{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "db2b2b85",
   "metadata": {},
   "source": [
    "# RAG — From Basic to Production\n",
    "\n",
    "### Five techniques. One travel corpus. Real numbers.\n",
    "\n",
    "| # | Technique | What it adds |\n",
    "|---|---|---|\n",
    "| 1 | **Basic RAG** | Vector similarity. The baseline. |\n",
    "| 2 | **Reranking** | A second model that re-orders top candidates. |\n",
    "| 3 | **Multi-query + Rerank** | Decompose the query, fuse the results. |\n",
    "| 4 | **CRAG** | Know when *not* to answer. |\n",
    "| 5 | **RAGAS** | Measure, don't intuit. |\n",
    "\n",
    "The corpus: **11,319 chunks across 305 WikiVoyage destinations**. Embeddings: `all-MiniLM-L6-v2` (384-d).\n",
    "\n",
    "A single multi-aspect query — *\"Tropical destinations with great snorkeling and vegetarian food\"* — is the through-line. Every technique is judged on whether it makes that query work better."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "17b0fdca",
   "metadata": {},
   "source": [
    "## Setup\n",
    "\n",
    "Run these cells once before the presentation. Dataset download is ~160 MB."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "ca00b626",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "✓ packages already installed (local execution)\n"
     ]
    }
   ],
   "source": [
    "import subprocess, sys\n",
    "subprocess.run([sys.executable, \"-m\", \"pip\", \"install\", \"-q\",\n",
    "    \"chromadb\", \"sentence-transformers\", \"huggingface_hub\", \"rank_bm25\", \"matplotlib\"], check=True)\n",
    "print(\"✓ packages installed\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "e1041b7e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "✓ dataset ready at /Users/psulin/repos/dev/iris-vector-rag/docs/datasummit/notebooks/data\n"
     ]
    }
   ],
   "source": [
    "from huggingface_hub import snapshot_download\n",
    "from pathlib import Path\n",
    "\n",
    "DATA = Path(snapshot_download(\n",
    "    repo_id=\"patjs/rag-workshop\",\n",
    "    repo_type=\"dataset\",\n",
    "    local_dir=\"./data\",\n",
    "))\n",
    "print(f\"✓ dataset ready at {DATA}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "97418e71",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/psulin/repos/dev/iris-vector-rag/.venv/lib/python3.12/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Warning: You are sending unauthenticated requests to the HF Hub. Please set a HF_TOKEN to enable higher rate limits and faster downloads.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      "Loading weights:   0%|          | 0/103 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      "Loading weights: 100%|██████████| 103/103 [00:00<00:00, 14363.58it/s]"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "✓ 11,319 chunks in vector index\n",
      "✓ 11,319 chunks in BM25 index\n",
      "✓ ground truth: 21 destinations for Query B\n",
      "\n",
      "Query A: What's there to do in Iceland?\n",
      "Query B: Tropical destinations with great snorkeling and vegetarian food\n",
      "Query C: Cheap nightlife in Bangkok with live music\n",
      "Query E: What does it cost to travel in New Zealand in October?\n"
     ]
    }
   ],
   "source": [
    "# All implementation lives here so the slide cells stay clean.\n",
    "import chromadb, json, numpy as np, matplotlib.pyplot as plt\n",
    "from collections import defaultdict\n",
    "from dataclasses import dataclass\n",
    "from chromadb.utils.embedding_functions import SentenceTransformerEmbeddingFunction\n",
    "from rank_bm25 import BM25Okapi\n",
    "\n",
    "# ── Vector store ─────────────────────────────────────────────────────\n",
    "EF = SentenceTransformerEmbeddingFunction(model_name=\"all-MiniLM-L6-v2\", device=\"cpu\")\n",
    "collection = chromadb.PersistentClient(path=str(DATA/\"chroma_db\")).get_collection(\"workshop\", embedding_function=EF)\n",
    "\n",
    "# ── Queries and ground truth ─────────────────────────────────────────\n",
    "QUERY_A = \"What's there to do in Iceland?\"\n",
    "QUERY_B = \"Tropical destinations with great snorkeling and vegetarian food\"\n",
    "QUERY_C = \"Cheap nightlife in Bangkok with live music\"\n",
    "QUERY_E = \"What does it cost to travel in New Zealand in October?\"\n",
    "\n",
    "with open(DATA/\"queries\"/\"ground_truth_b.json\") as f:\n",
    "    GROUND_TRUTH_B = set(json.load(f)[\"ground_truth\"])  # 21 destinations\n",
    "\n",
    "# ── Data model ───────────────────────────────────────────────────────\n",
    "@dataclass\n",
    "class Chunk:\n",
    "    text: str\n",
    "    doc_id: str\n",
    "    section: str\n",
    "    score: float\n",
    "\n",
    "# ── Core retrieval ───────────────────────────────────────────────────\n",
    "def vector_search(query: str, n: int = 20) -> list[Chunk]:\n",
    "    r = collection.query(query_texts=[query], n_results=n)\n",
    "    return [Chunk(d, m[\"doc_id\"], m.get(\"section\",\"\"), round(1-s, 4))\n",
    "            for d, m, s in zip(r[\"documents\"][0], r[\"metadatas\"][0], r[\"distances\"][0])]\n",
    "\n",
    "# ── BM25 (used in helpers, not a separate slide) ────────────────────\n",
    "_corpus = []\n",
    "for f in sorted((DATA/\"corpus\").glob(\"*/chunks.jsonl\")):\n",
    "    slug = f.parent.name\n",
    "    for line in f.read_text().splitlines():\n",
    "        if line.strip():\n",
    "            ch = json.loads(line)\n",
    "            _corpus.append(Chunk(ch[\"text\"], slug, ch.get(\"section\",\"\"), 0.0))\n",
    "_bm25 = BM25Okapi([c.text.lower().split() for c in _corpus])\n",
    "\n",
    "def bm25_search(query: str, n: int = 20) -> list[Chunk]:\n",
    "    sc = _bm25.get_scores(query.lower().split())\n",
    "    idx = np.argsort(sc)[::-1][:n]\n",
    "    mx = sc[idx[0]] or 1\n",
    "    return [Chunk(_corpus[i].text, _corpus[i].doc_id, _corpus[i].section, round(float(sc[i])/mx, 3)) for i in idx]\n",
    "\n",
    "def rrf(lists, k: int = 60) -> list[Chunk]:\n",
    "    s, b = defaultdict(float), {}\n",
    "    for rl in lists:\n",
    "        for rank, c in enumerate(rl, 1):\n",
    "            key = (c.doc_id, c.section)\n",
    "            s[key] += 1/(k+rank); b[key] = c\n",
    "    return [Chunk(b[k].text, b[k].doc_id, b[k].section, round(s[k],4))\n",
    "            for k in sorted(s, key=s.get, reverse=True)]\n",
    "\n",
    "# ── Recall ───────────────────────────────────────────────────────────\n",
    "def unique_slugs(chunks, k):\n",
    "    seen, out = set(), []\n",
    "    for c in chunks:\n",
    "        if c.doc_id not in seen:\n",
    "            seen.add(c.doc_id); out.append(c.doc_id)\n",
    "        if len(out) == k: break\n",
    "    return out\n",
    "\n",
    "def recall_at_k(chunks, gt, k):\n",
    "    return len(set(unique_slugs(chunks, k)) & gt) / len(gt) if gt else 0.0\n",
    "\n",
    "# ── Display ──────────────────────────────────────────────────────────\n",
    "def show(chunks, gt=None, n=10, title=\"\"):\n",
    "    if title:\n",
    "        print(f\"\\n  {title}\\n  {'─'*max(60, len(title))}\")\n",
    "    print(f\"  {'#':>2}  {'score':>6}  {'destination':<22}  section\")\n",
    "    for i, c in enumerate(chunks[:n], 1):\n",
    "        hit = \" ✓\" if gt and c.doc_id in gt else \"\"\n",
    "        print(f\"  {i:>2}.  {c.score:>6.3f}  {c.doc_id:<22}  {c.section[:18]}{hit}\")\n",
    "\n",
    "# ── Visual styling (presentation-grade) ──────────────────────────────\n",
    "plt.rcParams.update({\n",
    "    \"figure.figsize\": (12, 5),\n",
    "    \"figure.dpi\": 110,\n",
    "    \"font.size\": 14, \"axes.titlesize\": 18, \"axes.titleweight\": \"bold\",\n",
    "    \"axes.labelsize\": 14, \"axes.spines.top\": False, \"axes.spines.right\": False,\n",
    "    \"axes.grid\": True, \"grid.alpha\": 0.25,\n",
    "    \"xtick.labelsize\": 12, \"ytick.labelsize\": 12,\n",
    "    \"legend.fontsize\": 12, \"legend.frameon\": False,\n",
    "    \"lines.linewidth\": 3, \"lines.markersize\": 9,\n",
    "})\n",
    "ACCENT, WARN, GOOD, NEUTRAL, PURPLE = \"#0E63B0\", \"#D8511D\", \"#1E8E50\", \"#666666\", \"#7B5BA6\"\n",
    "\n",
    "print(f\"✓ {collection.count():,} chunks in vector index\")\n",
    "print(f\"✓ {len(_corpus):,} chunks in BM25 index\")\n",
    "print(f\"✓ ground truth: {len(GROUND_TRUTH_B)} destinations for Query B\")\n",
    "print(f\"\\nQuery A: {QUERY_A}\")\n",
    "print(f\"Query B: {QUERY_B}\")\n",
    "print(f\"Query C: {QUERY_C}\")\n",
    "print(f\"Query E: {QUERY_E}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3ed995b2",
   "metadata": {},
   "source": [
    "---\n",
    "\n",
    "# Basic RAG\n",
    "\n",
    "### *Vector similarity. The baseline you build everything else on.*"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9458d74d",
   "metadata": {},
   "source": [
    "## How it works\n",
    "\n",
    "```\n",
    "query ──► embed ──► cosine similarity ──► top-k chunks\n",
    "```\n",
    "\n",
    "The query becomes a **384-dimensional vector**. Every chunk in the corpus has been pre-embedded into the same space.  Retrieval is one matrix multiply: the chunks whose vectors point in the same direction as the query come back at the top.\n",
    "\n",
    "**No keywords. No rules. No LLM in the retrieval loop.** Pure semantic similarity, and that's already enough to feel like magic on the right kind of question.\n",
    "\n",
    "The whole talk is about *which* questions break that and what to do about them."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "274ef73e",
   "metadata": {},
   "source": [
    "## Two queries — one easy, one hard\n",
    "\n",
    "We'll judge every technique against the same pair.\n",
    "\n",
    "| | Query A | Query B |\n",
    "|---|---|---|\n",
    "| **Question** | *\"What's there to do in Iceland?\"* | *\"Tropical destinations with great snorkeling and vegetarian food\"* |\n",
    "| **Constraints** | one — a single named destination | three — geography · activity · diet |\n",
    "| **Expectation** | basic RAG should nail it | basic RAG should struggle |\n",
    "\n",
    "If Query A doesn't work, nothing later will. If Query B works on basic RAG, you don't need the rest of this talk."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "77f25a81",
   "metadata": {},
   "source": [
    "### Query A — *\"What's there to do in Iceland?\"*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "8ad70ddf",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "  Query A — \"What's there to do in Iceland?\"\n",
      "  ────────────────────────────────────────────────────────────\n",
      "   #   score  destination             section\n",
      "   1.   0.671  iceland                 Sleep\n",
      "   2.   0.669  iceland                 Overview\n",
      "   3.   0.640  iceland                 Sleep\n",
      "   4.   0.615  iceland                 Do\n",
      "   5.   0.611  iceland                 Respect\n"
     ]
    }
   ],
   "source": [
    "results_a = vector_search(QUERY_A, n=10)\n",
    "show(results_a, n=5, title=f'Query A — \"{QUERY_A}\"')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7b8c677e",
   "metadata": {},
   "source": [
    "**All five top results are Iceland chunks.** The retriever is alive. Top-1 score is 0.671 — a strong match. The question names the destination directly, the embedding model has seen \"Iceland\" in similar contexts a million times, and the right chunks come back.\n",
    "\n",
    "This is the case where basic RAG is *enough*. Plenty of production RAG systems ship at exactly this level of sophistication and serve their users well. The trouble starts when the question stops being so direct."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "43d7d8f4",
   "metadata": {},
   "source": [
    "### Query B — *\"Tropical destinations with great snorkeling and vegetarian food\"*\n",
    "\n",
    "Three constraints rolled into one sentence: **geography · activity · diet**.\n",
    "\n",
    "The query is still one vector. **No single chunk in the corpus was written to satisfy all three at once.**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "ead2e66b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "  Query B — \"Tropical destinations with great snorkeling and vegetarian food\"\n",
      "  ───────────────────────────────────────────────────────────────────────────\n",
      "   #   score  destination             section\n",
      "   1.   0.578  galapagos               Do\n",
      "   2.   0.575  phuket                  Other destinations ✓\n",
      "   3.   0.571  fernando-de-noronha     Do\n",
      "   4.   0.570  salar-de-uyuni          Eat and drink\n",
      "   5.   0.556  south-africa            See\n",
      "   6.   0.548  maldives                See\n",
      "   7.   0.536  seychelles              Do\n",
      "   8.   0.534  koh-phi-phi             Do ✓\n",
      "   9.   0.522  melbourne               Eat\n",
      "  10.   0.517  madagascar              Do\n",
      "\n",
      "  Recall at top-k against 21 relevant destinations:\n",
      "    R@5  = 0.05\n",
      "    R@10 = 0.10\n",
      "    R@20 = 0.14\n"
     ]
    }
   ],
   "source": [
    "results_b = vector_search(QUERY_B, n=20)\n",
    "show(results_b, gt=GROUND_TRUTH_B, n=10, title=f'Query B — \"{QUERY_B}\"')\n",
    "\n",
    "print(f\"\\n  Recall at top-k against {len(GROUND_TRUTH_B)} relevant destinations:\")\n",
    "for k in [5, 10, 20]:\n",
    "    print(f\"    R@{k:<2} = {recall_at_k(results_b, GROUND_TRUTH_B, k):.2f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a2175b49",
   "metadata": {},
   "source": [
    "The top-10 looks tropical-ish — Galápagos, Maldives, Seychelles, Madagascar — but **the destinations the query actually pointed at are missing from the top 20**. Bali, Indonesia, Kerala, Sri Lanka, India: nowhere to be seen.\n",
    "\n",
    "Recall at top-20 against the 21-destination ground truth is **0.14**. Three out of twenty-one. Worse, the top-1 score is **0.578** — well below the 0.671 we saw for Query A.\n",
    "\n",
    "The retriever isn't broken. It's behaving correctly. **It's the question that's harder than the embedding model can represent in a single vector.**"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "27710119",
   "metadata": {},
   "source": [
    "## The diagnostic"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "4b6ee8b8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABZMAAAINCAYAAABLQeFNAAAAOnRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjEwLjksIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcvJkbTWQAAAAlwSFlzAAAQ6wAAEOsBUJTofAABAABJREFUeJzs3Qd0FFUXB/CbHhIg9N5BpBcRRHpTbKig8oGK2BCwo4IgNhBQRLEAYkMBaYKCXVGqAtIsdKT3EnpIQvp+5/9gltnZ2ZpNMpv8f+fsSbI7u5mdN/XOffeF2Gw2mxARERERERERERERuRHq7kUiIiIiIiIiIiIiIgaTiYiIiIiIiIiIiMgrzEwmIiIiIiIiIiIiIo8YTCYiIiIiIiIiIiIijxhMJiIiIiIiIiIiIiKPGEwmIiIiIiIiIiIiIo8YTCYiIiIiIiIiIiIijxhMJiIiIiIiIiIiIiKPGEwmIiIiIiIiIiIiIo8YTCYiIiIiIiIiIiIijxhMJiIi8sH9998vISEhDo99+/bl6jJ85JFHHP7/W2+9JflBtWrVHL4X/s5vpk6d6rT+vPrqqw7T2Gw2adiwocM0q1atEitatmyZfR47dOiQ17NDREHCuB/k/oMKKpxDGrcHnGvmpj179kh0dLT9/5cpU0YuXLiQq/NgdVZop4JKf67p6tyZ3Lvlllsclt+sWbPEEsHkrKws+fbbb+XRRx+VJk2aSLly5SQyMlKKFy8utWrVkttuu03ee+89OXr0aCD+HfngwIEDEhoa6rTx4fHJJ59wWRJZ0JkzZ2ThwoUycuRIufnmm6VUqVK86CK7f//9V6ZMmWL/u1ixYtK/f3+vgt488QoOaKshQ4Y4PPfkk0+qIDP5b/HixfLMM89I8+bNpWLFiurCuWjRolKjRg3p2rWrvPHGG7Jr1y4u4ly+aWR8oF1w3Lvqqqvkvvvuk7lz50pGRgbbhSgb1+pbtmyRzz77TJ0vNG7cWMLDw522PQRsqGB67rnnJDU11eGco1ChQnk6T0QUOM8//7zT39m9YZTtYPL3338v9erVk9tvv10mT54sGzZskOPHj0t6erqcPXtWdu/eLd999508/fTT6mT9qaeekuTk5Oz+W/IS7ji4uvgMxN2I/A6BF55o5b6CfvcRF9A33HCDvPLKK/LTTz/JqVOn8nqWLKsgbqPDhw9XF4aaxx57TIoUKZKn80SB17t3b6latar977/++ku+/vprLmo/rFy5UgWQu3TpIu+8846sX79ejhw5oi6cz58/L3v37pVff/1Vhg0bJnXq1FEBzJMnT3JZ5xG0C457//zzj3zxxRfyv//9T+rXry9bt25lm5DfkPlsPF8oKLAdNWjQQB566CH5+OOPZePGjZKZmZnXs2VZBS1rft26dbJgwQL73zinxLklEeUfbdu2lVatWtn/PnTokEyaNCnvgskIdCDr+L///vNq+pSUFHn//feldevWauYp57kLGP/+++9sByILYvYhuYKAIm4waNDzhCf8+ROyxgYOHOjwHHorkG/QC6tjx44qgOwNBFgQeEHwefPmzVzcFrFjxw65/vrrVfCfiHzD80pyx3huce+996oe5kRWUalSJZWUqn+0bNkyr2cr6KDHgd6bb76pYrT+Cvf3jQgKm13UREREqLt3NWvWVJnJS5Yskfj4eKcuurfeequq/4eubJQzNm3apB6uILNt9uzZMnjwYDYBkYUDSldccYVs27Ytr2eFLODdd991+Lt9+/ZSvnz5PJsfylnIyBw6dKj9bxzTUaqhc+fOXPReQAk2dOk2BlJwEwaJDXXr1lW95f744w/Zv3+/U23Em266Sf7++29VcoFy1h133KEuFiEhIUFdPxjb5PDhwzJ9+nTeQCPKprJly6qf6E1MBdvOnTvlxx9/dHiuV69eeTY/RGZQOtd4DUS+69atm8TGxkpSUpL6+8SJEzJz5kzVayXXMpPRzQx1dYyaNWumspTRVRAlLxCoRAay2bTouvbSSy/5NdPkHW/KWLDUBZE1d/Tjxo1TAQ5cVOszUangwg3aefPmOTzHE/78X1vWmHnB8Q68gzIVffv2dQokV69eXWX4o3fWRx99pLKQUeZi/PjxTt3eDx486JQdTjnj8ccfVxeKeKCuK26goq6r0YoVK9gERH5m9eHaHPu7Y8eOqXJqRJ9++qnDcRLjCaA7PBHlPzExMSrOEKjrCr+CyWPGjFE1kfUw4ucvv/yiTtKNmcoIiqC7hBECzsZaoGb1LzHyeSBG08RdtwceeEBloqDrRlRUlFSoUEENuoJ6Ie5SvN3VcF20aJHKqMDOF98XgyGNHTvWaXpjEEAPA+0Yp0cgyV84KOCEQa9w4cJy4403OmWJ5+cadOiuirstaB9kd6IGFNoIWUZYD1A/EYPxoA4lBj0zDg4zYsQIp89Ed1l3NVqNA8vgb0B9RtxYwf9FW+C1b775xunzcQMG61a7du3UYJZYTzFAEO7I3XPPPWo90tdL9aWG7JdffqnWd2yv+Fxsr+gij3nzBBdwWI7IZkCPAtRAHzBggH2wIrNadNhOfaENGIZlbIS28LaOMrozoxtH06ZNVVujzUuWLKkuTPF9V69e7VddvcTERBk1apQaaBRtgsc111yj9h+BrD03YcIEta60adMmVwa/MH5XbZ3FzUHsM3ERgvm48sor5cUXX1RBTQ2+N+rvtWjRQi2PuLg4dRKK7c4VLEfU23/55ZdV5h/q+GFfjPUK/wfrPT4DAwO4ysjOzjZqXK8RxECd6tKlS6vBY/GzYcOGKhCF/ah299YTbJeff/65yhbG+qYtM2SW6peZv1DPTj84CmC8gkBCHUWsewhgYlvHfqJEiRJqbISePXuqIM/p06fdjgb+wgsvqKxPbTBevB/bDD43EIOb4bzhq6++UssV+3DMm7avxEkS1iVk7mLdwCC0rrg7j9i+fbvKaMU+Eusl2hPHT9ws9wR3+fFdcczBe7EcMZAmzpH80b17d6dsW46w7hmCkufOnXN4DusI2gHrox7aftCgQaoeuRHOD4znSTgv9fZ45GvtTeyTEMBu1KiRWu+wDWEdwn4F55bu9iXu1mnUxEQtaNThxnLAazgnME6Pc3ZXPvjgA6fpEYzPCdh/ml07uNv/BBvcDNQvS7SLcR+PY6BxmRt7fOJv4zQ4rpnBtQ4uHHv06KGOozgfxX4Tv6NOu9l5aaCP377WO0cWlf67oWcBjkXe0G8Ty5cvd3rdbABIMzgPwPqPnrVVqlRR84TlVrlyZbV/R49ddyVY3F1H4sYWjq+4jsQ6gM/EdhvI6zMcK7FPxDqnnePlJFfX8zhP+vDDD9U5I67JcL6F6xJcS+vhGIzzTyxrrFuYZ1xzoHeCK+i5g89++OGH1bk5eknjel+7Nq9du7Zazjg/Nbvm1+/XjbDueLsvx3HnvffeU+sK9rfaNobzCbzntddeU+da3sJ506OPPqq+j6/nI96YM2eOw9+Yb1fbgdmxT4vTYH5wvoLecliPce2AY4435VCxXuAcF9OjnXAtoR37cC2EBERjT5XsCnQ75UT8CfEc7DtwXEZPNVzXYl6x7WD5YF3Ac/369VO9efy5tsVywPaKa2QsdzzvSwYwrhf0n6kf60Pz888/O0yD60XjdfPatWud5u+tt97yeTwlV9sp9h2I+eC6DO2MfQLOq4zxstyIqWRmZqr/e/fdd6v4DpYHtm3s+3Fdhx5Y7uIK7rZDxImwf8CxOCwszOl81+y6Ys2aNeomo19sPjp//rwtPDwct68cHm+99Zbb9x0+fNgWFhbm9L5JkyY5TPfKK684TfP55587fd7evXudpuvbt6/p/961a5ft6quvdpre+KhYsaJt5cqVpp+xdOlSp+kxr4MGDXJ6Pi4uznbixAlbVFSUw/M33HCDy+VTo0YNh2mrV69uy8rKsvnrjz/+cJqv//3vf7YZM2Y4Pf/CCy/Y8iO0QbNmzTy2u/Z47bXX7O+tWrWq1+/DA+uHq/fib7xevHhxp/ctWLDA/r7MzEzbq6++aouIiPD4/+rVq2fbvHmz6fc224a+//57tf65+ryyZcvatm/f7nJZjh492hYSEmL63tjYWNs333xja9++vdNr2E59gW3Yl+WO76p39uxZ2x133OHVe2+77TbbmTNnTOfD7Lvs2LFDbZeuPq9ly5bq/+cEs/0d5jFQjJ+NdRbrZqFChUy/a926dW1Hjx61JSQk2Lp06eJymTz55JOm/w/ro7dtHBoaanv++eedPiM72ygcPHjQ1rFjR7/ea7ae/vPPP7Z27dq5/IzatWvbjh07lq12uvfeex0+E+ujO2bzadxmNKdPn7bdddddXi0Ps2Nyenq67ZlnnjE9zusfOH94+eWXXR7fzPafRhMmTPC63SMjI23vvfee6f9ydR6B8xK8z9VnvvPOOy6X+V9//WUrVaqUy/c+/vjjavl52y6wbNkyp+kXL15sswr9+VEg90vZVa5cOdPl705ycrKtRIkSTu8bPHiww3S+tKFxOlfL6Pjx47auXbt6XKdxLqE/d/BmnX777bdNt834+Hin5YT9uyudOnVymLZw4cK2xMREm6/M9t/G/Szg2sI43X333WfLLz744AOn77dixQqHaT755BOnaYztP3/+fKdp5s2b5/T/fvnlF9Ptwvho1aqV7ciRIzl2/PZ22/j777/VNZV+GpyLTpkyxetlbLZNeHoY4Ry3ZMmSXm2bX331lU/XkbgGc3V+jePQ9OnTbTnF7BzBbDv0h9m1CNZ34z5E367asRXLMCYmxnS6MmXKuLxe8eWaD+dkGzZs8Lhfd/cw25d/+OGHar/o63td7btx3e7qXNzT+Yg3ECMxfua0adNcTm+2jLCPuv/++13OI64TEZdwZevWrbaGDRt6XGY4f3zxxRfV9XJ2Bbqdcir+hLibL+skrvV9ubbdsmWLrVKlStlar4YMGeL0/kOHDjlMM3z4cNPrJr3x48c7TbNu3TqP+1AjszbEMbJIkSIul9tTTz2VazGV9evX22rVquWxLXEutm3bNp+2Q7PruMaNGzu9f9++fU7T+XJcdVjevr7hp59+Mv3CuDD3xOzCvWfPnjkaTEYAyOzCwNUDAWCzHZ7ZCmy28eGBEx+zC3+cVBk3LsDGZPwMXGxnx4ABA5w+c+7cuSrYZbxI9hSQCFZ9+vTxaQecU8Fk7LyMJ8NmFwQPPvigT/8TB0HjiZCrbQgnX54+79prrzVdjlOnTvVquylfvnyeBpNxwG3UqJFP769fv74KiHpzwK1cubLHz8NJcnZuAlklmIyT+OjoaLffFUH7Hj16eFwmCxcuzNbFqPZ44403HD4jO9vonj171MHen/e6Wk+NNwTNHr17985WO1WpUsXt8dPfYPKpU6dsderU8Xp5GI/JOLG/9dZbfWqPxx57LFeCydpjzpw5Xm1XpUuX9uqCxuxmHm4WuAsk628GetMu+n0bzh8CeY6Q34PJuDg1W/buLmg1DzzwgNP7WrRokaPBZAR13d2sND6wPnz55ZderdOuzlXxwEUnLs6Nz69atcr0Br0xIO1vYNebYHJKSoppcGj27Nm2/AIX8sbvN3bsWIdpzII0zz33nMM0zz77rNM0WKf0cA1g3I+4e6CN0OY5cfz2ZtvARbRxf+xrIDkQwWRvzoGND7N5NLuO9Oa8EttcTt08zO1gsqfrcXxXBESMyVjeXq/4EkzWgng4/wlUMBk3Hf19r9l6im3Q0zbr6nzEW1hXjZ+J46crZsvIm2tMfJfU1FSnz9u0aZOtaNGiPi337N5QDHQ75WT8yddgsrvkRbNrW1fnB74Ek3/88Uen9xvPTzp06OA0zcSJEx2mMSaDYb3IyMjIdjAZ7eAp0cXV9WqgYyorVqzwuH/TP4oVK6bWJW+2Q1dtaRZMBuP38He78nkAPpRFMEIXBG3QDHeuvvpqWbp0qcfPCxR0mUC3fGOXOKSjoxs0ugigO7x+HtC9DN0IMGo0ujC5g3IEgM/BCNPoTo8uGNrnoUvKjBkzHOYHKejGbpTormvUp08fP7+1qBIkxpIa6HqG7mj4TujqpK/BirR2DIbYqlUryS/MlgG6D6DrM7pM4XV0Q0CXKNRENHrwwQfVeoNyCEj9dzVIjMbd+q/v9oau3+gSiG6qWOYarCdmXfbQ1R7tghIc6Caj726ProaYF3SDQxcud7QukejqUadOHdWN1jhK/Z9//qkGGkJ3fw26vjz99NNOn4euQLfccov6iS7D2BaOHj0q2YXtCN1O8HnoWqyHLmvG2qH6vzGfxu5I6PbRqVMn1WVq9+7d8ttvvznUBduyZYs88cQTpqV0jLCe4PtiO8I8olsiuuHpoYsRuqagPEIww4BUgC472F9gOzGWBtG3D9oNXbZ++OEHVadUb+LEiep1M9h3ovsN9p3oqoVtFNsd9svYBxvLK6HrN7oCZWcbxX4Y3XvNBp1B6RZ0p8N+EuURtJrV3kB5B3S9RQ1CdFNCdz9jlyHsk9B1DMcgX6F7lrFkA7q2BQK63xnXZUBXRewz0C0P2zeWh7F94e2335bvvvvO4Tl0rcL+FusQ9rXYT+jLY6FrH9at7JTpwDqDfSR+Yh3C/8T8YV9m3K+j9MZdd92l2shTiQrAcQLrLUpqoNu3fr+RkZGhau2iW7OxXJXZ8sH5BrrVoasnzoF87bqM/Q66wem3CXxHcs3s3BJtrz++uTtXRbkavQ0bNuTo4kaXduP+Avu66667Tq3fOF7rzxmwH8NgKTg/8HT+rZ2rohsmPg/TY3vWusSinMvrr7/u0KUS5yPXXnutw+dgOzB2u8zOuaoRjhVaiQWcN2FbMS4T7DPuvPNOyS9Qogfd/LX9jlbWQc/4t1ndaOPf2udqcG2CdcxYIg3drlFuAO2KrrvauqK9B+UCXJW9yM7x25vSFFhX9csF53Oo7Ypjvy/wP1ErWLveMpZJ0F4zg/NGbB9G6D6MYz32KQsXLnT6THxXnEvg3NMd7TiFYznOcXFeguszfakTtM0jjzyijh84xgUzrB/4DigLgvNolGzSlyLCd0V3fW0Z4zoDg8MZy5OYXa8Y12scw7FeoswWlifWZ5zD6EtEod1wHNe6ymO70dYHlD/QQwkS474Hx2UNSjSYlQhCKQKcA+D8EufWmG+cU3tDK+vgz/mIt4znEtiOcZ3oC+0aE8dOPHC8NH4uvguuEXD+rZ93LFPjeTZKEKCbPvZhOMc3Hs9xnYVl6qm8qZmcaKfciD9hMHbsb7FdYL3GZyAmgPNJlIfQrxM438b+3FNJLdD2+fhsHAvwmZg3X2Bfh+1af36A4xZKymjtjHk0wnELZSH07zH73OzS2gHrE/Y9WNbz5893Kinl7no1EDGV8+fPq/XC+H9RFgTfFesgPgv7PA3iRVgvMM6Hq9IzxrbEMRnHT6wj+CxX5ZfQ5vpSOX5fV/gafUY3VmPEG3cCvYGupsb3ottQTmUmI3vBrGv7hQsXHKYbMWKExzsyZndDtO9u7LqMu0iapk2bOkyPtHYjY0aYq7sZ3jLLGkAGobu7kI8++qgtP0FZFeN3xJ0zM8hSRHubZfmYrY+e7tqbZdzgLtQPP/zgMB0yWJOSktTvZt0dnn76aYcsV3SXMcumNG4fZvOMx8cff2yfBttA69atPa73ZhmA2GZ37txpnwbfAV0izf6nr5nJvt591OD/GO86IoMF3Vr0vvvuO6fuhHgflq2nu7dY9rt377ZPg7ulxt4HeDRp0sQW7JnJ2r5NWz+RedqmTRvT6fQljtDF3+yuqln3mt9//12VRnAF67/xs8y6dvu6jWI7d9W90th1Dt8fz6NLkqdsHqxHP//8s30aZLtcccUVXn0Hb6xdu9bpsz799NNsZyavWbPGtF2R+WbMIklLS1NZWuihpF9GxuwL7CM2btzo8F78bexiZna33JvMZGTzoeubq66OWK/Myt0Yu9S5ylhD6RaUO3C3HzTOOzL4zEpj6Pe7gL/N/qe7/RsY99foVWEVVsxMfv/9902PXd749ttvTdtIf24XyMzkP//802maa665xqmrqlmGpLFrpqt1ulq1ag7HbcA6rm1D3bt3d5ge26q2/9cYu3Uiq8/f7sa+9izBo3///k7zlB/ceeedDt8T2bgaZBebLQvsa7RrGbSjsTzawIEDHf4Hlp1ZtrD+HBOfd+ONNzpNh1ITOXX8Nts2UEKrZs2a2c5INmN2bufOQw895DQ99sX6fQHWSbMSV8ZrUlfXkcYSOjiPQlkA43QotRHsmcl46K+1fv31V5f7q5MnT9qnM+v5ZJY5icxCXP+5gv2jsaeoq6w9T/ttI3RJN74Hn212HYRzkZEjRzrNW6DOR3xx0003OWU3uuMqextlDDQ4Lph1tzcer7744gunabD/+++//+zTYB+F81HjdP6WA82JdsrJ+BPOu3FOYtaLVoN9q6dl7Wr/h32r8TwV9Ps4bxjLeeBvDc7XzdYZ9MxwV27F2EvH38xkPFBGRd8LwSxWZna9GsiYyuuvv24af9MfSxFbePjhhz0eP11thziGG9cVV215zz33OLwXxx1/+BxMNvuCCDJ4Axe+Zt0zciqYbDxBQ1cRs4MMGs5YjwgrhacVGMED48m5kVmtM9Q/dNfFDTWlsgNdqY2fOWvWLPvrOEAb615j5+3uxDDYYENyt+F7K1DBZHeBAnTxMU6Prgdm3YHMLpBvv/12j/NsdsPHbN001rg1nmTggYOfES4s8jKYbHZyhQO3GbPSDO+++67HAy7+h1mdWbPuKrgYCvZgsj4w6mq9wkmnvgsSmHXVxnIygxNOBElxbEC9cFxU4QQID7N2eumll7K9jZqd4Pbr1y/bF2AIXhqZ1QhDTTB/4GaU8bOMN0u8mU/jdmTW3e/666/P1nzhhMnMI4884jSt8RjqTTBZg2MW9j2TJ09W3wtdvbX1x6yWtzEY4erizRgIR3d7480qdL/zdJNCfzKt17x5c5/2b9CtWzeXAae8ZsVg8qhRo0z3V95YtGiR6XqhL1MWyGCy2YUyAsxmUOfTGND1Zp32dN7y22+/Ob0HwWsNAtvGgCXqJPrLn2AyLnRwnMhvzM5ftGAKAojac0g80d+4w74Pli9f7jZgh4CLsTvrlVdeaTovKG/iLkgU6OO3WVCnQYMGORJI9jWYjOVmVvbIeFMScMwwToe20t9sMTuvRXk4BIuMzI7LuCEQ7MFk3CQzHsPNxmH66KOPHKZDDV9vx+TQbsJg2xk3bpyqSY3xjbR1s0KFCk6xAbM28LTf1jO7lscNH1+ugQJ5PpKdICDKBbpjduzDemy8ZjU7puCmgJ7ZTX+zmwT4bLNa7//++69P3zWn2ikn4096KHeAeA7GMUItety0wzqNbcE4T23btvVq/+eq3rOvjKWWsF1rN39xja0fX0if1HXgwAGX2ziSXQIVTEbw2AjnT56uVwMZU7nasK3hxha2ZyOMV2D8LAR+PW2HCIa7ut4288QTTzh9hj837H0uc4GUaSNvRxXXd9HXd2XIKcaUenQ7QDcVb9+L6d11icVoxfouLmYwSuPgwYMdRt9G90GMHmlW4gIp7khn9xeWMboN6aFrI7oKadA9ol27dg6jfqIrGVLd0YXfn+4DI0eOlJyE5YjuF76sp+hqo++qgW5i6MqEMhPocoyuCZgGXRDQfSQnoRu5Kxhh3QilGbAuGKF9nnzySY/vN1t+Zl3YjYxdIcy6Cpt1S8e2gG4VZl28c4PZMkAXKTNYhuje4un93nweuv9jvUTXOT108UJXJMC24W70eXSVxMNK0K0IXbGM5YyMsB8zdkHC9zZ2TcZ6hWWlwbEe3YnQtdqX8ijo4pddZl2tsG/ILn+3MW/pjyHujseBWB5aN1N/3z9s2DD18Aa6MHo6jhqhXMbo0aNVN1Sz5ZKd9QdlM/AwHkOx/ur3b8Z2NCuD4Kq7HLZ3b/Y5esbu4b5874Io0OeqOXm+arYNGUtMuIJu2uja6K7URbVq1Tx2d0VJGnTJ15cnwLmqVrIJZWz0ZWrgvvvuk9yEdsE5HEa+x7HDn+U8a9YsyUkvv/yy6lbvC+16QA/dfdEe+m6/6AaL4yVKngG6w+Lcy1jiwviZKBmhdcvVoGSCpy6zGmOJq5w8fhv3o/6WtggELDd9mQ3t/AZdg41wzMA2qC8TgvM+lL9yd3zDdmdWpg7HCGNXfP2y8WZdRkktq0GZGj1ce2F7Ma6fuAbS086nPZ1LoUv3s88+q7YRY0kXVzAdjqf6sjCB2Iejmzn2vdnh7/mIL4znEv6cV6JUhfGa1ZvzX2+v3fDZWHf0pUO19zdu3DjP2ymn408o0Ygyar6UhvX2esldfMIXOOag5J0GpS1QghDXkvrjGOJRWA9QZhJw/Ordu7fTcQzroTdlybwRFxenylsYYR01ligyXq8G6novMzNTlb3QQ4kflJXx5zhsBvEZT/OuZ1Z2CvsDX891fY6gIWBkhHqI3jA2mKsDRKAYTwJ8gZNm1KlF4NWVRo0aefwcNAhOxvW1lxBAxokYNhRjXVis7L6eiOqhlpJW71R/QWs8OKBmiz6YDDNnzvQrmIxaR8baUoGm1fLxxTvvvKMOErjw0K+reOhrxGCdRjAJtaxz4mIRGyvq4fiynqLmqhmzC0Zv1nOz/2+2AzOefJkdjFxdtOL5vAomZ3cZGk9kzfjyefrlgAtyre6ZGdSNs1owGSfVOGH1tL6YLROzmyDG9Qr1rf2p72bct/nDrK19rQ8XyG3MW1hPjLJzARGo5ZGd4yyY1a72BLWPjTdNA7X+uNpXG9tSX5/On32lr/R1JV2tD+T+XBXndDgf8HTybnauinEncEGSEwKxDblbp7w5V0XQDnVeBw0aZH8ON0lRM7ZmzZpO56qo8Vq/fn0JJNRI1oLeqCmIQNxbb73lNJ4ELlhRZ9HX7Qi1JXP6XBXHNl/P4ZHggGsN/T4EF9+4yNdfhGPMDX0wWXvNWGcS+2/9zd9A76Nz8/iNcTH8qYua2+eVYAwma8dXd8Fkf88rvVmXrRhMNvu+3pxbenNeuWvXLnUTzp+kg+yeW+bmeaU35yO+MJ5L+HNe6e/5b25cu+VGO+Vk/AkxIyQZ+nrt4O067c35gTdwYxMBcP184tiEYLJ+vAccxzD+ixZMxjQIJhuPY5guUIl+WKfMbp76e43mz/p+6tQpv6//vL1W8rUtjdcV/l5buB+JxoTZHSCc3HjzJc0Kenvzxc0WfiAy1DzBAGfueLvAcYKuX4mxgc+ZM0cd+IwDhmU30wMBYSPslHDyp3/oN2wNLs5dZeQEI2Rx4C4Q7iBhACNXcIKGwUEQYM8JVrjoN7sp4m9Re1xUm/E0sBUFD2MgGcwOxGbTeYJ93oQJE/yar+ycMAfTNuZtcMxdxnuw8HScNfr+++/9CiR7u/64uoGc1wMfGds6O1lUBYHZuSrOJb0ZSM/sXBWBU0/HuGA/V0XQTn9DHdsLArn4fP0N+NzISsaxBYNoISvVmKWNi21j76JghmMregvq4RwdAXUMuqO/uMZDPw2ynYyD5phlOgdq/crt4ze+2+OPP+7X/6OCdW75/PPP+72/teq5ZW6cjxjPLf05r8zp89+CRtvnIrsX+z9/gpDertOBilHgc4znXThGIZlKu0GP4DASA/XHMQSREafatm1bjh3HAr0d5cX6nujFtZKvbWnc1nH+509SZbg/dx6MIzZqd070IzIaHTt2TH7//Xen540ZeWYn62ZdE42j2pvBqJn6jEDcNTAbldcVT109vO0ihpIK6Mq0aNEi+3M4QTce9LBy+pMZrL8Thq4QRug6YNYNzlWJDLP0/WCFixEE2LFDxqis2FmhGxR+//nnnx3u3GH0U+z4MDp6IHlaT8yCAtoIz0bGrAdX7w8UrJPGngf4G5lK3s5zbsjuMsS+whN8ntn3Nvs8s8AfXQ4GGk9ysI9EplPz5s3tXXQQvOjatWvAF5vxuADbt29XJW+srEKFCk7PZTfjTFsexpM4LA9ky3n7fqNbb71Vqlev7tX7MYK9L9Dd3gjBpjfffFN1B9WyRz/++GOfjvfZZXZyabZvcPe8O8a2NlsfyDH4i3XTmImEc1V36xwyl81uVuT0uapxG3z00UdNu7+b8dR91ttzVVyI4PwPAVzN9OnT1bLU9/DCBSEyiXID5h3nZMaAqbGUUrDDhfOCBQsc9sE4BmqjvmMdueKKK1R2HtYLLXsN6zN+6hlLmpjto5Eta9bt14w+0zqnj984buD76LveT548WZ3jjRgxQqx6XunvuaUv56k8r3QN13g//fST0/NDhw6Vhx9+WGUmatnNZvuT7DJrZ2zDwcB4LoHYBLZxb48b2d3GjMdIbBPYpwTq2i0v2ilQ8SeUiTAmayI5Dr0SEC/C8kNMLi0tza8bMBDIdsZx7J9//rH/je1MXwYSPZoQrNQHkxGPQfzFeFzxVJor2JQsWdIpcxvrSSDPpXxty0BdV4T702UfGZxz5851eP6NN96Qe++912U3QOzQsbM3+yw9swxSsy6HyOz1BMEB/caMk2GUM6hXr57H9yJYHsi7DLgw0AeTUfvEuGNENwazrjzeQpsYl7GvEHj1NZiMWkNWvaurv/jBTgwPzeLFi53qd6FWkT6YbLYOGG+kZBdOwI1QggQHB+P6gAC4N+8PZHkRYzAZ67ExqIo7i4EIbPm73LEMcNGrh4OT2cHb32WIzzPeMMMFj1kNLv3dWdTdI/cXT6h/a6wva7ZcA7Gu4K64MZj8ySefWD6YjIshBI30x0Nf6qe5Wx7Lly93Wh6ogecNs+WGCwEEd3PiOGu2/iBTrlmzZn6tP4Fi1ssKN3exbhstXLjQ54wElBvwp6ZuQYZMW+N6iMDUE088IVWqVDF9D85ljTf6sY4aM3EDfa5q3AYR6PMmuSAnzlX1wWScpyLjTw/7arMa+jnFrGuysX6zt+tDXpVM8MSYhYVzan3NXO28FD3DUEcSQQYw288aPwtdchF00J+jIXkEJUS86Uasv/gN9PHbCNvl559/rj4P58AajD2BC293SUvZOV8wex7XNsaxQJAYhd4Nxgy8zZs3O13TIQhfo0YNt/OD6xCsy8YbR2bHCP3/tPK6nBfQRvobXoAb4sba6kggQlt5wxj48XReaXb8x/l/oOvxBhpKyehLCeGmKGqq50T5B7NrL2MwGddaxmAythF9DEX/fl/kVjsFKv5ktr/Fdm+sIZ/b57uu4NijL6+Da2Qkdmi0IDJiCDiHQKAc3xfHIr3Y2FjLX5P5KiwsTMWg9L2NsHwwDoQ3GcWBjj2ZXUP6e13hV7901JY1HnhxEEVRbePJNHYAOBGdNm2a0+c899xzTgvQLJsJQVL9yT12KN50cevevbvTcwiUusoUQbo3irujvouxyHt2IVPLWO/HeOLRp0+fbP2PQAwsgmyCvKp7G2jY2SJrw1X9J7PSLFoWiLsLRtQqCySc8BhrquHiCTdg9EF61A8cNWqU0/tvu+02ySlmAyHgwkG/nePkDPMaSL4ud+x7jPskZJfpM320rBrjvgPv0w9Q6cprr72m2kC/Y0fJGOM6gxP+nKwFH+zMbpjhzrQe7mYjqJMT64pZORsET3Gn39iVDG07adIkh4N/XkLPIE/d8X1ltjxwHMBgH8aADdb52bNnO9yQwfHSeBzHyaRxcFn9Z6CXEso/mQ3mGYj1B2MHTJ06VXITBgwyBgQwOIyx3iv+9vXEH+ufcd0MdFf2/OiZZ55x2j8giIYsY2MmMI61yK40G0wY414gK9TTuSougvXHCJQE+PDDD/06V8WNUK2moNnNBRzHcF6JEl2BhIsdBBdy8lzVFwjcGes1g9UDNP7cjDIOnKPP6NJncul/Nw7og0H7jAMBIVvJeJ6I8oQ4R3bVdRaZ3+PHj1cBJf01U6CP32awb8M1ozHLCoNPGxOZfOXL+YLZcgMEtPXLDefAZqU4unXr5rE0DtrhxRdfdLrA/+CDD5ymtdrYGlZitl4iEUYfO9ACet7WBDauKyhN6eomFoKE6Amrh5shOMcxGzMF65zZ9VxeMJbYCdS5pTfMzgFxjYkexMa4E27kGI/BvtaIza12ClT8yWy9xnmBPjaA9Rw3ga2yLhn32/4cxwJZL9lKuhvWC6x7PXv2dBl3w/Hho48+UoF1/XIMBKzvxiRAf68r/GopbLy424eLTT2UUsDdBlxU4UQPEXdkWJoF7RD9NgtAYQUyltFARg4WJLpMIYiFC1lvIvS9evVSOyX9RQNOTHGnGAsM84gNFQcbTIOH9rnaCNaBgu+EgxhGejaDCxXjCbwvcFfNWAcZJ6YISrrbIJFer8+cQWYzTtassmPKDgREkOGAC3x008SdTiwT7ISxTi1btszpPTgR1zMbOAM3R7CuI2CInSZO3I2ZO7565ZVXnC7QMIAgbpxgm0C3vx9++MGppjXm75577pGcgnnCOqvvRok7pVieCMCiuwouoANd4gL7ESxb/QETgWFkq6CNtHUa2zfuYGJbxt3aKVOm2KfHexEo00apR5tjnTBm0aNHhVn5CiPsxxAoRsYYgmcIiBkDEvDUU09l89uLGqATJ676QS6N8DqC2Xq4oHE32IsVmI3OiwAOtinMO9oJxw1va4T5uo1i8DYcv/S1U7FOaIMKIWCLdQrHGhy8cZKHgaGsACea+v01lhWOX+4GivUEXf6xLWP/ooesOPRUQVczrO84lqAHAk5usF/VYFnhXOCFF16wP4cLLixnBCKwzSC7CxdwCFBguWvrsz8nLlh/cFNIr1+/firghC5aONE2DuSRG5D5h3OOL774wuH5hx56SC1HHH+Q7WMc+NYbWhaivsskM5M9Q+YLgvc4WdfDfhslUXDhg2MDMrGwPzfrRYJjA4JqZjeBkXmorzmH4yTaBQEo7ZhtzJYzg2M8jm36usQI4GIe8XnYxyGwgc9El1ys41rWZqBGO9fD+Z+rkcPRozCnbmDjuIcbQYDvh8A8zoGM5/vYp3tzAziYIOiIY49ZGR+zi3CzddLdPhX7ZwQp9Osj9lW46Y73YN+J5YzzHASHXfWoCvTx2xXsS3Feqb/OxGfinBTn8Rhc2x9m5wu4tkSgVhvZHu2g3WQdNmyYWm76pAEcX7DfwHuwLiKL2JhIhWtL/THRHWSXIzsSx2IcZzHAojFJAdetCE5nB84ZsY3pmd3Y1G+H2jKzet1q7IuRga8PCGLfjP002hfrNnp/+HKtgu+tD3IhmIlMWFyra4NtYZBMLWMcCSfGHl0438G6grgI2hDHGjyHz8V2Z7yRkBdwboLtX98LFesFro1yGrZzBGtxbqRBgAu9YnGthXNH3Jwy64WHa1N/SjTkRjsFKv5ktr/F9QjicOitgiAk9h2BGIw7UNshzluM44G5Oo65SgzNr8kSTzzxhEpc0gdx0X7oeYprLW2gQLyO8zzcVMmpnv/G6wrtGtMvtmx4/vnn8Q19fjRv3tx2+vRpl5/bu3dvj59RqlQpp+f69u3r9Fnbtm2zFStWzOd5/Pzzzx0+Z+nSpU7TvPLKKz4tr6NHj9oiIiJM/9/IkSNt2TF69Ginz7zvvvs8vu/LL790el+rVq1s+UHFihV9avOqVavakpOTHT7j3LlztpiYGLfvq1+/vsN78DnGz/XGAw884NP8Fi5c2LZhwwanz8F6aZwW66+R2Tpttg1NnTrVq3mpXLmy0/N79+61+at169Ye/++JEyfs058/f97WqFEjn5Yh2i4hIcHpf7dv395p2nr16nn8PLwvKyvL7+/s7v978zBrZ2+YbQtG2Cd6sw80m3f9epCYmGgrW7asx+9y/fXXe7V++rON7tq1y1a6dGm/lyvmw5t13dtl5q2TJ086HUM+/fRTl9ObzafZ/8fnXnHFFX4fHzMyMmw333yzz+sr1hUjT/vPAwcO2AoVKuTX+mP87mgzb9Yxs/nCw+wYX7JkSY/zZravdLdetGzZ0mHanj172qxEfywxa9O8NmHCBFt4eLjP62etWrVs+/btc/m5w4YN8+tc1WwZHTt2zHQd8/TIzjrtSkpKiul84/Hggw/aAsGf76o9+vfvb8uP3n77bdPvGx0dbUtNTXVYV1wtm5kzZ7r8/FmzZtlCQkJ8Xt45efz2tG089thjpueba9eu9WsZ//bbbx7nHf/T13Ng42PKlClenXN7c14ZGhqq5ju7zP6/Nw9/9+lm1yLGcwdvj63eXK+MGzfO43fBdaHZMjc7fxs+fLjHz5s3b57De5555hm/l2ugz0d8MWjQIIfPqlKlistrGbPzWrN2Nfs+ZuvSpk2bbEWKFPFpnfQmvuFObrRToOJP3pxbm+1vza7jzK7PAu2JJ54wncfq1as7TLd69WqX32flypWmn+1tLM6b9c6b69WciKksX77cFhkZ6fN6Yfyf3m6HrvTq1cvhvS1atLD5y68yFxp0Y8JdBWM2pysoDo67zLiDbezOZewi6+4z0Q3P29HckRmFO2y+1NVB1wljl8ZAQJZcjx49nJ7HXYicKHFh9r/MyhgYi7bjLmB+qPPqyx1L1GlDRgjq0ekhUyG3Bv1AnULUzvFm0B101UE7+drFxx+4S4o7rK6WJ5YRsgKRFWDk74AA2v7Fl/cjcwuZB96s94Auwsg89TTQpgYZ2MbuUcZaXMiezo1BK4IZMlnnzZtnzwIyg14c3pZO8WcbRbYhuvGZde+zMmQgG9dvb2qyevO5yEQ065rnbc8bnAugrIC39VvRq6FNmzY+/y/cuUcXaFfjC2D7Q+ZJbg0QZjzGo+eUu0xxzBd6ongLmVbGLFEMKETeQ1YdapN6m8WLLFGUAMA+wuy4pnnppZfcZojjXNJYC9ldFjUyRXzpyo71LSfOAXDcRUa9GWPt6NyEdkHWtDG7Mr9wlY2Fnpn6/R3WFVe9qdxldGHfg0HKjGUw3EEWpv5YHejjtyfoLWTsCo8SE6gnrs9m9BZ6qvma1Y5zYJzbedMDCNe2KPFkrGnqCnrwjB071mU5DFwPoMedcXwXcjZo0CDTsl36/SViB94OWo7Pc1Vb35W3335bjeGA7SSYoHeX/toF5RiMvZ1zCrLHcY6DjFZP0CsVJS/0veP8kRvtFKj4E7Z/d/Wr0aMhENcBgeLqGKTPSgacjxljLtq1QU6OBZXX2rVrp+Kg3sZOtW0kkINuozSTsYdndq4rshVMBlx8Ij0fO2ic5CHQix222YUeDuA4aHoaZA6DLOCkGicj6H6BE1sM7IeUftSvxMbpy0LFhon3oAshdphoFBzwcdGrdZHHCTwCeQjQoZuWcaUPFLPyEbigzk79N3SdMNbWw/fyZiRlBNKMJym4qROI+st5Dd1iUN8T3QrQbQ01q7Hjwkkblg9OxtFdE8XhcVLq6qIMtb1xIonlqY2cmhMwXwgyoGsnuu9gnrEt4GQSgVLMLy4GUIYE3RCxHucWdNdDN2Bs75gnbJM46KHuKdY/dNE1DtSX3dGnsV3gBANlPHBC583glOiOj8A2tncEEND9DN1ucAKCbR4nK5hnnCRhn+XuppZZEAu1S1GjEl2w0CZ44KCHCx50ffTl8woyrNsY8VcbZRvrONYrrEfY1lAjypegvD/bKNYpBHrwwH4Z6wouFrGu4Ce2L9zkQ4kCK53YIGBr7PJmVkrKV9hOEBBGu+AiCt8ZyxJtg+0KJ7O48MVNL9yIMcL2iRN0HD+xH8MJJc4FsK/Aa2hfBCYGDBigghHoMupvzUDMB7ZxdCXE/8A84pwA+yd0sc7LrqNYbqizh2WIfTa+P/aDCGLghB/HVl+OIcaLBKyX/nbxLshwAo/9N8omPPvss+rmH9YZrbuyHl5DrWNXA0prcD6BIDXK5uD8AX/jmICLJDyHgZ68GXBHHyTEzQgc93Degs/BuoN9Ei6wcA6D9Qgle/B/UQrD25unvsJ2agxwIbCemzfg0DbYvrEvwTkIzntwDZAfaykCzivM1jn9oNAas2sUdMvHIK3u4FoH5YZQUx7lX9DtGtcB2Cfhf+OGOQJy6IqLLra4JsKxISeP3+5gHcQ+01gGEN278f+M9bw9wXzhOIdjFYIw+M7ezKtWUxU3MnAtq11PYB3FMkdiDhKhMI27gKYZJFkhsQHtgX0Sjpf4ifMPdLPnQHvewTqMcwsE33CTD/ti7DcRR8A5Iq4JjQP1uoPzQIx7gHMubBf4LG/g2gPlNFCqEKUasK7gvVhXsA/F/gwJENjGrALfz3gjMzcDlDhOon2wbaK8BvZl2C9h34LzUOwDEURGqRacN3qqRW6VdgpE/AnnBfgMxAWwnDB/OCfHTUbMO/YdVrr2NKubbHYcQ9uaXVthOm+S6oJZ80vXCYhXYD+PfRSORVgvsN5jvcF1FpLqEGPBuY8vwWdPEEjWl03FeWZ2ytqEID1ZcgjqTBkzbrETwA6hoMJFv3FwLgz+xEwjCmbY0RmD8Tg5CfRghbkFtYuMGWU5uKsk8gkuXJEpr8FJZm71oKDcgzEMcFGlrwOJm4kIplsJxh/Qaq3hAsxsPAIrw/wiQI/lrcGFn34U8oIGNYsRIDx37pz9OZy7W2XQKKJgo99PanDjFYEkoryGm5j6njbofYAMZU83VYkouCBZTz+2DOr2Dx482O/Py/6tHTcQ5UYGiLFboKtR3gsCDGqkh7tgVrswJNJDIAO9BHDH1Ayexyi1RtkdLISIzKHsjD4zA9lS+tHlKX9Azxp9IBmZqsaBYygwNw+R4WO8yY/sxYIKgXR9IBmZRoEemJqIiKwB2f/6kjIYKBk9QYgo//jjjz8cAsnoWfPYY49l6zNzNJgMKGuBrkj67D7UXEMtuoIAXTKefvpp1Y0aqf/GixMsC971IyvDqLbYjpEhh2xjdC1H9yDUxcM6je4Z6M6rh+5lxhtJRBQYCCrqazJi1HJ0Lab8A+dKyBbQQzkd1mTPGTimGXuIofu5sa5cfoV9CM5VsRzQ7fapp55yeB0lhHJiLBEiIrIGxCj0Y9WgxAOuAYkofxg7dqzD37jO8LaEjys5XngM9T9Qdwc16BBYBeyYUAsENWBQoyY/Qz0vV/V2UBelIJf8oOCzfft29fC0zaPGHbrIElHOQOYkHpQ/IWiM8kGUe5CFhdJM2sBDWVlZqtcNBktBTfX8DFlors5VcUxHbwgiIsq/UEM9JSUlr2eDiHLIDz/8EPDPzJVRLFAYHAM5kOPJOUYjze/BdCp4JyJYr3NzkB4iIqLswsBX+u5/dDlbDb0hiIiIiIg0+XNIZAtnGmF0WBS+HjZsmMrWJrI6lLH4/fff1d2sNWvWyOHDhyU+Pl71MMAADRh1GqPKorcB6iTjRgkREREFJ4wWj3PUZ555RpW4ICIiIiLSC7GhMB8RERERkY+WLVsmHTt2VL+3b99e/U1ERERERPkXg8kFQEZGhhw7dkzKlSsn4eFMRiciIiIiIiIiIiLfhfrxHgoyCCRXrlxZ/QxWmZmZcvLkSfWT2BbEbcJKuH+yDraFNbAdrINtYR1sC2tgO1gH24KIKHgxmExEREREREREREREHjGYTEREREREREREREQeMZhMRERERERERERERB4xmExEREREREREREREHjGYTEREREREREREREQeMZhMRERERERERERERB4xmExEREREREREREREHjGYTEREREREREREREQeMZhMRERERERERERERB4xmExEREREREREREREHjGYTEREREREREREREQeMZhMRERERERERERERB4xmExEREREREREREREHjGYTEREREREREREREQeMZhMRERERERERERERB6Fe56EiIiIiIioYLLZbHI6MU0SUzOlcFSYlCgcKSEhIXk9W0RERER5gsFkIiIiIiIig7PJ6TJ71QH5ZMle2R2fZH++ZplY6depuvRuVUWKxURwuREREVGBwjIXREREREREOos2x0u9wQtl6JzNsufE5UAy4G88X3/wQjUdkZUhi/7+++/P69mQqVOnqnlZtmxZXs8KERFlE4PJRERERERElyBA3PP91XIhLVP9bbM5Lhrt7+S0TDVdXgWUk5OTZdy4cXLNNddIsWLFJDo6WmrUqCH9+vWT7du3S372/vvvq8Bk4cKF5fz582I1mLdXX31V/Y5AbrVq1Rxex3OY5tChQ1KQzZ49W8LDw2Xfvn1Oy42IiKyLwWQiIiIiIqJLpS3um7xW1UnOMgSRjfA6pus7ea16X27as2ePNG3aVIYMGSKlSpWSESNGyMSJE+W2226Tr776Sho1aiTTp0+X/OqTTz6RWrVqSVJSkgpIEhERUe5hMJmIiIiIiAiZkqsOSFJqpsdAsgbTYWC+OasO5NryS0lJkW7dusmuXbvkyy+/lB9//FGeeuopefjhh+Wdd96RrVu3qkDrgw8+KL///rvkpZzIGl69erVs3rxZZbA2a9ZMBZaJiIgo93AAPiIiIiIiyncGTPlbth/1PpiJLONth/0Lfr781VaZ/edB1U3fxYdLRmamhIeFoS+//ek65YvIhw9d5dP/+uyzz1TAGAHknj17Or1evnx5mTVrlj1zGcFXQCmB6tWryyuvvOJUSgD1bB944AFZunSpdOjQwSEY/MYbb6hsZ7w/NjZW2rVrJyNHjlTZzxrUwe3YsaN8/vnnkpqaKpMmTZIdO3ZIr169pEGDBjJ48GAV9L7pppuc5hefc+LECTlw4IBERHge0BDB46JFi0qPHj3k3Llz8thjj8mGDRukcePGEuwSExPlrbfeUst79+7dqnQJbgz07dtXHn/8cft0aWlp8u6778qMGTNk586darm1aNFCXn75ZdU+nuAmBDK6//nnHzl+/LjExMSo97/44ovSpk0bh2mxPqDtsR6hHX/66SdVYqV58+Yyfvx4ufrqqx2mT09Pl1GjRsm0adPk2LFjap178sknpVChQgFcUkRElJcYTCYiIiIionwHgeR/9p3Nlf+VmpEl/+4/lyv/a968eernwIEDXU7TpEkTufbaa+XPP/+UgwcPSuXKlX3+PwkJCSqwiAxoBDMRrD1z5owK5uKz//jjD7nqKsdA+HvvvaeCk6jbXKlSJSlSpIh06dJFBSk//fRTp2DymjVrZNOmTTJs2DCvAskIbiMQes8996jgZO/eveXZZ59V84QyH8EMgfG2bduq5YHMc2SWY5ng7/nz59uDyRkZGWo5Ll++XH3/AQMGqOAuAsudOnWSb775Rm655Ra3/wvLqnjx4iqbHTcfsI5MmTJF3RDA57Zq1cphepQTwbwhE/y1115TbYws+BtvvFGVXEE7a/r06aPaCPPyzDPPyKlTp9QNjCpVquTQkiMiotzGYDIREREREVGQQHARwbsrr7zS7XQI/CGYjKxdf4LJCABiID8EjTHInwZB7IYNG6ogLjKZ9ZDBum3bNilXrpzD83fccYfMnTtXZarqX0MQGNncCD57AxnXCGwiixoQEL399ttl5syZajBCq2S/Istdn/XtjeHDh6u2ffvtt1UQVi8rK8v+O7K+Fy9erALM3bt3tz+PTHW0E7KAPQWTf/nlF5Vlrod2RRb56NGjVRa53smTJ9U8IeivqVu3rgpmI8P5kUceUc8tWbJEBZIxX19//bU9Ux8DDtavX99pPvD+Rx99VMKQsW9YbkREZF2smUxERERERBREGaxxcXEep9Om8aduMYJ6yHRFBnLNmjVVMFF7IDP2+uuvV0HmCxcuOLwPGczGQDIgexbv0wdWMV9z5sxRmcsoheANBJ/r1KkjLVu2tD+HDN6zZ8+q0hDBCsFiBMpr1KghTz/9tNProaGXL9u/+OILqVatmsoU1rcL1otbb71V9u7dq0qMuKMPJKMdkD0cHh6ugtHIFjf7/4MGDXJ47rrrrlM/9f8LAWRA0Flf8gXti2xyIiLKH5iZTERERERE+Q7qEftCq5mMkhW+igoPlboVi/hVM9lXCBKjBIUnCC5C2bJlff4fWoASA/iVLl3a7XT6rOfatWubTofAJzJTUeri+eefV8sJGa3IMtayWj1Bfd+//vpLBVtRekNTtWpVKVWqlAo0o8SCO5mZmapEA+o6a9mwvipcuLB6BBKWI0qIoD6xPnBsBpnfKGvhrl3wHV21BWzcuFHVV0YmsfFmg9k6XKFCBVW/Wa9kyZLqJwLRGtR5hnr16jl9hllmMhERBacCGUxGl6A333xTdSOKjIxUJzdjxoxR3Xq89ffff8vrr7+u7sjjwI8TGHQle//999WdYv1AFO7gjr/+Lq3LE9BLXdp8mUciIiIiooLK14HtYPKi3TJ0zmaf3zfyznoyoEtNt0FMXDOgLIO/QUwNSkzgOgMZoe4Chgi8AgZw83Sdgaxhs7IKGMztpZdecvk+Y0ATA7m50r9/f1WCAaUxUE8XwV8Eum+77TbxBqYHDDyHhxGuy/777z+35T9QGzi711NmAxjmJrQNvqO7GtHuvuOhQ4dULWwExJFBjExvZCojiI3rWwSYjdytsyxNQURU8BS4YDIGFsBAAzjAjh07VlJSUmTChAlqkIGVK1eqkzNPcBcdd70xsAW6++AkKj4+XtatWyenT5+2B5NRRwrdkMwOuKhJhZO2rl27Or2O4LbZHXp/ap0REREREZF3ereqIqMWbJPktEzJ8qJ8a2iISExkmPRqlXuDi911110qmPzhhx/K+PHjXWaeol4yriu0gc9KlCihfuJ6xQiDqOnh+qZYsWIqAI4yFIFw3333ydChQ1VQGPOyfv169bc3A+8hExdlIFB2w1hPGJBdi3IXyHxG7WRXUIID5TAQSPU3qI9SFIGGxCTcaEB9awSL3WUn4wYCguLIYkZpCn8Sq7C8MFAfgvrGus3ZgZIosHXrVmnevLnDa1u2bMnWZxMRkXUUqGAyToZw8oGRhRE4Llq0qHq+Z8+eqisOBi0wuxOrhwwAnKjcfffdquaXuwM97rTfe++9Ts9jwAR06erVq5c6cTA7QTF7HxERERER5ZxiMREybWAL6fn+agkVm9uAMgLJyPad/mgL9b7cgmsRDMKGHpGtW7dWg9sZSxzgWgWBxhEjRtifx6B95cuXV9c7SG7RMpVRpuCzzz5z+Axc4+B6BNmv06ZNU7WQjfB/fCmhgfIc//vf/1RQWPv/3g68h8H7ULbjoYcekjvvvNN0GiwTzCt6nLoKUKNUQ/v27QOSIR5IWN5oM3wHJDrhulRPH2BGUP65555TA+UhS9rXdnE12N3PP/8sa9euzdb36NGjh3zwwQcqw1k/AB/qOGOQRCIiyh8KVDD522+/VfXFEFDWAsmAu/U4KcHJB+7yussAxp1uZBQjCwAHdNwlx4kaymV4S+ui5e7kKT09XWVN46SPiIiIiIhyR5cGZWTuky2l7+S1kpiaqYLG+qCy9jcykhFI7ly/TK42DQKi33//vdxwww3qGuaWW25RA+KhxASyP5HwgoHxPv74Y6eSeygzgdIG6B3ZvXt3OXHihLo2wQBpCELqIVi5atUquf/++1UWK7KcUQ7hwIEDKjmmUKFCqmSFLzAQ3+effy5ffvmlynj2NssX84hrrttvv91txjYynXHN5yrgbGWjRo1SGeeoCY3liqA3rjHRpijfgWUOCDTjd5TaQE1rtD0yvXEdi/ZClrkx01zvxhtvVO2InraPPfaYSm5CCUcEe9FLF6UV/dW5c2e17JH9jfZFCRNkwk+ePFklb2mlV4iIKLgVqGCyNjItSloY4TkEk1Gqwl0w+ccff1R1pVavXi1DhgxRAyAgqNyiRQt1BxbdjdzBnX+cjKF2mat6yjj4opYyaqvhDj5OEHFyoZXP8AQBc/2gHEePHlU/8Xl4BCPMN+7IB+v85ydsC2tgO1gH28I62BbWwHawDraF/zrWLSmb3ugic1Yfkk+W7pM98Un216qVjpV+HatJr5aVJC4mwqvz00C3BQadQ5kIZIEiA/TFF1+0n/+jhAOCdqira/x/SKpBhu/06dNl+fLlcsUVV9izl3F9o79eQMAR07z33nsyb948WbhwobruQXYzrn2QuaxNq/309B0xxsxVV12lgpcoPejN8kDJBARJEZxE6Q1X70FWLILJCKIjUG7lbULLCtYvbyQRoe4zxvZBKQpkCuMGAdoImeHadMj2xfUkAuxoR1wnItkJJTyaNm2qMrON3w//T3sO685PP/2k1hn8L7x29dVXq+dQEhLBZP379fPq6rvoX8M8oRQHyj0OHjxY3ahA3W18F2SWa9/ZKm1BRIFnpZ4flDNCbAWoYn63bt3khx9+UCckqGesh4PnzTffrE6WcMfeDE68cAKDO784WcMgErj7unPnTnXnHlnKixYtUneRXXnnnXfUSdwbb7yhRjI2Qm0pdFXDARijDK9YsUKdKODkAqU5EMj2BHep9V3aNKjBhZF4gxFOMhITE7NV34zYFvkJtwnrYFtYB9vCGtgO1sG2CAxcLp1NzpCk1EyJjQqTYjHhbgezy8u2wHUASiVo5QaseN6MjGpkN+PaxJt6yYHEbcI62BZE+ZdZOVfKXwpUZjKCvRAVFWXaXUw/jRkMVADoqoPuYbjrq7/LjrvleB53zl3BoBA4aUJ3MTPIjNbr3bu3yky+6aabVJenX375xeP3RLAad/r1mcnIHkCWM+qDBeMJ/ImEFDmXkSoxEbFSsmi0zyfwFNgTPyx/3Fix4gVKQcF2sA62hXWwLayB7WAdbIvAuTR2neXbAokxuNZBST6UoUBJCXdjvOQ21ORF1jSSasqUyd3yIMBtwjrYFkREwatABZPRtQaQ8WuE+sT6aczghEzzwAMPOLyGDGXUXkYpDQSkzT4HQWZkRSNTwJfBKlDX6pprrlG1sTCfWuDbFdSD1teE1uDENZiCf2eT02X2qgPyyZK9slvXtbBmmVjp16m6Gm07Nwc7octwURJs61N+xHawDraFdbAtrIHtYB1si4LXFm+//bZ6WAkG/cMgbBh/Bj0ukSSTV+eR3Casg21BRBScrHObOhdUqlRJ/Tx06JDTa9pz2jRmUN4CtcMAtcKM8BzqPp09e9btwHuPPPKIz/OOWlOohYWs6IJg0eZ4qTd4oQyds1n2nLgcSAb8jefrD16opqPczRI/lZgmh06nqJ8FqEoOEREREflp5MiRqkQggodz585V9X2JiIgoOBWoYDJKPcCff/7p9Jr2HGoWu4KuadpnYLRcIzyHUYYRdDZCjWWcOGHAg+uuu87ned+xY4cqj1GyZEnJ7xAg7vn+armQdnEwBmO8Uvs7OS1TTceAcu5kiU9etFuaDV8sVzz7q7R/8y/1E3/jebxORERERGRm2bJlKjEGvTRRM5mIiIiCV4EKJt9+++2qWxUyhLXRjgEDQGCE4g4dOkjlypXVcyhVsX37dlVvWA8j6QIGttDDiLpHjhxRdZPNylDMmjVLfSZGsHVVt+zUqVOmz8+ePVuNeIwTL7N6z/kJgpL3TV6rMl6zPCS94nVM13fyWgYzcxCzxImIiIiIiIiIqMDVTMbgc6jTNWDAAGndurXqaoX6yRMmTFBZx++++67D4BAdO3ZUweOpU6fan+/Tp4/MnDlTBZPj4+PVNLt375aJEyeqAe5c1SdDABt1wR588EGX8zdq1ChZuXKldOrUSdVfTktLU39//fXXqoSGfv7yK9RIxkjZ3kJAOTE1U8Z9/5/0alVZIsNCJTIiVP2MigiViEs/8XdoKAft8zdLXCtn4SlLfO6TLaVLg9wfTIWIiIiIiIiIiHJegQomAwLIKBWBoPKQIUMkMjJS2rZtK6NHj5ZGjRp5fD+yir/77jsZO3aszJgxQ2UkY7A7ZD2jFljt2rWd3oOsYjy6desmFStWdPnZCEwjGxrB6pMnT6oAXrVq1WTQoEF5NuJxbsL3xWB7ISHOQUtPJv62Wz3cCQ8LuRhsDr/4iMLPS4Fm/XMRhp+RJg/T190Est39H/wdHhYa9FnioXIxS3zLuK4cGJGIiIiIiIiIKB8KsXEErXwPgwuifAdqOrsbYDCvnTqfKjUG/SIFEZKmvQ5mG4LVDkFrj0HvMIkMD7H/jAoPk4hLP43PT/9jv7zy9Vafv8vYXg1kQJeaObKc6LLMzEw5c+aM6nGRV6OhE9vCarhdWAPbwTrYFtbBtrAGtoN1sC2IiIJXgctMJutCuYqCCpm9KelZ6hHsQfGPl+yV/p1rqNIxRERERERERESUf1ivbz0VWIWjmF2ZH4Liu+OT5ExSel7PChERERERERERBRgzk8kyShSOlJplYmXPiSSfaiYj/7VU0SgZduuVkp6RJWmZNklNz5R0/c+MTEnLsEma4SeeT89w/br6eelzPNUNpsvOp2So9iQiIiIiIiIiovyDwWSyDJRF6Nepugyds9mn9yHG+9xNV8hDHapLTsrMuhhURnA57dLPi8HqLEnNyFKBbPxM0z30z2s/XU3r7j321zOzJC394s/U9Cz7/87ItFaku0g0dy1EREQUXI599IKk7d/u13sjq9aRcv3HBHyeiIiIiKyGER+ylN6tqsioBdskOc27TGDU6I2JDJNerark+LyFhYZITFS4xKi/IsRKsrJsFwPNLoLSnoLWzq9lyqdL98nZZN/LVZQqEilFC3HXQkRERMEFgeQL29fn9WwQ5YmpU6fKAw88IEuXLpUOHTrk6P+qVq2aVK1aVb7++usc/T/B6v7775dp06aJzZfuuh4sW7ZMOnbsKJ9//rn6fLIWbHP79u1TD6JgwIgPWUqxmAiZNrCF9Hx/tYSKzW1AGYFkZDNPf7SFel9BFhoaItGhYRIdEbi606WKRPmcJQ4nz6dJ17Er5N0+jaVh5biAzQ8RERERXZacnCyTJk2Sr776Sv777z9JSUmRChUqSOfOneXZZ5+VOnXq5Ltgy/Llyx2ei4uLk8qVK8tdd90lTz75pBQrVkzyGoJB1atXtwfttMAsgnn6wO3Zs2fl6aefzsM5JSJP3n33XbVfyW8BeP1+SbuRtHfvXnWjh8gbHICPLKdLgzIy98mWKuNYCxrraX/j9XlPtZTO9cvkwVwWjCxxDIpoXP7eWL/njLR/bbm8NG+LJKVm5MTsERERERVYe/bskaZNm8qQIUOkVKlSMmLECJk4caLcdtttKrjcqFEjmT59uuQ3oaGh8sUXX9gf+N6lS5eWV155Rbp06SJZWVkSDBC8QZDKSvr06SMXLlyQdu3a5fWsEFkGtlNsrznt119/VTcFiYIFM5PJsgHlLeO6ypxVB+TjJXtld3yS/bXqpWPlkU7VVWmLgp6RbJUscVc1pt9fuEu++euIjL+nkVzXsGxOzSoRERFRgYEM5G7dusmuXbvkyy+/lJ49ezq8jgAzspMffPBBlWWWl8HB8+fPS5EiRQL2eeiVeO+99zo8h4zk5s2by19//SUbN26UJk2aSH6TlpamAuXR0dE59j/CwsLUIxggSxQZlfmtJECgt5dAw/aHGzevvvpqXs9KvpCZmSmpqakSExMjkZEcvJ6CCzOTydLBzAFdaspfozvLrrevl+VDmqmf+BvPM5BsrSzxBpWLmn7GgZPJcud7q+WBj9bLsbMpOT7PRERERPnZZ599Jlu3bpUnnnjCKZAM5cuXl1mzZqlABQLLGgTeEAwyCwQh8w6v6UsxaMGt4cOHy5VXXilRUVFSokQJuf3221XQVg/vw/vxOR999JHKjEbgE/P41ltvqdd++ukn0++DaTHP6em+j9UB+Oxy5cqp34MhIIP5RbmO/fv3q9+1h7bsESjF36dOnZJHHnlELZtChQrJ6tWr1esoj/HMM8+oUhpok7Jly0rv3r1l586dDv9H397IVr/qqqvU5+DzUF4jMTHRq3UA7fLOO+9Is2bNJDY2VgU70WYIKmoQ6B4zZozqOo/PRztUrFhR+vbtKwcOHBCrOXTokFq2WIZYT5Hdj++H72C2/H7++Wdp2bKlWn7IhO/fv78kJV1OdtIcPnxYHn74YfXdsQwqVaqk/s/Ro0e93l5cQTtg3UB2/htvvGF//vjx4+p9uHGE/4n1ATdcrBRox/pRpUoVqV27tunrK1asUMvjpZdecngeNbXbt28vRYsWVcsevTE+/fRT089ATwUsR2wTWO7Y923bts10n4da1J988om0aNFCrdN4tGrVSr755hun9sd2iu1Vv61qy3bt2rXqph32j9rn4MYWStwYYR7wXuy7MW+oGY55nTt3rnod246xxIQ/n79jxw55+eWX7Z9ft25dmTlzppvWIfIPM5PJ8rBTLFE4UkLSo6U4fob4UXeBcjxLPK5QuMz586C8MHeLnE5Mc/qc+esOy+LNx+WVO+rJA+2qqTrPRERERDnl6PuDJO2gY4DNndTDu/z+Xyn7tsr+wbe4fB0dvDIyMiQxPFz0Z0CRla+Q8k++49P/mjdvnvo5cOBAl9MgO/faa6+VP//8Uw4ePKjqCvsqISFB2rRpozKgERRs3LixnDlzRgVh8Nl//PGHClDqvffeeyq41a9fPxXQQeAR5SdefPFFFQS66aabHKZfs2aNbNq0SYYNGyYREd71ODx58qT999OnT8u3334rv/zyi8rGrlevnlgdgl6jR49W3wNBWg2CPnpYbiVLlpShQ4eqYBwC5gjut27dWgWkEEBG++zevVs++OADtQxWrlzptAx++OEHGT9+vDz66KPy0EMPyeLFi1U7/fvvv+p3d9nICGDeeOONajoE9RCkQmAPQTqshygzomVOjx07Vnr06CE333yzqmWNGw648YH34nfciLACbIfXXXed2i6wDaG2OALr27dvlyVLlsgLL7zgMD0CySghgwAygrn4Ph9//LG6Jv3www8dAskI9MXHx6uAMraXDRs2qO0FbbNu3ToV6PW0vbjaFu+44w75/fffVWAQbQ/4DgiCYv7RtgjWYj4mT56syiasX79eBXHzGgLg9913n1rvsY5iHdbTykjo6xLjZsXIkSPVoIH4HcHkhQsXqmWFfZI+oI7a8Y8//rhqSwRVEVSfPXu2040RDeoDowwQygLdc8896rn58+dL9+7d1bIbMGCAummAbXXQoEHqZgNuqmnwGixYsEA2b94sd955pwrenjt3TgWHEQA+ceKEw808Df5feHi4PPbYY1K4cGEVKHbFn8/HvhrrJnpsYLlj34CbCzVr1lQ3RIgCxkb53sGDB3H+qn4Gq4yMDNuJEyfUT8o7WVlZtvizybZ//juofuJvo5MJKbaBU/6yFX3oG5ePLmOW27YcOpcn3yG/4DZhHWwL62BbWAPbwToKelvse+5m2/bulS39wDz6qmTJkrYiRYp4nO7xxx9X1wDff/+9+nvv3r3q71deecVp2s8//1y9tnTpUvtzTz/9tC0iIsK2evVqh2nPnDljq1Spkq1Dhw725/A+vL9YsWK2o0ePOn3+3XffbQsPD3d67aGHHrKFhITY9uzZ4/H7tG/fXv0Ps8cDDzxgS0lJCZptAt+latWqpq/17dtXfadevXo5nWu/9NJL6rXRo0c7PL9s2TL1fOfOne3Pae2N5btmzRqH6Z988kn1Gtrd3Towbtw49RymN85LZmam/Xe8lpSU5PRdfvvtN/X+N9980+F5fPd27dr53RZYRq6WnycbNmxQ8/TGG2+4nU5bfoUKFbLt3r3b4bWuXbuqbSMxMdH+XJ8+fdT0M2fOdJh22rRp6nms695uL9o6ALiGb9iwoZoW7ax3++2324oXL+40f5j3woUL2+6//36n/6lvc1+52n94Y+fOner9Dz/8sMPzWG+KFi1qa9u2rf25v//+W623WO/M9muhoaH274z9UWxsrK1GjRq2hIQE+3TYHzRv3txpnr/55hv13Pjx450+u1u3bmpe9J+D9Qzbqxl9++u3C3yXuLg4W1pamv15zAP+b5s2bRyed7dP8Ofzb7zxRodt88CBA2pd7d27t+l3IPIXy1wQkc9Z4pVKRF/MFjfJEi9ZJEo+ePAq+f65VlKrbKzp56zdfUbajlwmI+ZvlQtpmWwBIiIiIi8hOw2Zn55o0yCb1VeIG82YMUNlICOjDVm02gOZnddff73KTMaAbcasOK3khB4y/fA+/UBWmK85c+aoDFyUG/AGMu1+++03+wM1o5Hhh3lF9h1Ke+QXzz//vNO5Nrr9IzMYZS70kDWMDE5k1iJ7XA9ZuOjOr6dlWeLz3MFyRdd6lH8wzgvaQoPXUPcVkEWNUhxYV5Ahj/UQGej+Qtatfv3DA3Vm8X+Mzxu/u7vtYunSpXLs2DGP0yNbtUaNGk7LFFnbe/futX9nlEhAlundd9/tNLAhtiFkvl6Mx3reXjTI6EY2KTKTV61apdpZvx/47rvvVLY/1gn9ckDGK96HTF5/GZet1iMgOTnZ6Xk850mtWrVUJj0ya/X7DSwXfD99VjKyr7GskG1t/F+33nqrWt6LFi1S0yIDGyVHkHmvz+xGiQdkFRsh2xhZzv/73/+cPhslfDAv6NHhDWwbGnwnlKZBb4kbbrhBtY/ZgHrPPvus170w/Pl8fGf9toleKVgvUf6CKJBY5oKIckS7OqVl5asd5Z2fdsr4n3dKWobj6NoZmTYZ/9NOVf5i/L2NpXP9MmwJIiIiIi+CYQh4eIJgAxi71ntDC66gW73WpdvVdPoSGq5qorZt21bq16+vSl1oQVJ0Q0cQCDVlvYX3Ifish7rRqNOLUhrTpk1TXcBdQTkGlBVAMNLfweaw/BGMymlmy3LPnj1qOZoNxNewYUMVIEWAs3jx4vbnzUp/lClTRpXQQLkAdxCAQukAfVDLFQRT33zzTTUQIpazHgJg/kL5ArSrGeO6iVIAnmoFYxqUTXjttdekQoUKarkhyIlAIoLERsZAMmDZAYJ7gJIDuDnSoEED03UWbYbAL4Ld+nIfrrYX/XaDkg0okYF13Ng2CKoi8OqqJq4+qOgrV9v9uHHj1EPP20H5UF4CAWIEkLXyErjBhBsRd911l306lFEBlApxBduxtk0A1lMjY+kY7bMRmEVda0+f7Qn2fyj9gnXfWBfb1Xrvqc2z+/mu1lfUfiYKJAaTiSjHREeEybDb6sgdLSrK019skJU7Lp5w6e07kSw93vlT7rqmkozpWV/KxOXcKNVERERUcKAesS9QM9mW4jnDzkxIdIxEVazlsWZyuEnNZF8h+IVaoAgmuQtMIKinZQSqeXQz7gjmTQ9BKmjXrp3ToFjuAk5adqoZ1JxFHU8EPDt16qRqySLQjbql2YXsTASTkaXoLpiM7E5jMNpXGPxKn0WZU9wtS6tB3Wpk8F599dWqPjPq9GoB9169etnXJ3+gLiyyzvUQzEQ9YmRO63kb5EfgE4FN1ENGhj0ytFFbFusi6tTqtxV3Nx2MmcaBbmMEXFHDF0F6fX1t0JYpbqagjnCgIfPfCMF2ZFqj/rGnAKYZBIwxWCBuDuC7oeYz9gdoX31WsfbdUO8bGcZmvP2fRvhs3BDCoJSuIPjvCdq+a9euquY7vhPqZeMmDtYXDDaK9jJb773drv39fFfra3bXVSIjBpOJKMfVLl9EfhzcWmauPCAvztsiZ5KcR+uet+aQ/LbpuIy8s770aVOFA/QRERFRtvg6sN2BF3rIhe3r/fpf0dXqSZUx812+jvILyErUggHZgYAMgskY/AuBO1fd49FVG5mN2gBcWkakWTablt2nDxIXK1ZMzXN2g68aBKAwmByCyJgXDA6Gv73t8u0OSg6Ap4xtZDoiiIQyAP62gzeBJk/8HVAc5RKQTYzMamOQDQN14XONJUMwWJ8RBolDVi3KmLiDmxW4aYEMcnfZyQgOIlt6+fLlDsEyvM+b0hPuILPamF2NIDK6+Gdn3USGMsqvaCVYcIMAGb74Dh06dPDps7C9IBi6ZcsW0yAense2r88Y9wYC3Fieb7/9tiolgcCylm2Mm0T4HVm2gdpG9Vx9JoK4/v4/LCMMJIjlfOjQITUIHgKixpszWO8waCGysY2DfJrND2AARQz+qKdlOBs/G9M2bdrUnmHuz7aKIO/ff/+tbrZhoEBPgXhf5fTnE2UXayYTUa7AgfjeNlVl/ajO0uvaSqbTnE1Olyen/ys3jVsh24947r5JREREVNAg8xbBtffff9+05i26aKNuK7KgR4wY4RDIQXAGdXX1WWoIKn722WcOn4EgFbIFEdBwVWLA267gGmQDok4pMj/feOMNdW4YqIxKLcuwWbNmbqdDMA91ZxEM8/dhLDfgDwSzEWT1NVuwR48eqnzJhAkTHJ5Hdi3aFRnfxoAlAk9r1651eG706NH2z3MH6wACwmbZ6fqsSATm0Z7GTEmUkshOVnJOwPLTbj5osK1oJRW00hW+wPaCMhkIUhozXhE43b17t1rW/txEeOutt1Spg48//ljVWNbqgiMQioz8H3/8UWX3BmIbzQ3ICMc6gUAy9i3VqlVzCt4j+xmGDRvm1FZaG+KGCqB+OwLuCLzr68PjdWM2N2hZ1ch4N9v+jMsM26rZDTjtZpTxM44cOaLK+WRXTn8+UXYxM5mIclWpIlHy0UPNpNe1leWZGRtlT3yS0zR/7jwtbUYsk6dvvEKeu7m2KpdBRERERKIyQL///ns1CNOdd94pt9xyiz2gggxI1CBFtiKCTxiUTQ9lJhCgQfdplCVArVdkCiOb1RhEQcARZSGQNYianchyRnbqgQMHZPHixaqkgKsglivIAkWZCAych8Csr13VtYEBNQjIIpMUNViRafrUU08FxSqCwdHQhR/1gFu1aqUCRwgEo5axO4MHD1Y3EPATZR7wXgQqEUhDsB43GIyQgYlljQHKkKWOgcsQ0EedYGO5AiMsTwQrEZT7559/5MYbb1SDvSFbGSVFkA2tZcsjiIpAPdYXtBMGf0NWdKlSpcRKsM7iJgbWfwxMhgx8zCcy/VFH19+sWwxSiGXbu3dv9T9QjgZthO0LdcW1AL4/cFMI2x7qjWPbRr1xZPRjntGOKD+BG0gohYDANurjohQCyo7oB720AgSOEUAeO3as6kmAesvGIDvme9SoUap0DepQY5lWqlRJZdTjBhfKqqDN8Dlov9dff12tqxhoEgF31JnGMtKyuPWfj8xotD/aBe2DmwAYBBFBWpQGwnLTB7CxrU6ZMkXdUEENZnxmt27dVI1mzBtKkGCQSPRYQL3yjz76SPUgyE6dcMjpzyfKLgaTiShPdKxXRla92lHe+nGHvPfLTknPdLzrir/H/bBDvl57WN7p01g61HU9+AsRERFRQYIgLIJ7kyZNUkE8BF20Eg/IpPv3339VoMwIQUhk7yHAhCDsFVdcobJHYfXq1Q7TImi4YsUKeffdd1XwF8FBBFKQmXvNNdd4DESaQbAH3dbRfduXgfc0yGjUshYBQSMESFFTFMvAaoFLVwYNGqRKi6DtEBDE90IA0lMwGdnlaBN0e0dAGO2CIDJq/SLgaFZDG4EvBKIQ7ES3fwTfsLwQ3PRU6gMBS7Q71gEE8RH4w3O4+aAfMA11exHwQtAZGZ+YTwQ4kTGNYKeVIAMZN2EwuCSWHwKHCCJjYDjMO5anPxDsRAY46jGjbXAzB+2JngRoG38GwtTDvCGgjLZDABQ3FTDf2JYQcMQNn7lz56ptAs/j5g++k9UgsIuAL5aJ9ruZ4cOHq6AybpBMnDhR7d9QTgT7NQSaEQDW3yRDu6GWNtZR7AcQXMc6in2VsZa2dqMNP5H5jQA92gfBW2PWP7YTBG6xrz179qy6UYKgLgLZuNGCAP+sWbPU/GHe0BbYTyIDOzuwbebk5xNlV4iNlbjzPdQjwt1QFLjHQS4YBbLOHFmvLbYdTlAD9K3e5foOa+9rK8uonvVVZjNxm7AS7p+sg21hDWwH62Bb+ObYRy9I2v7tfi3ryKp1pFz/MXneFggWIziCTL4vvvjCkufNyPTbt2+fujYJRL1kXxSkbQLLGEFfBNcQ4LSagtQWlDfmzZunbnTMmTNHldghosBhZjIR5bm6FYvKz0PayPQV++Xlr7bKuWTn2liz/zwov2w8JqPuqi/3tK7i98AlRERERGbcBYODBTLzAAFlBOhQk1Tr6m0FyNxcs2aNKrWR24FkIsqfUlJS1ICU+utD1EzG/hD7GWO5HyLKPgaTicgSQkND5P521eSmxuXkhblbZN6aQ07TnElKl8em/iuzVx2Ud+9rLFeUK5In80pERERkVQigaEFlq8DgcOgajvlCCQR0SyciCgSUfnnsscdU+RKUnzh69KiqmYwBETF4oafyMUTkOwaTichSysRFy6f9tAH6Nsj+k8lO06zYcUpavbpMnrnxCnnmpiskigP0EREREVkWavwi4IOavqjrqq93SkSU3RryqHc8ffp0NahoeHi4qhP+2WefsbYwUQ5hMJmILKlLgzKyekRHefOHHTLh112SYRigLy0jS974/j/5et1heefextK2TnAMuEJERERU0CxbtiyvZ6HAQYYmh0eighJMxoCERJR7rFNAi4jIICYqXF69o578/lIHaVGzuOny2XksUW55a6U89vk/cjoxjcuQiIiIiIiIiCiHMJhMRJZXv1JRWfh8Wxl/byOJK2TeoWLGygNy9YuLZc6fB5mFQURERERERESUAxhMJqKgGaDvoQ7VZe1rnaX71RVMpzmVmCb9p/wtt41fJbuPJ+b6PBIRERERERER5WcMJhNRUClXLFqmDmgu855sKVVKFjKdZvm2k3LtK0tl3A//qdrKRERERERERESUfQwmE1FQur5RWVk9spM82bWWhIWGOL2empElo77ZLm1GLJU/d57Kk3kkIiIiIiIiIspPGEwmoqAVGxUur91VX5a92F6aVS9mOs1/RxPlhrEr5Ilp/3KAPiIiIiIiIiKibGAwmYiCXqMqcfLbsHYy7u6GUiTafIC+6X/slxYvLZF5aw5xgD4iIiIiIiIiIj8wmExE+QJKXTzSqYasfa2T3NqsvOk0J86nysOf/CU93v1T9sQn5fo8EhEREREREREFMwaTiShfqVC8kHwxsIXMefwaqVTCfIC+JVtOyLWvLJG3f9oh6Rygj4iIiIiIiIjIKwwmE1G+dGOTcrJmZCd57LqaYjI+n6SkZ8nI+duk3WvLZM2u03kxi0REREREREREQYXBZCLKtwpHh8uY/zWQpS+2lyZV40yn2Xr4vFz/xh8y6IsNcjY5PdfnkYiIiIiIiIgoWDCYTET5XpOqxWTxC+3kjV4NpHBUmOk0ny3fJy1eWizz1x3mAH1ERERERERERCYYTCaiAiE8LFQGdqkpa17rLDc1KWc6zfFzqfLAR+vlrvdWy74THKCPiIiIiIiIiEiPwWQiKlAwKN/sx6+RmY+1kArFo02n+W1zvLR8Zam898tODtBHRERERERERHQJg8lEVCDd0rS8GqBvQOcaEmIyQN+FtEx5+aut0mHUclm/hwP0EREREREREREVyGDy/PnzpWXLlhIbGyvFixeXW2+9VTZv3uzTZ/z9999y1113Sbly5SQqKkoqVqyoPmffvn0O03Xo0EFCQkJMHxMnTsyx+SMiz4oWipCxvRuqesoNK5sP0Lf5UIJ0ef0PeW7mRjnHAfqIiIiIiIiIqAALlwJmypQp8vDDD0uDBg1k7NixkpKSIhMmTJBWrVrJypUrpWHDhh4/Y/bs2dKnTx9p0qSJDBo0SEqXLi3x8fGybt06OX36tFSrVs1h+lKlSsk777zj9DktWrTIkfkjIt80q15clr3YTiYv2iNjvt0uyWmZDq/bbCKfLN0rP/xzVAWfb72qvLohRERERERERERUkITYbAiTFAxnzpxRgd6iRYvKli1b1E84cOCA1KtXTwV3lyxZ4vYzduzYIY0bN1ZZyVOnTpXQUPfJ3chMRrayMWM5p+bPzKFDh6Ry5cpy8OBBqVSpkgSjzMxMtXyQqR0WFpbXs1Og5fe2OHAqWWUhL9x43OU0NzYuJ+PubiiVS8ZIXsnv7RBM2BbWwbawBraDdbAtrINtYQ1sB+tgWxARBa8CVebi22+/lYSEBJX5qwVqoUqVKnLnnXfK0qVLVcDVnXHjxklGRoaMHz9eBZKTk5MlLS3N4//OysqSc+fOqZ85OX9ElD1VSsbIl09cI9MHNpdycVGm0/y84Zhc8/ISmfjrLsnIdL1NExERERERERHlJwWqzMWaNWvUT5SMMMJz06ZNU6UqkMXryo8//ih16tSR1atXy5AhQ2Tbtm0qqIys4ddff11lIhsdPnxYihQpogLPkZGR0qZNG3n55Zelffv2AZ8/QEAaD83Ro0ftd3/xCEaYbwTig3X+85OC0ha3NCkr7a7sIK8t2C6f/b5flbrQS0rNlOFzt8iXfx6Ud+5tJE2rFcvV+Sso7RAM2BbWwbawBraDdbAtrINtYQ1sB+tgWxDlX+w5m/8VqGAyyj2AWakH7TltGjPILEZgNjU1Vbp37y79+/eX0aNHy86dO9XP6667ThYtWuQQJEbZCgym16hRI4mJiZFNmzbJe++9J506dZKZM2dKr169AjZ/GmRNjxgxwnT+MahfsJ5sJCYmCqqycMfEtshNL9xYSW6sHycvzt8l248lO72+8WCCXPfGCrmvVXkZdH0VKRyVO7tVbhPWwbawDraFNbAdrINtYR1sC2tgO1gH24Io/8K4YZS/FahgMjKDISrKuet6dHS0wzRmzp8/r35ikL1hw4bJmDFj7K81a9ZMunTpop5ftWqV/XnUVda7/fbb5cEHH1TB5ccee0xuvfVWFWQOxPxpnnnmGVUqQ4MAODKn4+LiVH3VYD3ZwIBnxYoVYzCZbZHrOhUvLssbVFID9I39fodcSHcsbZFlE5m68qj8uvWMvNmrgdzUpFyOzxO3CetgW1gH28Ia2A7WwbawDraFNbAdrINtQUQUvApUMFkL2iKz2CglJcVhGjOFChWy//7AAw84vNa5c2dV2xilKhDwdfc5FStWlH79+snYsWNV4BlB6EDMnwb1lvU1lzXI6A3mrF6UEwn275BfFMS2wHcddNOV0r15JXl25kZZtDneaZojZ1Lk3snr5Zam5eTN3o2kYonL+4ycUBDbwarYFtbBtrAGtoN1sC2sg21hDWwH62BbEBEFpwI1AJ+7UhHuSkxoSpQoYS8TUb58eafX8RxqmJ49e9bjvFSvXl39jI+PD9j8EVHOq1Y6Vr56qqV89sjVUqao+QB9P/xzTFq8tFg+XLRbMpG2TERERERERESUDxSoYDJKPcCff/7p9Jr2XPPmzV2+H2UWtM84ePCg0+t4Ljw8XAWdPdmxY4f6Wa5cuYDNHxHlDuwL7mhRUdaN6iwPtK9mOk1iaqY8P2ezdB7zu2w44PkGExERERERERGR1RWoYDLqFRcpUkQ++eQTSUhIsD9/4MABmTdvnnTo0EEqV66snkOpiu3bt6t6w3p9+/ZVPydNmuTw/DfffCNHjhxRJSu0+sbIUEYtKLNA8scffyxlypSRVq1a+TV/RJT3isVEyLt9GsuvQ9tK3QpFTKf5Z99Z6fDachn+5WZJTMnI9XkkIiIiIiIiIgqUAlUzGYPPjRs3TgYMGCCtW7eW/v37q/rEEyZMUJmG7777rn3atWvXSseOHVXwWD+IXp8+fWTmzJkqmIwSFZhm9+7dMnHiRDXA3dtvv22fdtmyZfL0009Lt27dpEaNGqre8aZNm+Tzzz+X9PR09Tla4NnX+SMi67imVgn5/eUOMuHXXfLm9/9JiskAfRN/2y3f/nVE3rqnkdzQOOcH6CMiIiIiIiIiCrQCFUwGBGhLliypgrZDhgyRyMhIadu2rYwePVoaNWrk1SAB3333nRo8b8aMGSojGYPdIat45MiRUrt2bfu0V155pVxzzTXyyy+/yLFjx1RguGzZstK9e3cZPHiwNG7cOODzR0R5IzI8VJ69qbZ0v7qiPDNjgyzdesJpmoOnL8j/JqyR25pVkLG9G0j5Yjk7QB8RERERERERUSCF2Gw2jg6Vz2HwPpTHQE3nYB3AD+VCzpw5o7K3w8LC8np2CjS2hWfYrX619rAMnbNJTp5PM52mSHS4vNKjrjzYobqEhYawHYIYtwnrYFtYA9vBOtgW1sG2sAa2g3WwLYiIgleBqplMRJQbUJbmrmsqybrXOst9bauaTnM+JUOem7VJrn/jD9l08BwbhoiIiIiIiIgsj8FkIqIcUqJwpEzo20R+GtJaapcrbDrN+j1npP1ry+WleVskKZUD9BERERERERGRdTGYTESUw1rXLiUrXukgw2+rI1HhzrvdzCybvL9wl7R8Zan8tuk424OIiIiIiIiILInBZCKiXBAVESZDul0pq17tKO3qlDKd5sDJZLnzvdXywEfr5fi5FLYLEREREREREVkKg8lERLmoVrnC8t2zreTDB5uqMhhm5q87LM1fXCyfLd8nWVk20wH+TiWmyaHTKeonx1ElIiIiIiIiotwQniv/hYiIHAbo692qilzfsKyqlTxz1UGnpXPuQoYM+mKDzFl1UN69r7HUq1hUziany+xVB+STJXtld3ySfdqaZWKlX6fq6jOLxURwSRMRERERERFRjmAwmYgoj5QsEiUfPHiVCgI//cW/suv45QCxZs3u09J25DK59arysnDjcUlKzZSQEMdp9pxIkqFzNsuoBdtk2sAW0qVBmdz7EkRERERERERUYLDMBRFRHmtbp5SsfLWjDO12pUSaDNCXkWmT+euOqEAy2AyVL7S/k9Mypef7q2XR5vhcmW8iIiIiIiIiKlgYTCYisoDoiDAZdlsdWflKB2lTu6Rfn4Hyyqif3HfyWlUSg4iIiIiIiIgokBhMJiKykNrli8gPg1vLpPubSPHYCL8CyompmTJn1YEcmT8iIiIiIiIiKrgYTCYisuAAffe2qSrrXuskhaN9L20fGiLy8ZK9KkuZiIiIiIiIiChQGEwmIrKo0JAQSUzJ8Cs7eXd8kpxJYqkLIiIiIiIiIgocBpOJiCwK5Sqy47wfgWgiIiIiIiIiIlcYTCYisqjCUWHZen9KGoPJRERERERERBQ4DCYTEVlUicKRUrNMrISE+Pf+DqOWy+hvtsm5ZJa7ICIiIiIiIqLsYzCZiMjCA/H161Rd/B1HLzktS978YYc0eWGRTPx1l6SkZ69sBhEREREREREVbAwmExFZWO9WVVS5i1A/s5PhdGKaDJ+7Ra56YZF8sWK/ZGRmBXIWiYiIiIiIiKiAYDCZiMjCisVEyLSBLVSWsqeAMl5HSYxwFxMePpMij0/9V659Zal899cRsfmb8kxEREREREREBRKDyUREFtelQRmZ+2RLiYm8OCCfMVas/Y3Xv376Wvnn9S5yb+sqLoPPO44lSp/J66TzmN9l+bYTOT37RERERERERJRPMJhMRBQkAeUt47rK2F4NpHrpWIfX8Deex+ud65eRKiVjZNIDTeXPER3llqblXX7mX3vPyq1vr5Lbx6+Sf/adzYVvQURERERERETBLDyvZ4CIiLwveTGgS03p37mGnExIkcPHT0nFsiWlVNFoVQbDqE6FojLzsRaybvdpGTF/m/zx30nTz1269YQs3bpcbr+6grx0e12pVa4wm4SIiIiIiIiInDAzmYgoyCBwXKJwpFQqEa1+mgWS9ZrXLCHfP9dK5g+6VhpViXM53Tfrj0iLl5fIU9P/lSNnLuTAnBMRERERERFRMGMwmYioAEDAGSUwlr/YXj575GqpUcaxVIYmM8smU3/fL01fWCQvzdsipxPTcn1eiYiIiIiIiMiaGEwmIipAQkND5I4WFWXtyE7yTp/GUi4uynS6lPQseX/hLmky7Dd568cdkpSakevzSkRERERERETWwmAyEVEBFBEeKg+2ryb/jOkir95RT+JiIkynO3chQ15bsE1lKn+6dK+kZ2Tl+rwSERERERERkTUwmExEVIDFRIXLoBuvkA2vd1E/C0WGmU53/FyqPDtzozR/aYl8teaQZGXZcn1eiYiIiIiIiChvMZhMRERSPDZSZSj/M6azylgOCzUf1G/viSR56JO/pN1ry+S3TcfFZmNQmYiIiIiIiKigYDCZiIjsyhcrpGopr3utk6qt7Mqmgwly53ur5eZxK2XNrtNcgkREREREREQFAIPJRETkpGbZwvLZI1fL7y+3ly4NyrhcQit3nJLr3/hDek9cI1sPJ3BJEhEREREREeVjDCYTEZFLjasUk6+fvlZ+eK61NK9R3OV0P/17TFq9ulQGTPlb9p9M5hIlIiIiIiIiyocYTCYiIo/a1iklvw1rK7MeayF1KhQxnQblk2f/eVCaDV8kz8/eJCcSUrlkiYiIiIiIiPIRBpOJiMgrISEhcnPT8rLq1Y4y+YGmUrlEIdPp0jNt8uHiPdJ42G8y5tvtknAhnUuYiIiIiIiIKB9gMJmIiHwSFhoid7euIn+N7ixv9GogJQtHmk6XlJopY7//TxoPWySTft0tKemZXNJEREREREREQYzBZCIi8ktURJgM7FJT/n29iwztdqUUjgozne50Ypq8MHezNBu+WGas2C8ZmVlc4kRERERERERBiMFkIiLKlqKFImTYbXVkwxvXyaNdakhkuPmh5dDpC/LY1H/VQH3f/31EbCiyTERERERERERBg8FkIiIKiFJFouT1Xg1V+Yt7WlWW0BDz6f47mij3frBOOo/5XX7ffoJLn4iIiIiIiChIMJhMREQBVaVkjHzw4FVqoL6bm5RzOd1fe89Kt7dWSfd3Vsm/+8+yFYiIiIiIiIgsLlwsauPGjbJmzRo5efKk1K9fX2699Vb1fGpqqnoULVo0r2eRiIjcqFuxqMx6/BpZu/u0vPr1Vlm545TpdEu2nJAlW5ZLj+YV5cXb60jNsoW5XImIiIiIiIgsyHKZyTt37pTWrVtL06ZNZcCAAfLiiy/KN998Y3995syZUrx4cVm4cKHf/2P+/PnSsmVLiY2NVZ+FQPXmzZt9+oy///5b7rrrLilXrpxERUVJxYoV1efs27fPPs2ZM2dkwoQJctNNN0nVqlUlOjpaqlevLnfeeaf8888/pp8bEhLi8uHrPBIRWUGLmiXkx8Gt5eunW0rDynEup5u/7rA0f2mJPP3FBjl69kKuziMRERERERERBVlm8pEjR6Rdu3Zy/Phx6datm/p98ODBDtP07NlTHn30URUQ7tq1q8//Y8qUKfLwww9LgwYNZOzYsZKSkqICvq1atZKVK1dKw4YNPX7G7NmzpU+fPtKkSRMZNGiQlC5dWuLj42XdunVy+vRpqVatmpoOmdVPP/20dOrUSQXGy5Qpo4LlH374oZr/WbNmSa9evZw+v23btvLII484PV+5cmWfvy8RkRXghliXBmWlU70yMn/9YRm1YLvsPZHkNF1mlk0+X75PZq86IAM615Cnb7xCisdG5sk8ExEREREREZGjEJvNZhOLGDhwoHz00Ucq2KoFU0NDQ+X++++Xzz77zD4dsoovXLggGzZs8OnzkSmMQC9KZGzZssVeKuPAgQNSr149adGihSxZssTtZ+zYsUMaN26sspKnTp2q5s8VZClnZGRIrVq1HJ5HhvFVV10lJUqUUAF0/Wcg4NK3b1/12YFy6NAhFYg+ePCgVKpUSYJRZmamaj9kkoeFheX17BRobAtrCPZ2SM/Ikukr9svY7/+T4+dSXU4XFxMhT99QSwWWY6Isdf8z37RFfsK2sAa2g3WwLayDbWENbAfrYFsQEQUvS5W5+Omnn1RmsFlWrh4CwocPH/b587/99ltJSEhQmcn6mstVqlRRpSeWLl2qAq7ujBs3TgWIx48fr4LAycnJkpaW5nI+jYFkQFY0HsjARkazmfT0dDl//rzP35GIKBhEhIfKQx2qyz9jusgrPepKXCHzQPG55HQZMX+bNH1hkUxZtlcFoYmIiIiIiIgob1gqzevYsWMq69ibu5hJSc7doz1B2QlASQsjPDdt2jRVqsJdOYkff/xR6tSpI6tXr5YhQ4bItm3bVFAZWc2vv/66dOjQweN8ZGVlqe8aGRkpxYoVc3r9q6++khkzZqjvGRcXJ7fccouMGjXKXj7DEwTM8dAcPXpU/cTn4RGMMN9YbsE6//kJ28Ia8ks7RIeHyFNda8p9bSrL+wt3y0dL9kpKunPA+Ni5VHlmxkaZsHCXvHDbldK9WQUJDQ0RK8gvbZEfsC2sge1gHWwL62BbWAPbwTrYFkT5F3tr5n+WCiYjsOpNxvGuXbtU/WF/yj2AWakH7TltGjPnzp1TgdnU1FTp3r279O/fX0aPHq3qIOPnddddJ4sWLZL27du7nY9Jkyapz0E5CwzKp3f11VfLHXfcIbVr11b/Z8WKFfLJJ5/Izz//rGo6I5DtCbKmR4wYYTr/GHQwWE82EhMTBVVZuGNiW1D+3Cae7FhOel5VXCYuOShz1x2XTJMk5L0nkqXfp//IOz/ukOduqCrtahdT5YHyUn5si2DFtrAGtoN1sC2sg21hDWwH62BbEOVfpUqVyutZoIIUTL7mmmtk4cKFKtu3bt26ptOsXbtW1Uq+++67ff58lKSAqKgop9e0oK42jRmt7AQG2Rs2bJiMGTPG/lqzZs2kS5cu6vlVq1a5/Ixly5bJc889J9WrV1dBXyNkRuv17t1bZSbfdNNNajC/X375xeP3fOaZZ1QpDw0C18icRpYzanoG68kGAka44cBgDduC8u82gV3UpAfLyTM3J8qY73bIgvVHTKfbejRJHvx8q7SuXUJeur2utKiZd/u2/NoWwYhtYQ1sB+tgW1gH28Ia2A7WwbYgIgpelgomP/nkk/LDDz+orN/p06erAKjexo0b5b777lNlJZ544gmfPz8mJkb9RMavUUpKisM0ZgoVKmT//YEHHnB4rXPnzqr2MkppICBt9jnILO7WrZuULl1afv31VzUAnzduvPFGFWhfvHixmk9jNrMR6kHra0JrEOQI5kAH2j3Yv0N+wbawhvzcDrUrxMnUAc3l6f1nZeT8bbJ4i3l9+ZU7TssNb66Um5uUk5e615W6FZ33fbkhP7dFsGFbWAPbwTrYFtbBtrAGtoN1sC2IiIKTpQbgQ2bv8OHDZceOHXLttdeq0hPI9kKAuWbNmtK0aVP1Gko4ILjqK3elLNyVwNAg+KuViShfvrzT63gOdTPPnj3r9Nrvv/8uN9xwg8oMRnay2cB87iCTGQP/ISuaiKigaFK1mMwfdK18/1wrubqG6+zjH/89Jte+ulQGfva3HDjluocJEREREREREeWTYDK89tpragC6xo0by5EjR1QNypMnT8revXtV6Yu5c+eqgLM/tEznP//80+k17bnmzZu7fD8C29pnHDx40Ol1PBceHu6Ucbx06VKVXYy6McuXL/c5kAwIokdEREjJkiV9fi8RUbBrV6e0LBrWVmY82lyuLF/YdBqbTWTWqoPSbPhiGTpnk5w879wLhYiIiIiIiIjySTA5ISFB1SXu0aOH/P3333L8+HFVNgKBXmQOb968We68806/P//222+XIkWKqAHt8L80Bw4ckHnz5kmHDh2kcuXK6jmUqti+fbuqN6yHQfO0QfT0vvnmGxX8Rna1vgwFSlPcfPPNKmsZgWRkGLty6tQp0+dnz56tlgcym83qPRMRFQS4odftqgqy6tWOMun+JlKpxOXSQ3ppGVkyedEeaTz0N3nju+1yPiU91+eViIiIiIiIKD+yVM1kDF6EzGAEkAG1hfEIFJSYGDdunAwYMEBat24t/fv3V/WTJ0yYoIIU7777rsNAfx07dlTB46lTp9qf79Onj8ycOVMFk+Pj49U0u3fvlokTJ6oB7t5++237tOvXr1c1ktPT06Vfv36q1IUR6kNrpTNGjRql6ip36tRJ1V9OS0tTf3/99dcqGK2fPyKigio8LFTubVNV7rymkkxZuk/e+mmHnE5Mc5ouMTVTXv/uP/l4yV557uba8lCHahIVwZrGRERERERERPkimIxB46644ooc/R8IIKNUBILKQ4YMkcjISGnbtq2MHj1aGjVq5NUgAd99952MHTtWZsyYoTKSMd/Ieh45cqTUrl3bPi0yqS9cuKB+Hzp0qOnnoXyHFkxGYBrZ0AhWo7QHSnxUq1ZNBg0aJM8//7yUKVMmYMuBiCjYRUeEyWPX15Q+bavIxF93y8Rfd0lSaqbTdKcS02TYl5vlg992y7Db6kivaytLWGhInswzERERERERUTALsSFiaREI6mJ2VqxYkdezkq+gRAjKd6Cms7sBBq0sMzNTzpw5o7LLw8KYWci2IG4Tzk4kpMpbP+6QKcv2Snqm60NbnQpF5OXudeWmJuVUrxTun/IPbhfWwHawDraFdbAtrIHtYB1sCyKi4GWpmslPPPGErFq1StUZJiIi8kXpolEytndD+Wt0F+l9bWVxFSfefuS83D1prVz3+h+y4r+TXMhEREREREREwVjmomXLljJw4EA1YN1DDz0kt912m1StWlUKFTIfZAl1hYmIiPSqloqRDx+6Sp68oZa8tmCb/PTvMdMFtG7PGbl53Erp0qCMvNyjrjSuUowLkoiIiIiIiChYgsmoD4wuxyh18eGHH6qHK5guIyMjV+ePiIiCR72KRWX249fIml2n5dWvt8qqnadMp1u0OV497mhRUYbfVkdqli2c6/NKREREREREFAwsFUxu165dQOpXEhERaa6pVUJ+GtJaBYxHzN8qmw4mmC6cr9celm//OiL3takqQ7rVlvLFzHvFEBERERERERVUlgomL1u2LK9ngYiI8iHcqLyuYVnpXL+MzF93WF77ZpvsO5HsNF1Gpk0+W75PZv95UAZ0riFP3VBLisdG5sk8ExEREREREVmNpQbgIyIiykmhoSFy5zWVZN1rneXtexpJ2bgo0+kupGXKOz/vlMbDFqmfyanmZZVQlulUYpocOp2ifuJvIiIiIiIiovzKUpnJRllZWXLq1MUalyVLlpTQUMa+iYgo+yLDQ+XhjtWld6vK8uHiPfLezzvl3AXngPG55HRVb/nDRbvl+VvrSJ/WVSQiPFTOJqfL7FUH5JMle2V3fJJ9+pplYqVfJ3xuFSkWE8GmIiIiIiIionzFktHZn376Sa677jopUqSIlCtXTj0KFy4s119/vXqNiIgoEGKjwuXZm2rLv69fp0paREeYHxaPnUuVQV9skBYvL5ERX2+VeoMXytA5m2XPicuBZMDfeL7+4IWqRjMRERERERFRfmK5YPKQIUOkW7dusnjxYrlw4YJERUWpR0pKiixatEi99vzzz+f1bBIRUT5SonCkjLyzvvw9uov0bVdVwkLNB4PdE58k43/eKUmpmepvY1UL7e/ktEzp+f5qBpSJiIiIiIgoX7FUMHnBggXy1ltvSdGiRWX06NFy+PBhSU5OVo8jR47ImDFjpFixYmqab775Jq9nl4iI8pmKJQrJ+/c1kbUjO0n3qyv4/TlZtov1lPtOXqtKYhARERERERHlB5YKJk+aNEkiIiJkyZIlMmzYMClfvrz9NZS6GDp0qMpODgsLU9MSERHlhFrlCsvUAc1l2YvtpVP90n4HlBNTM1VtZSIiIiIiIqL8wFID8P3999/Svn17adq0qctp8FrHjh1l3bp1uTpvRERU8DStVkwWDGoly7bGS8/310hqRpbPnzFszmaZ9OsuKRMXLWWKRkupIpFSpmiUepQuGn3pJ56LVoP2hboosUFERERERESU1ywVTEaN5FKlSnmcrmTJkmpaIiKi3NCwcpxfgWRAGeWDp1PUw5PwsBApXSRKShW5GGy+GGi++Cij/6mmiZTwMEt1MCIiIiIiIqJ8zlLB5EqVKsnq1aslKytLQkPNL5AzMzNlzZo1UrFixVyfPyIiKphQriI3ZGTa5OjZFPXwJCREpERspNuAs/656Igwyc9Qo/pUYpocOZ0iWRFpKus7BAuJiIiIiIiI8mcw+aabbpKJEyfK448/LuPHj5fo6GiH11NSUmTQoEGyb98+efLJJ/NsPomIqGApHGW9QKzNJip4ise2I+c9Tl+0ULhDcBnBZtPgc1yUFI4KD5pALAY4RF3qT5bsld3xSfbna5aJlX6dqkvvVlVU+RAiIiIiIiLKvhAbUnks4vjx49KkSROJj4+X0qVLS48ePaR69erqtb1798r8+fPlxIkTajA+1FcuW7ZsXs9yUDh06JBUrlxZDh48qLK/gxEy0s+cOSPFixdXAzAS26Kg4zaRu3CobDZ8sew5kaSCuL5AYLZZ9WIq6BufkConz6eqwfmsrFBk2KWA88VazpcDzpH2Os9aILp4bESeBZ4XbY6X+yavlaTUTJWprW8b7W/cCJg2sIV0aVAmT+axIOL+yTrYFtbBtrAGtoN1sC2IiIKXpTKTERxesmSJ3HPPPfLvv//Khx9+aL9A1WLeCDbPmjWLgWQiIso1OBYhy3XonM0+v/el7nVkQJea9r8zs2xy+lJg+cT5VDmRkCrx51IlPiFFTp6/+Dx+x/MnzqdJmp+1mrPjQlqmHDiZrB7e1nk2y3bGo5Tu95KFA1fnGYHknu+vtp8fGIP82t/JaZlqurlPtmRAmYiIiIiIKD8Fk6Fu3boq6/j3339Xj8OHD6vnUSO5Xbt26kFERJTbUC5h1IJtKjjpTWZxaIhITGSY9GpVxeH5sNAQe41jTxAoPXchQ04kpFwMPiek2X/Xspy13xF8RoZubstOnWdjvWdjCY4oF3WeUdoCGclYPp7aAq+Hik36Tl4rW8Z1ZckLIiIiIiKi/BRM1jBwTEREVoK6uyiXgCxXBCfdBTERSEY28/RHW2QreInPwPvxuKJcEY/TJ6Ui8Hwp29lFwFl7DQFZq9d5jrtU59kx4BwtWw8n+BQ4R1thEMU5qw44ZIkTERERERFRPgkmExERWQ3q7qJcArJcEZxE0FgfVNb+RkYyAsmd6+dund7YqHCJLR0u1UrHepwW5TMultJwDD7rA87az1OJeVPnGVnZeOw6fnlgPX+hbT5eslf6d64RNIMLEhERERERWY2lgsmohfziiy/K5MmTpWvXrqbTLFy4UAYOHChvvPGG9OzZM9fnkYiICjYElFEuAVmuCE7ujr8c6KxeOlYe6VRdlbbITkZybogMD5WKJQqphyf6Os/22s7nHGs7a79jmvRM640wiGA42uqeD9ZKnfJFpHqZWKlRJla1Wbm4aAlFtJmIiIiIiIiCJ5g8Y8YMOX/+vHTq1MnlNB07dpSEhAT54osvGEwmIqI8gUAxyiUgy/VkQoocPn5KKpYtKaWKRufLrFd9nef6UtTttKhjjBIarspr2H+ez5s6zz/+c0w99KIjQlU2txZcrl46RgWb8XuVkjESER6YQQOJiIiIiIiCnaWCyZs2bZJGjRpJRITrbK7IyEhp3LixbNy4MVfnjYiIyAiB4xKFIyUkPVqK42c+DCT7CsugeGykenhb51nVdjaW2rgUjNaX4DiXQ3WeU9KzZPuR8+phFkivVKLQxSCzPZs5xv43SosQEREREREVFJa6Ajpx4oS0b9/e43TlypWTVatW5co8ERERUc5BMLZ6aTy8q/Mcfy5Frh+7Qo6cviC5UUwDJT72n0xWj2XbTji9jkEB7RnNlwLN2t/qRgNvMBARERERUT5iqWByXFycHDhwwON0hw4dkthYzxedRERElH+gznOlkjHyxPU1ZeiczT6/v0zRSDmXnCGpGVkBmycta3r1rtNOrxUtFG4on4GA88Ws5orFC7FOMxERERERBR1LBZObNWsmixYtku3bt0udOnVMp8FryEp2V1eZiIiI8q/erarIqAXbJDktUw2s5wnG1ouJDJN1o7pI0ehwOXo2RfaeSJI98Unq517dz3MXMgI2nwkXMmTjgXPqYRQVHipVS12uzXw5uzlG1WmOiggL2HwQERERERHly2Byv3795JdffpFu3brJrFmzpHnz5g6vr1+/Xnr37i1ZWVny8MMP59l8EhERUd4OgDhtYAvp+f5qCRWb24AyAskoNTH90RbqfVCxRCH1aHNlKafBA88kpV8OLhsCzsfOpQbsOyA7esexRPUwQmWMyiUKqaxmLdhsDziXiZEi0a7HliAiIiIiIiowweTu3bvLPffcIzNnzpSWLVtK3bp1pXbt2uq1HTt2yLZt29SFHgLKd911V17PLhEREeWRLg3KyNwnW0rfyWslMTVTBY31QWXtb2QkI5DcuX4ZrwdUxKNZjeKmgwXuO5HskM2sBZsPnrqg6isHgs0mcuDUBfX4fftJp9dLFYk0ZDNfDjbjtbys04zztFOJaXLkdIpkRaRJ6aLRrBtNRERERJSPWCqYDNOnT5f69evLuHHjZOvWreqhKV68uAwePFief/75PJ1HIiIiskZAecu4rjJn1QH5eMle2R2fZH8NwdVHOlWXXq2q2DOSAzFYYP1KRdXDKD0jSw6evmAPLu/TBZv3nUyWC2mZEignz6epx7o9Z5xeKxwVpoLL1RyymS/+XqlEIQlDlD0HnE1Ol9mrDsgnhnaoWSZW+nWqrkqTBKodyDsM7BMRERFRTgix4UzTgtLT0+Wvv/6yD8hXpUoVVVM5IoIXIr7CgIWVK1eWgwcPSqVKlSQYZWZmypkzZ9QNhbAw1pFkWxC3CetgW1gDTmdOJqTI4eOnpGLZklLKQhmxmLdj51Jkb3yy7IlPlL2G7GaU1sgNEWEhF+s0G7KZ8bNq6RiJ9rNO86LN8XLf5LWSlJqpSnTozyy1vxHkRmkS3ACgnMXAvjXxWGENbAfrYFsQEQUvywaTKXAYTKZA4omfNbAdrINtYR3B2hZnktIuBZcdg8x4HDmTkivzgKBvhWLRlwPNulrN+N1VVjECyahdjdNJb2pXozQJA8o5h4F96wrW/VN+w3awDrYFEVHwslyZC1d+/fVX2bRpk8pQ7tGjB0/CiIiIKF8oHhupHldVc67TjPIY+y4Flu3B5kvlMw6cSpaMzMDVaT58JkU9Vuw45fQ66khXL305q7lG6VgpXTRK7pu8zmMgGfA6BktEjWuUJmHJi8DTB/a1NtXT/k5Oy1TTMbBPREREREEfTP7oo4/k7bffls8++0zatGljf/7ee++V2bNn2/9u1aqVLF68WCIjI/NoTomIiIhyXqHIMKlbsah6GGVkZsmh0xd0AwEmq6zmfZf+RtAwUE4npqnHX3vP+v0ZCChjsMQ3v98udzSvJCGhIiESIqGhyFzGbyKhoSEXs5gvPY+fyJrG85df136/NJ3udWQ/X57W+XX13pDLP61SCiUQpS1QaoSBfSIiIiIqUMHkBQsWqO5f1157rf25pUuXyqxZs6RkyZLSs2dP+f3332XVqlUydepUeeSRR/J0fomIiIjySnhYqBpoD4+O9RxfQ1AxPiHVXjLDHmy+VEbjVGJaXs22TPptj3pYhatgs1aaQ/upD0IbX9cHp929rp/O3esX36+i6s6v64Pql17fE5+salb7Gth//5edMqBzDSkWGymR4aE5upyJiIiIKH+wVDB5+/bt0qBBA4cSFnPmzFEn2vPmzZMOHTrI+fPnpWrVqjJt2jQGk4mIiIhM4NypbFy0erS8oqTT6+eS0+0lM/bqsprx9+EzF5xKJORnCKxm2cuFFKAvLiJv/7RTPSAmMkziYiJUCZJisfgZqX53fO7i83EOf0eoDPr8kuWdE3BzBzdwjpxOkayINCltoQFCiYiIiII6mBwfHy+tW7d2eA6ZyGXLllWBZChSpIgqgbF+/fo8mksiIiKi4IZgYJOqxdTDKCU9Uw6cTL6UzXw52LwnPlH2n0yW9ADVaSZrQVkUPI6e9X3QR2Q1xxVyDDAj2zkuJtwelNZeuxicjrT/XSQ6PN8GVlF+ZPaqA/LJkr2yOz7J/nzNMrHSr1N16d2qCuuHExERUdCxVDA5JiZGEhMT7X+fOnVKduzYIXfddZfDdHFxcaocBhEREREFVnREmNQuX0Q9jDKzbCpzecO+s3Lv5HVc9KSkZWTJifOp6uErlOkwBpgdM6IvB6MvB6wvZ02H4QMsOiAi6lij/IgxVr7nRJIMnbNZRi3YJtMGtpAuDcrk1WwSERERBXcwGeUrUA85LS1NDa43f/589Xznzp2dMphLly6dR3NJREREVDAhcFelZIxULlFIZVciKOZLSQzE1EoXjZJXetS9XGLCZlOfgZ/4Wy791AaTM75u0/20eXhde7+YvS42ycq69H7MS9bFn7Ys96+rz8nC3F/6/Czz19X78Tm6583m0/5/Xb1u+J7G74eBGE+cz7sa2NmF73AmKV095ITv748rFK6CynFugtHFXQSjI3KoTjQCyT3fX63aCIzbiPY3MsEx3dwnWzKgTEREREHDUsHku+++WwYPHizt2rWTVq1ayWeffSZRUVFy66232qfJysqSv/76Sxo2bOj3/0GQ+s0335RNmzapoHXbtm1lzJgxql6zt/7++295/fXX5Y8//lBZ0qVKlZJmzZrJ+++/L9WqVXOYduPGjTJ8+HBZsWKFCpRj3ocOHSq33357js0fERERUU5BWQJ000d2pS8QQ3v2pivk3jZVc2zeChoELJsNX+xXYL9E4Uh5sH1VSbiQoUoyqEcSfqaputr4OyVdRc4t69yFDPWQUxd8fm9slK5OtC4Yffm5y5nRxnrSqBNtBssMGclasN8dvB4qNuk7ea1sGdeVJS+IiIgoKFgqmDxw4EBZsGCByk5eu3athIaGyrvvvqtqJmsWLVokp0+flvbt2/v1P6ZMmSIPP/ywCsyOHTtWUlJSZMKECSp4vXLlSq+C1LNnz5Y+ffpIkyZNZNCgQSpLGtnS69atU/OmDyZv2LBB1XhGUPzZZ59VQecZM2ZI9+7d5fPPP5f7778/4PNHRERElNNQ7xXd9JFd6SloBqhGgEHeerWqwsaxSGB/yC21ZUCXmm6nQw1tLcCMn+cuaAFnLfh8OfBsDEYnpmaKlaEEBR5HzvheJzoKdaINAWb8fehUsvpMb2HbwXKas+qAx7YgIiIisoIQm9b/yiIwOxh078SJEypYW6tWLYfXly5dqjJ9u3XrJjVq1PDps5FBjEBv0aJFZcuWLeonHDhwQOrVqyctWrSQJUuWuP0M1HBu3LixquM8depUFfB2B1nWyEhGcPzqq69Wz6Wnp8s111wje/fulf3799vnIxDzZ+bQoUNSuXJlOXjwoFSqVEmCUWZmplo+xYsXl7Aw80wQYlsUJNwmrINtYR1si9yn787vLqCMQDKCnvOeaimd67M+bKAhiFt/8EKfA/s5nQ2bnpGlCz6nydnkDHtQWgs+XwxEOz+XcCHdp0zrYFc4Klx6t6okxVGW41I2tPp5KVta+x01zcl/PE5YB9uCiCh4WSozGXCh4S7ruGPHjurhj2+//VYSEhLkmWeesQdqoUqVKnLnnXfKtGnTVMAVgVdXxo0bJxkZGTJ+/HgVSE5OTpbw8HBVjsJo3759qgxGhw4d7IFkiIiIkCeffFIeeOAB+e677+Tee+8N2PwRERER5RYMHIZ6r+imj+xKBCn1wUztbwQupz/agoHkHIJgIwZyQ2AfZRO8CeyjPXIykAyoSVyqSJR6+AqDPSKgrGU7OwSlLwWeL2dEOz+H9weTxNQM+WTpPo/TRUeE2oPNCC7bfzcLPuumQb3onKoRHSxw0+tUYpocOZ0iWRFpUrpotNoWiIiIKMiDyTlpzZo16idKRhjhOQRrUarCXbD2xx9/lDp16sjq1atlyJAhsm3bNhVURtYwaigjcOzt/wNkLGvB5EDMHyAgjYfm6NGj9ru/eAQjzDfqZQfr/OcnbAtrYDtYB9vCOtgWeaNj3ZKy6Y0uMmf1IRUM2xOfZH+tWulY6dexmvRqWUmVAOBxPGfbYc7jzeX+j/5SZRZcBfZR63fagKulQ52Slm+PotFh6lGlRLTPQUPc3DinL7uButAO2c8oxaGrFa0LRKdlWLdONGpYHz2boh6+Khwdrgs8X64DXVw3KKF+EMOLz0dI0UIREooVKEihXWf/eVA+XbbfYf9Uo0ysPNyhqvS+trJaFpS7eMwmyr/Ymzz/K1DBZJR7ALNSD9pz2jRmzp07pwKzqampquZx//79ZfTo0bJz507187rrrlM1nbXMal//X3bnT4Os6REjRpjOf2xsrATryUZiYqK6OOCOiW1B3CashPsn62Bb5K2eTYvJXU0ay6nEVDlx+ryULlFEShaOUpl/WamJciY1j2ewALiqQoSsGHq1zP87Xr5YdVT2nboccERAtk+r8tLjqjJStFC4Kh+W38WGiMQWFqlYGKUh8PAuQxp1oi8O7JepBifEQxvoT/s94dJrjs+ny4V062ZEJ6ZkqMeh074NVojk3aLR4RJXKFziYi7+LFooTIoVirj8OwYp1D8fc/H32MiwPM3+/X3HGXlsxnZJTstSg07q7Y1PkhfmbpVR32yXSffWkXa1i+fRXBZMPGYT5V8YK4zytwIVTEZJCsBgeEbR0dEO05g5f/68+olB9oYNGyZjxoyxv9asWTPp0qWLeh4DCPrz/7I7fxqUycAgfhoEwJE5HRcXp2oOB+vJBk5EixUrxmAy24K4TVgK90/Wwbawhri4TCmBLvY8ZucJnOoNqlBanr65npxMSJEjJ05LhdIlpBS79PukvB/LHkkPV7+0VPadSFYDHPoCgU6rhqFRu1oLmstp394bHhrikO1sL8OBAQtRgkP/HMpx6LKis1sfevGWeHl46jbVLup7GL+XLtsb0yGznzXdcw+P2UREwatABZNjYmLUT2QWG6WkpDhMY6ZQoUL231HvWK9z586qtjFKVSDgi8/x9f9ld/40qLesr7msQUZvMGf1opxIsH+H/IJtYQ1sB+tgW1gH28Ia2A7WUDqukIRnxUjxuEI8f8ol/TvXkKFzNvv8vjd6NZAHO1SXs0lpcsZejuPS70kYvDD90vPac7rfLVyaIyPLJifPp6mHr1Af2rEWNAYn9K4+dFJapir54mlwUMDrqDX+wEd/5figlOSIxwoiouBUoILJ+lIRdevWdXjNXYkJTYkSJVSZiKSkJClf3jlfAc8dOHBAzp49q4K+7kpTmP2/7M4fERERERHlnd6tqsioBdskOS3TYxATUIoYA1T2alVFIsNDpUxctHr4AgHTC2mZHoPQCEDbf9eeT0rzaj7zAjKGj51LVQ9fRYaFSFqm918MywB1tuesOiADutT0+f8REREVJAUqmIxSDx9++KH8+eefqr6xHp6D5s2bu3w/yizgM5YuXSoHDx50CvjiufDwcBV01v6f/rPN/p82TSDmj4iIiIiI8g6yWqcNbCE931+tsl3dBWoRSMb1xfRHW2QrGxafERMVrh4VS1zuSeltIPp8SobrrGezgDQGK0y6OKihVfkSSNZ79eut8ve+M1KheIxUKB4t5YtFS4XihdTvZYpGS1gQD0RIREQUKCE2rYhUAYCBRqpWrapqB2/ZssVeCgLZxPXq1VOBWgSKAaUq8Dym1WchT5s2Te6//3557LHHZOLEifbnv/nmGzUo3w033CA///yz/fk2bdqoGsrr1q1TdZUhIyNDrrnmGtm9e7fs379f/Q9f588XyGquXLmyCnYHa2Yzamph+aDmM8tcsC2I24SVcP9kHWwLa2A7WAfbIu8s2hwvfSevVdmuiD/qg8ra34WjwlQgOVjr9GZm2eScCjSbZUBffE498PulrGnt+aTUTAk2CCSXjYtSAebyxQo5BJsv/rz4fOHoApWv5Tfun4iIgpeljnS1atWS/v37q3rEOTH6IwKR48aNkwEDBkjr1q3V/0J94gkTJqg7+u+++6592rVr10rHjh2lb9++MnXqVPvzffr0kZkzZ8qkSZMkPj5eTYOgMALLCAK//fbbDv/z/fffl3bt2knXrl1l0KBB6nt98cUX8vfff8uUKVPsgWRf54+IiIiIiKypS4Myqv4uyiZ8vGSv7I5Psr9WvXSsPNKpuiptEcz1eRFcLVE4Uj18hRrPZz0Eoc+YBKExbV7Vh0bw/MiZFPUQOetyurhC4VJeCzAj8GwINlcsHi2likRJKLOciYgoSFkqMxkF+BE0jYyMlB49eqigatu2bQP+f7766isVtN20aZP6X/gfo0ePlkaNGtmnWbZsmWkwWRsMb+zYsTJjxgyVWYwM4i5dusjIkSOldu3aTv9vw4YNMnz4cFmxYoWkpaVJw4YN5fnnn1ff0d/58wUzkymQmEVgDWwH62BbWAfbwhrYDtbBtrAGXG6dTEiRw8dPScWyJaVU0Wh1zUP+LUvUhzYGnPVB6KNnLsiMlQctvXgjwkKkXBwCzeZZztrvhSLz78Dj3D8REQUvSwWTt27dqmoGI0iLQexwkoW6xMjQve+++xyyeMl7DCZTIPHEzxrYDtbBtrAOtoU1sB2sg21hHWyL3IPL22bDF8ueE0ni65VuuMoWtkkeJT87KR4bIRWKFboUdHYuqYGfJQtHBt3NCbTRiYQUOXL8lFQoW1JKW/gGS8Pnf5UDpy6o3z94oKnc07pKXs8SEVGes1SZC9QFRlmIN998U+bMmSMfffSRrFmzRp5++mkZOnSo/O9//1OBZdQbJiIiIiIiItJDULJfp+oydM5mnxfM6J715ZFONeRUYpocOXtBjqKsxdkUle188SfKXFz8HfWicxrKeuCx5XCCy2kiw0MdA8zFnDOe8YiKyPssZ2SOz151QD4xlH6pWSZWtVnvIC/9UpCg7Mu0P/bLrJUH5L+j51X5maqlYuSWpuXlqRuukDgf2xHrxnu/7JQf/zkq+08mq/W6ToUicnerKnJf26oOg1++/u12eeP7/zx+5t2tKsvkB6+y/306MU0+WLRbFm48LvtOJKna7YUiQqV6mVjp0qCsPHF9TSlZJMrHJUFUMFkqmKyJjo5Wg9zhgVIPyFZGnWKUm8AAeCj3MHDgQLnnnnskNjY2r2eXiIiIiIiILAJByVELtklyWqbD4IeuIE4VExmm6lijlnHpolHq0dhNEmpyaoYcPXuxhvLRsxcu/UyxB6HxOx4IuuUkBPEQfMPDHWQwI7PZ1cCB+B2Z0DmVIYxBKe+bvFYF8Iz/AlnkCP6jzaYNbKFqjpN1ZWRmyT2T1sovG487PP/f0UT57+hO+eavI/Lj4NZqvfIGbtDcPG6l7NHdYEhJz5K1u8+ox2+bjssXj7ZwCCh7Izws1P57woV06Tzmd4f/ARgkddPBBPX4eu0hWf5SB7/qwBMVNJYMJuuhvjAGu3vrrbfk5ZdfVgPcbdy4UQWTBw8eLA8++KDKWi5btmxezyoRERERERHlMWS3IijZ8/3VEio2twFlxKcQQJ3+aAufsmJjosKlZtnC6uEKAsknElKdspz1QWj8nnAhQ3Iasq3x2HTwnMtpUKPZ1cCBWhAatZ4jwi8H6bwNJKMttAqbxvIj2t8I/mO6uU+2DOqAcmJKhhSOtnyoxW/vL9xlDyRjnXntrvpSqkikuhmw63iS7D6eJE9N36Da0RuPT/3XHuStVTZWXuxeV06dT5MX521RNdJ//PeYTPpttzzZtZaa5t42VaRDvdJOn7PjaKI8Of1f+993tqho//2b9Ufs/yM8LERG3FFPGlSOk9U7T8nr313MckY5kwXrD8tDHapna/kQFQSW38Olp6fL119/rbKT//jjD/VcyZIlpU2bNvLrr7/Ke++9J9OnT5effvqJ5S+IiIiIiIhIBSMRzOo7ea3KPkTQWB9U1v5GRjICyZ3rBz54iUzKcgjAFosWqeZ6uvMp6aYlNfQZz8fPpXiVZZ0dCNwh4GbM3tRDVnHpIlEqsFyxhOOAgfrgc9FC4SpIj/IFyEhGINnT/ON1BP/RZlvGdc3VkhcIKo5csE3+3ndWoiNC5fqGZWXknfVdTj/ws79l1qqD9nIKd7SoJGO/3y6bDyZIzbKxsuKVjvbSCgiE/rLhmOw9kSQZmTa1PrS9spQ80bWm1KlQ1P6ZyC5vNPQ3+99rR3aSL1cflHlrDsmxc6lSpWQh6dexuvTvXMOrDPLU9Ez1fbxVq2xhlZHvDm6QfLh4j/3vZ2+6Qs0TlI+Llq5jV6jfUUpix9HzUrt8Ebeft/1IgizeEm//e9L9TaXlFSXV7xhYc9Q329XvH/y2Wx6/rqbqOVC5ZIx6GM1YecD+e9NqxaR93csBZ6yHmgaVisrj118MTHeoW1qm/r5fbWOQbpWC6UQWZ9lg8q5du+Tjjz9WpS1OnTqlDj4tW7aURx99VHr27CmRkZFy7tw5GTdunLz++uvy7LPPyooVF3dcREREREREVLAhoIyg5JxVB+RjQ53e6qVj5ZFO1VVpi7yu01skOkKKlI9wG3hDaYH4hFQVZD5sz26+9FOX8YwyEjkJWcSYDzw2HHCd5RwbdTHLGQFiX+YJ0yP4jzYb0KWm5IY/tp+UHu/+qUqGaEH1L1cfkj93npLzKZ6zxlftPCVz/jzoFCzffTxRlW/QApWai2VJDqgg8af9msmtzSqYfm7fD9fJtiPn7X8j6/f5OZvl4KkLMvp/DTzOFwLQN1wK7nrDmwEGtx1OkOPnUu1/X3sp8AtX1yiusn4RMIdl2054DCYv23rC/jvei8/QtKx1+bOxDLcfPS/1Kl4OvusdO5sic1cfsv/91A0Xg8UaBI0Rf8f6u/lQgkz8dZc9M1lrnyLR4XLLVeZtQUQWDiZjlOMFCxaogfeWLl0qWVlZEhMTIw899JAKIjdp0sRh+ri4OBk1apSsW7eOgWQiIiIiIiJygEAxgpLI5jyZkCKHj5+SimVLSqmi0TlWHzgnoP7rxZrHhaSZXA646SEBCyUzHGo4n3H8HT9PnE91KjURaAggI/jpD2SNI/jvbQZudmCZPf3Fv/ZAcpVSMfJy97pqALg3vtuuSh94su9Eshos7pkbr5BSRaPk8OmL7+n36V/2QGWlEoXkxdvrSNFCETLx190qAJ2akaUynBGQNcsIxk2DcXc3VAPbTft9vyr3ABN/2y09r60kjasUk9yG7Gq9snFRDutoidhIdaMB3GW4X/68ZIe63vo6x/rP1j7PVTAZA+tpbVijTKzcZggKN6oSJ1P6XS0vf7VFDp2+IMPnbnF4HQMHvnpHXdVORBRkweRKlSpJfHy82qFfeeWVqi4yBuErWtR8h6GpXLmypKQ43u0jIiIiIiIiAgQlMbBWSHq0FMfPIAokewvfKS4mQj305ROM0JX/2DktwHy5nIY90/nScxgELS8gwxdZ5GeS0nN8MLSNB885BL0/uL+ptK1TSv1ep3wRafHyEo+fgVIp3z/bSsrERduf23o4Qf7ae9Yh61cru9CqdklpMORXlYGNxzfrD0u/TjWcPvflHvXsJSQ61iujSmBowen5aw97DCYjCH3u09skkJINWeYRuuAvIAivrx3tSVLq5WkiDZ9lrM2d5OLzMLje1OX77H8/0bWWKodhVK5YlFQrFaOCyUZ//HdSlmw5IVeUc59JTUQWDCafPHlSunfvrrKQO3Xq5PX7hgwZIn369MnReSMiIiIiIiIKdgjSuao7q0GCF4K5CF4aM531Gc8Y1C+noMRETgeTdx27HEjG/YUWNS9nfV9ZoYjKbNfX2zVzTa0SDoFk2K4rT4HPbVmrhP3v4rGRUrdiUVm358zFaY9enlbvWt17EKRtVr2Y/PDPxexkb7K+c6JmckxUmMPfWjbw5f95+W9vBiGMjbo8DTK1HT7bcDMj1sXnfb58n5y7NIhlmaJRqoa10Zpdp+XWt1epEhxYp2Y82lyuqlZcNuw/K3dPWqvW4yGzN6nSLK7KjhCRRYPJ+/fvlwoVfN9wa9eurR5EREREREREFJhMbjzqVyrqNmB5MeB8OaNZG0zwwKkkWb/H+2CmEWrYBoNyhkCyVeREzWTUGtdD/WStLjIy3k8nXb65gHITnlQvffmGBt6L2uBaqQtkz+uZfR7+p35AQJRGiY5wDHjDZ8v32ms5/69lJWld+2L2OQb769mykkxedPEzFqw/wmAykRcstXf+9NNPVV3kW2+91e1033//vfzzzz/y8ssv59q8EREREREREdFlURFhUq10rHqYZTc3G75Y9pxI8qlGMyoUIGhZPDbnB0asWfbyfGMekS3c5sqLgcYdR897zEoGs4opV+oGnsPnrtl9WtrVuVjmAp+JgezMptVbveu0GiROC5rqs4z1852bkFGNWsbaIHwYpFArC4L5zdSNQtj+0vd1Ryv9AQj2rt19RpUBgVU7TtlfKxcXpcqOGM1dc0jdxNBuPjx8qSyI0YmEy0HuBEObntP97U1pDiKyWDD51VdfVTWSPQWTv/vuO/nss88YTCYiIiIiIiKyaHZzv07VZeiczT69D/HIRzpVz5W61o2rxKnA7O5LZSMem/qPvHR7XVUK5M3v//P7c5HNfVW1YvYA8GOf/yPDb68rRQuFy6Rfd6tayVA4Kky6X13R9DNGLtgmCM1WLRkj0/7Ybw+aQo/m5u/J6ZrJYaEh8kinGvLagm3q7/E/71Q1yEsViZQRX198Dq5rUEaVCdFgoMFZqw6q31GGYvKDV9mD0x3rlZalW0+ovx+f9o+81L2unDyfJu/+svPy+7vUdKqDjJsVExbusv/dt11VVZbETIPKRWXxlnh7ALpWucLStFox+Xf/OZm35pB9OrQZEQVZMNlbmZmZ+XLABCIiIiIiIqL8onerKjJqwTZJTstUQWJPEC/EgHa9WrkvtxAoiCu8c29juePdPyU90yb7TiTLQ5/8Za+/G1co3F6P11efPNxMbnlrpSoBcuDUBek/5W+H16PCQ1VQ1VWdYgwW99zMjU7PD+xSQ5pUzbug51Nda8nqnafkt83xciEt02keUY7i/b5NvP68ifc3lZveXCH7TyaroP79H653eP2GRmXl8etrOr1v4cbjsu1SbWrUlH7sOudpNE9cX0uVsDhwMlm184j5lwPfmlplY2Wgm88gossch8cMErt375ZixXjHiIiIiIiIiMiqkCk6bWALFbQ1JJY6weuYbvqjLVxmmOYElFr47tnW0uqKklIoMkziYiLk9qsryG/D2qrf/YXs11WvdpRnb7pC6lcsqoLkCHpWKVlI7m1dRZa/1N5tfd4pjzST57tdqabH+5BBPaZnA3n9fw0kLyFre84T18jb9zSSq2sUV9nV0RGhUrtcYRl04xWyZHg7qVC8kNefV6lEIbUs8N4ryhVWQXZ8Jj4b/2PW49fY6yjrva/LSr6rRUW3/xMB++UvtpfBt9RW2egoiYH1Df8Hfw/tdqUsGd4+V9c7omAWYkPfgDw0cuRIhzIXqJl8++23m06bkZEh27Ztk/nz58sNN9wgP/74Yy7OafA6dOiQVK5cWQ4ePCiVKlWSYM1GP3PmjBQvXlzCwpwL6hPboqDhNmEdbAvrYFtYA9vBOtgW1sG2sAa2Q95ZtDle+k5eq8o7IIinz1LW/kZgD4HkzvXLSEGEzNxGQ3+z/73xjetUqQoiIqvJ8zIXCCDj7iNi2vj577//qoc7ZcuWlTFjxuTaPBIRERERERGRf7o0KCNbxnWVOasOyMdL9sru+Is1igGD7aFGMkpbMDOUiMj68jyY/Pnnn6ufCCY/+OCD0qZNG3nooYdMp42MjJQKFSrItddeq34nIiIiIiIiIutDoHhAl5rSv3MNOZmQIoePn5KKZUtKqaLRHBOJiCiI5HkwuW/fvvbfp02bJjfeeKPDc0RERERERESUP6BHconCkRKSHi3F8TPEQzFlIiKylDwPJustXbo0r2eBiIiIiIiIiChXoT7yuU9v41InIstzHhKTiIiIiIiIiIiIiMhKmcmokYwuLRhMD4Pq4W9v4X1TpkzJ0fkjIiIiIiIiIiIiIgsEk6dOnaqCws8//7wKJuNvbzGYTERERERERERERFRAgslajeQqVao4/E1ERERERERERERE1pKnweT27du7/ZuIiIiIiIiIiIiIrMFSA/D16NFDHn300byeDSIiIiIiIiIiIiKycjD5p59+klOnTuX1bBARERERERERERGRlYPJlStXltTU1LyeDSIiIiIiIiIiIiKyepmL5cuXy5kzZ/J6VoiIiIiIiIiIiIjIqsHkl19+WWrWrCldu3aVdevW5fXsEBEREREREREREdEl4WIh3bp1k4iICFmzZo20bNlSypYtK1WrVpVChQo5TRsSEiKLFy/Ok/kkIiIiIiIiIiIiKmgsFUxetmyZ/XebzSbHjh1TDzMIJhMRERERERERERFRAQwm7927N69ngfLQsY9ekLT9201fs9lEMjIyJCk8XMzuI0RWrSPl+o/J+ZkkIiIiIiIiIiIqoCwVTEZJCyq4EEi+sH2922kycm1uiIiIiIiIiIiIyLID8BERERERERERERGRNTGYTERERERERERERETBVeZCM3fuXPn6669lx44dkpCQoAbjMxuAb/fu3Xkyf0REREREREREREQFjaWCyZmZmdKjRw/54YcfTAPIWhDZ1WtUcNkyM8WWmSEhYZZapYmIiIiIiIiIiPINS0XeJkyYIN9//73ceOON8t5778lrr70mM2bMkJSUFNm1a5f6/Z133pFBgwbJ6NGj83p2yUJSdv4jO+6qISHRsRIWW1RCY4pc/KkecRIWU0T9rr128fe4iz8v/a0eEVF5/VUs4dhHL6gBEc3gXk5GRoYkhYdLSIjz65FV60i5/mNyfiaJiIiIiIiIiKjgBpNnzZolcXFxMnv2bClatKiEhl4s6RwRESF169ZVAeS2bdvKzTffLPXr15e77747r2eZLMaWkiQZKUkip4769f6QiChd0BkB5iISpn5qv8eZBKUvToufIdExKns+2CGQfGH7erfTZOTa3BARERERERERkRVYKpi8fft2adWqlQokgxaUQ/mLsLAw9fsNN9wgzZs3l4kTJ/odTJ4/f768+eabsmnTJomMjFQB6jFjxkiDBg08vnfq1KnywAMPmL7WrFkzWb/+cgBu3759Ur16dbefN2rUKBk+fLj972rVqsn+/ftNp0XW9i233OJxHsl/tvRUyTx7Qj38EhpmGmi2/+6QCW3ImMajUBEJuXQThYiIiIiIiIiIyEosFUxG0LhEiRL2v2NiYtTPs2fPSsmSJe3P16xZU9VV9seUKVPk4YcfVoHjsWPHqhIaKK+BIPbKlSulYcOGXn3OCy+8oLKl9fTzCKVLl5YvvvjC9P0IIB84cEBuvfVWp9fq1KnjEGDWNG3a1Kt5ozyUlSmZ58+oh79UIPpSpjOyoe2/259zDEo7BKxjikhIRGRAvxIREREREREREZHlgskVKlSQo0cvlyeoWrWq+rlx40bp2LGj/fm9e/dKeLjvs37mzBl55plnpFKlSipwrGVA9+zZU+rVqydPPfWULFmyxKvPuu6666RDhw5up4mNjZV7773X6XnUfz548KC0bNnSNHhdtmxZ0/eRa+Ely0vh5tdJZlKCZCUnSFZSwsXf8TP5vCp/ESyyks+rR8bJw369PySqkK5m9OXsZ315Du01fW1prZxHSGR0wL8TEREREREREREFP0sFkxs1aiR//vmn/W8EkG02m7z66qty9dVXS5EiRWTmzJmyevVqadeunc+f/+2330pCQoIKKGuBZKhSpYrceeedMm3aNBXkrVy5slefl5iYqOo5R0X5Nmjbp59+qr5Xv379XE6DAc6Sk5PVd84PNXhzWkTpilL2kVEuX7dlZqgA7eUA88WfWUnnL/1+7lIg+vzlQLT993PqdzXyXBCwpV6QTDzOxPv3AeERQfNdiYiIiIiIiIiogAaTb7rpJlmwYIEsXrxYOnfurALIeO6nn35SJSQQWEXJCwRXX3nlFZ8/f82aNeonSloY4TkEk9etW+dVMPm2225TgWmoVauWCgwjSO0pYxpBYvwfBLP/97//uZxPZDWnpaWpUh9dunSRkSNHSuPGjb38pmQUEhYuYUWKq4c/bFlZkpWSpAs0O2Y/OwSqLwWfLwet8do5kcwgGbIuIz1bb7+wc4Psffo6+/IOK1Ls4s/Cl/8Otb+G5+NU+xARERERERERkbVZKoJzzz33qNIPKPOg+fLLL+X555+Xr7/+Wk6fPq1qHb/44osOZS+8dejQIfUTZS6MtOe0aVxBcBdlMRDgLV++vBw+fFjVRcY8/vHHHyr7OdTNAGoYRO/YsWMyYMAAFTA2QrkNDPCHesz4HAzoh8EGf/31V/XAYIGeIMitBbpBKx2CmtR4WFV2kmHx3hz/blExEopHiXI+vxWZ6La0FMfAMwLNl4LODtnQWta0frqkBPX+oJCZLmkH/vPpLarMRuHiEorAc+Fil35eDDar31XQWQtCX3wtJDom32ftx3/8ostlaRObujmVGB4uIeK8HCKrXCll3GTrU+Bg35OVlWXp/WtBwbawBraDdbAtrINtYQ1sB+tgWxDlX2FhYXk9C1SQgsnR0dFSv359h+cQcEUwFY/sQtkIMCtLgf+tn8YVBJLx0HvkkUfk7rvvljlz5sjcuXOlV69eLt//ySefqJ+uSlwgC1sP5Td69+4tLVq0UP9n27Zt4sn48eNlxIgRTs+fO3fONIBtFQiMZee9qIlteSGRIrElLz4MsLt1t8u1ZaSL7UKi2FAD+gKCzonq58W/L/+eZfj74nsSxZaSKFalBdTl+H7v3xQeISGo+xxbTEIK42ec+hkSW0xC1U/ja8UkBAMUBlEWdPKeLZKx+1+302QG+zaRTy6GUPYIN4144sS2IG4TVsL9k3WwLayB7WAdbAui/KtUqVJ5PQuUw4InqhIAyCqG1NRUp9dSUlIcpvGFVnYDweQffvjBZTAZ9ZgXLlwoV111lXp4C+UtunfvrrK0MXgfymq4g3IbDz/8sENmMoLRcXFxUry4f2UeckN6jfqS5qJMiJaFGe4mC9PK3y1wyvj9ThuyJy8kOpbguFSG42ImNLKjL2ZGJ637TbISz4qlIbh+7qRknjvpfxa0IeM5tLC1sqCTwsPF31ss2FYKxjZhjYshrB/FihVjMJltQdwmLIX7J+tgW1gD28E62BZERMGrQAWT9aUsUEbC2xIY3qhevbr6GR/vetCzzz77THWFRoZxdj7fUzAZ9Zj1AwxqkDFn5ay58gNfd3uygSxLBMes/B0sDcstsoRIXAmPkx54oYdc2L7er3+DLODIijUl8/wZ9cDghlYa0M+fLOiQ8MjLwedcrAWdnfg13sttJXCOffSCpO3fbvoaVm/c7LqAm10mbRZZtY6U6z8mgHND7qBElNWPdwUB28E62BbWwbawBraDdfy/vfuAj7q+/zj+uZHLIkAII2xB9kZZIshQ1DoQrVrFgVoBrXXXhXthqaNYrBXRitW6S2vFv7UsEXCwKqIyZIcRRggJ2bnxf3y+IUcu6wIkd7/LvZ6Px/dxd7+7+90v9+VC7n2f+3yZCwCITGENk0ePHn3c99UqMF2o71hode4rr7wiX3/9tYwZMybgOt2mBg4ceFzHs3HjRnOamlp5P10NkTVM1jYT2hKjtvcPWEVs287SbuqcwIporXo+Ei57Dh8qOc05dDRwLn/d4UxL9Yj2uYvEk7nPjGOugi4fQJuq6PKB9NGgOhp6QUciDZKDfcASIUtsAgAAAAAQmWHyF198cdz3PZ6wZdy4cXL77bebvsV33HGHv3p3x44d8uGHH8rIkSOlbdu2/t7Jul1bQ+hCe6UyMjIkJSWw361WpD3wwAP+x6iMLp6n+7vhhhskKSmp0ttUtm9VurBf7969pWPHjsf8cwPhZNMKwSNB6bHwFhYcDaBzKg+cy1+2ahV0cfoxVEHHxJqez/p8uQ/srtPjAwAAAAAAiJgweevWrSF9PG2R8Oyzz8pNN90kp59+ukyePNn0T54xY4YJp6dPn+6/7fLly2XUqFEyYcIEmT17tn+7BrrDhg0zpxoy79692/RK1oXxtFey9jaubuG96lpcvPXWW/Lqq6/Kueeea9pa6Nd+Vq1aZbZrL+fXX3+9Vp8PwMrssXFij20pMU2PfpgTTMUq6NIwOityqqCLC4+rCrq8/A2rZdN1/Uw4bYtxBZzaXXo+NvA6V6zYnS5zenRbXEmLD93mv+7I9WZb2X2VeYzS21NhDQAAAABAvRLWMLl9+/Yhf0wNkLX6V0Ple++9V1wulwwfPlyefvpp6dOnT9D7a4uKxYsXy8KFCyUrK8u0rdD7vfHGGyZ4riw82bt3r3zyyScmgB48eHCV+9YWG1qtPWfOHNm/f7+peG7durWpZr7//vv9fZMBVC6aq6Ar8HnFk30wrIdQIbA+EjTbKwm4Nbi2V9hWEkqXBNa6Le5oUH1kW8m+AgPuso8hDu1jTNsQAAAAAABqQ1QtwFfq0ksvNaM62vLCV0lQ9Nxzzx3z47Vo0UKKioqC3k6rpXUA4aYLhlWldLExZzWLjUWaaK2CDkWVtY6wstuPhs8mmI6rYcAdFxBYF2ekh/fnAAAAAADAAqIyTAZQvdTJU6u8zuPxSGZmpmkb43A4ovapDEUVdOHWH8MfxkY6r9cE+KUhvifcxwMAAAAAQAQLa5isi8np14/nz59vWjgcy+Jyer/NmzfX6fEBQDiroHdMuUTy1688rsdxNGwi8T2HHKkOLio5LSo57y2/zV1yirrhcxfz1AIAAAAA6oWwhsnbtm0zp8XFxQGXAQAnxtWqo7S+55Ua39609XEXVwyazfkC8QaEz0XlgunCCtcF7EdPy4TWlV5Xet4dvCVQpCnYtEa23z9OGgw8SxoMGCOudl3o4wwAAAAAiEhhDZO9Xm+1lwEAoWEWqYtxiUMXrQsjn7alcJcPszWALg2fS7dXUWVd5nbeSsJv//ZyAXf568TjrtWfq2DjajMO/P0PEtOirSQOOEsaDBwjCd0HlSwUCAAAAABABKBnMgDAMmy6YJ4rTkRHYviOQxdYPBpCF8jOZ34thZvX1sq+i/emyaFP3zDDnpAkif1HSoMBZ0niKaPEkdS4Vh4DAAAAAIC6QJgMAEAlCyzaHPEisfHmsj0mtk6eI2/eYTm87BMzxO6Q+O4DTcWyhsuuVh2YFwAAAACApVg2TPZ4PHLgwAEpLKx6Uah27dqF9JgAIJRc7btVeZ1pcex2i9PpFO1QcSz3RWjZExuJr6jAtNGoltcj+T9+Y8b+2U+Kq3UnSRx4pgmX47ucagJuAAAAAADCyXJh8uLFi+WJJ56QZcuW+Rfmq6q/pwYpAFBfpU6eWu0HbpmZmZKcnCwOQkZLi23bWdo8/JbkrlkiOSvnS+7KBeLJzgh6v6Jdm8zI/NdMcSQlS+Kpo0vaYfQfIfb4BiE5dgAAAAAALBsm/9///Z+MGzfOhMRNmjSRDh06SIMGvGEGAEQ2e3yiJA051wztx1yw6TvJWTFPclbMl6K0jUHv7zmcKdlf/MMMm9Ml8b2G+NthxDRrHZKfAQAAAAAAS4XJDz/8sKm2e+mll2Ty5MlU2wEA6l3LEW1XEd/1VDOaXX2/FKVvl9yV8024nPfTchFP9d+60YUB87770ox9sx6W2JN6lFQsDxwjcSf3NosYAgAAAABQ78Pkn376SYYNGya/+c1vwn0oAACEpOWIK7W9uC74tSRf8Gvx5GZJ7v8Wm4rl3NULxZubHfT+hdt+MiPjoz+Jo3EzaTDwLBMuJ/QZJvYjCwgCAAAAAFDvwmR9I966NV/XBQBEJ0diI2k4bKwZPnex5K9feaQdxjwpTt8e9P6eQ/sla967ZthccSZQLmmHcaY4k5uH5GcAAAAAANRflgqTzz77bFm6dKn4fD6zwB4AANHK5oyRhF6nmdHsuofNYnymYnnFPMnfuFrE6632/r6iAtM+Q8deEYnr3M/fDiO2fTf+nwUAAAAARHaY/MQTT8igQYPkd7/7nUybNs30nwQAINrpB6yxbTqbkXLxzeLOyjBtMEy4/N2X4ivIDbqPgp+/M+PAu8+Js2nrknYYA8dIfM/BYo+JDcnPAQAAAACIbJZKa9u1ayfLli2TsWPHyr/+9S8ZNWqUtGnTRuyVLCakb6x1wT4AAKKNs1GKNBp1mRne4kLJ/+GbknYYK+eL+8DuoPd3H9glhz570wxbXKIk9h9R0g7j1NHiSEoOyc8AAAAAAIg8lgqTdRGjP/zhD7Jhwwbxer2ydevWKm9LmAwAgJiqYg2DdTSf+KRZjK+0HUbB5u+DPkVa1Zzz9f+ZIXa7xHcbYNphaLjsan0yTzEAAAAAwJph8pNPPimzZs2SlJQUufrqq6VTp07SoEGDcB8WAAARQT9ojevQ04yml98u7oPpkrNygalYzvt+qfiKCqvfgdcr+T8tN2P/36ZKTMsOJcHyoDEmZLY5LPVnAwAAAAAgxCz1rvDNN9+U5ORk+e6776R169bhPhwAACKas0mqND77KjO8BXkmUNaq5ZxVC8RzaH/Q+xfv2SqZn8wyw96gkSSeMtr0Wk7sN0IciQ1D8jMAAAAAAKzDUmHy3r175ZxzziFIBgCgltnjEqTBoLPN8Hm9UrBpjalY1l7LRdvXB72/NydLDn/5TzPE4ZSEnkNKguUBZ4mrRTvmCwAAAACigKXC5JNOOsn0TQYAAHXHpr2Ru/Q3o9n4e6R4X1pJxbK2w/jxGxF3cfU78LhNlbMOef0xcbXr6u+zHNe5n9k/AAAAAKD+sVSYfN1118lTTz1lKpRbtGgR7sMBACAqxDRvK8nnX2+GJ++w5H232FQs56xaaCqSgynasUEO6pjzZ3E0aioNBpwpiQPHSGKfYaYiGgAAAABQP1gqTP7d734nK1eulFGjRsmMGTNk9OjRZjEhAAAQGo6EJEkaeoEZPo9b8tevKgmWV8wzPZSD8WQdkKwF75thi4mVhN6n+9thxKSkhuRnAAAAAABEQZjcqVMnc7p9+3Y5++yzJSYmRlJTU8VeyddlNWTevHlzGI4SAIDoYDO9kQeb0fy6h6Ro1+Yj7TDmSf76lSJeb7X39xUXSu7qhWbIzCkSe3JvfzuM2A49+cAYAAAAACKMpcLkbdu2BVwuKiqSHTt2hO14AADAUa7WJ0sTHeMmi+dwpmmDkbtyvuT+b7F483OCPlWFm9eakfH+H8WZ0tIEy4kDz5KEXqeJ3RXHUw0AAAAAFmepMNkbpMIJAABYgyMpWRqN/KUZvuIis3Bf6SJ+7v07g97fnbFHDn3+lhm2uARJ7HuGqVhOPHW0OBulVHm/9JlTpGj7+kqv8/lE3G635DqdUlmXLFf7bpI6eeqx/aAAAAAAAGuGyQAAIPLYYlyS2O8MM5rf+LgUbl9vKpa1z3LBz98Fvb+vIE9yvv2PGZoCx3U5xfRZ1nDZ1aZzQDsMDZJNi41quGvlpwIAAAAAlEeYDAAAao0Gv3EndTcj5dJbxX1w79F2GGuWiK+ooPod+HxSsGGVGQfeniYxLdqVVCxrO4zug5gpAAAAAAgjwmQAAFB3f2g0aSGNx1xphrcwX/LWLjMVyzkrF4gnc1/Q+xfv3SGZc183w57QUGxO/nQBAAAAgHAJ6zsyu91uxk8//SRdunQRh8NxTJVP2hcRAABEBntsvFl0T4fP65WCzWv97TAKt/0U9P7evOyQHCcAAAAAwIJhcrt27UwoHBMTYy63bds2oC8iAACon2x2u8R37mtG0yvvluJ9OyVn1QITLOf98LWIuzjchwgAAAAAsFKYvG3btmovAwCA6BDTvI0k/2KCGZ68w5L33ZeSo32WVy0Uz+HMWnkMb0GuePNzxB7foFb2BwAAAADRhsaDAADAUhwJSZI09HwzfB6P5G9cZSqWc1cskKJdm457v4Xb1snPV/UQR3JzcbU+uWS06ug/jWnWRmzH0HILAAAAAKJNRIXJGRkZ0rhx42PqrQwAACKXhrsJ3QeZIdc+KNt+d74Ubll7QvvUhf/ydWg7jbKPFRMrManty4TMHcXVSkPnjuJo0PgEfxIAAAAAiHyWCpNXr14t//nPf+Tiiy+W7t27+7fPnTtXJk2aJHv37pWGDRvK1KlT5eabbw7rsQIAgNCzu2LrbN++4kIpSttoRnmOhikB4XJpRXNMi3Zic5as/QAAAAAA9Z2lwuSXXnpJ3n77bbnxxhv923bu3CmXX365FBQUSHx8vGRlZclvf/tb6devn5x22mlhPV4AABAdPNkZkq9j3YrAKxxOEyiXbZdRGjo7GqWwsDAAAACAesVSYfI333wjffv2lebNm/u3zZ492wTJDz/8sDz++OOyaNEiGTNmjLz44ouEyQAAoMaczdtKYr8zpGj3Finatdm0uzhhHrcU795iRu7K+QFX2RMblqlkPnoa07K92F1xzBwAAACAiGOpMDk9PV1GjRoVsG3evHkSFxcn9913n7ms1w8ZMkRWrVoVpqMEAACRKKZJC0m96Rn/ZU/eYRMCF+3a4g+YzenuLeIrKjjhx/PmZkvBz/8zI4DdLjHNWpcEzLrwX+uTJdYsCNhRHMktqGYGAAAAYFmWCpO1Ajkm5mjfQbfbLStXrpTBgwdLQkKCf3v79u3lf/8r98YMAADgGDgSksTRqa/EdeobsN3n9Yo7Y8/RcNmEzZvNqfvArhN/jr1eKd6bZkbu/74IuMoWl1hpb2Y9tcfGM78AAAAAwspSYXKbNm1k3bp1/stLliyR/Px8GTFiRIXQuWy4DAAAUFtsRyqHdWhbjLK8hflStHurP1zWsLn4SOjszc854cf2FeRK4ea1ZpTnbNqqXG/mklPdrscMAAAAAFEVJg8bNkzefPNNeeGFF+Sss86Shx56yHzV8/zzzw+43Y8//iitW7c+7seZM2eO/OEPf5C1a9eKy+WS4cOHy9SpU6VXr15B76s9nK+//vpKrzv11FNNJXVZ1113nfmZKnP33XfLc889V2H7l19+KY899pisWFGyyM/AgQNNv2g9TgAAopmrfbcqr/P5Sr7V5HQ6xWY7tvvWlFYHx3XoYUbgY/tMD+aAdhlHKpqL96WZauQT5T6w24y875cGbLe54sTVssPRnsxlKpu1+roupM+cIkXb11c7D7nVzEPq5Kl1clwAAAAAoihM1r7I77//vtxzzz3+N2Znn322CVNLbd68WTZu3CiTJk06rsd4/fXX5cYbbzTB8bRp00yV84wZM2To0KGybNky6d27d432M2XKFOnevXvAtpSUlCpv/9Zbb1XY1qNH4BtR9fnnn8sFF1xgwvJHH31UYmNj5dVXX5XRo0fLZ599ZkJ2AACiVXUhpMfjkczMTElOThaHwxHS49IPv51NWpiR0Ou0gOu8xYVSnL49oF1GaejszTl0wo+t/Z0Lt68zozxH42YVKpn1NKZ5G7E5jv/PQA2S89cHfoBenvu49w4AAADAqiwVJnft2tUEutOnT5cDBw6YSt977723woJ8ffv2lQsvvPCY969vMO+66y7TTkMfp2HDhmb75ZdfboLd22+/XRYuXFijfY0ZM0ZGjhxZ48e++uqrg95G3wTfdNNNJkDW6uR27dqZ7ddee6307NlTbr75ZtmwYYPY+SorAAARwx4TK7Ftu5hRnjv7oAmVdSHAwl2bpLg0aE7fLuI58TjWc2i/5Ov48ZuA7TanS2JS25XpzXzktPXJ4khKPuHHBQAAAFA/WSpMVv369TOtJKqiYauO4/Hxxx9Ldna2CZRLg2Sloe2ll15q2lGkpaVJ27Zta7S/nJwcs2Cghr/BaJX14cOHJTExscpqKe0RvW3bNtMaozRIVo0aNTLV1Nrq4quvvjLtQAAAQORzNmxihnQ/+i0s5XMXm/YYpmVGaUXzkdYZnqwDJ/y4PneRFO3cZEZ5joZNjlYwl61obtFObDGuE35sAAAAAJHLcmFyXfr222/Nqba0KE+3aZisfYprEiZfdNFFJphWnTp1kokTJ5qQWvs0VqZx48bm9hokDxgwwLT0uPjii4/p+NTy5cuDhsn6OKXHpvbs2eOvfNYRifS4vV5vxB5/fcJcWAPzYB3MhXXUq7mw2cXRor3E6zhldMBVntxsKT4SLpdWMhfv3irF6VvFV1x0wg/tyT4o+TrKt7GwOySmRVsTMBftTTvu/WtP5XoxRxGgXr0mIhxzYQ3Mg3UwF0D9Fep2cwi9qAqTd+7caU61zUV5pdtKb1OVhIQE0xZDexe3bNlSdu3aZfohazislcVa/Vy2DUWLFi3k1ltvNQGyBsra71l7NF9yySVmEcDS/tC1dXxKFzDUKubysrKyTGV0pP6xoZXgWuHNLybmArwmrITfT9YRVXPR9CQz7H1GS5yIGT6vR7wH08WTvk08e7eLJ327ePZuE2/6dvEe2nfij+n1SPGebWacCF2cT1uPoe5F1WvC4pgLa2AerIO5AOqvpk2bhvsQUMeiKkzOy8szp5W1pYiLiwu4TVU0SNZRli4GOH78eHnvvffkgw8+kCuuuMJ/nS7yV5626ejfv788+OCD5ralldC1cXxKK6S1LUbZyuRBgwaZdhm6KFGk/rGhixtpIM+bIeYCvCashN9P1sFc6GrATUU696rw3Hjzc6Voz1bTm7mkknmL/7yvMD+k86Tf4orUv0ciDa8J62AurIF5sA7mAgAiV1SFyVpVrAoLCytcV1BQEHCbY6Eh56OPPmrC5Llz5waEyZVp0KCB3H333WZBvc8//9wf/NbW8Wk/6LI9oUtpCBvJQaxWfEf6z1BfMBfWwDxYB3NhHcxF5RwNGkpM574iOsrQilV3RnpJT+bS3sx6umuzuA/sKulJUcvcGbukcMNKie9yiticMbW+fwTiNWEdzIU1MA/WwVwAQGSKqjC5bKuI7t27B1xXXYuJmujQoYM53bdv33HfvrpWFid6fAAAAJV9IB7TtKUZiX0C12TwFhYcqWYuFzTv3iLevMPH/WS6D+yRtIcuE1tcoiT0HCwJfYZJYt/h4mrbxRwPAAAAAOuKqjBZWz288sor8vXXX8uYMWMCrtNtauDAwNXUa0p7IavU1NTjvr0eX+mx6IJ+lR1f6W0AAADqkj02TuJO6m5G+WrmHfeNlYJNa05o/76CXMldtdCM/Vo93biZJPYdJgl9hpuAOSalZn9TAQAAAAidoyvFRYFx48ZJUlKSzJo1S7Kzs/3bd+zYIR9++KGMHDkyoH/x+vXrTb/hsjIyMipdSOaBBx7wP0ap3Nxcf3uKsvbv328W39PeyOeee65/+xlnnCHt27c3fZfT0o6ukq7H+tprr0nHjh1l6NChJ/w8AAAAHC+tHq6L9hSeQ/sle/E/JX3GXbJl4iDZetto2fvaI5KzYp54TqASGgAAAEDtiarKZF3s5dlnnzUL4J1++ukyefJk0594xowZ5o3R9OnT/bddvny5jBo1SiZMmCCzZ8/2b+/du7cMGzbMnLZs2VJ2795teiWvW7fO9Eq++OKL/bf9+eef5ZxzzpGLLrpIOnfubBaP04rkv/71r2YV85deeklatWrlv732An755Zdl7NixMnz4cLntttvE5XLJzJkzJT09XT799FPTVwoAAKC+K9q5yYxD/zdbxO6QuM79TDuMhL7DJL5zf/otAwAAAGEQVWGy0gA5JSXFhMr33nuvCWs1uH366aelT58+Qe8/fvx4Wbx4sSxcuFCysrIkMTHR3O+NN94wwXPZXn/awkLD5KVLl5pqY61U1sfWCuQ777zTnJZ33nnnybx58+SJJ56QRx55xN96Y/78+TJixIhafjYAAABCy9WuqyQNPldy1yyRgk3fiXi9we/k9UjBhlVmZHww3d9v2YTLfYbRbxkAAAAIEZtPG9+hXtPF+7R9h7bOiNQF/Dwej6nm1upyreAGcxHteE1YB3NhHcxF6OyYconkr195XPeN7zZA2k2dY857crMl78dvJG/NUsn7fokU7dp8XPt0JDc3CwhqsEy/5aN4TVgHc2ENzIN1MBcAELmirjIZAAAA1uBIbChJg842QxUf2C153y+T3O81XF5q+ijXhCdzn2QvnmOGcrXtbEJlDZjjew4RR0JSnf4cAAAAQLQgTAYAAMAxcbXvVuV1+p03XZzY6XRKme5fNbpvTNNW0mj0ZWbol+eKdmwoCZbXLDEVzL7C/BodX1Haz2Yc+vQN0285vkv/kqpl+i0DAAAAJ4QwGQAAAMckdfLUOv/qsq5DEdu+mxlNLrxRfMVFkr/xfyZY1oD5WPota0sOHf5+y72GlLTF6DtcXG06B6x5AQAAAKBqhMkAAACwPFuMyyy6p6Pp+N+JJzdL8n78tqRq+fulNe637CvIldyVC8wo3285se8wcTZJreOfBAAAAIhchMkAAACIOI7ERhX7La9ZerTfctaBE+u33He4JPQcIvb4BnX6cwAAAACRhDAZAAAAEc/0Wz7zcjO033Lh9vUmVDbjePstO5wS37nfkX7Lw815mzOmzn8WAAAAwKoIkwEAAFCvaA/kuJO6m9Fk7MQj/ZZX+yuXa9xv2eOuvN+yVi33GUa/ZQAAAEQdwmQAAABEQb/lIWb4+y3/8I2/cvl4+y07m7QoqVrWthh9TqffMgAAAOo9wmQAAABEX7/lweeYcSL9lt0H90r2F/8wo7TfcmKf4ZLQdxj9lgEAAFAvESYDAAAgqgX0W/Z6pXDHhhPqt5z56V+P9ls+0hKDfssAAACoDwiTAQAAgCNsdntAv2VvcaEUbPzfifVbfv+PYo9vIPE9td9ySVsMV5vOprczAAAAEEkIkwEAAIAq2GNia6Xfsjc/R3JXzjejYr/lYeYyAAAAYHWEyQAAAMDx9lvev0vyvl9WC/2Wu5hQ+Vj7LafPnCJF29dXep3PJ+J2uyXX6ZTKiqBd7btJ6uSpNXocAAAAQBEmAwAAAMcpplnryvstr1kieT99ewz9ljea4e+33KV/SdVy3+ES16mv2Jwxld9v+3rTSqM67uP6yQAAAICKCJMBAACAuu63vGaJFGxeU/N+y+tWmFGh33Lf4eJq3Yl+ywAAAAgLwmQAAAAglP2W1ywxbTGKd285sX7LfYeLz13M3AEAACBkCJMBAACAsPVbXlrSb3nNUvFkZxxXv2VYA/2rAQBANCBMBgAAAMLWb/lXZvj7LWuvZQ2Xj6HfMqyB/tUAACAaECYDAAAAVuq3fNEkf79l7bWsVcs17rd8DArTfpZdf5gs9th4scXGi90VV3LqvxwrNld8FdfHHb3sihdxxtDHGQAAIAoQJgMAAAAW7rcs4+8p6be89mt/W4ya9luujjc3S3K++axWjlfsDrG54kzQXBo2l172B9LmcplAWi+74stcjj0aXPvvp/sps8+YWEJrAACAMCJMBgAAACKh3/KQc80o2295/1vPiCf7YLgPT8TrEV9Brnh01OXj2GzlQurSsLm0SrpsWK0BdGy5wPrIff23LR9ol+47zlSLAwAAIBBhMgAAABCh/ZazFrwv+VYIk0PF5zO9pD0h6CetldIVq6fLVFeb7Uev14AfAACgviNMBgAAAKKR0yWOhAbiLSooWezP5wv3EVmKr6jQDK9khftQAAAALIMwGQAAAIhC8Z36SLupc8x5n1b8FheKr7DAHy57C/PFV5QfeLmS60u26eWS7b6igiO31fuWXj56H22JgUCe3GzxFuSJPS6BpwYAAFgaYTIAAAAQ5WxHehGLK04cdfxYPndxYNhcWHjktDSgLhtGF/jD6orXl26veL1u87mLJFIUpW2Un6/tLfFdTpGEPqdLYp9hEtepr9icMeE+NAAAgACEyQAAAABCRgNSh4akiQ3r9HF8HvfRqmhTOV0uoK5QPX0kmK5BdbX/enNdYe0csLtY8n/61oyM914QW1yiJPQcLAl9hklC79Mltn03E/oDAACEE2EyAAAAgHrH5nCKLb6B2OMb1Onj+Lxe0yIk7ZFfScHP39XefgtyJXfVQjOUo1FTSeg91ATLWr3satGu1h4LAACgpgiTAQAAgAjlat+tyut0PT232y1Op1MqK2it7r6oOZvdLrbYeBNe1yVP1gE5vPTfZqiYFu1MqGzC5d6ni7NRSp0+PgAAgCJMBgAAACJU6uSpVV7n8XgkMzNTkpOTxeGo607IOBHOZm3E1aqD5K9bYVpo1ETx3h2SNU/Hu+ZybPvuJS0xNGDuMVjs8YlMCgAAqHWEyQAAAAAQRjEpqdL20b+Lt7hQCjaslrzvl0nu90ulYNMaEa+nRvso3L7OjMxPZok4nBLfuZ8/XI7v3F9sMa46/zkAAED9R5gMAAAAABZgj4mVhF6nmdF0/O/Ek3dY8n/8xh8uF6VtrNmOPG7JX7/SjIwPpostLkESug8qaYnRd5ipYtb2HAAAAMeKMBkAAAAALNi/2pGQJA0GjjFDuTP3Sd7aryRv7TLJXbNU3Ad21ejYfAV5kvu/L8ww+23YpCS0Nov5DZOY1PZiq+zAAAAAyiFMBgAAAIAI6F/tTG4uDc8YZ4bP55Pi9O2S9/1SEy5ryOw5nFmj/XiyD8rhrz41w+y3aWtJ7DvsyGJ+Q83jAAAAVIYwGQAAAAAijFYSu1qeZEbjc64Wn9dreiZrSwwTMP/0rfgK82u0L61wzlrwvhnK1baL6bWc2GeYxPccYiqkAQAAFGEyAAAAAEQ47YEc16GnGU0umiS+4iLJ//l//nA5/+fvTC/lmtDezDoOffqGiN0hcZ36+sPluK6nmN7OAAAgOhEmAwAAAEA9Y4txSUKPwWbIFXeJNz9H8n5aXhIur10mhdt+qtmOvB4p2LjajIMfzRCbK07iuw80LTES+5wusR16ie0EWncAAIDIQpgMAAAAAPWcPb6BNDh1tBnKfeiA5P1QspifBszFe3fUaD++ogLJW7PEjAO63waNJKHXUH+4HNOqI4v5AQBQjxEmAwAAAECUcTZuKg2HjTVDFe3d4a9aNov5ZWlUHJw3J0tyvvnMDLPflJamJUZpuOxsklqnPwcAAAgtu0ShOXPmyJAhQyQxMdGsqDx27Fj54YcfanTf2bNnm0/aKxsDBgwIuG1BQYG89tprcskll8jJJ58s8fHx0rZtWznvvPNk0aJFle7/pJNOqnL/c+fOrZWfHwAAAADKcrVoJ43HXCmt7npJTv7rKjnpj/+VZtc/IokDzhRbXGKNnyx3xh7JXvSRpP/pTtl84yDZeuto2TvrYTn87efiyc3iSQcAIMJFXWXy66+/LjfeeKP06tVLpk2bZgLfGTNmyNChQ2XZsmXSu3fvGu1nypQp0r1794BtKSkpAZe3bdsmEydOlNNOO02uvfZaEySnpaXJzJkzZfTo0ebx77333gr77tatmzz44IMVtvfv3/+Yf14AAAAAOBZayBLbvpsZTS68UXzuYinYtEZyj1Qu529YJeIurtG+inZtMuPQZ2+K6CKBHXv7K5fjuw0Ue2wckwMAQASx+Xw+n0SJzMxMU/nbsGFD+fHHH82p2rFjh/To0UMGDRokCxcuDFqZfP3115vK4pEjR1Z724yMDNm+fbuccsopAdv37NljwuycnBxJT0831dGl9Ph0fPHFF1Jbdu7c6Q+y27RpI5HI4/GY+dPnysECH8wFeE1YCL+frIO5sAbmwTqYC+uob3PhLciT/HUrJHftUtMao3DrjyLH8bbSFhMr8d1OlYTew0zAHHdyb7E56q7eqb7NQyRjLgAgckVVZfLHH38s2dnZctddd/mDZNWuXTu59NJL5c033zSBqwavNaFhcExMjMTGxlZ6vVYql69WVi1btpQRI0bIP//5T9mwYYNpuVGe2+2WvLw8SUpKYgELAAAAAJZhj0uQxP4jzFCew5mmz3Ju6WJ+e7bWaD++4kJzPx3yjog9oaEk9BpiqpY1XHa16cx7IQAALCaqwuRvv/3WnGpLi/J0m4bJK1asqFGYfNFFF5lgWnXq1Mm0s9CQ2ums2VO6a9cuc9qiRYtKj1P7ORcVFUlCQoKcddZZ8sQTT0jfvn1rtG89rtJjK62ELv30V0ck0uP2er0Re/z1CXNhDcyDdTAX1sFcWAPzYB3MhXXU+7nQEHjwuWao4v27JP+HryVv7VLJ18X8Du2v0W68edmSs/y/ZihHcnNJ6DVU4rUlRu+hEtO0VdB97Hv1ISnasaHS63ziM0U7OU6n2MRW4XpXu67SfNJTNTpWnJh6/5oAohjf/Kj/oipM1nYPqrJWD6XbSm9TFQ13L7/8chPwaoWxhsJvvfWW3HfffbJkyRJT/Wy3V7+u4SeffCLLly831ckdOnQIuE7bbWgbDe3HrPtZuXKlvPTSS/Lf//7XjOHDhwf9OV944QV5/PHHK2zPysoyIXUk0j8ytBJcu7Lwi4m5AK8JK+H3k3UwF9bAPFgHc2EdUTcXzgSRfmdKbL8zxeXziWfPVilev1yK1y0X94aV4ivIqdFuPJn75PCSf5mh7M3bSUz3QRLTbZDEdB0g9gaNK9wnb8uP4t78XfX7rWK7Bs3aBgN1L+peE0AUadq0abgPAXUsqnomn3nmmaYn8ubNm6Vjx44B1+l2vf6ZZ56R+++//5j2q0/h+PHj5b333pN3331Xrrjiiipv+9NPP8npp59uvq6lQXH546jMmjVrTD9nve26deuOqzJZ768LAkZyz+RDhw5J48aN+WODuQCvCUvh95N1MBfWwDxYB3NhHczFUT6PWwq3rDWtLbRquUDD5eKiY39SdZHAk3qaimUzug8Se2y87HzoMrPP4xHXdYC0eerD47ovjg2vCaD+4gOi+i+qKpO1qlgVFhZWuK6goCDgNsdCg+FHH33UhMlz586tMkzW/sgaWOvXeT7//PMaBclK21tcfPHF8v7778umTZtMW43qaD/osj2hy76gI/lFrZXakf4z1BfMhTUwD9bBXFgHc2ENzIN1MBfWwVwc4XCIs9sASew2QOSy28RbWCD561dI3pF+ywVb1op4vcGfUJ9PCrf+YMahf78q4oyR+K6nijtj93HPkc1GCBJKvCYAIDJFVZhctpWFtpGoaQuMmihtV7Fv374qK5I1SNbQet68eaZS+Hj3HyxMBgAAAIBIYI+Nk8S+w81QnpxDkvfjt5L3/VITLhft2lSzHbmLJf/Hb+r2YAEAQHSFyRrgvvLKK/L111/LmDFjAq7TbWrgwIHHte+NGzea09TU1ArX/fDDDyZI1h5cCxYskFNOOaVW9w8AAAAA9YGjQWNJGnyOGao4I72kalnHmqXiPpge7kMEACCqVb9SXD0zbtw4SUpKklmzZgX0FN6xY4d8+OGHMnLkSGnbtq3ZlpeXJ+vXrzf9hsvKyMiosF8NiR944AH/Y5T1/fffy6hRo0xrC+3LXF2QXNm+VenCfr17965xawwAAAAAiHQxKanSaOQvpeWtL0jHWd9KhxmLpPnEp6TBkF+IvUGjWn2soj3b5MB7z8uh//5dclYukIItP4j70AHx1aTtBgAAUSKqKpOTk5Pl2WeflZtuusksgjd58mTTP3nGjBmm7/H06dP9t12+fLkJgSdMmCCzZ8/2b9dAd9iwYea0ZcuWsnv3btMrWRfG017J2tu4bEg9evRoExLron5r1641oyytkG7RooU5/9Zbb8mrr74q5557rmlroT2kVq1aZbZrL+fXX389JM8TAAAAAFiNvmdztT7ZjORfXCs+j8f0TM79vqRyOX/dcvEVVVwfp6Y8WQck44MXK17hjBFn4+bibNKizEgVZ0qqOR+j55u0EHt8gxP7AQEAiABRFSYrDZBTUlJMqHzvvfeKy+WS4cOHy9NPPy19+vQJev/x48fL4sWLTZVxVlaWJCYmmvu98cYbJnjWP3BKbdmyxV9t/Pvf/77S/S1atMgfJmuLjS+++ELmzJkj+/fvNxXPrVu3lhtuuMGE0aV9kwEAAAAg2tkcDonr1NeMlEt+I97iQtl+zwVStGND7T6Qu1jcB3aZUR0NkwPC5spOk5uLzRlTu8cHAEAI2Xw+ny+UD4jQ08UFtX1HWlracS8wGG4ej0cyMzNNdbnD4Qj34UQ15sIamAfrYC6sg7mwBubBOpgL62AuQmfHlEskf/1KsSybTRwNU45UNmu43KKkwrncqSMpOaBQqb7hNQEAkSvqKpMBAAAAAAgLn8+009ChLTqqYnO6KrbVKH+akir22HixqvSZU6Ro+/pKr9OSNv0mbq7Tqfl6Ba723SR18lQJt82Th4p7/05zPvW3z0uj0ZeF+5AAIOwIkwEAAAAAUS+u66nS6u6XxX0wXdwH91Z56s07XOfPlc9dJMX70syojj2hYeUVzmVD58ZNxeYI/Vt/DZKDVYm7Q3Y0qE3arzxr/ruStegjKdr5s/n3GtO8rTQYfK40GXeTOBIbHtP+PLlZcvCfr0jO8s/Nv3n9MMXVtos0GnWpNDrzCtPSpjI5qxdJ1n/fkfyNq8WTc0gcCUniTGkl8V1PkaZX32culzr4yWvm32Ph5rUBrys+JACOHWEyAAAAACDqaVuJmKYtzaiONz9X3JkaLu8Vd0Z6yfkjp8V6qtsz95pey3XNm5ctRTrSfq76Rna7OBo19S8W6PAvGhhY7Wxv0Khet9ZA7fB53LJr2kTJXbkgYHvRzk1ycOdLcvirT6Xdk++bf1c1oa+ZtEcul+I9244+RlGhFGxYZUbu6kXS6p6ZAYGyz+uV9JfvleyFHwTsy5N90Ayt+m8ybnJAmJzx/nTzegFw4giTAQAAAACoIXt8orjiO4qrVccqb6NLE2moVV2Fsw5td1HnvF7xZO4zQ6syq2JzxQZWNvvPl2m3kaytNeLq/pgjmH7YoP9G6quDH8/0B8k2V5w0m/CQOBs1kf3vPCfFu7dI8Z6tkv6X+6XNg7NrtL/0P9/jD5JjWnWUZuN/J+7sTNk/+0nxFRVIzvL/SuYnr5lwuFTGhy/6g2Rd+LLxL66VuC6niN0VJ8UHdkv+uhVii4kNeJzY9l3Nazb25D6S8f4fQ/PaA+opwmQAAAAAAGqRVvg6G6WYIR16VHk7X3GRuA/trzxwziipcNbKTV9Bbp3Pj1aDFu/dYUZ17A0aB1Q0H612PtrfWSuhI0neuhVy4O9/kIJNa0xA2uDUUdLsmilV3n7PjLske9FH5nzDUZdKw2EXyYEPpkvhtp/E1bKDnPTCf8x1nsOZpr1C7or5UrR3u648KM4mzSW+52nS5KJJEtu2i3+f2nphy02n+y+f9KcFkr34n5L95b/MBwHO5m0k+RfXSuPzrq9RBbm3uND8PDXlatnRtEQJ1t4ic+5f/ZdTLr3VHJNyJLeQtAd/ac7nrloohTs3SWybTtXurzBto+R9t9h/OfWWZyWh+8CS4885JAfeedacz5z7miSPnSg2u108hw/JwX+94r9P6ylvSELPwQH7bXzWFRUeq93T//CfP/jPv1R7XACqR5gMAAAAAKgXdOG2qpQu+uasZtG3ULPFuCSmWWszquPNzylpoVHaUqNchbM5n7lPxFP3XYg15CvSsWND1TeyO0x7jUiQ98PXsvOJa0zfX6XVsBri5v20wjzvweT/tFyyF88xFeBlFe3eato36PyUVbw3zYzDSz6Wlnf+SZKG/KLS/e5+9mYpStt49H67t8i+1x+T4n27pPn1Dwc9Lg2g0x68VGqqJr2DNfz1HNrvvxx/JPg157v0F9He3Ef+DeZ9vzRomKy38XM4S/ZRur9uA/zn9TnUVi5aXZyzeqH4CvPNdmezNpK3ZonsfeV+Kd63UxyNUiTx1NHS9Fd3irNxsxr/7ACODWEyAAAAAKBeSJ08tcrrPFpVmZkpycnJ4qhiQS+r0q/yazBXXTinfWQ92RlVVjiX9nXW9ht1zuspGRan7UjSX3nAHyTrInJNx99jQv4D7/9R3Pt3Bt2HVnK72naWJpfcIs5GTU2bBbXnxdv9QbKzaStpeuXvxJ7YUDL//aoJoH3FhbLnT3dJfLeBlVYEuzP2SPMbnzDHpIvdabsHlfnJLGk44mKJ69hLQq04fXvA5bKBrS7y6EhK9ofNZXsgV6WozP70vmUXinSUC4OL0reZMLlw2zr/Np2fjI/+dPTygd2S9fnbpjK6/e//VeO+zQCODWEyAAAAAAARTlsAaLhnAr5qgkZtf6BVq1rp7Cm7aGC5ELq0+rM+K9z6o6n4LZX62+ckoddp5rwGxNtuOzPoPmyx8dL28fcCgtXC7Ruk4Ofvju731uclsXdJC4uEHoNk86TTTOsSHYe//lSSfzGhwn6bXnWfv4VEYt/hsuXm0/3h9OFlnwQNkzWE7jqn+pYlx8pbmBe4wRkTcNFW5rK3oNxtK+ErOPpvzOZ0Vbmvkv2VtHrx5mYFbNd5ajr+XvHmZMm+N580pxoq7//7s9Ly1udr8mMBOEaEyQAAAAAARAl7TKzYm7c1YWN1FbvevMPVVjibU61CjYAK5KoUlQmStfdJXNdT/Bdj23QWe4NGJpysjrZjKN9SoXDnxoD9lm3Z4GjQWGLbdZWCjatLjiHt58r3W6aFhFZKx3XqJznLPz9y3FvD0jPZHptQoed3VZftcYG3rYwtLv7ofY9Uh1d12R5XsqihLSZwAcjmEx6WxFNGmvPFWqn8wXRzPnf1oqCPD+D4ECYDAAAAAAA/XeDNkdjQjLKLxFW2IJsn64A/dC4+uFc85jRdcr79j3hzs+v9s+pMbi5WVBc9k2NS2wc+hn6YcKT1is9dbBYc9N+25UlBH9NVZn96X5/H7W91YXqAB9y2ZH8xzdsEHlOLtpWej4Z/e0C4ECYDAAAAAIBjZnM4xNmkhRnl7di9RfLXr7T0s+pq2eHoBZ9PCjb+TxJ6DjEXC3duClqVbFSymqNWNQfsd/0qSeg91Fz05GZJYZnFC11V9MHW5y7upO4lu3AXB1QZu2oQ1NYF/WBBexmX9kXO+2m5vy2ImesyVeoJfUraelQn4UjrD8PjlvwNqyShx+CS/f34rf8qR3Jz087C3KdXyfyU0oX3XK1P9p+vLFgGULsIkwEAAAAAQNSJ7dhLYlp2kOI9JW0j0l/6XckCfM4YyfjgxePfb/tuEteprz8A3qP7vfJucSQkycFPXjO9kpUtLlGShl5Q6T4O/H2aCaK1Ejdr/num+rtU0ukXBj2GuuiZrB8eJJ93vRx45w/m8sE5fzYL5zkbNZH9b0/z3y7xlFEBgfqeGXdJ9qKPzPmGoy6Vlre+YM5ru4+EvsMlb80Sczn9z/dKs6vuEXfWQTn4r7/47598wa9NT3AVd3If0wIkf90Kc3nfm09J0+JCE/xnfvqG/z5Jw8cFHHvud1+K90gf8LL9wAu2rDULIyoNpatb5BJACcJkAAAAAAAQle08Wtw0VXY+ea2Iu1iK9+6QPX+81VynFbj2hIbizTu+dgkt73hR0h75VUnP6f07Jf1PdwY+dkystLzthSr7FMe0aCf7Zj1UYXvy+TdI3Mm9JVyajJss+etXmJ7EvqKCCseo7S1a3Hw0WA4m9ZZnJe3hy6R4b5oJ9Xc/95uA6xMHnClNxk4MvM+tL0jaw5eLO2OPFO3YILt/f2PA9VoF3uTimwK2pf/lfjMP5R36v9lmqJTL75DYK+6q8bED0arkox0AAAAAAIAok9j7dGn72DsS32OQ2Fxxpko1aej50u6Zf/orVo+Hq1VHOemP/5Umv/ytuNp3E1tsvNicLnE2ayMNR18u7Z+dK0lDflHl/VvdOUNSLr/d3F7vpxXUza5/RJrd8KiEk1Ztt37gdWk+6WmJ69zfVFfbXLHiat1JmlzyG2k/7d8Sk5Ja4/3FNG0l7Z/91NxXK4M1ZNd96r71MVrf95q/j3LZXsvtn/s/SR47yTwvR+/TT5r/+nFp88jbZqFJAHXD5tNlWlGv7dy5U9q2bStpaWnSpk1gs/pI4fF4JDMzU5KTk8XhcIT7cKIac2ENzIN1MBfWwVxYA/NgHcyFdTAX1sA8hNaOKZccd8/k+G4DpN3UORJNivelyZabjvYQ7vjKMtOqAgCshjYXAAAAAACgVmk1blW0pM3tdovT6axs/bpq7wsACC/CZAAAAAAAUKtSJ0+t8jqqxAEgctEzGQAAAAAAAAAQFJXJAAAAAAAAYaT9kbvO2cEcALA8KpMBAAAAAAAAAEERJgMAAAAAAAAAgiJMBgAAAAAAAAAERZgMAAAAAAAAAAiKMBkAAAAAAAAAQJgMAAAAAAAAADhxVCYDAAAAAAAAAIIiTAYAAAAAAAAABEWYDAAAAAAAAAAIijAZAAAAAAAAABAUYTIAAAAAAAAAICjCZAAAAAAAAABAUITJAAAAAAAAAICgCJMBAAAAAAAAAEERJgMAAAAAAAAAgiJMBgAAAAAAAAAERZgMAAAAAAAAAAiKMBkAAAAAAAAAEBRhMgAAAAAAAAAgqKgMk+fMmSNDhgyRxMRESU5OlrFjx8oPP/xQo/vOnj1bbDZbpWPAgAGV3mf79u0yfvx4adasmcTHx0u/fv3ktddeq/IxvvzySxk9erQkJSWZoeeXLFly3D8vAAAAAAAAAJwop0SZ119/XW688Ubp1auXTJs2TQoKCmTGjBkydOhQWbZsmfTu3btG+5kyZYp07949YFtKSkqF2+3cudME11lZWXLHHXdIhw4d5OOPP5aJEyfKrl275NFHHw24/eeffy4XXHCBtG7d2lwXGxsrr776qgmUP/vsMznrrLNO8BkAAAAAAAAAgGMXVWFyZmam3HXXXdKmTRsTHDds2NBsv/zyy6VHjx5y++23y8KFC2u0rzFjxsjIkSNrFDqnp6fLP/7xD7nkkkvMNg2StRr6qaeekmuuuUY6duxotns8HrnppptMgKzVye3atTPbr732WunZs6fcfPPNsmHDBrHbo7KgHAAAAAAAAEAYRVUqqRXB2dnZpjK5NEhWGtpeeumlsmjRIklLS6vx/nJycqSwsLDK6/Py8uSjjz4y1cilQXIpDbXdbre88847/m3aymLbtm1y2WWX+YNk1ahRI3PMmzZtkq+++uoYfmIAAAAAAAAAqB1RVZn87bffmlNtaVGebnvzzTdlxYoV0rZt26D7uuiii0wwrTp16mSqjTUgdjqPPqVr166V/Px8Oe200yrcX7dpn+Xly5fX+PiU3n7YsGHVHpseV+mxqT179vgrn3VEIj1ur9cbscdfnzAX1sA8WAdzYR3MhTUwD9bBXFgHc2ENzIN1MBdA/eVwOMJ9CKhjURUma/9ipW0uyivdVnqbqiQkJJi2GNq7uGXLlqbv8VtvvSX33XefqSzW6ufSNhTVPZ62smjatGnA49XG8akXXnhBHn/88QrbtW+zLjoYqX9saCW4z+fjFxNzAV4TlsLvJ+tgLqyBebAO5sI6mAtrYB6sg7kA6i/NulC/RVWYrG0nSoPc8uLi4gJuUxUNknWUNWnSJBk/fry899578sEHH8gVV1wR9PFKH7Ps49XG8SmtkNa2GGUrkwcNGmTaZSQnJ0uk/rGhldyNGzcmTGYuwGvCUvj9ZB3MhTUwD9bBXFgHc2ENzIN1MBcAELmiKkzWqmJVWZ/jgoKCgNscCw05H330URMmz5071x8mV/d4pY9Z9hOb2jo+7Qddtid02a8aRPLXDbTiO9J/hvqCubAG5sE6mAvrYC6sgXmwDubCOpgLa2AerIO5AIDIFFUL8FXXKqK6FhM1oYvsqX379tXo8TQwPnDgQMDj1eXxAQAAAAAAAMCJiKowWVs9qK+//rrCdaXbBg4ceFz73rhxozlNTU31b+vdu7dpT1HZ433zzTem/2/pMdX0+MreHgAAAAAAAABCJarC5HHjxklSUpLMmjVLsrOz/dt37NghH374oYwcOVLatm3r7028fv1602+4rIyMjAr7dbvd8sADD/gfo5S2pPjlL38pW7dulTlz5gTc5/nnnxen0ylXXnmlf9sZZ5wh7du3N32X09LS/Nv1WF977TXp2LGjDB06tFaeCwAAAAAAAAA4FlHVM1kXn3v22WflpptuktNPP10mT55s2k3MmDHD9D2ePn26/7bLly+XUaNGyYQJE2T27NkB1cbDhg0zpy1btpTdu3ebXsnr1q0zvZIvvvjigMecOnWqzJ8/X6655hpZtWqVaYfx8ccfm97KDz/8sJx88sn+22ov4JdfflnGjh0rw4cPl9tuu01cLpfMnDlT0tPT5dNPPzV9pQAAAAAAAAAg1KIqTFYaIKekpJhQ+d577zVhrQa3Tz/9tPTp0yfo/cePHy+LFy+WhQsXSlZWliQmJpr7vfHGGyZ41lC6rHbt2pkWFVOmTDGhcE5OjnTp0sWcnzRpUoX9n3feeTJv3jx54okn5JFHHvG33tBAesSIEbX4TAAAAAAAAABAzdl82rgX9Zou3qftO7R1RqQu4OfxeCQzM9NUl2sFN5iLaMdrwjqYC+tgLqyBebAO5sI6mAtrYB6sg7kAgMhFzwQAAAAAAAAAQFCEyQAAAAAAAACAoAiTAQAAAAAAAABBESYDAAAAAAAAAIIiTAYAAAAAAAAABEWYDAAAAAAAAAAIijAZAAAAAAAAABAUYTIAAAAAAAAAICjCZAAAAAAAAABAUITJAAAAAAAAAICgCJMBAAAAAAAAAEE5g98Ekc7tdpvTPXv2SKTyeDySlZUlubm54nA4wn04UY25sAbmwTqYC+tgLqyBebAO5sI6mAtrYB6sg7kA6rfU1FRxOokc6ytmNgrs37/fnA4aNCjchwIAAAAAAIB6LC0tTdq0aRPuw0Adsfl8Pl9d7RzWUFBQIGvXrpVmzZpF7CdDWlWtYfjy5culZcuW4T6cqMZcWAPzYB3MhXUwF9bAPFgHc2EdzIU1MA/WwVwA9RuVyfVbZCaLOCZxcXEycODAevGsaZDMp1vWwFxYA/NgHcyFdTAX1sA8WAdzYR3MhTUwD9bBXABA5GEBPgAAAAAAAABAUITJAAAAAAAAAICgCJMRERo2bCiPPvqoOQVzAV4TVsLvJ+tgLqyBebAO5sI6mAtrYB6sg7kAgMjFAnwAAAAAAAAAgKCoTAYAAAAAAAAABEWYDAAAAAAAAAAIijAZAAAAAAAAABAUYTIAAAAAAAAAICjCZAAAAAAAAABAUITJAAAAAAAAAICgCJNhWb///e/lV7/6lXTu3Fnsdrs4nc5wH1JU+vnnn+Wxxx6T008/XVJTUyUxMVF69Oght912m+zZsyfchxdVDhw4IDfccIP07dtXUlJSJC4uTjp06CBXXHGFrF69OtyHF9W8Xq8MGTJEbDabnHXWWeE+nKiiz3lV44cffgj34UWd7Oxseeihh6R79+4SHx8vTZo0kcGDB8vbb78d7kOLGvp/dnWvCx27du0K92FGzevhySeflN69e0tSUpL5v3vgwIHy5z//WYqLi8N9eFFl//79csstt0j79u3F5XJJq1atZOLEiZKenh7uQ6uXjvV9nNvtlmnTpknXrl0lNjbWzM/NN98sGRkZITtmAEDN2Xw+n+8Ybg+EjL7Zady4sfTv31/WrVtn/gjUPzQQWvfff7+89NJLcuGFF5qwTMOBb775Rv72t79Jw4YN5auvvpJu3boxLSGwadMmufbaa+W0004zb4Y02N+2bZvMnj3bvBmaO3eunHPOOcxFGLzwwgvy6KOPSk5Ojpx55pkyf/585iGE/1cMHz5cJk2aVOE6/b3VqFEj5iJENKAcNWqU+eDruuuuk549e0pubq5s3LhRWrZsKQ8++CBzEQLff/+9GeVt377dBP2nnHKKrFq1irmoY/o3q/5/rR/2TpgwwXyoUlhYKP/4xz/kyy+/lKuuuooPWUJE30Po869/M5X+HbV161YT6jdt2lS+/fZbad68eagOJyoc6/u4a665xrweLrjgAhk7dqyZn+nTp8vJJ59s3nfo37wAAAvRMBmwok2bNvnPjxgxwudwOMJ6PNFqxYoVvszMzArbZ86cqR9E+S677LKwHBeO2rVrl3l9jBo1iqclDDZv3uxLSEjwTZ8+3bwmzjzzTOYhhPQ5nzBhAs+5BYwePdqXmprq27FjR7gPBZV46KGHzOvllVde4fkJgXnz5pnn++677w7Y7na7ff369fPZ7XZfdnY2cxECd9xxh5mLqVOnBmxftmyZz2az+SZOnMg8hPF93IIFC8z8jB07NmD7Rx99ZLY//vjjzA8AWAxtLmBZ+kk0wm/AgAGmsqA8ba2gKqt+Qmi1aNHCVIwfOnSIpz4M9GuyWoF566238vyHkX5l/PDhw8xBmCxbtkwWLlwo9913n7Rt21Y8Ho+p1Ic16Hy88cYbprpv/Pjx4T6cqJCVlWVO9ev6ZTkcDtM2TE+13QLqnv5uUtdff33A9qFDh5o2DO+++64UFBQwFWF6H6ffdlR33XVXwPZf/vKXctJJJ/mvBwBYB2EygONS2m9Rg0yEPjTTr5Fra4vly5ebYEBDG/1qIEJr1qxZ5uvKeqo9AREeH330kflARVvv6IdfV199tfk6M0Ln008/9QcIGgDofGiPWA3SnnrqKRNmInw+++wz8/+29jDVeUHd07UmNLzX3rEffPCB7Nixw7R80dfD559/Lo888ojpDYu6p+1FVEJCQoXrdJv+DUWP/fDRNiP6N5S20ytPW5Js3rxZDh48GJZjAwBUjhXNAByXhx9+uNIqD4SmAlD7kpbSnrBaDahvTBE6u3fvlnvuuUfuvvtusygiwvftCQ0vu3TpYgKDpUuXmnBfwzN9rdDTPTS0J6b69a9/bRYGfe2110zPzJdfftn8f6FB2quvvhqio0F5+ppQlfUWR93Q6uOPP/7YLCKmIX4pXTz39ddf5++nENJvD23YsMFUKI8bN86/XReSXr9+vTmvv6P0/xOE3s6dO03v6so+XGnTpo3/NrqgKwDAGgiTARyzqVOnmgVk9A9yXVQGoaXB5bx580xwplVOb731lvl6v14Otlo2ao8GBPrmRxfeQ/isWLEi4PKVV15pqvTPO+88ueOOO+Q///lP2I4tmpS2GNFKTK3WLw0FNETr0aOHCZf1g5euXbuG+UijjwZmWjneu3dvswgZQkc/7NV/8yNHjpSzzz5b8vLy5M033zTtkfTDFl2oEnVP2yeUBvv6t5JWwOqClPqBsNfrNbfRuUF46HOfnJxc6XX64UvpbQAA1sF3cgEckxdffFEefPBB88bo73//u3kzhNDSP7jPOussOf/88+XOO++UBQsWyCeffCKXXnopUxEi7733nvz73/+WV155xXydH9byi1/8woRm+tqgD2ZolL4OtO1O2eoy7Ql71VVX6YLPsmjRohAdDcrSXsnaZkQDTITOmjVrZNiwYaYqVqvy9f/oa6+91nwYPHDgQLnllltk//79TEmIWo58+OGHpk+1rvmhfXj179j27dvLjTfeaG6jbZIQHtpqpLQVSXml/4dX1qIEABA+hMkAauyFF14wlX5nnnmmqXLiDzvrhMtjx441FZj0ia17+obntttuM1Vm+oZ006ZN/qHy8/PN+b1794bgaFAVbbXgdrvpsxgiuuieatmyZYXrSrfR8zL0NMTXlgoa9l9zzTVhOILo/vBd/7+47LLLArZrb1gNlrXSUtc9QGhcfPHFphpZF45evHixpKWlmV7WpYF+9+7dmYow0VYWuhZIZYGytrcovQ0AwDoIkwHUyLRp08xXlM8991yZO3cuQbLFaICpMjMzw30oUfFc65vP//73v2YV+LJDffXVV+b87bffHu5DjWraAiYmJkZSUlLCfShRoXThJA1oyivdxoKtoafV+Vu2bDHhpS5OidAvVFzZ4pP6QVfZU4SGViZru5czzjhDWrdubcJL7aNc9v9whN6gQYNMuxFdiK+8r7/+2izsSr9kALAWwmQANeqRfP/995s+pP/617/8/csQWlVVumo1ss6L9maksqbuaU9Y/bpsZUPpG1U9ry1IULcyMjIq3f7uu+/K6tWrzYdflS3og9p30UUXmbCytId7qZycHNMjVoN9reZHaGmvakWLi9DT9hZq9uzZAduLi4vlnXfeMcEmC76F15QpU8z/Iw899FCYjyS6lX5r4vnnnw/YPmfOHPM3Lt+qAADrsfn0+2+ABekbUv06mtKvaOoqy48//rj/ev7wC40///nP8tvf/tZUlD3zzDMmECirQYMGAStjo+5oixHttagLi2l7Be1XvW7dOvnb3/7mD2yuvvpqpiCMdE60Dcz8+SVQFb0AAAaESURBVPOZhxDQwH7ZsmUyevRoadeunRQVFZnLukBoamqqLF26VDp27MhchIj+LtJFWXXBsV//+tfm9fDXv/7V/J56+umnTXCD0NGvjWv1pb4GdA4QWvp366mnnmrm4cILL5RzzjnHtLZ4++23TasFXRSufHiGutOtWzfTEqxTp07mG0b//Oc/TbuL3/zmN+ZvXYT3fZz229cPgrVwRT+c3Lp1q/zxj380Lau0YlnfbwAArIMwGZalC2PoH3lV4XOQ0NCVxjWkrIouXkKf3tDQgHLmzJmycuVK2bdvn/l6rPYi1YVltKWCfk0Q4UWYHFq6COJf/vIX+eGHH0xgo/8v6ActujjlfffdJ82bNw/xEeGzzz6T3//+97Jq1SrztWWt1NfQXxe9QujXOdD2VBpYanCJ0NO/j5588knzQfCePXvMgpS9evWSSZMmyQ033MAixiH+e1Y/YNT2IzoP/fv3N4sglu9pjfC8j9OK/eeee84sGKohtLa10PBfP4hs2rQp0wIAFkOYDAAAAAAAAAAIip7JAAAAAAAAAICgCJMBAAAAAAAAAEERJgMAAAAAAAAAgiJMBgAAAAAAAAAERZgMAAAAAAAAAAiKMBkAAAAAAAAAEBRhMgAAAAAAAAAgKMJkAAAAAAAAAEBQhMkAAAAAAAAAgKAIkwEAAIBaMnv2bLHZbPLYY4/xnAIAAKDeIUwGAAAAAAAAAARFmAwAAAAAAAAACIowGQAAAAAAAAAQFGEyAAAAIpr2KD7ppJOkuLhYpk6dKj179pT4+Hjp16+fud7r9cq7774rV111lXTt2lWSkpIkMTFRevfubXob5+bmVrtfvf/06dPNfuPi4qRFixZy3XXXyd69e4/pON9//32JjY2VZs2ayfLly2vlZwcAAABCyRnSRwMAAADqgAa+F198sSxYsEBGjBghvXr1kqKiInNdXl6ejB8/XpKTk6V79+4mZD58+LCsXLlSHn/8cZk7d64sWbLEBNCVueaaa2TOnDlmv926dZOvv/5a3nzzTVmxYoWsXr3aBMTBvPjii3LnnXdK+/bt5fPPP5cuXbrU+nMAAAAA1DXCZAAAAES8tLQ0cTgcsm7dOlNNXJbL5TJh8Pnnn2/Ol8rPz5dbbrlF3njjDRP23n///RX2u337dnE6nbJ+/XoTBKvs7Gw566yzTJj83nvvyYQJE6o9Nt3vtGnTpE+fPvKf//xHWrZsWWs/NwAAABBKtLkAAABAvfDMM89UCJKVBshatVw2SFZaifzSSy+ZsPgf//hHlfv905/+5A+SVcOGDeWee+4x57/44osq7+d2u007DA2SzzjjDPnyyy8JkgEAABDRqEwGAABAvXDRRRdVe71WLWuLic2bN5s+ydoaQ2nIvHHjxkrvo0Hz2WefXWG7trtQu3fvrvR+uv+xY8fKZ599Jpdccom88847NWqHAQAAAFgZYTIAAAAiXvPmzavseawVwpMnT5a//vWvx7xfbUmhgXJ5uoifKiwsrPR+umCfPq5WJH/44Ydit/OFQAAAAEQ+/qoFAABAxKsqSC4NdjVI7tGjh/z73/821cS6OJ/P5zOjuh7GxxsCn3feeWbBP13YT3syAwAAAPUBYTIAAADqtY8++sicvv/++3LhhRea8DgmJsbfjiI9Pb3WH7N///4yf/58EyhPnDhRZs2aVeuPAQAAAIQaYTIAAADqtYMHD5rTtm3bVrhOexlrdXJdOOWUU2TBggXSpEkT02Zj5syZdfI4AAAAQKgQJgMAAKBeK10sb8aMGQHbV65cKQ888ECdPna/fv1k4cKF0rRpU7n55pvl5ZdfrtPHAwAAAOoSYTIAAADqtfvvv9+0tXj44YdNuHvllVfKiBEjZPDgwXL22WdL+/bt6/Tx+/TpYwLlZs2ayS233CIvvfRSnT4eAAAAUFcIkwEAAFCvDR06VL755hs599xzzeJ7ugjfoUOH5Pnnn5e33347JMfQq1cvWbRokaSmpsqtt94qL774YkgeFwAAAKhNNl9dNYkDAAAAAAAAANQbVCYDAAAAAAAAAIIiTAYAAAAAAAAABEWYDAAAAAAAAAAIijAZAAAAAAAAABAUYTIAAAAAAAAAICjCZAAAAAAAAABAUITJAAAAAAAAAICgCJMBAAAAAAAAAEERJgMAAAAAAAAAgiJMBgAAAAAAAAAERZgMAAAAAAAAAAiKMBkAAAAAAAAAEBRhMgAAAAAAAABAgvl/9GqT5m/xni8AAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 1320x550 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visual: score curves side by side\n",
    "fig, ax = plt.subplots()\n",
    "ranks = np.arange(1, 11)\n",
    "ax.plot(ranks, [c.score for c in results_a[:10]], marker=\"o\", color=ACCENT, label='Query A — \"Iceland\"')\n",
    "ax.plot(ranks, [c.score for c in results_b[:10]], marker=\"s\", color=WARN, label='Query B — \"tropical + snorkel + vegetarian\"')\n",
    "ax.set_xlabel(\"rank\"); ax.set_ylabel(\"similarity score\")\n",
    "ax.set_title(\"Query A — strong top-1 match (Iceland)   |   Query B — weak top-1 match (no clear winner)\")\n",
    "ax.set_xticks(ranks); ax.legend(loc=\"upper right\")\n",
    "drop_a = results_a[0].score - results_a[9].score\n",
    "drop_b = results_b[0].score - results_b[9].score\n",
    "ax.text(10.2, results_a[9].score, f\"  drop = {drop_a:.3f}\",\n",
    "        color=ACCENT, fontsize=12, fontweight=\"bold\", va=\"center\")\n",
    "ax.text(10.2, results_b[9].score, f\"  drop = {drop_b:.3f}\",\n",
    "        color=WARN, fontsize=12, fontweight=\"bold\", va=\"center\")\n",
    "ax.set_xlim(0.5, 12.5)\n",
    "ax.set_ylim(0.49, 0.69)\n",
    "plt.tight_layout(); plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "97d347b5",
   "metadata": {},
   "source": [
    "## What the chart actually says\n",
    "\n",
    "The blue line (Query A) sits a full **0.10 above** the orange line (Query B) at every rank. Same retriever, same corpus — only the question changed, and the entire retrieval distribution shifts down.\n",
    "\n",
    "We'll come back to that **top-1 score** in the CRAG section. Treated as a confidence signal it lets the system tell good queries from bad ones *before* anything is sent to the LLM.\n",
    "\n",
    "For now, the simpler observation: the retrieval scores tell you whether your retriever found something good, *independently* of how the LLM phrases the answer."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "69644902",
   "metadata": {},
   "source": [
    "## Takeaway\n",
    "\n",
    "> **Basic RAG bottlenecks on retrieval, not generation.**\n",
    "\n",
    "Query A's top-5 is a single destination. Query B's top-10 spans **ten different destinations** with no clear winner. Hand the LLM the wrong context and it will produce a confident, fluent, *wrong* answer — and the LLM is not the problem.\n",
    "\n",
    "Three constraints in one vector compresses the signal beyond what the embedding model can represent. Each technique that follows attacks that compression from a different angle."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d8eef772",
   "metadata": {},
   "source": [
    "---\n",
    "\n",
    "# Reranking\n",
    "\n",
    "### *Two models in series. The fast one for recall, the slow one for precision.*"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "33c39bd8",
   "metadata": {},
   "source": [
    "## The mechanism\n",
    "\n",
    "```\n",
    "query ──► bi-encoder ──► top-N candidates ──► cross-encoder ──► top-k re-ordered\n",
    "            (fast)                              (slow, accurate)\n",
    "```\n",
    "\n",
    "The bi-encoder we used in the previous section embeds the query and each chunk **independently**. It never reads them together — that's why it's fast and pre-computable, and also why it loses nuance.\n",
    "\n",
    "A **cross-encoder** reads each (query, chunk) pair **jointly** and outputs one relevance score. It can't pre-compute anything because it needs the query, but for the small handful of candidates we ask it about it can spend serious capacity.\n",
    "\n",
    "The pattern: **bi-encoder for recall, cross-encoder for precision.** Two-stage retrieval is one of the highest-ROI changes you can make to a RAG system."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9f65eeaa",
   "metadata": {},
   "source": [
    "## When reranking helps — Query C\n",
    "*\"Cheap nightlife in Bangkok with live music\"*\n",
    "\n",
    "Bi-encoder top-5 returns chunks that *sound* nightlife-shaped. The cross-encoder reads each one and ranks them by whether they actually answer the question."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "7a488068",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      "Loading weights:   0%|          | 0/105 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      "Loading weights: 100%|██████████| 105/105 [00:00<00:00, 11868.43it/s]"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  Bi-encoder top-5 (similarity):\n",
      "    1. 0.665  bangkok        Drink     RCA, the **Royal City Avenue** strip, is home to much more than nightc…\n",
      "    2. 0.660  bangkok        Do        Of a completely different nature are Bangkok's famous transvestite sho…\n",
      "    3. 0.631  singapore      Drink     For a night out Singapore style, gather a group of friends and head fo…\n",
      "    4. 0.618  thailand       Drink     thumb|Go-go bar in Soi Cowboy, [[Bangkok]]  Thailand is infamous for i…\n",
      "    5. 0.582  chiang-mai     Get in    In the following some specific destinations: Bangkok thumb|Feet up on …\n",
      "\n",
      "  Cross-encoder top-5 (relevance):\n",
      "    1. ce=+3.95   thailand       Drink     thumb|Go-go bar in Soi Cowboy, [[Bangkok]]  Thailand is infamous for i…\n",
      "    2. ce=+2.65   bangkok        Drink     RCA, the **Royal City Avenue** strip, is home to much more than nightc…\n",
      "    3. ce=+2.07   sydney         Drink     Many pubs are called hotels, but only very few can ever offer you a pl…\n",
      "    4. ce=+1.05   bangkok        Drink     thumb|220px|Sirocco at State Tower, [[Bangkok/Silom|Silom]]  Bangkok's…\n",
      "    5. ce=-0.79   jakarta        Drink     Jakarta may be the capital of the world's largest Muslim-majority coun…\n",
      "\n",
      "  Mean cross-encoder relevance of top-5:\n",
      "    Basic order:  -1.40\n",
      "    Rerank order: +1.79\n",
      "    Lift:         +3.18\n"
     ]
    }
   ],
   "source": [
    "from sentence_transformers import CrossEncoder\n",
    "reranker = CrossEncoder(\"cross-encoder/ms-marco-MiniLM-L-6-v2\")\n",
    "\n",
    "candidates_c = vector_search(QUERY_C, n=30)\n",
    "ce_scores_c = reranker.predict([(QUERY_C, c.text) for c in candidates_c])\n",
    "results_c_rerank = [c for _, c in sorted(zip(ce_scores_c, candidates_c), key=lambda x: x[0], reverse=True)]\n",
    "\n",
    "print(f\"  Bi-encoder top-5 (similarity):\")\n",
    "for i, c in enumerate(candidates_c[:5], 1):\n",
    "    snippet = c.text[:70].replace(\"\\n\", \" \")\n",
    "    print(f\"    {i}. {c.score:.3f}  {c.doc_id:13s}  {c.section:8s}  {snippet}…\")\n",
    "\n",
    "print(f\"\\n  Cross-encoder top-5 (relevance):\")\n",
    "for i, (s, c) in enumerate(sorted(zip(ce_scores_c, candidates_c), key=lambda x: x[0], reverse=True)[:5], 1):\n",
    "    snippet = c.text[:70].replace(\"\\n\", \" \")\n",
    "    print(f\"    {i}. ce={float(s):>+5.2f}   {c.doc_id:13s}  {c.section:8s}  {snippet}…\")\n",
    "\n",
    "# Mean cross-encoder score: how relevant is each top-5 according to the cross-encoder?\n",
    "top5_basic_ce = [float(ce_scores_c[i]) for i in range(5)]\n",
    "top5_rerank_ce = sorted(ce_scores_c, reverse=True)[:5]\n",
    "print(f\"\\n  Mean cross-encoder relevance of top-5:\")\n",
    "print(f\"    Basic order:  {np.mean(top5_basic_ce):+.2f}\")\n",
    "print(f\"    Rerank order: {np.mean(top5_rerank_ce):+.2f}\")\n",
    "print(f\"    Lift:         +{np.mean(top5_rerank_ce) - np.mean(top5_basic_ce):.2f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "cb08d62b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1320x550 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visual: cross-encoder scores in basic order vs rerank order, top-10\n",
    "fig, ax = plt.subplots()\n",
    "top10_basic = ce_scores_c[:10]\n",
    "top10_rerank = sorted(ce_scores_c, reverse=True)[:10]\n",
    "x = np.arange(1, 11)\n",
    "width = 0.4\n",
    "ax.bar(x - width/2, top10_basic, width, color=NEUTRAL, label=\"Basic ranking\")\n",
    "ax.bar(x + width/2, top10_rerank, width, color=ACCENT, label=\"Rerank ranking\")\n",
    "ax.axhline(0, color=\"black\", linewidth=0.8)\n",
    "ax.set_xlabel(\"rank\"); ax.set_ylabel(\"cross-encoder relevance score\")\n",
    "ax.set_title(\"Reranking lifts answer-bearing chunks to the top\")\n",
    "ax.set_xticks(x); ax.legend(loc=\"upper right\")\n",
    "plt.tight_layout(); plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "750c5c95",
   "metadata": {},
   "source": [
    "## What the chart says\n",
    "\n",
    "The right-hand bars descend monotonically — that's reranking working as designed. The cross-encoder placed the most answer-bearing chunks at the top, with a clean falloff into negatives.\n",
    "\n",
    "The left-hand bars are jagged. The bi-encoder put a chunk worth +4 at rank 4 and a chunk worth −9 at rank 7. **Same chunks, same pool, completely different orderings.** Mean relevance of the top-5 went from −1.40 to +1.79 — a lift of **+3.18**. That's a meaningful improvement in what the LLM actually sees as context."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2c51d910",
   "metadata": {},
   "source": [
    "## When reranking *cannot* help — Query B again\n",
    "\n",
    "Now the same cross-encoder on Query B's candidates — **the same 20 chunks**, just reordered."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "0edf543a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  Recall at top-k against 21 relevant destinations:\n",
      "      k     Basic R@k    Rerank R@k\n",
      "      5          0.05          0.05\n",
      "     10          0.10          0.05\n",
      "     20          0.14          0.14\n",
      "\n",
      "  R@20 is identical: same 20 chunks, just reordered — the candidate pool is the ceiling.\n"
     ]
    }
   ],
   "source": [
    "# Same 20-chunk pool. Rerank can only change ORDER, not membership.\n",
    "ce_scores_b = reranker.predict([(QUERY_B, c.text) for c in results_b])\n",
    "results_b_rerank = [c for _, c in sorted(zip(ce_scores_b, results_b), key=lambda x: x[0], reverse=True)]\n",
    "\n",
    "print(f\"  Recall at top-k against {len(GROUND_TRUTH_B)} relevant destinations:\")\n",
    "print(f\"    {'k':>3}    {'Basic R@k':>10}    {'Rerank R@k':>10}\")\n",
    "for k in [5, 10, 20]:\n",
    "    rb = recall_at_k(results_b, GROUND_TRUTH_B, k)\n",
    "    rr = recall_at_k(results_b_rerank, GROUND_TRUTH_B, k)\n",
    "    print(f\"    {k:>3}    {rb:>10.2f}    {rr:>10.2f}\")\n",
    "print(f\"\\n  R@20 is identical: same 20 chunks, just reordered — the candidate pool is the ceiling.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "404a8db8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1320x550 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visual: recall side by side (same 20-chunk pool)\n",
    "fig, ax = plt.subplots()\n",
    "ks = [5, 10, 20]\n",
    "basic_rec  = [recall_at_k(results_b, GROUND_TRUTH_B, k) for k in ks]\n",
    "rerank_rec = [recall_at_k(results_b_rerank, GROUND_TRUTH_B, k) for k in ks]\n",
    "x = np.arange(len(ks))\n",
    "width = 0.38\n",
    "b1 = ax.bar(x - width/2, basic_rec,  width, color=NEUTRAL, label=\"Basic\")\n",
    "b2 = ax.bar(x + width/2, rerank_rec, width, color=ACCENT, label=\"Rerank (same pool)\")\n",
    "for bars, vals in [(b1, basic_rec), (b2, rerank_rec)]:\n",
    "    for bar, v in zip(bars, vals):\n",
    "        ax.text(bar.get_x()+bar.get_width()/2, v+0.01, f\"{v:.2f}\",\n",
    "                ha=\"center\", fontsize=12, fontweight=\"bold\",\n",
    "                color=NEUTRAL if bars is b1 else ACCENT)\n",
    "ax.set_xticks(x); ax.set_xticklabels([f\"R@{k}\" for k in ks])\n",
    "ax.set_ylabel(\"context recall\"); ax.set_ylim(0, 0.35)\n",
    "ax.set_title(\"Reranking can shift small-k recall, but R@20 is the pool ceiling\")\n",
    "ax.legend(loc=\"upper left\")\n",
    "plt.tight_layout(); plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "15ec89b7",
   "metadata": {},
   "source": [
    "## The pool is the ceiling\n",
    "\n",
    "R@20 for basic and rerank is **identical at 0.14** — because we're measuring against the *same* 20 chunks. Rerank changed the order, but the set of distinct destinations in the pool can't change.\n",
    "\n",
    "At smaller k the bars *can* differ either direction. Here R@10 actually drops from 0.10 to 0.05: the cross-encoder pushed a relevant destination *out* of the top-10. Reranking is a tool, not a magic wand — when the cross-encoder's training (MS MARCO web Q&A) doesn't match your query intent, it can hurt."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8abf2d7b",
   "metadata": {},
   "source": [
    "## Takeaway\n",
    "\n",
    "> **Reranking reorders. It does not expand the candidate pool.**\n",
    "\n",
    "A document that was never retrieved is still absent after reranking. The cross-encoder can shuffle good chunks to the top — sometimes raising recall at small k, sometimes lowering it — but **R@(pool size)** is fixed.\n",
    "\n",
    "On Query C the lift was clear (+3.18 mean relevance). On Query B the cross-encoder is no help because **no chunk in the pool answers all three constraints anyway**.\n",
    "\n",
    "To break through the pool ceiling, we need to retrieve **differently** — not just reorder."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a570339b",
   "metadata": {},
   "source": [
    "---\n",
    "### Intermezzo — three reranking paradigms\n",
    "\n",
    "We just used a **cross-encoder** to rerank. It's the gold standard for accuracy and the slowest option — it can't pre-compute anything because it has to read the (query, chunk) pair *together*. There's a third design point worth knowing.\n",
    "\n",
    "```\n",
    "Bi-encoder       query ─► [ 1 vec ]              chunk ─► [ 1 vec ]               dot product\n",
    "Cross-encoder    ( query , chunk ) ──►   joint encode   ──►  1 score\n",
    "ColBERT          query ─► [tok₁ tok₂ tok₃ …]    chunk ─► [tok₁ tok₂ tok₃ …]      MaxSim sum\n",
    "```\n",
    "\n",
    "| Approach | Speed at query time | Accuracy | Pre-compute chunks? |\n",
    "|---|---|---|---|\n",
    "| **Bi-encoder** *(Section 1)* | fastest | lowest | ✓ one vector per chunk |\n",
    "| **ColBERT** | medium | high | ✓ one vector *per token* |\n",
    "| **Cross-encoder** *(Section 2)* | slowest | highest | ✗ has to see the query |\n",
    "\n",
    "**ColBERT** keeps token-level embeddings on both sides and scores by *late interaction* — for every query token, take the max similarity to any chunk token, then sum. This recovers most of the joint-encoding accuracy of a cross-encoder while letting you cache chunk embeddings, the way a bi-encoder does.\n",
    "\n",
    "> **The production sweet spot when cross-encoder reranking is too slow at scale.** Same place in the pipeline as the previous reranker — different tradeoff curve.\n",
    "\n",
    "(No demo here — model is ~440 MB and the tradeoffs land without the numbers. The repo's `pylate_colbert` pipeline implements it for production use.)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "eaccfece",
   "metadata": {},
   "source": [
    "---\n",
    "\n",
    "# Multi-query + Reciprocal Rank Fusion\n",
    "\n",
    "### *Stop forcing one vector to do three jobs.*"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "96bf1c78",
   "metadata": {},
   "source": [
    "## Why one vector wasn't enough\n",
    "\n",
    "The embedding model was trained to compress a sentence into 384 numbers that capture its meaning. That works beautifully when the sentence has *one* meaning.\n",
    "\n",
    "Query B has three:\n",
    "- **a geography constraint** (tropical)\n",
    "- **an activity constraint** (snorkeling)\n",
    "- **a diet constraint** (vegetarian)\n",
    "\n",
    "When we average them into one vector, we get something that's a little bit each of the three — and there is no chunk in the corpus that's *exactly* one-third of each, so nothing scores particularly well. **The recall ceiling we hit in Section 2 is a direct consequence of trying to represent three concepts with one vector.**"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9d4517bd",
   "metadata": {},
   "source": [
    "## The fix — decompose\n",
    "\n",
    "```\n",
    "                     ┌── \"snorkeling coral reef tropical\" ─────► results₁ ──┐\n",
    "query ──► [LLM] ──►  ├── \"vegetarian Hindu Buddhist food\" ─────► results₂ ──┤──► RRF ──► rerank ──► top-k\n",
    "                     ├── \"Southeast Asia island beaches diving\"─► results₃ ──┤\n",
    "                     └── original query ─────────────────────────► results₄ ──┘\n",
    "```\n",
    "\n",
    "Decompose the query into **focused sub-queries**, one per concept. Each sub-query is now narrow enough that a single 384-d vector *can* represent it.  Each gets its own retrieval pass with its own steep score curve.\n",
    "\n",
    "In production an LLM does the decomposition. For this demo we wrote them by hand so the mechanics are visible."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b1be3e8d",
   "metadata": {},
   "source": [
    "## Reciprocal Rank Fusion in one paragraph\n",
    "\n",
    "You now have N ranked lists of chunks. How do you combine them?\n",
    "\n",
    "**RRF** scores each chunk by `1 / (k + rank)` summed across all the lists it appears in (`k` is a constant, typically 60). A chunk that appears at rank 3 in list 1 *and* rank 5 in list 2 beats a chunk at rank 1 in list 1 alone. **Documents that show up consistently across decompositions float to the top**, regardless of any individual list's absolute scores.\n",
    "\n",
    "It's two lines of code, parameter-free, and works."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "fb34b317",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sub-queries:\n",
      "  · Tropical destinations with great snorkeling and vegetarian food\n",
      "  · snorkeling coral reef tropical islands\n",
      "  · vegetarian Hindu Buddhist food culture\n",
      "  · Southeast Asia island beaches diving\n",
      "  · Indian vegetarian cuisine traditions\n"
     ]
    }
   ],
   "source": [
    "sub_queries = [\n",
    "    QUERY_B,\n",
    "    \"snorkeling coral reef tropical islands\",\n",
    "    \"vegetarian Hindu Buddhist food culture\",\n",
    "    \"Southeast Asia island beaches diving\",\n",
    "    \"Indian vegetarian cuisine traditions\",\n",
    "]\n",
    "print(\"Sub-queries:\")\n",
    "for q in sub_queries:\n",
    "    print(f\"  · {q}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "d2d7d746",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "  Query B — multi-query + RRF + rerank\n",
      "  ────────────────────────────────────────────────────────────\n",
      "   #   score  destination             section\n",
      "   1.   0.031  cook-islands            Do\n",
      "   2.   0.029  bora-bora               Do\n",
      "   3.   0.015  melbourne               Eat\n",
      "   4.   0.015  seychelles              Do\n",
      "   5.   0.016  phuket                  Other destinations ✓\n",
      "   6.   0.016  salar-de-uyuni          Eat and drink\n",
      "   7.   0.026  beijing                 Eat\n",
      "   8.   0.031  fernando-de-noronha     Do\n",
      "   9.   0.015  koh-samui               Do ✓\n",
      "  10.   0.015  vietnam                 Eat\n"
     ]
    }
   ],
   "source": [
    "# Stage 1: multi-query + RRF\n",
    "fused = rrf([vector_search(q, n=20) for q in sub_queries])\n",
    "\n",
    "# Stage 2: rerank the fused pool with the cross-encoder\n",
    "ce_scores_mq = reranker.predict([(QUERY_B, c.text) for c in fused[:40]])\n",
    "results_mq = [c for _, c in sorted(zip(ce_scores_mq, fused[:40]), key=lambda x: x[0], reverse=True)]\n",
    "\n",
    "show(results_mq, gt=GROUND_TRUTH_B, n=10, title=\"Query B — multi-query + RRF + rerank\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "d5dd34f7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  Recall at top-k against 21 relevant destinations:\n",
      "\n",
      "      k       Basic      Rerank      Multi-query+rerank\n",
      "      5        0.05        0.05                    0.05\n",
      "     10        0.10        0.05                    0.10\n",
      "     20        0.14        0.14                    0.24\n"
     ]
    }
   ],
   "source": [
    "# Compare all three pipelines on Query B (same comparison point)\n",
    "print(f\"  Recall at top-k against {len(GROUND_TRUTH_B)} relevant destinations:\\n\")\n",
    "print(f\"    {'k':>3}    {'Basic':>8}    {'Rerank':>8}    {'Multi-query+rerank':>20}\")\n",
    "for k in [5, 10, 20]:\n",
    "    rb  = recall_at_k(results_b, GROUND_TRUTH_B, k)\n",
    "    rr  = recall_at_k(results_b_rerank, GROUND_TRUTH_B, k)\n",
    "    rmq = recall_at_k(results_mq, GROUND_TRUTH_B, k)\n",
    "    print(f\"    {k:>3}    {rb:>8.2f}    {rr:>8.2f}    {rmq:>20.2f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "4925c9ac",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1320x550 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visual: recall climbs visibly\n",
    "fig, ax = plt.subplots()\n",
    "ks = [5, 10, 20]\n",
    "def vals(rs): return [recall_at_k(rs, GROUND_TRUTH_B, k) for k in ks]\n",
    "basic_v, rer_v, mq_v = vals(results_b), vals(results_b_rerank), vals(results_mq)\n",
    "x = np.arange(len(ks))\n",
    "width = 0.27\n",
    "b1 = ax.bar(x - width, basic_v, width, color=NEUTRAL, label=\"Basic\")\n",
    "b2 = ax.bar(x,         rer_v,   width, color=ACCENT,  label=\"+ Rerank\")\n",
    "b3 = ax.bar(x + width, mq_v,    width, color=GOOD,    label=\"+ Multi-query + Rerank\")\n",
    "for bars, vs, col in [(b1, basic_v, NEUTRAL), (b2, rer_v, ACCENT), (b3, mq_v, GOOD)]:\n",
    "    for bar, v in zip(bars, vs):\n",
    "        ax.text(bar.get_x()+bar.get_width()/2, v+0.01, f\"{v:.2f}\",\n",
    "                ha=\"center\", fontsize=12, fontweight=\"bold\", color=col)\n",
    "ax.set_xticks(x); ax.set_xticklabels([f\"R@{k}\" for k in ks])\n",
    "ax.set_ylabel(\"context recall\"); ax.set_ylim(0, 0.45)\n",
    "ax.set_title(\"Multi-query breaks the recall ceiling\")\n",
    "ax.legend(loc=\"upper left\")\n",
    "plt.tight_layout(); plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "87253069",
   "metadata": {},
   "source": [
    "## What changed — and what didn't\n",
    "\n",
    "R@20 jumped from **0.14 → 0.24** — a 71% relative improvement at exactly the same evaluation point where rerank-on-the-same-pool was stuck at the ceiling. Multi-query is *the* technique that genuinely moves recall on hard queries.\n",
    "\n",
    "R@5 and R@10 are roughly flat. The decomposition helps **breadth**, not pinpoint-precision at the very top — and that's the right shape. Diverse sub-queries pull in destinations that no single retrieval would have surfaced; they don't all get to rank 1.\n",
    "\n",
    "Production note: an LLM writes the sub-queries from the user's original. Costs one extra LLM call per query, returns hard-to-replicate retrieval improvements."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fab47d65",
   "metadata": {},
   "source": [
    "## Takeaway\n",
    "\n",
    "> **One vector cannot represent three concepts. Three vectors can.**\n",
    "\n",
    "Multi-query decomposition shifts the work upstream — the LLM (or, here, hand-written sub-queries) reframes the question into pieces the embedding model can actually answer. RRF fuses the answers without trusting any one signal too much.\n",
    "\n",
    "This is the technique that pays for itself the moment your queries get hard."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6b0d4353",
   "metadata": {},
   "source": [
    "---\n",
    "\n",
    "# CRAG — Corrective RAG\n",
    "\n",
    "### *Knowing what you don't know.*"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "de85437b",
   "metadata": {},
   "source": [
    "## The hallucination tax\n",
    "\n",
    "Every technique so far has been about **finding the right answer**. CRAG is about a different problem: what if there *is* no right answer in the corpus?\n",
    "\n",
    "The default behaviour of basic RAG when asked something outside its knowledge:\n",
    "\n",
    "1. Retriever returns nearest neighbours regardless of how poor the match is\n",
    "2. LLM — trained on the entire internet — happily produces a confident, fluent, fabricated answer\n",
    "3. User has no way to tell\n",
    "\n",
    "Hallucination on a corpus gap is the **most expensive failure mode** in production RAG. A wrong answer costs trust; trust is hard to rebuild."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e11e1bd6",
   "metadata": {},
   "source": [
    "## The mechanism\n",
    "\n",
    "```\n",
    "query ──► retrieve ──► top-1 score ──┬─► high   →  answer\n",
    "                                     ├─► medium →  refine + retrieve again\n",
    "                                     └─► low    →  refuse\n",
    "```\n",
    "\n",
    "CRAG evaluates retrieval *quality* before generating. The signal we use is the simplest one available: **the top-1 similarity score**. If no chunk is a strong match, the question is outside what the corpus can answer — and \"I don't know\" is the correct response.\n",
    "\n",
    "Three buckets, two thresholds:\n",
    "- **Confident** — top-1 ≥ 0.65 → answer using retrieved context\n",
    "- **Ambiguous** — 0.50 ≤ top-1 < 0.65 → query rewriting / web search / extra rounds before answering\n",
    "- **Disoriented** — top-1 < 0.50 → refuse\n",
    "\n",
    "Real CRAG implementations use a separately-trained relevance evaluator. The thresholds-on-similarity version below is the *minimum viable* form, and on a well-calibrated embedding model it's already useful."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "67ee9113",
   "metadata": {},
   "source": [
    "## The test case — a corpus gap\n",
    "\n",
    "**New Zealand was deliberately excluded** from the corpus. The query asks about NZ travel costs in October. Nothing matches.\n",
    "\n",
    "We compare its top-1 score against Query A (Iceland — present in corpus) and Query B (multi-aspect — present but ambiguous)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "cd8a87e1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  top-1 = 0.671  [ CONFIDENT]  →  What's there to do in Iceland?\n",
      "    nearest neighbours: iceland, iceland, iceland\n",
      "\n",
      "  top-1 = 0.578  [ AMBIGUOUS]  →  Tropical destinations with great snorkeling and vegetarian food\n",
      "    nearest neighbours: galapagos, phuket, fernando-de-noronha\n",
      "\n",
      "  top-1 = 0.499  [    REFUSE]  →  What does it cost to travel in New Zealand in October?\n",
      "    nearest neighbours: masai-mara, australia, zanzibar\n",
      "\n"
     ]
    }
   ],
   "source": [
    "def crag_decision(query, confident=0.65, uncertain=0.50):\n",
    "    chunks = vector_search(query, n=5)\n",
    "    s = chunks[0].score\n",
    "    if s >= confident:    state = \"CONFIDENT\"\n",
    "    elif s >= uncertain:  state = \"AMBIGUOUS\"\n",
    "    else:                 state = \"REFUSE\"\n",
    "    return s, state, chunks\n",
    "\n",
    "for q in [QUERY_A, QUERY_B, QUERY_E]:\n",
    "    s, state, chunks = crag_decision(q)\n",
    "    print(f\"  top-1 = {s:.3f}  [{state:>10}]  →  {q}\")\n",
    "    print(f\"    nearest neighbours: {', '.join(c.doc_id for c in chunks[:3])}\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "e781821e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1320x462 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visual: CRAG decision diagram\n",
    "fig, ax = plt.subplots(figsize=(12, 4.2))\n",
    "queries = [(\"Query A — Iceland\", QUERY_A), (\"Query B — tropical\", QUERY_B), (\"Query E — New Zealand\", QUERY_E)]\n",
    "labels, scores = [], []\n",
    "for label, q in queries:\n",
    "    chunks = vector_search(q, n=5)\n",
    "    labels.append(label); scores.append(chunks[0].score)\n",
    "\n",
    "CONFIDENT, UNCERTAIN = 0.65, 0.50\n",
    "colors = []\n",
    "for s in scores:\n",
    "    if s >= CONFIDENT: colors.append(GOOD)\n",
    "    elif s >= UNCERTAIN: colors.append(\"#D8AB1D\")\n",
    "    else: colors.append(WARN)\n",
    "\n",
    "y = np.arange(len(labels))\n",
    "ax.barh(y, scores, color=colors)\n",
    "for i, s in enumerate(scores):\n",
    "    ax.text(s + 0.005, i, f\"{s:.3f}\", va=\"center\", fontsize=13, fontweight=\"bold\")\n",
    "ax.axvline(CONFIDENT, color=GOOD, linestyle=\"--\", alpha=0.7, label=f\"answer  ≥ {CONFIDENT}\")\n",
    "ax.axvline(UNCERTAIN, color=WARN, linestyle=\"--\", alpha=0.7, label=f\"refuse  < {UNCERTAIN}\")\n",
    "ax.set_yticks(y); ax.set_yticklabels(labels, fontsize=13)\n",
    "ax.set_xlabel(\"top-1 similarity score\"); ax.set_xlim(0, 0.85)\n",
    "ax.set_title(\"CRAG: route on confidence, not assumption\")\n",
    "ax.legend(loc=\"lower right\"); ax.invert_yaxis()\n",
    "plt.tight_layout(); plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4bf435fe",
   "metadata": {},
   "source": [
    "## Reading the chart\n",
    "\n",
    "Three queries, three different routing decisions, all from one number.\n",
    "\n",
    "- **Iceland (0.671)** clears the confident threshold by a clear margin → green → answer\n",
    "- **Tropical (0.578)** sits in the middle band → yellow → refine before answering\n",
    "- **New Zealand (0.499)** falls below the refuse threshold → red → \"I don't know\"\n",
    "\n",
    "The thresholds are tunable knobs, not absolutes. **Calibrate them on a held-out set of queries you know are answerable and queries you know aren't**, and pick the cutoff that minimises the failure mode you fear most.\n",
    "\n",
    "For a customer-support bot, lean toward refuse — false answers cost more than declined questions. For an internal research tool, lean toward answer — the user can verify."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bbe6dc0d",
   "metadata": {},
   "source": [
    "## Takeaway\n",
    "\n",
    "> **\"I don't know\" is the correct answer when your corpus doesn't have the information.**\n",
    "\n",
    "A fluent wrong answer is worse than silence in a production system — it costs trust, and trust is expensive.\n",
    "\n",
    "Sections 1–3 helped you find the right answer when one exists. CRAG protects you when one doesn't. **Qualitatively different category of improvement, qualitatively different category of failure prevented.**"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "64bb70ac",
   "metadata": {},
   "source": [
    "---\n",
    "\n",
    "# RAGAS\n",
    "\n",
    "### *Measure, don't intuit.*"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d093191b",
   "metadata": {},
   "source": [
    "## \"It seems better\" is not a metric\n",
    "\n",
    "We've shipped four pipeline changes. Each one *felt* like an improvement — and on Query B, three of them produced different recall numbers. But which is actually better for *your* application?\n",
    "\n",
    "You can't answer that question by looking at examples. You answer it by picking metrics, defining ground truth, and measuring.\n",
    "\n",
    "**RAGAS** is the de-facto standard framework for RAG evaluation. It gives you four standard metrics that decompose RAG quality along its two axes: **retrieval** (did we find the right chunks?) and **generation** (did the LLM use them faithfully?)."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3a42807e",
   "metadata": {},
   "source": [
    "## The four metrics\n",
    "\n",
    "| Tier | Metric | What it measures | Needs |\n",
    "|---|---|---|---|\n",
    "| **Reference-free** | **faithfulness** | does the answer contradict the retrieved chunks? | LLM judge |\n",
    "|  | **answer relevance** | does the answer address the question asked? | LLM judge |\n",
    "| **Reference-based** | **context recall** | did we retrieve all the relevant chunks? | ground truth ✓ |\n",
    "|  | **context precision** | are the retrieved chunks actually relevant? | ground truth ✓ |\n",
    "\n",
    "**Reference-free metrics** can be computed on any pipeline output without labelled data, using an LLM-as-judge to score faithfulness and relevance. Cheap to run, useful for monitoring production drift, but limited by the judge model's biases.\n",
    "\n",
    "**Reference-based metrics** need labelled data — query plus expected relevant chunks — and that's a real upfront investment. The payoff is much sharper signal: you measure against ground truth, not a judge.\n",
    "\n",
    "We have ground truth for Query B (21 destinations). **Context recall is computable right now, no API key required.**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "fadb9602",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  Context recall against 21 relevant destinations\n",
      "\n",
      "    pipeline                         R@5    R@10    R@20\n",
      "    ──────────────────────────────────────────────────────\n",
      "    Basic                           0.05    0.10    0.14\n",
      "    + Rerank                        0.05    0.05    0.14\n",
      "    + Multi-query + Rerank          0.05    0.10    0.24\n"
     ]
    }
   ],
   "source": [
    "# Final evaluation table — context recall across all four pipelines\n",
    "pipelines = {\n",
    "    \"Basic\":                      results_b,\n",
    "    \"+ Rerank\":                   results_b_rerank,\n",
    "    \"+ Multi-query + Rerank\":     results_mq,\n",
    "}\n",
    "\n",
    "print(f\"  Context recall against {len(GROUND_TRUTH_B)} relevant destinations\\n\")\n",
    "print(f\"    {'pipeline':<28}  {'R@5':>6}  {'R@10':>6}  {'R@20':>6}\")\n",
    "print(f\"    {'─'*54}\")\n",
    "for name, rs in pipelines.items():\n",
    "    row = \"  \".join(f\"{recall_at_k(rs, GROUND_TRUTH_B, k):>6.2f}\" for k in [5,10,20])\n",
    "    print(f\"    {name:<28}  {row}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "97dc41b8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1320x550 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visual: clean grouped bar chart, all pipelines, R@5/10/20\n",
    "fig, ax = plt.subplots()\n",
    "ks = [5, 10, 20]\n",
    "colors = [NEUTRAL, ACCENT, GOOD]\n",
    "x = np.arange(len(ks))\n",
    "width = 0.27\n",
    "for i, (name, rs) in enumerate(pipelines.items()):\n",
    "    vs = [recall_at_k(rs, GROUND_TRUTH_B, k) for k in ks]\n",
    "    offset = (i - 1) * width\n",
    "    bars = ax.bar(x + offset, vs, width, color=colors[i], label=name)\n",
    "    for b, v in zip(bars, vs):\n",
    "        ax.text(b.get_x()+b.get_width()/2, v+0.01, f\"{v:.2f}\",\n",
    "                ha=\"center\", fontsize=12, fontweight=\"bold\", color=colors[i])\n",
    "ax.set_xticks(x); ax.set_xticklabels([f\"R@{k}\" for k in ks])\n",
    "ax.set_ylabel(\"context recall\"); ax.set_ylim(0, 0.4)\n",
    "ax.set_title(\"RAGAS context recall — Query B across pipelines\")\n",
    "ax.legend(loc=\"upper left\")\n",
    "plt.tight_layout(); plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c0e40fd7",
   "metadata": {},
   "source": [
    "## Different applications, different metric weights\n",
    "\n",
    "The chart says multi-query+rerank wins on context recall for Query B. **That doesn't mean it's the right pipeline for your application.**\n",
    "\n",
    "| System | What it should maximise | Why |\n",
    "|---|---|---|\n",
    "| Medical Q&A | **faithfulness** | citing something the chunks don't say is malpractice |\n",
    "| Customer support | **answer relevance** under latency budget | speed competes with completeness |\n",
    "| Legal precedent search | **context recall** | missing a relevant case is the worst error |\n",
    "| Internal docs assistant | **context precision** | high precision keeps the LLM focused |\n",
    "\n",
    "The metric you care about most should drive the pipeline you ship."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "66b68100",
   "metadata": {},
   "source": [
    "## Takeaway\n",
    "\n",
    "> **RAGAS doesn't tell you which pipeline to ship. It gives you the vocabulary to defend the choice.**\n",
    "\n",
    "Reference-based metrics need labelled data, and labelling is a real investment. **The richer the ground truth, the more useful the evaluation.** That investment is itself a design decision — and now a defensible one.\n",
    "\n",
    "Once you have the metrics, you have something to tune *against*. Without them, every change is a guess that *seems* better."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8034858a",
   "metadata": {},
   "source": [
    "---\n",
    "\n",
    "# Where to from here"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "61de8177",
   "metadata": {},
   "source": [
    "## What we covered\n",
    "\n",
    "| Technique | What it fixes | What it can't fix |\n",
    "|---|---|---|\n",
    "| **Basic RAG** | Establishes a baseline | Multi-aspect queries, exact-term matches |\n",
    "| **Reranking** | Ordering of candidates | The recall ceiling |\n",
    "| **Multi-query + Rerank** | The recall ceiling | Requires good decomposition |\n",
    "| **CRAG** | Hallucination on corpus gaps | Doesn't improve retrieval itself |\n",
    "| **RAGAS** | How you know which pipeline to ship | Quality of the ground truth |"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e194fc2b",
   "metadata": {},
   "source": [
    "## One step beyond\n",
    "\n",
    "**Hybrid retrieval** — fuse vector search with **BM25 keyword search** when your queries contain rare or proper-noun terms. Vector embeddings drift toward related concepts; BM25 nails exact matches. Same RRF fusion technique we used in Section 3.\n",
    "\n",
    "**GraphRAG** — when the answer requires *relationships* the text never spells out (*\"diving spots similar to the Great Barrier Reef\"*). A knowledge graph stores `similar_to`, `near`, `same_type` as edges; vector search can't recover that.\n",
    "\n",
    "Both extend the toolkit. Both layer on top of what we built today."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3770c426",
   "metadata": {},
   "source": [
    "## The single most important thing\n",
    "\n",
    "> **The retriever, not the LLM, is where almost every production RAG problem lives.**\n",
    "\n",
    "Every technique in this talk attacks retrieval. Reranking, decomposition, fusion, refusal, evaluation — none of them touch the generation step. **Pick a strong open-weight LLM, give it the right context, and your system will be fine.**\n",
    "\n",
    "Spend your time on retrieval."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "name": "python"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}