQQFBZk6Rqknz8l48qyMU2TP6eIRCH0Zzm2HRv1h4E9gYXX/170Pvxz5hU/2fMLCXgvxcbl51ZacM9Gc6dmTam+/zctZnpxKTGfjxKBbDqMr7u8nTdMilVL+hZ2TFk5R6oSGhvLhhx+Sn59P7dq1/9MaK4QQQhSpnDTY8CHs/AFsKkDfb8BvBJiZduRhvj6f+cfm08K1xX+KTfhn/KZZs+Zs+SWK4a1rlardhW4kBacodYYMGcKQIUNMHUMIIUR5pxQc+RNWvW7YE73FKOjyFtiVjkXT15xdQ3xGPK+2erXQ85m7d2NexYWdufbk5utLbXc6lNOC05hZ2UKI4lGWh+kIIR4gSVEQNgHORED1pjBkAbi3MHWqAkop5hyZg4eTBx1rdiz0fOauXdi3bMm645dwtLGgpWfpKJQLU+4KThsbG5KTk6lcubIUnUKUMKUUycnJ2NjYmDqKEEIULjcTNn0C274BSzvo+Sn4PwFmpWtnnj0X93A0+ShvBryJmfbfrv28c+fIT0zE1r8l644n0rFeFSzNS+/iQ+Wu4HR3dycuLo5Lly6ZOopJZGdnyz/2RpJnZZy7fU42Nja4u7sXYyIhhLgHSsGJMFg5GVLPQdNh0O0dcCidywj9cuQXKllXom+dvoWevz5+81ytBiQdvEDXUtydDuWw4LS0tMTT09PUMUwmIiKCZs2amTpGmSDPyjjynIQQZd7laFg5CaJWQdVG8PhKqN3W1Klu6UzqGSLiInim6TPYWBT+C3/m7t2YV67M2nRbzM00gupXKeGUd6fcFZxCCCGEEADkZcPWr2DL52BmAcHvQ+un4Ia9yEujeUfnYWVmxZD6hU+gvb7+pp2/P2uPJ+JfuxIV7Uy7fNOdlN7OfiGEEEKIexW1Fr5vAxEfQP2e8NxuaPtcqS82k7OS+evUX/T17ktl28qFvibv/Hny4+PJ9fHjeEJaqe9OB2nhFEIIIUR5khoH4a/Csb+gsjeMXAZ1Opk6ldEWn1hMrj6XkY1G3vI1mbsM4zf3VPAAMujaSApOIYQQQojil58LO76DjR+D0kPnN6HteLCwNnUyo2XnZ7Po+CKC3IPwquB1y9dl7t6NecWKrLhqjVcV8HSxL8GU90YKTiGEEEKUaRWvHIIfXoGkE4bu85CPoFJtU8e6a3+d/osrOVd4tPGjt31d5q5dWDZvwY7oKzweWDYmSkvBKYQQQoiyKS0BVr+B36ElULEWDFsM9UNMneqe6JWeeUfn0bhyY/xdC92OHIC8CxfIO3+ehOAB5F1RZWL8JkjBKYQQQoiyRpcPu2fChvchP5uY2oPxGPE1WNqaOtk92xi7kZirMXzc4ePbblxzff3NCBt3KtpZ0rxWxRJKeH+k4BRCCCFE2RG7C0JfgoRDUKcz9PyUmEOxeJThYhNg7tG5VLevTrfa3W77uozduzFzcuLPFGs6NaiKRSneXehGUnAKIYQQovTLSIa1U2DffHB0g4fnQqN+oGlArKnT3ZfDSYeJvBjJRP+JWJjdvjTL3LWb3MZNuZylo0vD0rlLUmGk4BRCCCFE6aXXw965sO5tyEmDts9Dx0lg7WDqZEUiT5/Hl3u/xMHSgYF1B97+tRcvknfuHMeadcHSXKNDvdK9u9CNpOAUQgghROl0YR+EvgznI6F2IPT6DKo2NHWqIqPT63h186vsjN/JW23ewsHq9kX09fU3Q82q0dqzMk42pXsR+xtJwSmEEEKI0iXrCqx/D3bPAvsqMOBH8B18rfu8fNArPW9te4tVMauY4D+BQfUG3fE9mbt3g4MDm/WVeLMMdaeDFJxCCCGEKC2UggOLYPUbkHUZWo2FTq+BbUVTJytSSik+3Pkhy08vZ1zTcYxqPMqo92Xu2sVlr0boNbMysxzSdVJwCiGEEML0Lh6B0AlwbhvU8IeRf0L1pqZOVeSUUnyx9wsWnVjEY40f4+mmTxv1vryEBHJjYtjToRX1XR2p6WxXzEmLlhScQgghhDCdy9Gw/VvY8zPYVIA+X0OzkWBWNpb7uVs/HvyR2YdnM7jeYF5q8dJt19y8UdKMGWBuzu82XvQpY93pIAWnEEIIIUqaUnBuu6HQPB4KZubQfCR0eQvsnE2drtj8cuQXpu+fTt86fXk94HWji82cqChSFv/GlW59iLV1oWujstWdDlJwCiGEEKKk6PLg6HLYPt0wA922ErR/CVqOAafqpk5XrJacXMInez6hW+1uvN32bcw041twL077GDMHB/707YlLfA5+7hWLL2gxkYJTCCGEEMUr6wpEzoVdP8LV81DZG3p9Dk2HgVXZGot4L1acXsG729+lfY32TGs/7Y6Lu98offNmMrZsofLEiayKzSKkcTXMzMrebH0pOIUQQghRPJJPw84fYN8CyMsAzw7Q+wvw7lZux2j+29qza3lz65u0rNaSz4M+x9Lc+LUzVX4+F6dNw7J2LbY37Uxa1BGCG1crxrTFRwpOIYQQQhQdpeDsVtj+HZwIAzMLaPIwtBkH1ZqYOl2J2hy3mYmbJuLj4sM3nb/BxsLmrt6fsmQJuadO4/rVV3yy7gyNqjvRpUHZmzAEUnAKIYQQoijk58KRpbDjW4g/ALbO0GECtBwNjmWzVe5+7E7YzYsRL1K3Yl2+6/oddpZ3N3RAd/Uql77+BruWLfnTtg5xV44z/0nfMtmdDlJwCiGEEOJ+ZF6GyNmw6ydIiweX+tD7S2g6FCxtTZ3OJA5cOsCz657F3cGdGd1m4GTldNfXSPphBrqUFOxeeJlv/j5Nh3pVaFfXpRjSlgwpOIUQQghx95JOwY7v4MBCyMsEr07QdzrU6fzAjM8szPHLx3lm7TO42LrwY/CPVLKpdNfXyD13jivz5lGhf39mXrTianYek0MaFEPakiMFpxBCCCGMoxTEbDasn3kyHMytDHucB4wD18amTmdyZ1LOMHb1WOwt7ZkZPJOqdvc23jLx08/A0pL8x59i9i9HGdjMnUZud99KWppIwSmEEEKI28vPhcN/GArNi4fAzgU6ToaWT4JD2ZzEUtRir8YyevVozDQzZgbPxM3B7Z6uk7l7N2mrV+Py/Hg+3HcFDXg5uF7RhjUBKTiFEEIIUbiMZIj82TA+M/0iVGkIfb+BJoPB8u5mXJdnCRkJjF49mlx9LrO7z6a2U+17uo7S67n40TQsqlXjYveHWDpzD093rINbxbI/FlYKTiGEEELc7NLJf8Zn5meDd1cI+N4wPtPI7RgfFElZSYxZPYaruVeZ2X0mdSvVvedrpf71F9lHjlB92kc8uyGairaWPBNUpwjTmo4UnEIIIYQwjM88E2HoNj+1BsytoekQw/jMqg1Nna5USslOYeyasVzMvMgPXX+gceV7H8eqz8zk0udfYNOkCQfqB7B1+x6m9G6Ek43xC8WXZqWq4NQ0zR7YBLyllPrb1HmEEEKIci8/Bw4tMSzUnngE7KtA0Gvg/wQ4VDF1ulIrPTedp9c+zdnUs0zvMp3mrs3v63rJP88mPzGR6p99xjPhJ6jlbMeIgHvrmi+NirXg1DTtZ6A3kKiU8rnheAjwFWAOzFRKfXTt1CTgt+LMJIQQQgggIwl2z4LdMyEjEao2hn7fQZNBYGFt6nSlWmZeJs+ue5YTl0/wRacvaOPW5r6ul3fxIsmzZuEYEkK4WTWOJxzgm2HNsLIoP8tLFXcL5xxgOvDL9QOappkD3wLdgDhgt6ZpfwFuwFFARiELIYQQxSXx2LXxmYtBlwN1gw3d5l5BMj7TCLm6XF7Y8AL7L+1nWvtpBNUMuu9rXvriS8jPp8L/XuCzRSdo6l6B3r7V7/u6pYmmlCreG2iaB/D39RZOTdPaAFOVUt2vff7qtZc6APZAIyALGKCU0hdyvbHAWABXV9cWixYtKtb8ZU16ejoODg6mjlEmyLMyjjwn48mzMo48J+MV2bNSikpX9lMzdjnOV/ahM7Piomsn4tz7kmnvfv/XN7GS+p7SKR2zLs3iUNYhhlceToBDwH1f0+LsWSp/+BEZwcEs8uvF7yfzmNzKhgbO5kWQ+GbF/Zw6deoUqZTyL+ycKcZw1gBib/g8DmitlHoOQNO0x4CkwopNAKXUj8CPAP7+/iooKKhYw5Y1ERERyDMxjjwr48hzMp48K+PIczLefT+rvGw4uBh2fA+XjoGDK3R+A/MWT+BmX5l7Wymy9CmJ7ymdXserm1/lUNYhXmv9GsMaDLvvayqlODtyJLnOzni8/harvttN14ZVeXpgyyJI/F+m/LtnioKzsPb6gmZWpdSckosihBBClEPpif+Mz8xMgmpNoP8P4DNQxmfeA73S8/b2t1kZs5IXW7xYJMUmQNqaNWTtiaTa1Kl8vSuejNx8JpXxLSxvxRQFZxxQ84bP3YELJsghhBBClC8Xjxhmmx/6DXS5UK8HtBkHHu1lfOY9Ukrx8e6PWXpqKU/5PsUTPk8UyXX1ubkkfvIp1nW9Se0UwvyvtzKkZU3qujoWyfVLG1MUnLuBupqmeQLngaHAIybIIYQQQpR9ej2cXgfbpxvW0bSwheaPQutnwMXb1OnKvG/2fcOCYwsY2Wgkz/o9W2TXvTJvPnmxsdScOZPJ605jYWbGi13L/haWt1LcyyItBIIAF03T4jCsrzlL07TngFUYlkX6WSl1pDhzCCGEEOVOeiIc+h0i50DSCXCsDl3eghaPgZ2zqdOVCz8d/ImfDv3EoHqDmOg/Ea2IWonzL18m6fvvse/YgVO1G/P331t5vktdqjqV34V6irXgVEoVOshBKRUGhN3rdTVN6wP08faW39yEEEI8QPKy4HioYSLQqXWgdODWHAb8CI0HgIWVqROWG/OPzufrfV/Ty6sXb7R+o8iKTYCk6dPRZ2VRdeJEJoQdw8XBirEdvIrs+qVRqdppyFhKqRXACn9//zGmziKEEEIUK70ezm0z7Gt+9C/IuQpONSDwf9B0KFSpb+qE5c4fJ/9g2u5pdKnVhfcC38PcrOiWKMo5dYori3+j0pAhbMl1ZFf0Cd7t74ODdZksyYxWvr86IYQQoqxKisLzzHzYNx5Sz4GVAzTqZygya7cDs/KzC01pEnYmjLe3v01gjUA+7vAxFmZFWypd/PhjzOzsqDRuHB/9chgvF3uGtqx55zeWcVJwCiGEEKVFRjIc+dPQmnk+klqYQZ1O0GUKNOgFVnamTliurTu3jte2vEYL1xZ8EfQFVuZFO0QhffMWMjZtpuorr/DnmQxOJabzw4gWWJqX/18epOAUQgghTCk/B06GG7aajFoF+nxwbQLB77E9rQZtuw80dcIHwtbzW5m4cSKNKzdmepfp2FrYFun1VX4+iR9Pw7JWLWwGD+Hzr7bhX7sS3Ru7Ful9SispOIUQQoiSphTE7jK0ZB5ZCtkp4FANAp4B36FQzQeA3IgIk8Z8UOxJ2MMLG17Aq4IX33X9DntL+yK/R8rvv5MTdYoa33zNzB3nuZSWww8jWhTpZKTSTApOIYQQoqRcPgMHf4MDi+BKtGHNzIZ9DOMyvYKgCCenCOPsvbiX59Y/R3WH6szoNoMK1hWK/B66tDQuff0Ndi1bkt26PTM+jaCHTzVa1K5U5PcqrcpkwSnLIgkhhCgzsq4YWjEPLIbYHYAGnu2h4yuGYtO6fO4sU9rp9DpmHZ7Fd/u/o4ZDDX7q9hOVbSsXy72SZ8xAd+UKVSdP4v31UeTm65nY/cFaXaBMFpyyLJIQQohSLT8XTq2Fg4vgxErDNpMu9Q0Ls/sOhgrupk74QLuYcZHXtrzGroRd9PDowZtt3sTRqngK/9zYWC7P/YUK/ftzoWptFi7YxPDWtfCq4lAs9yutymTBKYQQQpQ6SsGFvYbu8sN/QGYy2LmA/5PQdAhU95P9zEuBiNgI3tz6Jjm6HN4NfJd+dfoV6zjKxE8/AwsLqrzwAs+FH8fGwoznu9QttvuVVlJwCiGEEPcjJdaw88+BRZAcBebW0KAnNB0GdTqDuaWpEwogR5fDZ3s+Y+HxhTR0bsi0DtPwrOBZrPfMjIwkbdUqXMY/x4EsC1YducjL3erh4mBdrPctjaTgFEIIIe5W9lU4utxQaMZsNhyr1Rbajjcszm5b0aTxxM3OpJxh4qaJnLxykpGNRvJC8xeKfI3Nf1N6PRc//AgLV1ecH3uMp+bux9XJmtHty/cWlrciBacQQghhDF0+nNlgWMroeCjkZ4NzHej0umFcZiUPUycU/6KU4s+oP/lo10fYWdrxbZdv6eDeoUTufXXFCrIPH8Zt2kesPnOVvedSmPZQE2ytHsyVCKTgFEIIIW5FKUg4ZOguP7QEMhLBthI0G2FYL9PdX8ZlllJXc6/y9ra3WX12NQHVA/ig3QdUsatSIvfWZ2WR+PkX2Pj4YNuzF9O+3Ew9Vwceav7gThaTglMIIYT4t6sXDAXmgUWQeBTMLKFed8O4zLrBYFG83bHi/uxP3M8rm17hUuYlXmzxIo81fgwzreS2j0z6/gfyL16kxmefsmhPHDHJmfz8mD8WD8AWlrdSJgtOWYdTCCFEkctJh+N/G4rMMxGAAvdW0OszaDwQ7JxNnVDcgU6vY+ahmXx/4Huq21fnlx6/0KRKkxLNkDxrFsk//kiFAQPQ+TTlq08iCPByplP9qiWao7QpkwWnrMMphBCiSOh1EL3JUGQeWwF5GVCxFnSYaNj9p3IdUycURkrISODVza+y5+Ieenj2YErAFBysSnaty6SffuLSZ5/j1LMH1d99h8/WnSY5I5fZPRs+MFtY3kqZLDiFEEKI+3LxqGFR9oNLIO0CWFeAJoMMRWbNADB7cLs+y6KDmQd5Y8Ub5OpyeS/wPfrW6VviBV7SDzO49OWXOPXqhdu0j7iYkc/MLWfo29QNX/eKJZqlNJKCUwghxIMhPfGfcZkJB0Ezh7rdIOQDqNcDLG1MnVDcpez8bD7b8xmLLi2ioXNDPu7wMR4VPEo8x6XvviPp629w6tMHtw8/QLOw4Is1R9Hp1QO3heWtSMEphBCi/MrLMixhdHAxnFoHSmfY8SdkGvg8BA4lM2tZFL3TKaeZuGkiUVei6OTYiU97flrsa2sW5tL0b0maPp0K/fpR/YP30czNmb/jLIv3xPJkO09qOtuVeKbSSApOIYQQ5YteD+e2GdbLPPoX5FwFpxoQ+LxhKaOqDUydUNwHpRS/R/3Ox7s+xs7Sju+6fIfulK7Ei02lFEnffEPSd99TYcAAqr/3Lpq5ObO2RPPu30fp3KCqtG7eQApOIYQQ5UNSlKG7/OBvkHoOrBwMu/74DgGP9jIusxxIzUnl7e1vs+bsGtpUb8MH7T/AxdaFiFMRJZpDKcWlL78iecYMKgx6iOrvvINmZsa3G07xyaoT9PCpxldDm2FlId9z10nBKYQQouxKu3hti8lFcD4SNDPw6gRdphj2M7eyN3VCUUT2XtzLpM2TSMpM4qUWLzGq8agSXVvzOqUUlz7/nOSfZlJx8GCqTX0LNI3P15zk63VR9PNz47OHmz7Qa24WRgpOIYQQZcvVC4YljI4uh7PbAAWuPhD8HjR5GByrmTqhKEI6vY4fD/3IDwd+oIZDDeb1nIePi49JsiilSPzkUy7//DMVhw2l2ptvgqbx0crjzNh0hsH+7nw40Bdzswd7CaTCSMEphBCi9Es5ZxiPeXQ5xO0yHKvaCIImQ8O+4NrItPlEsUjISGDy5slEXoykl1cv3mj9RomvrXmdUorEj6Zxee5cKj3yCK5vvoFS8M6Ko8zZFsOIgFq809cHMyk2C1UmC07ZaUgIIR4Al8/8U2Re2Gs4Vq0JdH4DGvaDKvVMm08Uq3Vn1zFl2xTy9fl80O4D+tTpY7IsSikufvghV36ZR6URI3B9/TX0Cl5feohFu2MZ3c6T13vJ4u63UyYLTtlpSAghyqmkU3B0maHITDhoOObWDLpONbRkys4/5V52fjaf7vmUxScW06hyIz7u8DG1nWqbLI9Siovvvc+VBQtwHvUoVSdPNqyv+ftBlu47z/jO3rzUrZ4Um3dQJgtOIYQQ5UjicUOBeXQ5JB4xHHNvaRiT2bAvVDJdsSFKVtSVKF7Z9AqnUk7xWOPHeL7Z81iaW5osj9LrSXj3XVIWLsL58cep+spE8vWKFxbtJ/RQPBOC6/Fc57omy1eW3LHg1DTtD+BnYKVSSl/8kYQQQpRrSkHC4X+KzKQTgAa12hgWZG/YByrUMHVKUYKUUiw5uYSPd3+MvaU9P3T9gcAagabNpNeT8PY7pCxeTOXRT1Ll5ZfJ1el5dsE+1h67yBu9GjK6vZdJM5YlxrRwfg88DnytadoSYI5S6njxxhJCCFGuKAXx++HoclpFLoKN8YYljGoHQqsxhiJTZpc/kFJzUpm6bSprz60l0C2Q99q9h4uti0kzKb2ehLfeImXJ71QeO5YqL75Adp6ep+ZHsunkJd7t15iRbTxMmrGsuWPBqZRaC6zVNK0CMAxYo2laLPATMF8plVfMGYUQQpRFShnWxrw+JjPlHGjmZFdsgl2XSdCgt2wt+YDbk7CHyZsnk5ydzAT/CYxsNNIka2veSOn1xL/5Jql//EnlZ56myvPPk5mr48m5u9kZfZmPH/JlcMuaJs1YFhk1hlPTtMrACGAksA9YALQDRgFBxRVOCCFEGaPXG5YtOrrcMMP8ahyYWUKdTtDhFWjQi4O7DhLkH2TqpMKE8vX5/HjwR2YcnIG7gzvze8ynsUtjU8dC6XTEv/4GqcuW4fLss7g89yxpOfk89vMuDsSl8uUQP/r5yXCPe2HMGM4/gQbAPKCPUir+2qnFmqbtKc5wQgghygC9Ds5t/6fITE8Ac2vw7gJd3oR6IWBb0dQpRSkRnx7P5M2T2Zu4lz5efXg94HXsLU2/I5TS6Yh/7TVSl/+Fy3PPUeW5Z0nJzOXRn3dxLP4q04c1o0eT6qaOWWbdtuDUNM0M2K+UGljYeaWUf7GkEkIIUbrp8iFms6HIPP43ZFwCCxuo2w0a9Ye6wWDjZOqUopRZe3Ytb217q1SsrXkjlZ/PhcmvcvXvv6nyv+dxeeYZktJzGDFzJ2cuZfDDiBZ0aehq6phl2m0LTqWUXtO0HsA7JZRHCCFEaZWfC9GbDGMyj4dC1mWwtId6wdCoH3h3A2vT7AIjSres/Cw+2f0JS04uoXHlxnzc4WNqOdUydSzgWrH5yiSuhoVR5cUXcXlqLBevZjN85k7irmQy6zF/2teVscb3y5gxnKs1TXsI+FMppYo7kBBCiFIkPwdObzC0ZJ4IhexUsHKE+j2gUV+o0wWs7EydUpRiN66t+XjjxxnfbLxJ19a8kcrL4/zEV0gLD6fqhJepPHo051OyGP7TDi6l5TDn8VYEeFU2dcxywZiC8yXAHtBpmpYFaIBSSpmsr0S2thRCiGKUlwWn1hrGY55YCblpYFMB6vcytGR6BYGljalTilJOKcVvJ37jkz2f4GDpwIyuM2hbo62pYxVQeXmcf3kCaatXU/WVV6j8xOOcS85k2E87uJqdx7zRrWleq5KpY5YbxiyL5FgSQe6GbG0phBBFLDcDolYbWjJProa8DLCtBI37G8ZkenYACytTpxRlREp2Cm9te4v1sesJrBHIe4GmX1vzRio3l/Mvv0zamrVUnTyJyo89xulL6Tzy0w5y8vUsHBOAT40Kpo5ZrhgzS10DhgOeSql3NU2rCVRXSu0q9nRCCCGKT/bVa0XmMohaC/lZYOcCvoMNLZke7aCUdH2KsmN3wm5e3fxqqVpb80YqN5e4F18ifd06XF97DedHR3IiIY3hM3cCikVjA2hQTSa8FTVjutS/A/RAZ+BdIB34FmhZjLmEEEIUh6wUOBluaMk8tQ50OeDgCs1GGIrM2m3BzNzUKUUZlK/PZ8bBGfx48EfD2po959O4sunX1ryRPjeX88//j/SICFzffAPn4cM5fD6VkbN2YmVhxoLRbfCuKhPfioMxBWdrpVRzTdP2ASilrmiaJv0qQghRVmReNswqP/aXYQKQPg+cakDLJ6FhX6jZGsxKTwuUKHsupF9g8ubJ7EvcR986fXmt9WulYm3NG+lzcoh7/nkyNm6i2ltTqDRsGPvOXeHRn3fhZGPJr2NaU7ty6cpcnhhTcOZpmmYOKABN06pgaPEUQghRWqVfMqyPeXS5YSkjpYOKtSDgacOYTLfmUmSK+5aclcyCYwtYeHwhCsWH7T+kt1dvU8f6D31ODnHPjSdj82aqvf02lYYMZueZZJ6YsxsXR2t+HRNAjYq2po5ZrhlTcH4NLAWqapr2PjAIeKNYUwkhhLh7aQlwbIWhyDy7FZQenL0g8HlDd3l1P9A0U6cU5cD59PPMOTyHpaeWkqvLpWvtrrzY/EVqOpW+Pcb12dnEjXuWjO3bqfbuO1R6+GG2RCUx+pfd1Khoy69jAnB1klUXipsxs9QXaJoWCXTBsCRSf6XUsWJPJoQQ4s5Sz/9TZJ7bDihwqQftXzYUma4+UmSKInPqyilmHZ7FyuiVaJpGH68+PObzGF4VvEwdrVD6rCxix40jc8dOqr/3HhUfGsj64xd5ev5evFzsmT+6NS4O1qaO+UAwZpZ6LSATWHHjMaXUueIMJoQQ4hZSzhnWyDy6HOKuLRhStTEEvWooMqs2MG0+Ue4cuHSAmYdmEhEbga2FLcMaDGNU41FUs69m6mi3pM/MJPaZcWTu2kX1Dz6g4oD+hB9OYPzCvTSo5sQvT7Sikr1MSSkpxnSph2IYv6kBNoAncAIoXVPPhBCiPLt8xlBgHv0LLuw1HKvmC53fNBSZLnVNm0+UO0optl3YxsxDM9lzcQ9OVk480/QZHmnwCBVtKpo63m3pMzKIffoZMiMjcZv2ERX69mX5/vO89NsBmrpXYPbjrahgK0t+lSRjutSb3Pi5pmnNgaeKLZEQQgiDpCjDGplHl0PCIcMxt+bQ9W3DtpLOpbMbU5RtOr2ONefW8POhnzl2+RhV7aoy0X8ig+oNws6y9G9jqkvPIPbpp8jauw+3adOo0Kc3S/bE8sofB2nl4cysx1riYG1Me5soSnf9xJVSezVNkzU4hRCiqCkFl45fa8lcDolHDcfdW0Hw+4Yis2It02YU5VauLpcVp1cw+8hszl49S22n2rzd9m16e/XGyrxsdD3r0tOJHfsUWQcOUOPTT3Dq2ZN5O87y5rLDtK/rwo8j/bG1knVmTcGYMZwv3fCpGdAcuFRsiYQQ4kGSlw1ntxi2k4xaBVdiAM2wAHuPj6FBb6hQw9QpRTmWmZfJkpNL+OXILyRmJdLQuSGfdfyMLrW6YF6GNgHQpacTO3oMWYcOUeOzT3EKCWHm5jO8F3qMrg2rMv2R5thYlp2vp7wxpoXzxr3U8zGM6fyjeOIYR9O0PkAfb29vU8YQQoh7czXesKXkyVVwJsKwb7mFDXh2hLbPG4pMR1dTpxTl3JXsK/x6/Fd+PfYrV3Ov0qpaK94NfJc2bm3QytjKBlpWFrFPjibryBFqfP45Tt2D+XbDKT5ZdYIePtX4amgzrCxk3VlTMmYM59slEeRuKKVWACv8/f3HmDqLEELckV4PF/YZtpSMWgXxBwzHK9SEpkOhXnfwaA9WpX98nCj7EjISmHtkLn9E/UFWfhadanbiySZP0rRKU1NHuye6q1ep9NXXZMXFUeOLz3Hs2pXPV5/g6/Wn6O/nxqcPN8XCXIpNUzOmS/2v251XSvUtujhCCFFOZF+F0+sNrZin1kDGJdDMDOMxu7xlKDKrNpI1MkWJOZN6htmHZ/P3mb9RStHLqxdP+DxBnYp1TB3tnulSUzn35GgsYmNx/+ZrHDp14sOVx/lx0xmGtqzJ+wOaYG4mf8dKA2O61KOBasD8a58PA2KAVcWUSQghyiTbzPOwbbqhFfPsNtDng00F8O5mKDC9u4Kds6ljigfMkaQjzDw0k3Xn1mFtbs3geoMZ1XgUbg5upo52X3QpKZx74klyoqJIeeopGgR14q2/jvDL9rM82qY2U/s0xkyKzVLDmIKzmVKqww2fr9A0bZNS6rXiCiWEEGVCfq5hC8mo1XAynNaXzxiOV2kIbZ4zFJnurcBclmARJUspxc6Encw8NJOd8TtxtHRkdJPRDG84nMq2lU0d775lHTjAhUmTybtwAffp33BBr+fVPw+xeE8sYzt48WqPBmVuHGp5Z8xPwSqapnkppc4AaJrmCVQp3lhCCFFKpV00FJhRq+D0BshNB3Nr8OzAycpdqdfzOahU29QpxQNKr/RsOLeBmYdmcjj5MC62LrzU4iUervcwDlYOpo533/S5uSR9M53kWbOwcHWl5k8/Ye3vz0/TV7M9PpbnO3vzYrd6UmyWQsYUnC8CEZqmXfvVHQ9gbLElEkKI0kSvh/j9Ba2YXNhnOO7oBk0eNrRienYAK3suRERQT4pNYQJ5ujxCo0P5+fDPRKdG4+7gzpsBb9LPux/W5uVjr/Csw0eIf3UyOVGnqPDQQFwnT0Zna8/zi/axPV7HxO71ebaTrF5TWhkzSz1c07S6wPXNeY8rpXKKN5YQQphQTpqh9TJqFUStgfSLgAbuLaHzG1AvBFx9ZMKPMLnMvEz+jPqTuUfnkpCRQP1K9fm4w8d0q90NC7PyMZRD5eaS9MMMkmbMwMLZGfcfvscxKIjsPB3PLYhk7bFEhjWwkmKzlDPqu/FagXmgmLMIIYTpJJ/+pxUzZivo88C6Anh3NhSY3l3B3sXUKYUAIDUnlYXHF7Lg2AJSclJoXrU5UwKm0K5Gu3LVnZx94iQXJk8m59gxnPr2odprr2FesSInEtJ4bekhIs9e4d3+PtTMjjZ1VHEH5ePXHyGEuFv5uXBu+z8LsCdHGY671IeApw1FZs3WYG5p2pxC3CAxM5FfjvzCkpNLyMzPpKN7R55s8iTNqjYzdbQipfLzSZ45i0vffou5kxM1vvkap27dSM/J56vQo/y8NQZHGwu+HtaMvk3diIiQgrO0u23BqRl+TXJXSsWWUB4hhCg+6ZcMa2KeDDd0medcBXMr8GgHrcZA3WBw9jR1SiH+4+zVs8w+PJu/Tv+FTukI8QjhCZ8nqO9c39TRilzO6dNcePU1sg8exDEkhGpT3sS8UiVWHLjAe6FHSUzLYWjLmrzSvQGV7MvGHu/iDgWnUkppmrYMaFEycYQQoggpBQkHDS2YJ1fB+UhAgUM1aNzf0Irp2RGsy/7sXVE+xebGMmHjBNacXYOFZsHAugMZ1XgUNR1rmjpakVM6HZfnzOXSV19hZmtLjc8/w6lnT04lpvPWrJ1sPZWMTw0nfhjRgma1Kpk6rrhLxnSp79A0raVSanexpxFCiPuVkw7RG69tI7kG0uIBDWo0h06vGVoxqzeVCT+i1FJKsefiHmYdmsXW+K3YW9rzWOPHGNloJC625XMccW5MDBdee52svXtx6NyZ6m9PJbdCJaaFH2fm5jPYWprzbn8fHmlVS3YOKqOMKTg7AU9rmhYDZAAahsZP3+IMJoQQRrsc/c9YzJjNoMsFayeo0wnqdoe63cChqqlTCnFbeqVnY+xGZh2exYFLB3C2caZPxT5M7jEZJysnU8crFkqv58qCX0n87DM0S0vcpn2EY58+rD56kXdmbuRCajaDWrgzuUcDXBzKx/JODypjCs4exZ5CCCHuhi4PYncaWjFProakE4bjlb2h1VhDK2atNmAh47tE6ZenzyM8OpyfD//MqZRT1HCoweutX6e/d392bNlRbovN3LjzxL/+Opk7d2Lfvj3V33uXOHMHxs/Zw6aTl2hQzZGvhzXD30O2gy0PjFmH86ymae2Aukqp2ZqmVQFkwJMQomRlJF+b8LMKTq2DnFQwswSPQPB/3FBkVq5j6pRCGC07P5ulp5Yy98hczqefx7uiNx+0+4AQzxAszcrv6ghKKVJ+W0LitGmgaVR79x2s+w7g642nmbFxL9YWZrzVpxEjA2pjYW5m6riiiNyx4NQ07S3AH6gPzAYsgflAYPFGE0I80JSCi4f/mfATtxtQYF8VGvUxdJXX6QTWjqZOKsRduZp7lcXHFzP/2HwuZ1+maZWmTG41mQ7uHTDTyneBlRcfT/wbb5KxdSt2AQG4vf8eG1MtmPrlJuKuZDGgWQ1e7dGAqk42po4qipgxXeoDgGbAXgCl1AVN00z6E17TtD5AH29v2VVAiHIlN/PahJ9VhjGZV88bjrs1g6DJ1yb8+IFZ+f5HWZRPSVlJzDs6j99O/EZ6XjqBboE82eRJ/F39y9Vi7YVRSpG6dBkXP/gApdPhOuVN0oP78vTfx1h3PJF6rg4sGhtAgFdlU0cVxcSYgjP32vJICkDTNPtiznRHSqkVwAp/f/8xps4ihLhPV87ePOEnPxusHAytl0GvGib8OFYzdUoh7llsWixzDs9h2all5OnzCPYI5kmfJ2lYuaGpo5WIvMREEqa8RXpEBLb+Laj8zrv8fCaf777cjIWZxus9G/JYoAeW0n1erhlTcP6madoMoKKmaWOAJ4CfijeWEKLc0uVD3K5/usovHTMcd/aCFo9Dve5Quy1YyIxUUbaduHyCnw//THhMOOaaOX3r9OVxn8ep7VTb1NFKhFKKq6FhJLz7Lio7m6qTJ3GwdQij/jjG2eRMevtW541ejahWQbrPHwTGTBr6VNO0bsBVoB4wRSm1ptiTCSHKj8zLhok+J8Ph1FrITgEzC0Nh2WyEYQF2FxkiI8qHfYn7mHloJpviNmFrYcvIhiMZ2Wgkrvaupo5WYvKTk0l4+x3SVq/Gpqkv5q9N5bWDGayaG4lXFXvmP9madnXL55qionDG7qV+CLAF1LWPhRDi1pSCxKM3TPjZBUoPdi7QoJdhLGadTmBTwdRJhSgSSik2n9/MrEOz2Ju4l4rWFXnW71mGNRhGBesH6/v86qrVJEydij49HecXX+SPOh35+vczaGi8ElKf0e28sLKQ7vMHjTGz1EcDU4D1GBZ9/0bTtHeUUj8XdzghRBmSlwXRm6/t8LMaUmMNx6s3hfYTDK2Ybs1kwo8oV/L1+ayOWc2sw7M4eeUkrnauTGo5iYF1B2JnaWfqeCUq/8oVLr73PldDQ7Fp1IiEcZMYdyCTM2tPEdK4Gm/2aUSNiramjilMxJgWzolAM6VUMoCmaZWBbYAUnEI86FLj/plRfmYj5GeBpT14BUGHiYaWTKfqpk4pRJHL0eWw/NRyZh+eTVx6HJ4VPHk38F16efbC0rz8rqF5K2nrNxD/1hR0V1KwGfM0n1dtw4q1l6hd2Y45j7ckqL7s9PWgM6bgjAPSbvg8DYgtnjhCiFJNr4O4Pf+0Yl48bDhesTY0f9Qw4cejnUz4EeVWem46v538jXlH55GUlYRPZR8m+E+gU61O5X4NzcLorl7l4gcfkrpsGVb16rH7qTd5/6Qe3ZVkXupWj7EdvLCxNDd1TFEKGFNwngd2apq2HMMYzn7ALk3TXgJQSn1ejPmEECZmkZcOh/+41pK5BrIug2Zu2Dqy27uGItOlHpTzdQTFgy05K5kFxxaw6MQi0nLTCKgewIftP6R1tdblfg3NW0nfvIX4N94gPymJ7CGP8qJDK44fzqZrw6q81acxNZ0frCEF4vaMKThPX/vvuuXX/i/bewhRXiWdurZPeThtz24DpQNbZ8OamPVCoE5nsK1o6pRCFLu4tDh+OfoLS6OWkqPLoUutLjzZ5El8XHxMHc1kdOkZJE6bRsqSJZh7evLnE1P56ZIt7nYaMx/1p2ujB2c2vjCeMcsivV0SQYQQJqTLg3Pbr80qD4fkU4bjVRsRW3MAtbuOBXd/MJOuMVH+5ehyWHd2HUtPLWVn/E7MNXN61+nN4z6P41XBy9TxTCpjx07iX3uNvPh4znd/iFccA0i/bM7znb0Y18lbus/FLRm7LJIQorzJvGzoIj8ZblgjMycVzK3Aoz20ftow4adSbaIjIqhdq7Wp0wpR7I4mH2Vp1FJCo0NJy03Dzd6NZ5o+w4C6A6hm/2DvdqXPzCTxs8+5smABejd3vuk3kXCq0tGrCm/3bYyHi8k3IRSlnBScQjwolIJLx691la+C2J2GtTHtq0Kjvoaucq8gsHYwdVIhSkxqTip/n/mbZaeWcfzycazMrOhSuwsD6w6kVbVWD+REoH/LjIzkwmuvkXf2HIcDQnjDpQOVKzjxQ5/GdG/s+sCOYRV3RwpOIcqz/ByI2fJPV3nKWcPxar6GZYvqdYfqsjameLDolZ4dF3aw9NRS1p1bR54+j4bODXmt9Wv09Oz5wC3Ufiv67GwuffkVl+fOJcfFlQ87P8feSp6Mae/Fc529sbOSEkIYz5iF3z8G3gOygHCgKfCCUmp+MWcTQtyL9ETDkkUnw+H0BshNBwtbQ+tluxcNRaaTm6lTClHizqefZ9mpZSw/tZz4jHicrJx4uN7DDKg7gAbODUwdr1TJOnCAC6++Ru6ZM2xv3JGPPYNp0cCNlX198K4qvSDi7hnz60mwUuoVTdMGYFiT82FgAyAFpxClgVKQcOhaK+ZKOB9pOO5UA3wHG7rKPdqDlSxRIh482fnZrD+3nj9P/cnO+J1oaLRxa8NLLV6iU61OWJvLmrE30ufmkjT9W5JnziTDsRLvtx3LBe8mfNq7Eb2aVJfuc3HPjCk4r2+Z0BNYqJS6LN9wQphYXhZEb/pnPObV84AGNVpApzcMrZjVmsjamOKBpJTi2OVj/Bn1J2HRYaTlplHDoQbj/MbRr04/3Bykhb8w2UePcn7yq+SePMkGr9b80LgPQzs14vkudXGwlu5zcX+M+Q5aoWnacQxd6uM0TasCZBdvLCHEf1y98M9YzOvbSFo5QJ1O0Ok1w6xyB9k+Tjy4UrJTCI0OZWnUUk5cOYGVmRVda3dlQN0BMgHoNlReHkkzfuTS99+TZuPApwFPYNamHb/396Geqyy5LYqGMetwTtY0bRpwVSml0zQtE8NuQ0KI4qTXQ/w+Q5F5YiUkHDQcr1hLtpEU4hqdXsfO+J38eepP1p9bXzAB6PXWr9PDs4dMALqD7BMniZs0ibzjx9lQszm/BQzmxYH+9PNzk+5zUaSMmTRkBzwL1ALGAm5AfeDv4o0mxAMoJx3ORBjGYp5cDRmJoJlBzdbQdaphPGaVBtJVLh54cWlxLD+9nGWnlpGQkUAF6woMrj+Y/t79ZQKQEVR+PkkzZ5H4zXTSLGz4pvUo6g3qw4pu9XCysbzzBYS4S8Z0qc8GIoG21z6PA5YgBacQRSPl3D+tmDGbQZcL1hXAu4uhwKzbDeycTZ1SCJPLzs9m3bl1LI1ays6EfyYAvez/Mp1qygQgY+WcOcPplyaiHT/KVjdfNvd6gteHtaGRm5Opo4lyzJiCs45SaoimacMAlFJZmrSzC3Hv9DqI23OtFXMVJB41HK/sDa3GGorMWgFgLq0MQiilOHrZsAPQjROAnvV7ln51+lHdobqpI5YZSqfjws9zuPzV12RqFswNfIxOzzzC3ObumJnJP+uieBlTcOZqmmYLKABN0+oAOcWaSojyJjvVsH3kyVWGNTKzLoOZBdRqA90/gLrdwcXb1CmFKDVuNQFoYN2BtKzWUiYA3aWcmBgO/28idicOs7taY+Kf/B8fPhRABTv5xVaUDGMKzrcwLPheU9O0BUAg8FhxhhKiXEg+fW3ZonA4uw30+WBbyTCbvF4I1OkMthVNnVKIUkOn17Ej3rAD0PUJQI0qN+KN1m8Q4hkiE4DugdLrOfHDbHK++xo9ZvzW9QkGThpDk5oVTR1NPGCMmaW+RtO0vUAAoAH/U0olFXsyIcoaXR6c2/HP2pjJUYbjVRpCm+egfg9wbwlm5qbNKUQpE5cWZ9gB6PTymyYADfAeQH3n+qaOV2alnDnLgecnUPXUYY5Wb4jlpNd5M7i5dJ8LkzBmlvoAYL1SKvTa5xU1TeuvlFpW3OFuk6kP0MfbW7oghYllXoZTaw1F5qm1hq5zcyvDckWtxkK9YKjkYeqUQpQ62fnZrD23lmVRywomALV1a8sE/wl0qtkJK3MrU0css/R6PZu/mInj7O9wALYNGMuA18fh7CCTqoTpGNWlrpRaev0TpVSKpmlvAcuKLdUdKKVWACv8/f3HmCqDeEApBUknDTPKT66C2B2g9GBfBRr0gfohhj3LrWWxZCH+7foEoMXJi3ntt9dIy/tnAlB/7/5Us69m6ohlmtLpiPojlHPf/UCNhGiiajSg5kfv82TLRqaOJoRRBWdhI7Nljyvx4MjPgbNb/9nl50qM4Xg1X2g/wTAe060ZmMkkBiEKk5Kdwt9n/mbpqaWcvHISS82SYM9gBnoPxL+av0wAuk8qN5dLS5cT8+0MHBPPY+7gQvQTL9LzpSexsJAhPKJ0MKZw3KNp2ufAtxhmqo/HsC6nEOVX+iXDbPKT4XB6PeSmg4WNofUy8H+GWeUVapg6pRCl1vUJQH9G/cmG2A3k6fNoXLkxb7R+A8cLjvRs39PUEcs8XXoGKUt+48JPP2NxOYmLFWqwe/D/GPzyo1SpYGfqeELcxJiCczzwJrAYw6Sh1Rh2HhKi/FAKLh7+Z8JP3B5AgaMbNHnY0Irp2QGs5Ie4ELcTmxZrmAB0ajkXMy9S0boiQ+oPob93/4IJQBEJEaYNWcblX77MlfnzSZq3ANKuctilDjt6DWfkcw/Tx7OyqeMJUShjZqlnAJNLIIsQJcpMl2PYPvJ6kXk1znCiRgvo9Jphr/JqvrKNpBB3cH0C0NKopexK2GWYAFSjLRNbTpQJQEUo7/x5kmfP4fKSJWg5OWyv7sPfrbrS55FgPguojYW5DE0QpZcxs9TrARMAjxtfr5TqXHyxhCgmV+MhahWcXEVg1DrQ54ClPdTpBEGTDWtkOrqaOqUQpZ5SiqPJR/kz6k9WRq8smAD0nN9z9PPuJxOAilBOVBRJP80k9e9Q9MA692aENe5Kx26t+KG9J1UdbUwdUYg7MqZLfQnwAzAT0BVvHCGKmF4P8fv/mfATv99wvEIt4qt3xb3TE1C7HVjKD2whjHEl+wqhZ0L589SfRF2Jwtrcmm61uzHAe4BMACpimfv24fTd95w5eJAcCyvCPNuyqWk3+ndvzpLWtXCykV2CRNlhTMGZr5T6vtiTCFFUcjPgTMS1rvLVkJ4Amhm4t4IubxnGY1ZtyKmNG3H3DjJ1WiFKPZ1ex/b47QUTgPL1+fhU9uHNgDcJ8QzBycrJ1BHLDaUUGZs3kzjjR3IiI8m3smNeg2COtApmeLAvL/u5YS0zz0UZZEzBuULTtHHAUm7YQ10pdbnYUglxt1LO/dOKGb0ZdDlg7QTeXQwFpnc3sJfB9ELcjcImAA2tP5QBdQdQr1I9U8crV1R+PlfDV3Fxxo/ook6SbFeRJU36cbppAC8+1Ib36leVHYJEmWZMwTnq2v8n3nBMAV5FH0cII+l1cD7ynwXYE48Yjjt7QcvRhgk/tduCuXQ5CXE3svOzWXN2DctOLWNXwi7MNDPaurXllZavEFQzSCYAFTF9djapS5eS8ONMiL9ArGNVljQfgmVwD0Z3rsfVMwcIaijjykXZZ8wsdc+SCCLEHWVfNayJeTLcsEZmZjJo5obCMvg9qNcDXGS7UyHullKKI8lHWBq1tGACkLuDO+Objadvnb4yAagY6K5e5crCRSTOnoOWcoUTlWryR5vHqdm7O6919MbTxR6AiDMmDipEETFqxyBN03yARkDBzAql1C/FFUqI//jtUTgeBvo8sKlomE1erzt4dwXbiqZOJ0SZdCX7SsEOQFFXorAxtzFMAKo7gBauLWQCUDHIS0zk8i/zSPp1IWaZGURWrU9Y5+H49+vC5209qeIo+52L8smYZZHeAoIwFJxhQA9gCyAFpyg5zl7QZpyhFdO9JZjL7qpC3AudXse2C9tYemrpfyYA9fDsgaOVo6kjlku5Z8+SOHMWqcuWQX4+m92aEtElhO79OzDTvyb21vIzTZRvxnyHDwKaAvuUUo9rmuaKYYkkIUpO16mmTiBEmRZ7NZalp5ay/PRyEjMTqWRdiWENhtHfu79MACpG2UePEv/DT2StWU2+ZsbqWv4cDOzNQ30CmN+kuizWLh4YxhScWUopvaZp+ZqmOQGJyIQhIYQo9bLys1h7di1LTy1ld8LugglAk1tNJsg9CEuZVFcslFJk7trNhe9/IH/HdjItrPnbuyPxXfsxomcLXvaujCY7mIkHjDEF5x5N0yoCPwGRQDqwqzhDCSGEuDfXJwBd3wEoPS9dJgCVEKXXk75+PXHf/gDHjnDF2oHljXuieg9gVPcmNHarYOqIQpiMMbPUx1378AdN08IBJ6XUweKNJYQQ4m5czr7M36cNE4BOpZySCUAlSOXmkvL335z/7kfM484Sb+fMiuaDqDxoIOM71ce9kp2pIwphcsZMGlqnlOoCoJSK+fcxIYQQplHYBKAmLk1kAlAJ0WdmkvzbEuJ/+hnL5ETOOlVnZeAoGg0bwFttPaloJ2uWCnHdLQtOTdNsADvARdO0SsD1ASdOgFsJZBNCCFGIW00AGuA9gLqV6po6XrmXf+UKl36ZT9Iv87DISON4ZS82dX+YtsP68GkLd2wsZetJIf7tdi2cTwEvYCguI/mn4LwKfFu8sYQQQtwoKSuJNWfXEB4dzt7EvZhpZgS6BcoEoBKUFx/P+R9nkfbH71jk5rC7WmMO9uxDyJDufNXIFXPZelKIW7plwamU+gr4StO08Uqpb0owkxBCCCA1J5W1Z9eyMmYluxN2o1d6vCt683yz5+lTp49MACohOadPc/bbGeSFh6GUIsK9GXHdB/HwoA484eFs6nhClAnGzFJP0DTNUSmVpmnaG0Bz4D2l1N5iziaEEA+c9Nx0NsRuIDwmnG3nt5Gv8qnlWIvRTUYT4hEiXeYlKOvAAc58/R1mWzeRbW7JWs82ZPUfyiP9WuFdVcbHCnE3jCk431RKLdE0rR3QHfgU+B5oXazJhBDiAZGVn8XmuM2Ex4SzKW4TObocqtlXY2SjkYR4htDQuaGs21hClFKkb9nC6a++w/rwfjIsbVnVOBiHYY8wKrgpVZ1s7nwRIcR/GFNw6q79vxfwvVJquaZpU4svkhBClH+5uly2XdjGyuiVRMRGkJmfSWWbyjxU9yF6ePbAt4qvLGVUgpROx5WVq4iZ/j22MadIs3Fiqf8APEY9wosdGuAgW08KcV+M+Rt0XtO0GUBXYJqmadaA/BQUQoi7lK/PZ1fCLsKjw1l7bi1puWlUsK5AD88e9PDsgb+rP+ZmMsO5JOlzcri45E/iZ/yE7aV4kh2qsLnjozR7Ygivt/DAykL+uROiKBhTcA4GQoBPlVIpmqZVByYWbywhhCgf9ErP3ot7CY8JZ83ZNVzOvoy9pT1danUhxCOEALcALM1khnlJ06WnEztnPlfmzsUmLYXYiu7s7/0sHZ94iLcbVpMhDEIUsdutw+mklLoK2AAR1445AznAnhJJJ4QQZZBSisNJh1kZs5JVMatIzEzExtyGjjU70sOjB+3c22Ftbm3qmA+k/KQkTn0/i+w/fsM6O5NjVepytt8z9BrVh341K5k6nhDl1u1aOH8FemNYg1PxzzqcXPvcqxhzCSFEmaKUIi43jq/2fsXK6JWcTz+PpZkl7Wq04+UWLxNUMwg7S9ni0FRyzp3j+Fc/YBb+N2a6fCLdfUkbMIyBQ7vxSGX5cxGiuN1uHc7e1/7vWXJxhBCibIlOjSY8OpyVMSuJTo3GPMGcgOoBPN30aTrX6oyTlZOpIz7QMo8e4+jn07HdugENMzZ7tcLikZEM6h+Is71sPSlESbldl3rz271R1uEUQjyozqefJzw6nPCYcI5fPo6GRgvXFrQyb8W4buNwtpHFwE1JKUXqzt0c/2I6FQ7sRrOwZm3jLrg+MYrHuzbD1komZglR0m7Xpf7Ztf/bAP7AAQzd6r7ATqBd8UYTQojSIzEzkdUxq1kZs5KDlw4C4FvFl0ktJxHsEUxVu6pERERIsWlCSq/n4qq1RH/zPRXPHAcre8ID+tNg7OM8F1BXtp4UwoRu16XeCUDTtEXAWKXUoWuf+wATSiaeEEKYzuXsy4atJaNXEnkxEoWigXMDXmj+At09uuPu6G7qiAJQeXlE/7aMxB9/osLFWHJsK7Gm20haPj2KFxq5yYxzIUoBY5ZFanC92ARQSh3WNM2v+CIJIYTpXM29yvpz6wmPDmdH/A50SodnBU+eafoM3T2741VB5kuWFvqsLI7Nmk/GvF9wTE0ixakaex8aR6dnhvO8u7Q0C1GaGFNwHtM0bSYwH8Ps9BHAsWJNJYQQJSgzL5OI2AhWxqxk6/mt5OnzqOFQg8d9HifEI4R6lepJK1kpkn/lCvu/mQVLf8M+K43zlT25/OQYQp58iBBne1PHE0IUwpiC83HgGeB/1z7fhGEvdSGEKLNydDlsidvCypiVbIzdSLYum6q2VRnaYCg9PHrg4+IjRWYpk3Uhnr2ffof96hXY5+ewv0Zj1LBH6TW8BxVsZfF8IUqzOxacSqls4Itr/wkhRJmVp89jx4UdhMeEs+7cOjLyMnC2caafdz9CPEJo7tpc9i8vhVJORLH/k+lU3raOCkqxt44/FR9/goH92mFtITPOhSgLjGnhFEKIMkun17Hn4h5WRq9k7bm1pOak4mjpSHDtYEI8Q2hVrRUWZvKjsDS6sCOS4198i+uBHVQ0M2evTwdqPTOWER2bYiYzzoUoU0rNT1lN0xpi6LZ3AdYppaTbXghxT/RKz8FLB1kZvZLVZ1eTlJWErYUtnWp2oodnD9q6tcXKXBb9Lo2UUpwKX0/s9B+ofvowjpY27Gnbhyb/G8uopnVMHU8IcY+MLjg1TXMClFIq7S7e8zOG7TETlVI+NxwPAb4CzIGZSqmPlFLHgKc1TTMDfjL2HkIIAYZC5ejlo4RHh7MqZhXxGfFYmVnRwb0DIZ4hdHDvgK2FraljiltQOh0HFi4jZeYsXBOisbZxJLLHCNr870lGeVQzdTwhxH26Y8GpaZo/MBtwNHyqpQBPKKUijbj+HGA68MsN1zMHvgW6AXHAbk3T/lJKHdU0rS8w+dp7hBDijk5dOcXKmJWER4dzLu0cFpoFbWu0ZXyz8XSq2QkHKwdTRxS3kXc1jd0zF6Jb8isuVy6id3Dh0NBxBD03ikAX2RZUiPJCU0rd/gWadhB4Vim1+drn7YDvlFK+Rt1A0zyAv6+3cGqa1gaYqpTqfu3zVwGUUh/e8J5QpVSvW1xvLDAWwNXVtcWiRYuMifHASE9Px8FB/oE1hjwr45TG53Qp7xJ7M/cSmRFJfF48Ghp1berSwq4FTe2aYm9umqVxSuOzKo3S09JwuJRExupN1Di0F2tdLtGV3Int2I3anVtgI1tPFpDvKePIczJOcT+nTp06RSql/As7Z0yXetr1YhNAKbVF0zSju9ULUQOIveHzOKC1pmlBwEDAGgi71ZuVUj8CPwL4+/uroKCg+4hS/kRERCDPxDjyrIxTWp5TfHo8q2JWER4TzpHkIwA0r9qcxz0fp1vtbrjYupg4Yel5VqWVLi2Ni38u58ysOVROPE+2uSX767ai+ohhdOsfhKXMOP8P+Z4yjjwn45jyOd2y4NQ0rfm1D3dpmjYDWIhh4fchQMR93LOwqYVKKRVxn9cVQpQzSVlJrI5ZTXhMOPsS9wHQuHJjJvhPoLtHd6rZy9i+0k4pRfbBg5yfv5Cs8HAs8nJIcXJjd9dHaTl6KKOaesh6p0I8AG7XwvnZvz5/64aPb98Pf3txQM0bPncHLtzH9YQQ5UhqTqph//KYlexO2I1e6fGu6M34ZuMJ8QihllMtU0cURtClpZHy1wouzP8Vi+jTZJlbsalmM9K79aamd0Ve7NfV1BGFECXolgWnUqpTMd1zN1BX0zRP4DwwFHikmO4lhCgD0nPT2RC7gZXRK9l+YTv5Kp9ajrUY02QMIR4heFfyNnVEYQSlFNkHDnBh/kKyVoVjnpdLTIUarPd/GJd+fRjVtTHuleyIiIgwdVQhRAkzZpa6NfAQ4HHj65VS7xjx3oVAEOCiaVoc8JZSapamac8BqzAsi/SzUurIPaUXQpRZWflZbIrbRHh0OJviNpGrz6W6fXVGNhpJiGcIDZ0bSldrGaG7epXLy5ZzYf4irM6dIcvcigj3ZsQFdqdD7/Z82KQ6NpYyPlOIB5kxk4aWA6lAJJBzNxdXSg27xfEwbjMx6E40TesD9PH2llYPIcqSXF0u2y5sY2X0SjbEbiArP4vKNpUZVG8QPTx74FvFV7aWLCOUUmTt38+5ub+St241Fnm5xFR0Z1vAUKoN6MdD7epSu7JpVgsQQpQ+xhSc7kqpkGJPcheUUiuAFf7+/mNMnUUIcXv5+nx2xe9iZcxK1p1dR1peGhWsK9DLqxc9PHrQwrUF5mbS+lVW6FJTSfxzGfELFmEbF0OmhTWbajbnSudedOrbgQ/qVsFctp0UQvyLMQXnNk3TmiilDhV7GiFEuaBXevZe3Et4TDhrzq7hcvZl7C3t6VKrCyEeIQS4BWBpZmnqmMJISiky9+3j9M/z0SLWYpGfR2xFdyLbD8dzcH8ebVOXyg7Wpo4phCjFjCk42wGPaZoWjaFLXcOwjJFRC78LIR4MSikOJR0y7F8es5rErERszG0IqhlEiEcI7dzbYW0uRUlZoktNJe63P7i0cDH2F86RZ2HNltotyQ3pQ9d+Helbs6KMsxVCGMWYgrNHsacQQpRJSilOXjnJyuiVhMeEcz79PJZmlrSr0Y4JnhPo6N4RO0s7U8cUd0EpRdruPZycNQ+rrRFY5ucRV7Emx7qOot7QgYxp5YWdlTH/dAghxD/u+FNDKXW2JIIIIcqOM6lnCI8OJzwmnOjUaMw1cwKqB/B006fpXKszTlayB3ZZo0tJ4fSvS7iyeAlOF2PRLKzZ5NUa8z4D6N6/A/2ryLaBQoh7J7+mCiGMEpcWR3hMOOHR4Zy4cgINDf9q/oxoOIKutbvibONs6ojiLimluLxjFydmzsNhx0YsdfnEV6rF9p6j8RnxEE81rYWluawaIIS4f2Wy4JRlkYQoGRczLrLh6gZ+Cv2Jg0kHAfCt4suklpMI9gimql1VEycU9yL/8hWOzF1E5p+/U/HSBawsbNhWry32Dw2ie9/29HeyMXVEIUQ5UyYLTlkWSYjicyX7CmvOrmFl9EoiL0aiUDRwbsALzV+gu0d33B3dTR1R3AOlFPGbtxM1ax6V9mzBSpdPtHNtDg54muajBvFUfTeZACSEKDZlsuAUQhStzLxMImIjCI0OZdv5beSrfDwrePKM3zM4JzozpNsQU0cU9yg3+TL7Zi4g/6+lOCfHY2dhw57GHXAe/DBd+7TDwVr+GRBCFD/5SSPEAypPn8eOCzsIjQ5l/bn1ZOVn4WrnyshGI+np1ZP6leqjaZrse10GKaWIXruJM7MXUHX/dpz0+Zx08eDMkGdp9cQQnqhdxdQRhRAPGCk4hXiAKKU4cOkAoWdCWX12NZezL+No5UhPz5708upFC9cWsrVkGZaRmMSeGfMwC1uOy5WLVLC04WCzTlQbNoQeIW2wspA/WyGEaUjBKcQD4HTKaULPhBIWHcb59PNYm1vT0b0jvbx60a5GO6zMrUwdUdwjvV7PkbANxM37lRqHdlJVr+NUVS8uPvoCbZ98mJausnqAEML0pOAUopxKyEggPDqc0OhQjl8+jplmRutqrXmm6TN0qdUFBytZV7EsS45LYPeM+diuXkHV1ERcLG053rIbtR99hF6d/DGT/cyFEKWIFJxClCOpOamsObuGsOgw9iTsQaFo4tKESS0nEeIZgouti6kjivugy9exa+kaEhctxvPobmorHdHVvTkz+FECnxhMy0qOpo4ohBCFKpMFp6zDKcQ/svOz2Ri3kbAzYWw+v5k8fR61nWrzTNNn6OnVk9pOtU0dUdync6fj2PvjfCquD8M17RIWVnacCQyh3hMj6NnWz9TxhBDijspkwSnrcIoHXb4+n10Juwg9E8q6c+vIyMugim0VhjYYSi/PXjSq3EjWVCzjsnLy2LpkFalLllD3ZCT1lY5z7vVIGPUkrUcNoqWjvakjCiGE0cpkwSnEg0gpxZHkI4SeCWVl9EqSs5NxsHSgW+1u9PLqRUvXlpibmZs6prhPRw9Hc3DWAlw3r6JGehIVrO05H9SLRmMfpXuzxqaOJ4QQ90QKTiFKuZjUGMKiwwiLDuPs1bNYmlnSwb0Dvbx60b5Ge2wsZBvCsi4lI4eNC8PIXvoHjc7so6nSc75WA9KeeobmIwZiYSt/xkKIsk0KTiFKoUuZlwiPCSf0TChHko+godGyWkue8HmCrrW74mTlZOqI4j7p9YqdkSc5MWcRtXesoV5GMuk29iR264fv2FE09Klv6ohCCFFkpOAUopRIy01j7dm1hEWHsSthF3qlp6FzQyb4TyDEIwRXe1dTRxRFIP5KJhsW/I3+72U0PXuA1kpPgldjdEP/R/PBfTC3kdZMIUT5IwWnECaUq8tlc9xmQqND2Ri7kVx9Lu4O7oxpMoaenj3xquhl6oiiCOTm64nYdowz8xdRf896mmVeJtPGgdSeAw2tmfVlxQ0hRPkmBacQJUyn1xF5MZLQ6FDWxKwhLS8NZxtnBtUbRE+vnvi6+MoM83LiZHwqG3/9G6vwFTSPO0RNpSepbhOsRrxM/QG9MbOSHZ6EEA8GKTiFKAFKKY5dPkbYmTBWxqwkMTMROws7utTqQi+vXrSu3hoLM/nrWB6kZecRvukwFxb+hu+BjbTLvEKmnSPZ/QbjM/ZRGnp5mjqiEEKUOPkXTohiFHs1lrDoMEKjQ4lOjcZCs6BdjXZM9J9Ix5odsbWwNXVEUQSUUuw5k8T6v/cT/e53tLxwBB+lJ6VBU5xGTaZ+rxBpzRRCPNDKZMEpOw2J0iw5K5nwmHDCosM4eOkgAM2rNmdEwAiCawdT0aaiaQOKIpOYls2KdQdJXvI7rY5uYVjWFbLsndA/PIy6jw/H2lNaM4UQAspowSk7DYnSJiMvg/Xn1hN6JpQd8TvQKR31KtXjheYv0NOzJ9Udqps6oigi+To9G44msOu3MKpvCad1wjHMlZ50n2ZcavUQ7V8YjyatmUIIcZMyWXAKURrk6fLYemEroWdCiYiNIFuXjZu9G4/7PE5Pz57UrVTX1BFFETpzKZ2/1h4gfdlS2p3cysCsFLIdKmD5yAg8Rw3HqlYtIiIipNgUQohCSMEpxF3QKz37EvcReiaU1WdXk5qTSkXrivTz7kdPz574VfXDTDMzdUxRRDJz8wndH8f+P1ZSZ+c6uiYcxRxFdlN/qo16hIpdu0iBKYQQRpCCUwgjnLh8grDoMFZGryQ+Ix5bC1uCagbR26s3baq3wdLc0tQRRRFRSrEvNoW/1+5HF/oXnc/sYERWCrlOFXF47HHcHhmCVa1apo4phBBlihScQtzChfQLhhnmZ0I5lXIKc82cNm5teL7583Su2Rk7SztTRxRFKDk9h6WR5zj8ZzhND0QwMOEY5ijym7eixqjhOHbuhGYpv1gIIcS9kIJTiBtcyb7C6pjVhEWHsTdxLwB+Vfx4rfVrdPfojrONs4kTiqKk0ys2nbxE2Lr92K4NpWvMTgKzUsmr4EzFJ5+k6rAhWLm7mzqmEEKUeVJwigdeZl4mEbERhEaHsu38NvJVPl4VvBjfbDw9PXvi7igFR3lzLjmTJTtjOLliNW2ObuKxxONogNYygBojh+EYFCStmUIIUYSk4BQPpDx9HkeyjhC+OZz159aTlZ+Fq50rIxuNpKdXT+pXqi/bS5Yz2Xk6wg8nsHr1HipvW0vw2d30zE4lv1JlnMeMofLgwVi51zB1TCGEKJek4BQPDKUUBy4dKJhhfjn7Mo4pjvT07Ekvr160cG0hM8zLGaUUh89fZemmo6SErSTwzC7GXz6L0jQsA9pQbfgwHDp2lNZMIYQoZlJwinLvdMppQs+EEhYdxvn081ibW9PRvSO1M2vzdPensTKXZW3Km5TMXJbvjuHQ0lXUP7iZ/glHsdTr0NX2xOXxl6nYtw+Wrq6mjimEEA+MMllwytaW4k4SMhJYGb2SsOgwjl8+jplmRutqrXmm6TN0qdUFBysHIiIipNgsR/R6xdZTl9iwfCO2EatpH7uXVrmZ5DlVpOKwYVR9aADWDRvKUAkhhDCBMllwytaWojCpOamsObuGsOgw9iTsQaFo4tKESS0nEeIZgouti6kjimJwIiGNVRv2k7ZiBS1P7uDh9EvkW1hh3r4DNYcOwj4wEM2iTP6oE0KIckN+CosyLTs/m41xGwk9E8rm85vJ1+dT26k2zzR9hp5ePantVNvUEUUxiE/NInT7Kc4tD6XRoS10SzoNQGZDX1yGPodzzx6YOzqaOKUQQojrpOAUZU6+Pp9dCbsIPRPKunPryMjLoIptFYY1GEYvz140qtxIuk3LodSsPMIPxHFw+Rpq7I6gTfxhAnV5ZLm6Yff0OKoPGiBrZgohRCklBacoE5RSHE46XLC9ZHJ2Mg6WDnSr3Y1eXr1o6doSczNzU8cURSwnX8eG45fYEr4Nu41r6HguEt+cNPLsHLDp04+aQx/CpmlT+QVDCCFKOSk4RakWkxpDWHQYYdFhnL16FkszSzq4d6CXVy/a12iPjYWNqSOKIqbXK3ZGX2b1psPkrDIsZTT8ajw6cwu0gLbUGDIIh6COmFnJhC8hhCgrpOAUpc6lzEsFM8yPJB9BQ6NltZY84fMEXWt3xcnKydQRRTE4Fn+VFbvOEP93OM1P7OChSycxV4q8eo2o8vyTVOzVE4tKlUwdUwghxD2QglOUCmm5aaw9u5aw6DB2JexCr/Q0dG7IBP8JhHiE4GovayaWR+dTsvhrbxyHwyKod3AzXS4cxC4/hzwXV8PuP/37Y+3laeqYQggh7pMUnMJk9ErPlvNbWHZqGRtjN5Krz8XdwZ0xTcbQ07MnXhW9TB1RFIPUzDxCD8WzZf0eqmxfT6fYSIKyUtDZ2GLfswdVBw3Azt8fzUx2fRJCiPJCCk5R4tJy01h+ajm/Hv+V2LRYnG2cGVRvED29euLr4isTQMqh7Dwd648nEr71OGYb1hB0dg/jU2LRm5lh2aoNrg8PwLFzZ8xsbU0dVfy/vTuPrrK+9z3+/mYGkuwwCMhkZBIQRSQiICqKAsoQEBQFq3WqrfXYnlprPbW1vZ57jmfde3q7Opy6bOux2oBDqGFUEZXggIwZEBLmQOYQphCSQJL9u39k20MtyEbZefZOPq+1ssje2cn+8OVJ+OT57ed5RERCQIVTWs3eo3tZWLiQxbsWU9dUxxUXXMFjIx9j4kUTiY3Stazbmma/Y92egyzZWET1yvcZt2c9D1UWEOP8uAGD6f7wk/imTSXmggu8jioiIiGmwikh5Xd+Pin7hL8U/IWPSz8mJiqGW1JvYd7QeQzvNtzreHKeOefYVl5D1uYStq36mJGFa5ldlk/iyTr8XbrS9b5vkpKeTsIlg72OKiIirUiFU0LieONxFu9azMLChRTVFNGtQzceueIRbh98uy4x2QYVH6pjSV4ZH2bn0j/3QyYWb+K24wfxxyeQdPNEusyaRaexY7BonStVRKQ9isjCaWbTgekDBw70Oop8wf6a/SwsXMibu97keONxLut2Gc9d+xyTLppEbLSWzduSw8dPsnxLOe98upNOn2YzsXgTzxzcizMjPu0qut72Q5JuvpnoxE5eRxUREY9FZOF0zi0FlqalpT3kdRZpWUZdW76WjIIMPiz5kGiLZlLqJOYPnc/lF1zudTw5jxoam1lVUMmSjfs49tHHTCjawBOV24htbsJSL6bbvT/AN30asRde6HVUEREJIxFZOCU81DXWsXT3UhYULmDP0T10SejCwyMe5vbBt9O9Y3ev48l54neOD3ceIGtzKTs/2sjY3et4oCyX5IZanC+FLvPuxDcjnYThl+oMAyIicloqnHLOio8V82rhq7y5802ONR5jWNdh/Nv4f2Ny6mTionW5wbbAOcdnpTVk5ZbyQXY5o3YvY3LJZh6sqcDFxJJ04w2kzJxJ4rXjsVi9VEJERL6cCqcExTnHuop1ZBRkkF2cTbRFc/NFNzNv6DxGXDBCe7baiP0H61icW8qKDbvplb+Om4o3cduBXUThiB85ks4zv0PylMlE+3xeRxURkQiiwilfqq6xjmV7lrGwcCG7juyic3xnHrzsQeZeMleXm2wjDh0/yfL8MrI2FePfvJGJ+zfyXMVnxDWdJLpPH45Ou5URjz1GXL9+XkcVEZEIpcIpp1VaW8prha+xaOciak7WMKTLEJ695lluufgW4qPjvY4nX1P9yWZWbqtgcW4ZRRvymbBvIz8syyWl7ggkJZEyexa+9Bl0GDmS7OxslU0REflaVDjlb5xzbKzcSEZBBh8Uf4BhTOw3kflD5zOy+0gtm0e4pmY/H+8+yOKcUj7dtJPRezZyR9lmLjpUAtHRJF53Hb70dBJvmEBUvH6pEBGR80eFU2hoamD5nuVkFGaw8/BOfPE+7rv0PuZeMpcLE3V6m0jmnCO/5Chv5pTyzuZ9DNqVw+TSzTxQWUiU30/C8OH4vnMvyVNvJaZLF6/jiohIG6XC2Y5VHK/g1cJXydyZydETRxnceTC/GPcLbr34VhJiEryOJ19DUfVxsnJLWZJTQsftnzGpZDO/L88n/kQ90T16kPLgg/jSZxA/YIDXUUVEpB1Q4WxnnHNsrtpMRkEG7+9/H4fjxr43Mm/oPNJ6pGnZPIJV155gWV4ZWbllHCjYycTiTfxreS6da6qxjh1JvmUyvpnpdBw9GouK8jquiIi0Iyqc7cSJ5hOs2LOCBYULKDxUSHJcMvdceg93XnInvRJ7eR1PvqK6k02s3FpJVm4pOVv3M744h4crckmt3ANRUXQaOxbfzB+RNHEiUR07eh1XRETaKRXONq7yeCWvbX+NzB2ZHD5xmIEpA/nZ2J8xrf80OsR08DqefAVNzX4+3FVNVk4p728pZXjxVqZV5PBE6VaimpuIHzQI3z1PkDxtGrE9dMUnERHxngpnG+ScI+9AHhkFGazat4pm18yEvhOYP3Q+o3uO1rJ5BHLOkVt8hKycUpblldGtZBe3lOfwSkku8fW1RHftiu8bd7e8LnPIEP0bi4hIWFHhbENONp9kXe06fr/892w7uI2k2CTmD53P3CFz6ZvU1+t48hXsOVBLVm4Zi3NLqd9fwqSyHH5TnkvnQ+VYfDxJN92EL30GncaNw2L07SwiIuFJ/0O1AQfqDvD6jtd5ffvrHGo4xMW+i3n66qeZPmA6HWP1ur1IU3WsgWV55WTllrJrbyXjy/J56kAeF5VsB6DjVVfhe/y7JE2eTHRiosdpRUREzk6FM4LlH8gnoyCDlUUraXbNXNfnOoafHM7DUx7WkmqEqT3RxDufVZCVW8raHZVcUbWDO6vzubwoj+jGk8SlpuL7/vdInjaduD69vY4rIiJyTiKycJrZdGD6wIEDvY7S6hqbG3ln3zssKFjAluotJMYmcueQO7lryF30S+7H6tWrVTYjRGOznzU7DpCVW8a7W8vpVV3CzAN5/GD/ZhKOHSHa5yP59jn40meQcPnl+ncVEZGIFZGF0zm3FFialpb2kNdZWkt1fTVvbH+D13e8TnV9NanJqTw1+inSB6bTKbaT1/EkSM45Nu8/TFZOGcu3lEP1AaZW5vFiRS6dK/ZDbAxJEya0XGLyuuuwuDivI4uIiHxtEVk425Ot1VvJKMjgraK3aPI3Mb73eOYPnc+4XuOIMp28O1LsqqplcW4pWbmlVFUd4fqqbfz7wXz67t2K+f10GDEC38PfJGnKFGI6d/Y6roiIyHmlwhmGGv2NrNq3ioyCDPIO5NExpiN3DL6Du4bcRaov1et4EqTKmgaW5pWRlVvK1pIjjDi4m0cPf8aluzcT3VBPbO/e+L79ML4ZM4hLTfU6roiISMiocIaRQw2HWpbNt79OVX0V/ZL68ePRPyZ9QDqJcToaORIca2jk7c8qWJxbxie7q+lztII7j2zlX/duIP5wNVGJiSRPn4pvxgw6jBqlS0yKiEi7oMIZBgoOFrQsm+99i5P+k4zrNY5nxj3D+N7jtWweAU42+cnecYCsnFJWFVSScLyGmYe28mjZZnzFuyE6mk7jryElPZ3EG28kKiHB68giIiKtSoXTI03+Jt7b/x4LChawuWozHWI6MGvQLOYNmUf/lP5ex5Oz8Psdm/Yf5s2cUlZsKef4sTomHtnB76rz6bUjF2tuJn7YUFKe+jHJU6cS062b15FFREQ8o8LZyg43HGbRzkW8WvgqlXWV9E7szRNpTzBz0EyS45K9jidnsaPyGFk5pSzOLaP0cB0ja/bz05qtDClYT1RdLTHdu+O775skz5hBwuDBXscVEREJCyqcrWT7oe0sKFzA8j3LOdF8gqsvvJqnxzzNtb2vJToq2ut48iUqjjawJK+UrJwytpXX0PVkLfcfL2D8jo+JKy/BOnQgedLN+NLT6Xj11Vi0/j1FREROpcIZQk3+JlYXryajIIONlRtJiE5gxoAZzBsyj4Gd299J6yNJTUMjb28JXPlnz0HM7+d2fwk/Kd1Il7z10NxEhyuvJOXR75A8ZTJRnXQuVBERkTNR4QyBoyeO/m3ZvPx4Ob069eLxUY8za9AsfPE+r+PJGZxoauaDwgMszi3lvcIqTjb5GRXfwG/q8xmwORuqKonu0gXfvfeQMmc28f31WlsREZFgqHCeRzsP72RB4QKW7V5GQ3MDo3uO5snRTzKhzwQtm4exPQdqeXntPv66uYSahiZ6djCejC9lzJ6PiMrZAECna8eT8pN/IemGCbr6j4iIyDlS4fyamv3NZJdks6BgAesq1hEfHc+0/tOYN3QegzvroJFw5fc78g808d8vrid7xwFio4353ZuYXrqRxLfexX/kCDG9LiTl0e+SMmsWsb16eR1ZREQkYqlwfkVHTxwla1cWCwsXUlpbSs9OPfn+ld9n9qDZpCSkeB1PzuBYQyOZm0p4ee0+9lafoF9CFf/ZcR+Xb1lD06J8iI2l08SJpMyZQ6exY3QAkIiIyHmgwnmOdh/ZzYKCBSzds5T6pnpG9RjF42mPc0PfG4iJ0jjD1e4Dtbz8SRGZm0o4fqKJ9LhDPLlvNamFubj6eqIGDqD7j5/EN2MGMV26eB1XRESkTVFDCoJzjjUla8goyGBt+VriouKY2n8q84bOY0iXIV7HkzPw+x3ZOw7w0idFZO84QNemOh5v2smYwo+xoj344+NJnj6NlNmz6XDFFZiZ15FFRETaJBXOIP0u93ccbDjIYyMfY/bg2XRJ0F6wcFXT0EjmxhJeXlvEvupaJtTu478P59FzyzpobCRhxOWkPPu/yE9K4tIpU7yOKyIi0uapcAbBzPjVDb/igo4XEBsV63UcOYNdVbW8vLaIRZtK6HDkIPfWfMb1u9cSe6CSaJ8P37y78M2e/bcrALnVq70NLCIi0k6ocAapV6KOUg5Hfr9j9Y4qXvpkH58UljOuqpBfHsyl7+58zO+n07ixpDz1IxJvuokonc5IRETEEyqcEpFqGhp5Y2MJr6wtoqmoiNvKN/GD4o3EHztKTI8e+B7+FimzZxPXp4/XUUVERNo9FU6JKJ8vmy9dt5tRRTk8WbGZ1LKdEBND0g0TWk5nNH68TmckIiISRlQ4Jez5/Y4Ptlfx0sd7qdiQyy37N/BSWQ5xJ+qJS00l5Ykf4ktPJ6ZbN6+jioiIyGmocErY+nzZPHP1Ngbmf8TdxRvod7gU4hPwTZlMyu1z6DBqlE5nJCIiEuZUOCXs7Ko6xp8/2sv2ldlcv+tT/qM8n9jmJuIvvZTO33uQ5KlTiU5K8jqmiIiIBEmFU8LC58vmmStzSMpeyZR965l3/CCuUyJd5t5BypzZJAwb5nVMERER+QoisnCa2XRg+sCBA72OIl/T0fpG3lhXRO6iFVy55UO+W1lAtPMTOyqNC+b+iKRJk4hKSPA6poiIiHwNEVk4nXNLgaVpaWkPeZ1FvppdVcfIXPIpTcuWMKFoPdc01NCU0oWuD9xP1zmziUtN9TqiiIiInCcRWTglMjX7HR/kFbPxL4u4aN17pFfvxm+GjbmGPt+4i8Rrr8VidSUnERGRtkaFU0LuaH0jy9/M5sgbmaTtWk96Yz313XrS6ZFHuXDuHGJ79PA6ooiIiISQCqeEzI7d5Xzyh4V0zX6bkYeLaYyOoWHs9fS6fz7JY67GoqK8jigiIiKtQIVTzqumZj8fZ71H+YLXuKRwPVc3N3KoRz/8j/6AS+++neiUFK8jioiISCtT4ZTz4lBJBZ8+/wod3l1Oz6OVJMUmcHDcRC576BsMuWqkTs4uIiLSjqlwylfmmpvZsWwVRS8voNe2jVzs/OzrNZAD8+9lzH23E5eU6HVEERERCQMqnHLOGopLyH3hFeztpSQfO0xyfCIF19zKpQ/ezZQxI7yOJyIiImFGhVOC4j95kqoVK9nz5wX4CnJJArb2Hord9TA33j+HsSmdvI4oIiIiYUqFU77UiZ072fvyQuqXLyOh7hgnO6Swemw6g74xl9uuv5yYaB1pLiIiIl9OhVP+gf/4cY6seIv9rywkfsc2Gi2aTb2Hc/z2qdx89zQm9O3sdUQRERGJICqcAoBzjoa8PKpee4OaFW8Rc6KeiqQerE27jb5zZzP7xuF07hTndUwRERGJQCqc7VzT4cPULFlC5cLXoWgPDdFxZPe+guKxN3HznIn8dFhPLZuLiIjI16LC2Q45v5+6Tz/l8BuZ1Ly7CmtqpLBzP9678nY6T72VeTcOY+iFyV7HFBERkTZChbMdaayo4Mhf/8rhzEU0l5VRG9+JVf3GkHPZtdx46zieTeurZXMRERE571Q42zjX2Mix1as5kpnJ8Q8/Ar+f3O6DeSvtbprHXcs3rhvEE0N7aNlcREREQkaFs406sWcvRxZlcjRrMc0HD1KT2JllA28ku//VjLv2cp4al6plcxEREWkVKpxtiL++noS1ayn64x+p37gJf1Q0uX2GkzVoJuWDr+Duay5msZbNRUREpJWpcEY45xwNW7dxJPMNapYtx1dbS0XXC8m6bBrv9L6SS4al8sC4i7lpaHctm4uIiIgnVDgjVPPRoxxdtowjmYs4UVCAPy6e/AGjyOg6gl09BzLryj4sHJfKkJ5aNhcRERFvqXBGEOccdes3cCQzk2MrV+JOnOBYvwEsHjOXrC7DSenehWt6NPGXuRNI6ahlcxEREQkPKpwRoLGqiqNZizmyKJPGfftxnRIpvGICLyReSmFSL8b278ovr0nlpqE9+HBNtsqmiIiIhBUVzjDlmpqoXfMhRzIzqc3OhuZmGoaNYMWUyfw5tj9RCfHMGtmHX427SMvmIiIiEtZUOMOUv66O0n/+ZywxiX03zeL5jsPIdcn0TunAD8ddxB1pfbUnU0RERCKCCmeYKqx1LHvg5/y5IoY6vzGuf1deGJfKxKE9iI4yr+OJiIiIBE2FM4w0NftZua2Slz4pYv3eQyTExnPbVX24d2wql/RM8jqeiIiIyFeiwhlGVm6r5JGMzfTp3IGf3DqUO9L64usY63UsERERka9FhTOM3DysB3+8J40bhnTXsrmIiIi0GSqcYSQ2OoqbhvXwOoaIiIjIeaVrHYqIiIhISKlwioiIiEhIqXCKiIiISEipcIqIiIhISKlwioiIiEhIqXCKiIiISEipcIqIiIhISKlwioiIiEhIhVXhNLOZZvYHM1tsZpO8ziMiIiIiX1/IC6eZvWhmVWb22Rfun2Jm281sl5n9GMA5l+Wcewj4JjA31NlEREREJPRaYw/nS8CUU+8ws2jgd8AtwDDgLjMbdspDng58XEREREQinDnnQv8kZqnAMufc8MDtscDPnXOTA7efCjz0ucDbu865VWf4Wt8CvgXQo0ePUa+++mqI00eW2tpaEhMTvY4RETSr4GhOwdOsgqM5BU+zCo7mFJxQz+mGG27Y5JxLO93HYkL2rF+uN1B8yu0S4Grgn4CbAJ+ZDXTOPf/FT3TOvQC8AJCWluYmTJgQ+rQRZPXq1WgmwdGsgqM5BU+zCo7mFDzNKjiaU3C8nJNXhdNOc59zzv0a+HVrhxERERGR0PHqKPUSoO8pt/sAZR5lEREREZEQ8moP5wZgkJldDJQCdwLzzvWLbNq0qdrM9p3vcBGuG1DtdYgIoVkFR3MKnmYVHM0peJpVcDSn4IR6Thed6QMhL5xmthCYAHQzsxLgGefcn8zsUeAdIBp40Tm39Vy/tnPugvMatg0ws41nesGu/D3NKjiaU/A0q+BoTsHTrIKjOQXHyzmFvHA65+46w/0rgBWhfn4RERER8VZYXWlIRERERNoeFc625wWvA0QQzSo4mlPwNKvgaE7B06yCozkFx7M5tcqJ30VERESk/dIeThEREREJKRVOEREREQkpFc4IZWZTzGy7me0ysx+f5uPpZpZvZrlmttHMxnuR02tnm9Mpj7vKzJrNbE5r5gsnQWxTE8zsaGCbyjWzn3mR02vBbFOBWeWa2VYzy27tjOEiiG3qiVO2p88C34NdvMjqpSDm5DOzpWaWF9im7vMiZzgIYladzezNwP9/681suBc5vWZmL5pZlZl9doaPm5n9OjDHfDO7MuShnHN6i7A3Ws5duhvoD8QBecCwLzwmkf95je7lQKHXucNxTqc87n1aTtM1x+vc4TorWs6nu8zrrBEwpxRgG9AvcLu717nDdVZfePx04H2vc4fjnIB/Af4j8P4FwCEgzuvsYTqr/0PL+b4BhgDveZ3bo1ldB1wJfHaGj98KvEXLpcbHAOtCnUl7OCPTaGCXc26Pc+4k8CqQfuoDnHO1LrBVAZ2A9nh02FnnFPBPwCKgqjXDhZlgZ9XeBTOnecBfnXP7AZxz7XW7Otdt6i5gYaskCy/BzMkBSWZmtOxMOAQ0tW7MsBDMrIYB7wE45wqBVDPr0boxveecW0PLdnIm6cDLrsWnQIqZXRjKTCqckak3UHzK7ZLAfX/HzGaZWSGwHLi/lbKFk7POycx6A7OA51sxVzgKapsCxgaW9d4ys0tbJ1pYCWZOg4HOZrbazDaZ2T2tli68BLtNYWYdgSm0/OLX3gQzp98CQ4EyYAvwPeecv3XihZVgZpUH3AZgZqNpudRin1ZJF1mC/v48X1Q4I5Od5r5/2IPpnHvTOTcEmAk8G+pQYSiYOf0KeNI51xz6OGEtmFltBi5yzo0AfgNkhTpUGApmTjHAKGAqMBn4qZkNDnWwMBTUz6mA6cDHzrkv2yPTVgUzp8lALtALuAL4rZklhzZWWApmVs/R8gtfLi2rVzm0z73BZ3Mu35/nRcgvbSkhUQL0PeV2H1p+8z0t59waMxtgZt2cc9UhTxc+gplTGvBqy0oV3YBbzazJOZfVKgnDx1ln5ZyrOeX9FWb2X9qmTrtNlQDVzrnjwHEzWwOMAHa0TsSwcS4/p+6kfS6nQ3Bzug94LvAyqV1mtpeW1yeub52IYSPYn1P3QcuBMcDewJv8vXPqEeeD9nBGpg3AIDO72MziaPlhveTUB5jZwMA3G4Gjz+KAg62e1FtnnZNz7mLnXKpzLhXIBB5ph2UTgtumep6yTY2m5eeHtqkvzAlYDFxrZjGBpeKrgYJWzhkOgpkVZuYDrqdlbu1RMHPaD0wECLwe8RJgT6umDA/B/JxKCXwM4EFgzam/LMvfLAHuCRytPgY46pwrD+UTag9nBHLONZnZo8A7tBy196JzbquZfTvw8eeB2bRsTI1APTD3lIOI2oUg5yQEPas5wHfMrImWbepObVP/OCfnXIGZvQ3kA37gj865056apC07h++/WcDKwB7hdifIOT0LvGRmW2hZCn2yna0sAEHPaijwspk103K2iAc8C+whM1tIy5lFuplZCfAMEAt/m9MKWo5U3wXUEdgrHNJM7ez/CxERERFpZVpSFxEREZGQUuEUERERkZBS4RQRERGRkFLhFBEREZGQUuEUERERkZBS4RQRERGRkFLhFBE5hZm9aGZVZtbq5840s2+a2W8D7//czH74JY+dY2arzCzfzNaY2b1f8tghZrbWzE582dcUEQkVFU4Rkb/3EjAl2AcHrtTRqj9Lzew5Wk6Y/qBz7nJgJnClmf2/M3zKIeAx4P+2TkIRkb+nwikicgrn3BpaCtoZmVmqmRWY2X8Bm4G+ZvaEmW0I7HH8xSmPvSdwX56ZvRK4b7qZrTOznMBeyh7B5jOzCcBFzrn5zrmiQOZDzrnv0XJVkatO83eqcs5tABqDfR4RkfNJl7YUEflqLgHuc849YmaTgEHAaFouPbjEzK6j5VrzPwGucc5Vm1mXwOd+BIxxzjkzexD4EfB4kM/7LeCZwHXa/wQMAN4FDgP/CdxPyzWnRUTChgqniMhXs88592ng/UmBt5zA7URaCugIIPPz61475z7fc9oHeM3MLgTigL3n8Lx9nHM7zOx7wDrn3F1m9mta9l5up6WAioiEFS2pi4ichZn1NbPcwNu3A3cfP/UhwL87564IvA10zv0pcL87zZf8DfBb59xlwMNAwjnE8Qf+HAK8HXj/rcCf3YGqc/haIiKtQoVTROQsnHPFp5TJ50/zkHeA+80sEcDMeptZd+A94A4z6xq4//MldR9QGnj/jEeXn0GlmQ2gZW/mpMB9k2lZsXoa+Ms5fj0RkZBT4RQROYWZLQTWApeYWYmZPXC2z3HOrQQWAGvNbAuQCSQ557YC/xvINrM84JeBT/k58IaZfQhUn2PEP9FytPkfgGvMbD1QC9wIZDvn3jvN36mnmZUAPwCeDvy9ks/xeUVEvjJz7nSrPSIiEq7M7De07NH8aeBgpGTgDuB151yNt+lERP6RCqeISAQys7uBB4FkWo6Gf8E594a3qURETk+FU0RERERCSq/hFBEREZGQUuEUERERkZBS4RQRERGRkFLhFBEREZGQ+v89Wc/CQUAJQAAAAABJRU5ErkJggg==\n",
"text/plain": [
"<Figure size 792x504 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"ds = datasets.DatasetDeep1B(10**6)\n",
"\n",
"sqrt_nlist = 128\n",
"nlist = sqrt_nlist**2\n",
"maxtrain = nlist * 50\n",
"\n",
"pyplot.title(\"Deep1M\")\n",
"eval_quantizers(ds, sqrt_nlist, nprobes=[1, 2, 4, 8, 16, 32, 64, 128, 256])"
]
},
{
"cell_type": "code",
"execution_count": 26,
"id": "81fcfb7c",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"flat\n",
"nprobe=1 1-recall @ 1: 0.2344 dis/q=102.55\n",
"nprobe=2 1-recall @ 1: 0.3555 dis/q=202.20\n",
"nprobe=4 1-recall @ 1: 0.5023 dis/q=398.57\n",
"nprobe=8 1-recall @ 1: 0.6451 dis/q=778.48\n",
"nprobe=16 1-recall @ 1: 0.7696 dis/q=1521.64\n",
"nprobe=32 1-recall @ 1: 0.8722 dis/q=2946.55\n",
"nprobe=64 1-recall @ 1: 0.9399 dis/q=5679.57\n",
"nprobe=128 1-recall @ 1: 0.9749 dis/q=10928.08\n",
"nprobe=256 1-recall @ 1: 0.9922 dis/q=20974.09\n",
"IMI\n",
"nprobe=1 1-recall @ 1: 0.278 dis/q=525.55\n",
"nprobe=2 1-recall @ 1: 0.412 dis/q=906.89\n",
"nprobe=4 1-recall @ 1: 0.5487 dis/q=1530.37\n",
"nprobe=8 1-recall @ 1: 0.6792 dis/q=2579.32\n",
"nprobe=16 1-recall @ 1: 0.7864 dis/q=4331.35\n",
"nprobe=32 1-recall @ 1: 0.8715 dis/q=7236.14\n",
"nprobe=64 1-recall @ 1: 0.9346 dis/q=12052.80\n",
"nprobe=128 1-recall @ 1: 0.9695 dis/q=20145.48\n",
"nprobe=256 1-recall @ 1: 0.9878 dis/q=33822.70\n",
"RCQ\n",
"nprobe=1 1-recall @ 1: 0.2385 dis/q=252.10\n",
"nprobe=2 1-recall @ 1: 0.3595 dis/q=443.67\n",
"nprobe=4 1-recall @ 1: 0.496 dis/q=784.66\n",
"nprobe=8 1-recall @ 1: 0.6345 dis/q=1364.02\n",
"nprobe=16 1-recall @ 1: 0.7551 dis/q=2400.98\n",
"nprobe=32 1-recall @ 1: 0.8565 dis/q=4255.88\n",
"nprobe=64 1-recall @ 1: 0.924 dis/q=7601.78\n",
"nprobe=128 1-recall @ 1: 0.9667 dis/q=13698.68\n",
"nprobe=256 1-recall @ 1: 0.9906 dis/q=24946.44\n",
"2-level clustering\n",
"2-level clustering of (100000, 128) nb clusters = 128*128 = 16384\n",
"perform coarse training\n",
"\n",
"assigning the training set\n",
"done in 0.07 s. Sizes of clusters 312-1678\n",
"done in 5.52 sing sub-cluster 127/128\n",
"nprobe=1 1-recall @ 1: 0.2407 dis/q=113.52\n",
"nprobe=2 1-recall @ 1: 0.369 dis/q=224.14\n",
"nprobe=4 1-recall @ 1: 0.5076 dis/q=440.81\n",
"nprobe=8 1-recall @ 1: 0.656 dis/q=863.22\n",
"nprobe=16 1-recall @ 1: 0.7759 dis/q=1681.53\n",
"nprobe=32 1-recall @ 1: 0.8705 dis/q=3260.76\n",
"nprobe=64 1-recall @ 1: 0.9381 dis/q=6289.80\n",
"nprobe=128 1-recall @ 1: 0.9757 dis/q=12082.95\n",
"nprobe=256 1-recall @ 1: 0.9931 dis/q=23143.32\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 792x504 with 1 Axes>"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"ds = datasets.DatasetBigANN(nb_M=1)\n",
"\n",
"sqrt_nlist = 128\n",
"nlist = sqrt_nlist**2\n",
"maxtrain = nlist * 50\n",
"\n",
"pyplot.title(\"bigann 1M\")\n",
"eval_quantizers(ds, sqrt_nlist, nprobes=[1, 2, 4, 8, 16, 32, 64, 128, 256])"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"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": "3.8.10"
}
},
"nbformat": 4,
"nbformat_minor": 5
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment