{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "0fbe4e92",
   "metadata": {},
   "source": [
    "# Building a grounded memory worker: an iterative distillation experiment\n",
    "\n",
    "**Project:** SDP 2026 memory engine · **Experiment date:** 8 October 2026\n",
    "\n",
    "**Scope correction:** the product is general-purpose memory. The coding-focused runs below are initial experiments. The automatic coding-only second stage was stopped before training; general-purpose corpus, training and evaluation are recorded from current receipts in Section 12c.\n",
    "\n",
    "This notebook reconstructs the actual local, AI-assisted experiment from saved artifacts. It records unsuccessful attempts and measured limitations as well as successful runs. It is a research record, not a claim that every experiment was originally conducted inside Jupyter.\n",
    "\n",
    "**Question:** Can a small model on an 8 GB laptop GPU extract durable facts with exact evidence, synthesize supported observations, and answer with citations?\n",
    "\n",
    "Run All refreshes tables and figures from local receipts. It does **not** start training, download weights, modify databases, or invoke a paid teacher. Outputs remain visible when the notebook is shared without large ignored datasets. Missing results are explicitly marked pending."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "f6005162",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-10-08T17:00:09.042998Z",
     "iopub.status.busy": "2026-10-08T17:00:09.042830Z",
     "iopub.status.idle": "2026-10-08T17:00:09.535004Z",
     "shell.execute_reply": "2026-10-08T17:00:09.534650Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Report refreshed: 2026-10-08T17:00:09.532810+00:00\n",
      "Python: 3.12.13\n"
     ]
    }
   ],
   "source": [
    "from pathlib import Path\n",
    "from collections import Counter\n",
    "from datetime import datetime, timezone\n",
    "import json, hashlib, re, ast, sys\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from IPython.display import display, Markdown, Image\n",
    "ROOT = next(p for p in [Path.cwd(), *Path.cwd().parents] if (p / 'Cargo.toml').exists())\n",
    "SLM = ROOT / 'slm-distill'\n",
    "DATA = SLM / 'data/research-v2'\n",
    "FIG = SLM / 'reports/figures'\n",
    "FIG.mkdir(parents=True, exist_ok=True)\n",
    "def read(path):\n",
    "    p = Path(path)\n",
    "    return json.loads(p.read_text()) if p.exists() else None\n",
    "def rows(path):\n",
    "    p = Path(path)\n",
    "    return [json.loads(line) for line in p.read_text().splitlines() if line.strip()] if p.exists() else []\n",
    "def show(value):\n",
    "    display(pd.DataFrame(value)) if isinstance(value, list) else display(value)\n",
    "def digest(path): return hashlib.sha256(Path(path).read_bytes()).hexdigest()\n",
    "def missing(label): display(Markdown(f'**Pending / unavailable:** {label}. No result is inferred.'))\n",
    "def show_acceptance(receipt):\n",
    "    identity=receipt.get('deployment_identity_before',{})\n",
    "    show([{'worker_model':receipt.get('worker_model',receipt.get('health',{}).get('worker_model')),\n",
    "           'passed':receipt.get('passed'),'served_gguf_sha256':identity.get('served_gguf_sha256'),\n",
    "           'started_at':receipt.get('started_at'),'finished_at':receipt.get('finished_at')}])\n",
    "    show([{'check':name,'passed':value} for name,value in receipt.get('checks',{}).items()])\n",
    "    graph=receipt.get('steps',{}).get('correction_graph',{})\n",
    "    assertions=graph.get('assertions',[])\n",
    "    observations=[a for a in assertions if a.get('kind')=='observation']\n",
    "    supports=[e for e in graph.get('edges',[]) if e.get('relation') in ('supports','derives')]\n",
    "    projection=receipt.get('projection',{})\n",
    "    show([{'assertions':len(assertions),'active_assertions':sum(a.get('status')=='active' for a in assertions),\n",
    "           'observations':len(observations),'support_edges':len(supports),\n",
    "           'observations_with_two_distinct_supports':sum(len({e['to_id'] for e in supports if e['from_id']==a['id']})>=2 for a in observations),\n",
    "           'projected_nodes':len(projection.get('nodes',[])),'projected_relationships':len(projection.get('relationships',[]))}])\n",
    "    examples=[]\n",
    "    for topic in ('location','preference'):\n",
    "        recalled=receipt.get('steps',{}).get(topic+'_recall',{}).get('results',[])\n",
    "        hit=next((r for r in recalled if any(source.get('quote') for source in r.get('sources',[]))),None)\n",
    "        if hit:\n",
    "            examples.append({'query_topic':topic,'recalled_statement':hit.get('statement'),\n",
    "                             'source_quote':next(source['quote'] for source in hit['sources'] if source.get('quote'))})\n",
    "    if examples: show(examples)\n",
    "    show([{'query_topic':topic,'answer':receipt.get('steps',{}).get(topic+'_reflect',{}).get('answer'),\n",
    "           'citation_count':len(receipt.get('steps',{}).get(topic+'_reflect',{}).get('citations',[]))}\n",
    "          for topic in ('location','preference')])\n",
    "    if receipt.get('recheck_provenance'): display(receipt['recheck_provenance'])\n",
    "    if receipt.get('error'): print('Acceptance error:',receipt['error'])\n",
    "    display(Markdown('Full jobs, graph and retrieval responses remain in the local acceptance receipt. Quoted examples aid inspection; the check results above determine acceptance.'))\n",
    "\n",
    "print('Report refreshed:', datetime.now(timezone.utc).isoformat())\n",
    "print('Python:', sys.version.split()[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "69bcab0c",
   "metadata": {},
   "source": [
    "## 1. Define the task and acceptance criteria before comparing models\n",
    "\n",
    "The unavailable original worker artifact meant we needed a reproducible replacement. The worker has three tasks:\n",
    "\n",
    "| Task | Input | Output and acceptance |\n",
    "|---|---|---|\n",
    "| Extraction | Timestamped events + earlier facts | Durable claims, exact source quotes, valid source indices, entity/slot/value and correction metadata |\n",
    "| Consolidation | Accepted facts | Useful observations supported by at least two distinct supplied facts |\n",
    "| Reflection | Question + retrieved evidence | Answer with valid citations, or explicit insufficient evidence |\n",
    "\n",
    "PostgreSQL is the evidence store; Neo4j is a derived projection. A queued retain response alone does not prove extraction or graph completion. Final operational checks must follow jobs through extraction, consolidation, projection, retrieval and correction.\n",
    "\n",
    "We distinguish **schema acceptance**, **semantic support**, **reference coverage**, and **live pipeline completion**. These are different measurements."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fea1f59c",
   "metadata": {},
   "source": [
    "## 2. Iteration ledger: observation → hypothesis → change → check\n",
    "\n",
    "| Iteration | Observation | Hypothesis and intervention | Evidence / remaining boundary |\n",
    "|---|---|---|---|\n",
    "| Bootstrap 1.7B | Original GGUF unavailable | Learn the output contract from deterministic examples | Excellent synthetic scores; not evidence of natural conversation quality |\n",
    "| Naturalistic refinement | Synthetic success did not transfer | Add teacher-labeled coding timelines and deterministic replay | Natural extraction remained weak; exact quotes and IDs frequently failed |\n",
    "| Memory optimization | Full output logits caused OOM | Checkpoint answer-only output-head chunks | Training completed within laptop capacity |\n",
    "| Structured decoding | Valid-looking claims had invalid source attribution/IDs | Bind source and quote together; constrain enums and existing IDs | Original blocked job completed with the 1.7B bridge worker |\n",
    "| Stronger base | Remaining errors were semantic as well as structural | Use Qwen3-4B-Instruct-2507 and NF4 QLoRA | 7,141-token smoke OOM; 4,091-token smoke passed after dtype correction |\n",
    "| Corpus expansion | Narrow examples could encourage memorization | Public traces + diverse stacks/domains/behaviors; family/repository splits | Audit, counts and leakage receipts below |\n",
    "| Label adjudication | Auditor incorrectly rejected valid null timestamps | Re-review flags against the actual worker contract | Keep original audit and revised decisions; exclude remaining rejected cases |\n",
    "| Stage 2 + comparison | Need evidence that fine-tuning helps the stronger base | Matched decoder/quantization comparison and fresh held-out subset | Results appear only when their receipts exist |\n",
    "\n",
    "Changes to base, data and decoder occurred together across generations. Cross-generation improvement cannot be attributed to model size alone."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c45a01f0",
   "metadata": {},
   "source": [
    "## 3. Bootstrap: useful contract learning, insufficient generalization\n",
    "\n",
    "The first adapter used Qwen3-1.7B, 1,260 training examples, 140 validation examples and 280 synthetic test examples. The matched 28-case comparison below uses identical prompts with thinking disabled. CPU/GPU differed, so its latency is not a fair speed comparison."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "e934f7e4",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-10-08T17:00:09.537486Z",
     "iopub.status.busy": "2026-10-08T17:00:09.537348Z",
     "iopub.status.idle": "2026-10-08T17:00:09.547097Z",
     "shell.execute_reply": "2026-10-08T17:00:09.546827Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model</th>\n",
       "      <th>n</th>\n",
       "      <th>accepted</th>\n",
       "      <th>exact</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>base</td>\n",
       "      <td>28</td>\n",
       "      <td>27</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>bootstrap</td>\n",
       "      <td>28</td>\n",
       "      <td>28</td>\n",
       "      <td>28</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       model   n  accepted  exact\n",
       "0       base  28        27     13\n",
       "1  bootstrap  28        28     28"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "first 28 identical synthetic held-out cases; exact slot/support/citation targets, not natural conversation accuracy\n",
      "Latency comparable: False\n",
      "bootstrap synthetic test\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'model': 'memex-extractor',\n",
       " 'dataset_sha256': 'd44e9139e24f0fe2fe8aa70106888948574fa898854bf470e9e8d461b8cecc50',\n",
       " 'scope': 'synthetic held-out project identities; no real-world accuracy claim',\n",
       " 'cpu_only': False,\n",
       " 'count': 280,\n",
       " 'counts': {'extract:total': 160,\n",
       "  'extract:accepted': 160,\n",
       "  'extract:first_try': 160,\n",
       "  'extract:semantic_exact': 160,\n",
       "  'tp': 160,\n",
       "  'fp': 0,\n",
       "  'fn': 0,\n",
       "  'secret_leaked': 0,\n",
       "  'consolidate:total': 40,\n",
       "  'consolidate:accepted': 40,\n",
       "  'consolidate:first_try': 40,\n",
       "  'consolidate:semantic_exact': 40,\n",
       "  'reflect:total': 80,\n",
       "  'reflect:accepted': 80,\n",
       "  'reflect:first_try': 80,\n",
       "  'reflect:semantic_exact': 80},\n",
       " 'seconds': 457.133}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "refined synthetic test\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'model': 'memex-extractor:local-v1',\n",
       " 'dataset_sha256': 'd44e9139e24f0fe2fe8aa70106888948574fa898854bf470e9e8d461b8cecc50',\n",
       " 'scope': 'synthetic held-out project identities; no real-world accuracy claim',\n",
       " 'cpu_only': False,\n",
       " 'count': 280,\n",
       " 'counts': {'extract:total': 160,\n",
       "  'extract:accepted': 160,\n",
       "  'extract:first_try': 160,\n",
       "  'extract:semantic_exact': 138,\n",
       "  'tp': 144,\n",
       "  'fp': 23,\n",
       "  'fn': 16,\n",
       "  'secret_leaked': 0,\n",
       "  'consolidate:total': 40,\n",
       "  'consolidate:accepted': 39,\n",
       "  'consolidate:first_try': 38,\n",
       "  'consolidate:semantic_exact': 39,\n",
       "  'reflect:total': 80,\n",
       "  'reflect:accepted': 80,\n",
       "  'reflect:first_try': 80,\n",
       "  'reflect:semantic_exact': 80},\n",
       " 'seconds': 1273.547}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "comparison = read(SLM/'data/eval/bootstrap-comparison.json')\n",
    "if comparison:\n",
    "    show([{'model': k, 'n': comparison['n'], **comparison[k]} for k in ('base','bootstrap')])\n",
    "    print(comparison['scope'])\n",
    "    print('Latency comparable:', comparison['latency_comparison_valid'])\n",
    "for label, path in [('bootstrap synthetic test','bootstrap-student-full'), ('refined synthetic test','refined-bootstrap-full')]:\n",
    "    result = read(SLM/f'data/eval/{path}/summary.json')\n",
    "    if result: print(label); display(result)\n",
    "    else: missing(label)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ada1b35d",
   "metadata": {},
   "source": [
    "## 4. Natural conversations exposed the failure\n",
    "\n",
    "The refined 1.7B adapter mixed teacher-labeled conversations with replay. Its naturalistic extraction result under the earlier unconstrained decoder was only 18 accepted windows out of 33. This set was subsequently used for diagnosis, so it is **development evidence**, not a fresh final test.\n",
    "\n",
    "The semantic judge's supported fraction is measured over emitted claims in accepted windows. Its earlier recall calculation also excludes failed windows; it must not be presented as end-to-end recall. Later comparisons retain failures in the coverage denominator."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "b40189b2",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-10-08T17:00:09.548658Z",
     "iopub.status.busy": "2026-10-08T17:00:09.548572Z",
     "iopub.status.idle": "2026-10-08T17:00:09.562845Z",
     "shell.execute_reply": "2026-10-08T17:00:09.562495Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>task</th>\n",
       "      <th>n</th>\n",
       "      <th>accepted_fraction</th>\n",
       "      <th>first_try_fraction</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>extract</td>\n",
       "      <td>33</td>\n",
       "      <td>0.545455</td>\n",
       "      <td>0.515152</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>consolidate</td>\n",
       "      <td>10</td>\n",
       "      <td>0.600000</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>reflect</td>\n",
       "      <td>32</td>\n",
       "      <td>0.968750</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          task   n  accepted_fraction  first_try_fraction\n",
       "0      extract  33           0.545455            0.515152\n",
       "1  consolidate  10           0.600000                 NaN\n",
       "2      reflect  32           0.968750                 NaN"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "{'planted_visible': 30,\n",
       " 'recall': 0.5,\n",
       " 'claims': 43,\n",
       " 'secret': 0.023255813953488372,\n",
       " 'supported': 0.6744186046511628,\n",
       " 'trivial': 0.023255813953488372,\n",
       " 'unsupported': 0.27906976744186046}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Representative validator failures:\n",
      "- every claim needs paired source indices and exact quotes: claim \"The readings SQLite table has meterId as a non-nullable text reference to meters.id, recordedAt\n",
      "- claim \"The wattwatch readings endpoint fails when SQLite compares a timestamp string to an ISO date because...\": quote \"src/routes/api/meters/[id]/readings/+ser\n",
      "- extraction schema: related.assertion_id must be a UUID\n"
     ]
    }
   ],
   "source": [
    "natural = read(SLM/'data/pi-corpus/eval/local-v1.json')\n",
    "if natural:\n",
    "    e=natural['extract']\n",
    "    show([{'task':'extract','n':e['windows'],'accepted_fraction':e['accepted'],'first_try_fraction':e['first_try']},\n",
    "          {'task':'consolidate','n':natural['consolidate']['items'],'accepted_fraction':natural['consolidate']['accepted']},\n",
    "          {'task':'reflect','n':natural['reflect']['items'],'accepted_fraction':natural['reflect']['accepted']}])\n",
    "    display(e.get('judge'))\n",
    "    print('Representative validator failures:')\n",
    "    for failure in e.get('failures',[])[:3]: print('-', failure)\n",
    "else: missing('1.7B naturalistic evaluation')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e7de7050",
   "metadata": {},
   "source": [
    "## 5. Fix the representation without weakening validation\n",
    "\n",
    "The canonical API uses parallel `source_indices` and `quotes` arrays. The constrained generation format emits `source_quotes: [{source_index, quote}]` first, binding each quote to visible spans from that event. A lossless codec restores the API contract. Consolidation identifiers are restricted to supplied facts. The Python teacher/training pipeline and Rust runtime share schema files.\n",
    "\n",
    "Exact text does **not** imply entailment: a model can quote a rejected proposal and still incorrectly describe it as a decision. Semantic grading remains necessary. Short teacher quotes are widened to the shortest permitted containing span; source index and claim statement stay unchanged."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "2eaa7b09",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-10-08T17:00:09.563750Z",
     "iopub.status.busy": "2026-10-08T17:00:09.563661Z",
     "iopub.status.idle": "2026-10-08T17:00:09.583484Z",
     "shell.execute_reply": "2026-10-08T17:00:09.582236Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>file</th>\n",
       "      <th>sha256</th>\n",
       "      <th>bytes</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>src/model/extract_schema.json</td>\n",
       "      <td>6da41f75e14a316c5335e3587db678cbe41d2ef8fe11ca...</td>\n",
       "      <td>3663</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>src/model/consolidate_schema.json</td>\n",
       "      <td>ac89fc0a3c4c993715ea9154affebf200912a13013cd4d...</td>\n",
       "      <td>1064</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>src/model/paired_sources.txt</td>\n",
       "      <td>f802c0c4790fcead4400a860aebb243bdd351fc2d4964c...</td>\n",
       "      <td>265</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                file  \\\n",
       "0      src/model/extract_schema.json   \n",
       "1  src/model/consolidate_schema.json   \n",
       "2       src/model/paired_sources.txt   \n",
       "\n",
       "                                              sha256  bytes  \n",
       "0  6da41f75e14a316c5335e3587db678cbe41d2ef8fe11ca...   3663  \n",
       "1  ac89fc0a3c4c993715ea9154affebf200912a13013cd4d...   1064  \n",
       "2  f802c0c4790fcead4400a860aebb243bdd351fc2d4964c...    265  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "schema_files = [ROOT/'src/model/extract_schema.json', ROOT/'src/model/consolidate_schema.json', ROOT/'src/model/paired_sources.txt']\n",
    "show([{'file':str(p.relative_to(ROOT)), 'sha256':digest(p), 'bytes':p.stat().st_size} for p in schema_files])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "097ea893",
   "metadata": {},
   "source": [
    "## 6. Research-informed base model and training method\n",
    "\n",
    "Selected **[Qwen3-4B-Instruct-2507](https://huggingface.co/Qwen/Qwen3-4B-Instruct-2507)**, Apache-2.0, revision `cdbee75f17c01a7cc42f958dc650907174af0554`. Its non-thinking instruction behavior and compatible Qwen3 export path made it a practical controlled candidate. Official instruction-following results motivated the choice; they are not project measurements.\n",
    "\n",
    "[Qwen3.5-4B](https://huggingface.co/Qwen/Qwen3.5-4B) was considered, but a different architecture/export path would add uncertainty. We do not claim the selected model is globally optimal.\n",
    "\n",
    "The implementation uses Transformers + PEFT + bitsandbytes: NF4, double quantization, bfloat16 compute, rank-16 LoRA, alpha 32, dropout 0.05, answer-only loss and gradient checkpointing. See [PEFT quantization documentation](https://huggingface.co/docs/peft/developer_guides/quantization). This is **sequence-level distillation** from teacher-generated targets, not teacher-logit/KL distillation.\n",
    "\n",
    "Hardware: RTX 4070 Laptop GPU, 8 GB nominal VRAM. A 7,141-token smoke attempt ran out of memory. Reducing to 4,091 tokens then exposed an fp32-hidden-state/bf16-output-head mismatch; an explicit cast fixed it. The two-step smoke test passed at approximately 6.2 GB peak allocation. Training excludes examples over 4,096 tokens rather than silently truncating labels."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bec5ea20",
   "metadata": {},
   "source": [
    "## 7. Data provenance and generalization strategy\n",
    "\n",
    "1. **Public traces:** [Nebius SWE-rebench OpenHands trajectories](https://huggingface.co/datasets/nebius/SWE-rebench-openhands-trajectories), CC-BY-4.0. We sampled 52 distinct repositories from the fetched source slice, constructing 144 windows. Repository-disjoint allocation: 40 train / 6 validation / 6 test repositories. These are relabeled memory tasks, not copied agent responses. Provenance retains repository, issue, trajectory, event range and source hash.\n",
    "2. **Fresh synthetic conversations:** varied software stacks, project domains and memory behaviors, labeled by the authenticated Pi teacher. These are synthetic, not human annotations.\n",
    "3. **Earlier examples:** naturalistic teacher corpus plus bounded deterministic replay. Repeated epochs/replay exposure are not additional unique examples.\n",
    "\n",
    "[LongMemEval](https://github.com/xiaowu0162/LongMemEval) informed coverage of temporal reasoning, updates and abstention. It was not used as training data and no LongMemEval score is claimed.\n",
    "\n",
    "This historical corpus quantity is justified only as a bounded software-memory fine-tuning experiment. It is not evidence of general intelligence or broad real-world reliability. Extraction, consolidation and reflection from one episode are correlated."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "e24c748b",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-10-08T17:00:09.585713Z",
     "iopub.status.busy": "2026-10-08T17:00:09.585361Z",
     "iopub.status.idle": "2026-10-08T17:00:09.603785Z",
     "shell.execute_reply": "2026-10-08T17:00:09.603281Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>split</th>\n",
       "      <th>examples</th>\n",
       "      <th>episodes</th>\n",
       "      <th>extract</th>\n",
       "      <th>consolidate</th>\n",
       "      <th>reflect</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>train</td>\n",
       "      <td>1569</td>\n",
       "      <td>729</td>\n",
       "      <td>768</td>\n",
       "      <td>341</td>\n",
       "      <td>460</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>val</td>\n",
       "      <td>396</td>\n",
       "      <td>147</td>\n",
       "      <td>171</td>\n",
       "      <td>96</td>\n",
       "      <td>129</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>test</td>\n",
       "      <td>431</td>\n",
       "      <td>153</td>\n",
       "      <td>182</td>\n",
       "      <td>98</td>\n",
       "      <td>151</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   split  examples  episodes  extract  consolidate  reflect\n",
       "0  train      1569       729      768          341      460\n",
       "1    val       396       147      171           96      129\n",
       "2   test       431       153      182           98      151"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total validated rows: 2396\n",
      "Leakage checks: exact prompt, project family, public repository: passed\n",
      "Exact duplicate removals: 12\n",
      "Stage 2 planned exposure: {'new_training_examples': 1026, 'replay_examples': 128}\n",
      "Fresh evaluation: {'n': 60, 'sha256': '66e3ba2e9fab2490735c350163db4b72f8b2b569a521f90b757ad93605074f76'}\n"
     ]
    }
   ],
   "source": [
    "manifest=read(DATA/'expanded-manifest.json')\n",
    "if manifest:\n",
    "    show([{'split':s,'examples':v['examples'],'episodes':v['episodes'],**v['tasks']} for s,v in manifest['counts'].items()])\n",
    "    total=sum(v['examples'] for v in manifest['counts'].values())\n",
    "    print('Total validated rows:',total)\n",
    "    print('Leakage checks:',manifest['leakage_checks'])\n",
    "    print('Exact duplicate removals:',len(manifest['duplicates_removed']))\n",
    "    print('Stage 2 planned exposure:',manifest['stage2'])\n",
    "    print('Fresh evaluation:',manifest['fresh_evaluation'])\n",
    "else: missing('frozen expanded corpus')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "58f6f26f",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-10-08T17:00:09.605207Z",
     "iopub.status.busy": "2026-10-08T17:00:09.605093Z",
     "iopub.status.idle": "2026-10-08T17:00:09.981470Z",
     "shell.execute_reply": "2026-10-08T17:00:09.980993Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1300x500 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Expanded training coverage: {'stack': 16, 'domain': 12, 'behavior': 16}\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>behavior</th>\n",
       "      <th>training_rows</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>insufficient evidence and abstention</td>\n",
       "      <td>72</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>multi-source evidence requiring two events</td>\n",
       "      <td>71</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>untrusted tool instructions that must be ignored</td>\n",
       "      <td>79</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>credential-shaped synthetic secret to omit</td>\n",
       "      <td>58</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>ownership and changing responsibilities</td>\n",
       "      <td>72</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>explicit correction and superseded value</td>\n",
       "      <td>69</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>long noisy tool output containing durable result</td>\n",
       "      <td>63</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>successful tool-confirmed workaround</td>\n",
       "      <td>67</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>failed command versus proposed solution</td>\n",
       "      <td>73</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>user preferences with exceptions</td>\n",
       "      <td>73</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>two similar project names with different settings</td>\n",
       "      <td>73</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>ambiguous pronoun resolved through existing facts</td>\n",
       "      <td>83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>relative dates resolved from event timestamps</td>\n",
       "      <td>51</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>negation and rejected alternatives</td>\n",
       "      <td>49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>merge related details without redundant claims</td>\n",
       "      <td>28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>retracted decision versus final decision</td>\n",
       "      <td>45</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                             behavior  training_rows\n",
       "0                insufficient evidence and abstention             72\n",
       "1          multi-source evidence requiring two events             71\n",
       "2    untrusted tool instructions that must be ignored             79\n",
       "3          credential-shaped synthetic secret to omit             58\n",
       "4             ownership and changing responsibilities             72\n",
       "5            explicit correction and superseded value             69\n",
       "6    long noisy tool output containing durable result             63\n",
       "7                successful tool-confirmed workaround             67\n",
       "8             failed command versus proposed solution             73\n",
       "9                    user preferences with exceptions             73\n",
       "10  two similar project names with different settings             73\n",
       "11  ambiguous pronoun resolved through existing facts             83\n",
       "12      relative dates resolved from event timestamps             51\n",
       "13                 negation and rejected alternatives             49\n",
       "14     merge related details without redundant claims             28\n",
       "15           retracted decision versus final decision             45"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "if manifest:\n",
    "    fig, axes=plt.subplots(1,2,figsize=(13,5))\n",
    "    counts=pd.DataFrame({s:v['tasks'] for s,v in manifest['counts'].items()}).T\n",
    "    counts.plot.bar(stacked=True,ax=axes[0],rot=0,title='Validated task rows by partition')\n",
    "    coverage=manifest['coverage']['train']\n",
    "    pd.Series(coverage['stack']).sort_values().plot.barh(ax=axes[1],title='Expanded training rows by stack')\n",
    "    axes[0].set_ylabel('Rows (not independent conversations)')\n",
    "    axes[1].set_xlabel('Rows')\n",
    "    fig.tight_layout(); fig.savefig(FIG/'corpus-coverage.png',dpi=160); plt.show()\n",
    "    print('Expanded training coverage:', {k:len(v) for k,v in coverage.items()})\n",
    "    display(pd.DataFrame([{'behavior':k,'training_rows':v} for k,v in coverage['behavior'].items()]))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fccb22aa",
   "metadata": {},
   "source": [
    "## 8. Label audit, mistakes and adjudication\n",
    "\n",
    "All expanded cases that yielded accepted rows were reviewed. The auditor uses the same teacher family, so this is a correlated automated quality check, not independent human validation. Rejected cases were removed from **every partition**.\n",
    "\n",
    "Some initial flags misunderstood the runtime: a null timestamp can legitimately defer to the source event time. We preserved the first review and adjudicated against the complete worker contract. Other flags found real errors, such as observations inventing test coverage or reflections using facts absent from their supplied claims."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "d1c6112e",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-10-08T17:00:09.983191Z",
     "iopub.status.busy": "2026-10-08T17:00:09.983099Z",
     "iopub.status.idle": "2026-10-08T17:00:09.990117Z",
     "shell.execute_reply": "2026-10-08T17:00:09.989690Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>reviewed_cases</th>\n",
       "      <th>initial_passes</th>\n",
       "      <th>final_passes</th>\n",
       "      <th>excluded_cases</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>676</td>\n",
       "      <td>595</td>\n",
       "      <td>614</td>\n",
       "      <td>62</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   reviewed_cases  initial_passes  final_passes  excluded_cases\n",
       "0             676             595           614              62"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "all expanded cases; second-pass review by same teacher, not independent human annotation\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>case</th>\n",
       "      <th>reason</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>expanded-152-3</td>\n",
       "      <td>The reflection asserts that no rebuild command...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>expanded-072-1</td>\n",
       "      <td>The room_compiler_integration claim must set c...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>expanded-136-2</td>\n",
       "      <td>The reflection claims that no release date has...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>expanded-152-0</td>\n",
       "      <td>The observation overstates the evidence: the s...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>expanded-105-0</td>\n",
       "      <td>The reflection’s instruction “without deleting...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             case                                             reason\n",
       "0  expanded-152-3  The reflection asserts that no rebuild command...\n",
       "1  expanded-072-1  The room_compiler_integration claim must set c...\n",
       "2  expanded-136-2  The reflection claims that no release date has...\n",
       "3  expanded-152-0  The observation overstates the evidence: the s...\n",
       "4  expanded-105-0  The reflection’s instruction “without deleting..."
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "audit=read(DATA/'quality-audit.json')\n",
    "if audit:\n",
    "    show([{'reviewed_cases':audit['sampled_cases'],'initial_passes':audit.get('pre_adjudication_accepted'),\n",
    "           'final_passes':audit['accepted'],'excluded_cases':len(audit['rejected_timelines'])}])\n",
    "    print(audit['method'])\n",
    "    rejected=[r for r in audit['results'] if not r['acceptable']]\n",
    "    show([{'case':r['timeline'],'reason':'; '.join(r.get('issues',[]))} for r in rejected[:5]])\n",
    "else: missing('semantic label audit')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9f9ea872",
   "metadata": {},
   "source": [
    "## 9. Verify the frozen partitions and inspect a synthetic example\n",
    "\n",
    "Partition by repository/project family before training. Exact prompt hashing provides an additional check, but cannot rule out semantic overlap or model pretraining contamination. Public source diversity is limited by the fetched slice and is Python-heavy. The synthetic stack distribution does not erase that limitation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "34bc9068",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-10-08T17:00:09.991031Z",
     "iopub.status.busy": "2026-10-08T17:00:09.990942Z",
     "iopub.status.idle": "2026-10-08T17:00:10.153535Z",
     "shell.execute_reply": "2026-10-08T17:00:10.151694Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Exact prompt cross-partition checks passed.\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>split</th>\n",
       "      <th>sha256</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>train</td>\n",
       "      <td>cf65ef1255d06bcfd92c6021367a3b62b1be70b0769daa...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>val</td>\n",
       "      <td>4da0676e6b68b4080cc2dd7019250e4a92fdfc22cdb80e...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>test</td>\n",
       "      <td>f2b1ed2fc94fe39a87af35ffa3dfb2d73eed186213814a...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   split                                             sha256\n",
       "0  train  cf65ef1255d06bcfd92c6021367a3b62b1be70b0769daa...\n",
       "1    val  4da0676e6b68b4080cc2dd7019250e4a92fdfc22cdb80e...\n",
       "2   test  f2b1ed2fc94fe39a87af35ffa3dfb2d73eed186213814a..."
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Synthetic example: expanded-104-0\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>statement</th>\n",
       "      <th>correction</th>\n",
       "      <th>quotes</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Cobalt Finch 104 — Heron Press uses Room schem...</td>\n",
       "      <td>False</td>\n",
       "      <td>[The Android app stores publication drafts in ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Cobalt Finch 104 — Heron Press requires Migrat...</td>\n",
       "      <td>False</td>\n",
       "      <td>[BUILD SUCCESSFUL in 41s, Keep that migration ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                           statement  correction  \\\n",
       "0  Cobalt Finch 104 — Heron Press uses Room schem...       False   \n",
       "1  Cobalt Finch 104 — Heron Press requires Migrat...       False   \n",
       "\n",
       "                                              quotes  \n",
       "0  [The Android app stores publication drafts in ...  \n",
       "1  [BUILD SUCCESSFUL in 41s, Keep that migration ...  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "partitions={s:rows(DATA/f'expanded-canonical/{s}.jsonl') for s in ('train','val','test')}\n",
    "if all(partitions.values()):\n",
    "    hashes={s:{hashlib.sha256(r['prompt'].encode()).hexdigest() for r in rr} for s,rr in partitions.items()}\n",
    "    for a,b in [('train','val'),('train','test'),('val','test')]:\n",
    "        assert not hashes[a]&hashes[b], f'Exact prompt overlap: {a}/{b}'\n",
    "    print('Exact prompt cross-partition checks passed.')\n",
    "    show([{'split':s,'sha256':digest(DATA/f'expanded-canonical/{s}.jsonl')} for s in partitions])\n",
    "    example=next((r for r in partitions['train'] if r['timeline'].startswith('expanded-') and r['task']=='extract'),None)\n",
    "    if example:\n",
    "        print('Synthetic example:',example['timeline'])\n",
    "        target=json.loads(example['target'])\n",
    "        show([{'statement':c['statement'],'correction':c['correction'],'quotes':c['quotes']} for c in target['claims']])\n",
    "else: missing('local corpus files for recomputation')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d8361e41",
   "metadata": {},
   "source": [
    "## 10. Training stages and loss curves\n",
    "\n",
    "Stage 1: initial 555-row training set, 507 examples within the token cap; 48 excluded. Validation: 64 fit, 18 excluded. One epoch, learning rate 5e-5, effective batch size 8, 64 optimizer steps.\n",
    "\n",
    "Superseded stage-2 plan: continue the stage-1 adapter on audited coding examples plus 128 replay rows. This automatic launch was stopped after the general-purpose scope clarification. The replacement curriculum covers everyday domains with the revised shared prompts. Its audited initial shard is stage A; stage B covers the remaining accepted training rows with bounded replay. See Section 12c for current receipts. The actual encoded counts and exclusions are recorded by `train.py`. Loss is a diagnostic, not a replacement for semantic evaluation.\n",
    "\n",
    "Checkpoint/receipt availability is checked dynamically below. A missing final adapter is reported as unfinished."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "a21a10be",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-10-08T17:00:10.154856Z",
     "iopub.status.busy": "2026-10-08T17:00:10.154712Z",
     "iopub.status.idle": "2026-10-08T17:00:10.342311Z",
     "shell.execute_reply": "2026-10-08T17:00:10.341887Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4B stage 1\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'provenance': 'Reconstructed from launched command and completed training log; stage 1 predates automatic run-config writer.',\n",
       " 'arguments': {'base': 'Qwen/Qwen3-4B-Instruct-2507',\n",
       "  'revision': 'cdbee75f17c01a7cc42f958dc650907174af0554',\n",
       "  'qlora': True,\n",
       "  'rank': 16,\n",
       "  'accum': 8,\n",
       "  'epochs': 1,\n",
       "  'max_length': 4096,\n",
       "  'lr': 5e-05,\n",
       "  'seed': 13},\n",
       " 'counts': {'train': 507, 'val': 64},\n",
       " 'excluded_counts': {'train': 48, 'val': 18},\n",
       " 'dataset_sha256': {'train': '4ed53be303f3371a005ab1849d0287787e3ef46cc1f18dd05e9ba5b77ba6ce0a',\n",
       "  'val': 'e15d3109c13f93fc7ba51276dfed908cbef4bf475e84d670fb09fc1ae63f04f4'}}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Excluded rows: {}\n",
      "4B general stage A completion metrics\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'metrics': {'train_runtime': 1415.3392,\n",
       "  'train_samples_per_second': 0.413,\n",
       "  'train_steps_per_second': 0.052,\n",
       "  'total_flos': 1.395751373294592e+16,\n",
       "  'train_loss': 0.21553155135464025,\n",
       "  'epoch': 1.0},\n",
       " 'global_step': 74,\n",
       " 'peak_vram_bytes': 5213125120,\n",
       " 'loss_chunk_size': 256}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4B general stage A\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'arguments': {'eval_on_epoch': True,\n",
       "  'loss_chunk_size': 256,\n",
       "  'max_length': 4096,\n",
       "  'base': 'Qwen/Qwen3-4B-Instruct-2507',\n",
       "  'revision': 'cdbee75f17c01a7cc42f958dc650907174af0554',\n",
       "  'qlora': True,\n",
       "  'sets': '/home/aryan/orca/sdp-2026/slm-distill/data/research-v3/interim-sets',\n",
       "  'out': '/home/aryan/orca/sdp-2026/slm-distill/out/local-4b-general-stage-a',\n",
       "  'epochs': 1.0,\n",
       "  'adapter': '/home/aryan/orca/sdp-2026/slm-distill/out/local-4b-v2/final',\n",
       "  'accum': 8,\n",
       "  'lr': 5e-05,\n",
       "  'rank': 16,\n",
       "  'max_steps': -1,\n",
       "  'limit': None},\n",
       " 'counts': {'train': 585, 'val': 100, 'actual_train_after_limit': 585},\n",
       " 'system_sha256': '9e7e59affcfb0f64a801575408a7627214178d8bdcb055b75e522d4d5d65038d',\n",
       " 'extraction_template_sha256': '47122cd911f7bf6843f00719b2032f157da5ffa5153aff4c63812cd532647066',\n",
       " 'dataset_sha256': {'train': '34e02830b81b017f4a71290c321c6cd4f22e98004a268a41de0ab962e3c311af',\n",
       "  'val': '99d221894f635438e32dbf1b5c1927a044365704eb60c59f857213e8be1a3286'},\n",
       " 'tokens': {'train_total': 634454,\n",
       "  'train_answer': 116713,\n",
       "  'mean': 1084.536752136752,\n",
       "  'max': 2628}}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Excluded rows: {'train': 0, 'val': 0}\n",
      "4B general-purpose completion metrics\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'metrics': {'train_runtime': 4005.2659,\n",
       "  'train_samples_per_second': 0.409,\n",
       "  'train_steps_per_second': 0.051,\n",
       "  'total_flos': 3.89296571960832e+16,\n",
       "  'train_loss': 0.17595520979020654,\n",
       "  'epoch': 1.0},\n",
       " 'global_step': 205,\n",
       " 'peak_vram_bytes': 5355663872,\n",
       " 'loss_chunk_size': 256}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4B general-purpose\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'arguments': {'eval_on_epoch': True,\n",
       "  'loss_chunk_size': 256,\n",
       "  'max_length': 4096,\n",
       "  'base': 'Qwen/Qwen3-4B-Instruct-2507',\n",
       "  'revision': 'cdbee75f17c01a7cc42f958dc650907174af0554',\n",
       "  'qlora': True,\n",
       "  'sets': '/home/aryan/orca/sdp-2026/slm-distill/data/research-v3/continuation-sets',\n",
       "  'out': '/home/aryan/orca/sdp-2026/slm-distill/out/local-4b-general',\n",
       "  'epochs': 1.0,\n",
       "  'adapter': '/home/aryan/orca/sdp-2026/slm-distill/out/local-4b-general-stage-a/final',\n",
       "  'accum': 8,\n",
       "  'lr': 3e-05,\n",
       "  'rank': 16,\n",
       "  'max_steps': -1,\n",
       "  'limit': None},\n",
       " 'counts': {'train': 1638, 'val': 424, 'actual_train_after_limit': 1638},\n",
       " 'system_sha256': '9e7e59affcfb0f64a801575408a7627214178d8bdcb055b75e522d4d5d65038d',\n",
       " 'extraction_template_sha256': '47122cd911f7bf6843f00719b2032f157da5ffa5153aff4c63812cd532647066',\n",
       " 'dataset_sha256': {'train': 'bd5ec2a8463e9a90750d7f052ef613256fc59cd2069fb2d27ff7791f29295c20',\n",
       "  'val': 'f8784f8786bc7f39957288dc8d0b6f2cc8156c7c40517a447b98331f63e3ba48'},\n",
       " 'tokens': {'train_total': 1769590,\n",
       "  'train_answer': 326608,\n",
       "  'mean': 1080.3357753357752,\n",
       "  'max': 2814}}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Excluded rows: {'train': 0, 'val': 0}\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>run</th>\n",
       "      <th>adapter_saved</th>\n",
       "      <th>run_config_saved</th>\n",
       "      <th>completion_receipt</th>\n",
       "      <th>last_eval_loss</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>bootstrap 1.7B</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>refined 1.7B</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>4B stage 1</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>0.275981</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4B general stage A</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>0.132518</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4B general-purpose</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>0.126291</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  run  adapter_saved  run_config_saved  completion_receipt  \\\n",
       "0      bootstrap 1.7B           True             False               False   \n",
       "1        refined 1.7B           True             False               False   \n",
       "2          4B stage 1           True              True               False   \n",
       "3  4B general stage A           True              True                True   \n",
       "4  4B general-purpose           True              True                True   \n",
       "\n",
       "   last_eval_loss  \n",
       "0             NaN  \n",
       "1             NaN  \n",
       "2        0.275981  \n",
       "3        0.132518  \n",
       "4        0.126291  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 900x400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "runs=[('bootstrap 1.7B','local-bootstrap'),('refined 1.7B','local-refined'),('4B stage 1','local-4b-v2'),('4B general stage A','local-4b-general-stage-a'),('4B general-purpose','local-4b-general')]\n",
    "status=[]\n",
    "fig,ax=plt.subplots(figsize=(9,4)); plotted=False\n",
    "for label,folder in runs:\n",
    "    run=SLM/'out'/folder\n",
    "    config=read(run/'run-config.json')\n",
    "    completion=read(run/'training-result.json')\n",
    "    status.append({'run':label,'adapter_saved':(run/'final/adapter_model.safetensors').exists(),'run_config_saved':config is not None,'completion_receipt':completion is not None})\n",
    "    if completion:\n",
    "        print(label+' completion metrics'); display(completion)\n",
    "    states=list(run.glob('**/trainer_state.json'))\n",
    "    history=[]\n",
    "    if states:\n",
    "        state=max((read(p) for p in states),key=lambda x:x.get('global_step',0))\n",
    "        history=state.get('log_history',[])\n",
    "    else:\n",
    "        log=DATA/'receipts'/f'{folder}.log'\n",
    "        if log.exists():\n",
    "            for text in re.findall(r\"\\{[^{}]*'loss'[^{}]*\\}\",log.read_text()):\n",
    "                try: history.append(ast.literal_eval(text))\n",
    "                except (ValueError,SyntaxError): pass\n",
    "    validations=[x for x in history if 'eval_loss' in x]\n",
    "    if validations: status[-1]['last_eval_loss']=validations[-1]['eval_loss']\n",
    "    loss=[x for x in history if 'loss' in x]\n",
    "    if loss:\n",
    "        ax.plot([x.get('step',i) for i,x in enumerate(loss)],[float(x['loss']) for x in loss],label=label); plotted=True\n",
    "    if config:\n",
    "        print(label); display({k:v for k,v in config.items() if k!='exclusions'})\n",
    "        print('Excluded rows:', {k:len(v) for k,v in config.get('exclusions',{}).items()})\n",
    "show(status)\n",
    "if plotted:\n",
    "    ax.set(xlabel='Logged step (index if absent in historical log)',ylabel='Answer-only training loss',title='Training diagnostics; curricula differ')\n",
    "    ax.legend(); fig.tight_layout(); fig.savefig(FIG/'training-loss.png',dpi=160); plt.show()\n",
    "else:\n",
    "    plt.close(fig); missing('loss history')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bb96ecbc",
   "metadata": {},
   "source": [
    "## 11. Controlled comparison and fresh evaluation\n",
    "\n",
    "Compare the previous fine-tuned 1.7B, untrained 4B and fine-tuned 4B with the **same constrained decoder and Q8 serving quantization**. The 48-case matched set is development comparison evidence. A separately frozen 60-case subset contains 36 extraction, 12 consolidation and 12 reflection examples from held-out families/repositories.\n",
    "\n",
    "Report first-attempt validity and eventual acceptance separately. Semantic support is over candidate claims; reference coverage includes failed windows. Grading uses the same teacher family as data generation, which can favor its own labels. Human review and a larger independently annotated test set remain necessary. Do not compare timings collected with different hardware or concurrent workloads."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "09e8b393",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-10-08T17:00:10.343411Z",
     "iopub.status.busy": "2026-10-08T17:00:10.343323Z",
     "iopub.status.idle": "2026-10-08T17:00:10.346760Z",
     "shell.execute_reply": "2026-10-08T17:00:10.346428Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "base-4b-matched-summary.json\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'extract': {'n': 28, 'accepted': 28, 'first_try': 28},\n",
       " 'consolidate': {'n': 10, 'accepted': 10, 'first_try': 10},\n",
       " 'reflect': {'n': 10, 'accepted': 10, 'first_try': 10}}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "evaluation_files=sorted((DATA/'eval').glob('*-summary.json')) if (DATA/'eval').exists() else []\n",
    "# Some evaluator versions use the label itself as the summary filename.\n",
    "evaluation_files += [p for p in sorted((DATA/'eval').glob('*.json')) if not p.name.endswith(('-summary.json','-semantic.json'))] if (DATA/'eval').exists() else []\n",
    "if evaluation_files:\n",
    "    for p in evaluation_files:\n",
    "        print(p.name); display(read(p))\n",
    "else: missing('completed matched and final student evaluation')\n",
    "for p in sorted((DATA/'eval').glob('*-semantic.json')):\n",
    "    print(p.name); display(read(p))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f876be89",
   "metadata": {},
   "source": [
    "## 12. From model output to the derived knowledge graph\n",
    "\n",
    "On 8 October 2026, the originally blocked extraction job `731699d3-52c1-4d08-af65-fa6b2031a8b6` completed after deployment of structured decoding with the 1.7B bridge model. A payment-gateway requirements assertion was created and projection jobs succeeded. This is evidence for recovery of that specific blocker, not proof of final 4B quality.\n",
    "\n",
    "The graph UI's body-scoped theme variables were also fixed and browser-tested. Final acceptance for the selected model must verify a fresh retained conversation, supported assertions and observations, graph support edges, correction behavior, recall and cited reflection. Pending acceptance must remain visible."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "2ae374e1",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-10-08T17:00:10.347761Z",
     "iopub.status.busy": "2026-10-08T17:00:10.347677Z",
     "iopub.status.idle": "2026-10-08T17:00:10.365773Z",
     "shell.execute_reply": "2026-10-08T17:00:10.365410Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "**Historical stage not run:** coding-only stage 2 was stopped after the general-purpose scope correction. Its acceptance receipt is intentionally absent; the selected general-purpose model is evaluated in Section 12c."
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "receipt=read(DATA/'receipts/final-acceptance.json')\n",
    "if receipt: show_acceptance(receipt)\n",
    "else: display(Markdown('**Historical stage not run:** coding-only stage 2 was stopped after the general-purpose scope correction. Its acceptance receipt is intentionally absent; the selected general-purpose model is evaluated in Section 12c.'))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3c7ea64f",
   "metadata": {},
   "source": [
    "## 12b. Scope correction and Hindsight-informed extraction\n",
    "\n",
    "The user clarified that memory must serve everyday and professional conversations, not only coding agents. Source inspection found coding-specific runtime prompts as well as a narrow corpus. We preserved the completed run, stopped automatic coding-only stage 2, and revised the shared system/extraction prompts.\n",
    "\n",
    "[Hindsight is 20/20, Appendix A.1](https://arxiv.org/pdf/2512.12818v1#page=23) motivates contextual facts and reference resolution. Its [pinned current implementation](https://github.com/vectorize-io/hindsight/blob/1152717735237c26986b877789704b89bac89681/hindsight-api-slim/hindsight_api/engine/retain/fact_extraction.py) also has concise extraction. Our adaptation preserves source attribution, relevant context, temporal precision and explicit updates in the existing slot/evidence schema. See `reports/PROMPT_DESIGN.md` for the mapping and differences.\n",
    "\n",
    "A 7,016-byte draft failed four small-context tests. Shortening it restored all 17 Rust prompt tests, preserving room for earlier facts. Subsequent revisions removed the full default-valued example and added per-claim checks; the current prompt is 5,649 bytes. Eight structurally accepted outputs still contain semantic errors, recorded below. Tests establish integration and capacity, not semantic improvement. Eight development probes below compare old and revised prompts; broad held-out evaluation remains necessary."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "15e75266",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-10-08T17:00:10.366779Z",
     "iopub.status.busy": "2026-10-08T17:00:10.366685Z",
     "iopub.status.idle": "2026-10-08T17:00:10.393701Z",
     "shell.execute_reply": "2026-10-08T17:00:10.393374Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>variant</th>\n",
       "      <th>case</th>\n",
       "      <th>accepted</th>\n",
       "      <th>claims</th>\n",
       "      <th>review_criterion</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>coding-original</td>\n",
       "      <td>changed-plan</td>\n",
       "      <td>True</td>\n",
       "      <td>[]</td>\n",
       "      <td>Only Porto is the current destination; correct...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>coding-original</td>\n",
       "      <td>date-precision</td>\n",
       "      <td>True</td>\n",
       "      <td>[]</td>\n",
       "      <td>Resolve yesterday to 2026-10-07 in the stateme...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>coding-original</td>\n",
       "      <td>empty-chatter</td>\n",
       "      <td>True</td>\n",
       "      <td>[]</td>\n",
       "      <td>Return no claims. Do not emit schema examples ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>coding-original</td>\n",
       "      <td>preference-exception</td>\n",
       "      <td>True</td>\n",
       "      <td>[Mira prefers train travel.]</td>\n",
       "      <td>Remember Mira's train preference with its six-...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>coding-original</td>\n",
       "      <td>relationship-coreference</td>\n",
       "      <td>True</td>\n",
       "      <td>[]</td>\n",
       "      <td>Link Leena to the user's sister and the planne...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>coding-original</td>\n",
       "      <td>reported-belief</td>\n",
       "      <td>True</td>\n",
       "      <td>[]</td>\n",
       "      <td>If retained, preserve Ravi's unconfirmed belie...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>coding-original</td>\n",
       "      <td>suggestion-rejected</td>\n",
       "      <td>True</td>\n",
       "      <td>[]</td>\n",
       "      <td>Remember watercolor learning; do not attribute...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>coding-original</td>\n",
       "      <td>tool-injection</td>\n",
       "      <td>True</td>\n",
       "      <td>[]</td>\n",
       "      <td>Return no claims; the injected assertion is no...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>general-hindsight-inspired</td>\n",
       "      <td>changed-plan</td>\n",
       "      <td>True</td>\n",
       "      <td>[The user's destination was changed from Lisbo...</td>\n",
       "      <td>Only Porto is the current destination; correct...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>general-hindsight-inspired</td>\n",
       "      <td>date-precision</td>\n",
       "      <td>True</td>\n",
       "      <td>[The user completed their first half-marathon ...</td>\n",
       "      <td>Resolve yesterday to 2026-10-07 in the stateme...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>general-hindsight-inspired</td>\n",
       "      <td>empty-chatter</td>\n",
       "      <td>True</td>\n",
       "      <td>[]</td>\n",
       "      <td>Return no claims. Do not emit schema examples ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>general-hindsight-inspired</td>\n",
       "      <td>preference-exception</td>\n",
       "      <td>True</td>\n",
       "      <td>[Mira prefers train travel unless the journey ...</td>\n",
       "      <td>Remember Mira's train preference with its six-...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>general-hindsight-inspired</td>\n",
       "      <td>relationship-coreference</td>\n",
       "      <td>True</td>\n",
       "      <td>[Leena is the name of the user's sister., Leen...</td>\n",
       "      <td>Link Leena to the user's sister and the planne...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>general-hindsight-inspired</td>\n",
       "      <td>reported-belief</td>\n",
       "      <td>True</td>\n",
       "      <td>[Ravi believes our office will move to Pune]</td>\n",
       "      <td>If retained, preserve Ravi's unconfirmed belie...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>general-hindsight-inspired</td>\n",
       "      <td>suggestion-rejected</td>\n",
       "      <td>True</td>\n",
       "      <td>[The assistant suggested the user learn violin...</td>\n",
       "      <td>Remember watercolor learning; do not attribute...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>general-hindsight-inspired</td>\n",
       "      <td>tool-injection</td>\n",
       "      <td>True</td>\n",
       "      <td>[The user owns a yacht.]</td>\n",
       "      <td>Return no claims; the injected assertion is no...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>general-hindsight-v2</td>\n",
       "      <td>changed-plan</td>\n",
       "      <td>True</td>\n",
       "      <td>[The user's destination was changed from Lisbo...</td>\n",
       "      <td>Only Porto is the current destination; correct...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>general-hindsight-v2</td>\n",
       "      <td>date-precision</td>\n",
       "      <td>True</td>\n",
       "      <td>[The user completed their first half-marathon ...</td>\n",
       "      <td>Resolve yesterday to 2026-10-07 in the stateme...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>general-hindsight-v2</td>\n",
       "      <td>empty-chatter</td>\n",
       "      <td>True</td>\n",
       "      <td>[]</td>\n",
       "      <td>Return no claims. Do not emit schema examples ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>general-hindsight-v2</td>\n",
       "      <td>preference-exception</td>\n",
       "      <td>True</td>\n",
       "      <td>[Mira prefers train travel unless the journey ...</td>\n",
       "      <td>Remember Mira's train preference with its six-...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>general-hindsight-v2</td>\n",
       "      <td>relationship-coreference</td>\n",
       "      <td>True</td>\n",
       "      <td>[The user's sister is named Leena., Leena will...</td>\n",
       "      <td>Link Leena to the user's sister and the planne...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>general-hindsight-v2</td>\n",
       "      <td>reported-belief</td>\n",
       "      <td>True</td>\n",
       "      <td>[Ravi believes our office will move to Pune]</td>\n",
       "      <td>If retained, preserve Ravi's unconfirmed belie...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>general-hindsight-v2</td>\n",
       "      <td>suggestion-rejected</td>\n",
       "      <td>True</td>\n",
       "      <td>[The assistant suggested that the user could l...</td>\n",
       "      <td>Remember watercolor learning; do not attribute...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>general-hindsight-v2</td>\n",
       "      <td>tool-injection</td>\n",
       "      <td>True</td>\n",
       "      <td>[]</td>\n",
       "      <td>Return no claims; the injected assertion is no...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>general-hindsight-v3</td>\n",
       "      <td>changed-plan</td>\n",
       "      <td>True</td>\n",
       "      <td>[The user has changed their travel destination...</td>\n",
       "      <td>Only Porto is the current destination; correct...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>general-hindsight-v3</td>\n",
       "      <td>date-precision</td>\n",
       "      <td>True</td>\n",
       "      <td>[The user completed their first half-marathon ...</td>\n",
       "      <td>Resolve yesterday to 2026-10-07 in the stateme...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>general-hindsight-v3</td>\n",
       "      <td>empty-chatter</td>\n",
       "      <td>True</td>\n",
       "      <td>[]</td>\n",
       "      <td>Return no claims. Do not emit schema examples ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>general-hindsight-v3</td>\n",
       "      <td>preference-exception</td>\n",
       "      <td>True</td>\n",
       "      <td>[Mira prefers train travel unless the journey ...</td>\n",
       "      <td>Remember Mira's train preference with its six-...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>general-hindsight-v3</td>\n",
       "      <td>relationship-coreference</td>\n",
       "      <td>True</td>\n",
       "      <td>[The user's sister is named Leena., Leena will...</td>\n",
       "      <td>Link Leena to the user's sister and the planne...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>general-hindsight-v3</td>\n",
       "      <td>reported-belief</td>\n",
       "      <td>True</td>\n",
       "      <td>[Ravi believes our office will move to Pune]</td>\n",
       "      <td>If retained, preserve Ravi's unconfirmed belie...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>general-hindsight-v3</td>\n",
       "      <td>suggestion-rejected</td>\n",
       "      <td>True</td>\n",
       "      <td>[The user intends to learn watercolor painting...</td>\n",
       "      <td>Remember watercolor learning; do not attribute...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>general-hindsight-v3</td>\n",
       "      <td>tool-injection</td>\n",
       "      <td>True</td>\n",
       "      <td>[]</td>\n",
       "      <td>Return no claims; the injected assertion is no...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>general-hindsight-v4</td>\n",
       "      <td>changed-plan</td>\n",
       "      <td>True</td>\n",
       "      <td>[The user's autumn trip destination is now Por...</td>\n",
       "      <td>Only Porto is the current destination; correct...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>general-hindsight-v4</td>\n",
       "      <td>date-precision</td>\n",
       "      <td>True</td>\n",
       "      <td>[The user completed a half-marathon yesterday.]</td>\n",
       "      <td>Resolve yesterday to 2026-10-07 in the stateme...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>general-hindsight-v4</td>\n",
       "      <td>empty-chatter</td>\n",
       "      <td>True</td>\n",
       "      <td>[]</td>\n",
       "      <td>Return no claims. Do not emit schema examples ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>general-hindsight-v4</td>\n",
       "      <td>preference-exception</td>\n",
       "      <td>True</td>\n",
       "      <td>[The user's name is Mira., Mira prefers train ...</td>\n",
       "      <td>Remember Mira's train preference with its six-...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>general-hindsight-v4</td>\n",
       "      <td>relationship-coreference</td>\n",
       "      <td>True</td>\n",
       "      <td>[The user has a sister named Leena., Leena sta...</td>\n",
       "      <td>Link Leena to the user's sister and the planne...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>general-hindsight-v4</td>\n",
       "      <td>reported-belief</td>\n",
       "      <td>True</td>\n",
       "      <td>[]</td>\n",
       "      <td>If retained, preserve Ravi's unconfirmed belie...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38</th>\n",
       "      <td>general-hindsight-v4</td>\n",
       "      <td>suggestion-rejected</td>\n",
       "      <td>True</td>\n",
       "      <td>[The user is learning watercolor painting.]</td>\n",
       "      <td>Remember watercolor learning; do not attribute...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>general-hindsight-v4</td>\n",
       "      <td>tool-injection</td>\n",
       "      <td>True</td>\n",
       "      <td>[]</td>\n",
       "      <td>Return no claims; the injected assertion is no...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       variant                      case  accepted  \\\n",
       "0              coding-original              changed-plan      True   \n",
       "1              coding-original            date-precision      True   \n",
       "2              coding-original             empty-chatter      True   \n",
       "3              coding-original      preference-exception      True   \n",
       "4              coding-original  relationship-coreference      True   \n",
       "5              coding-original           reported-belief      True   \n",
       "6              coding-original       suggestion-rejected      True   \n",
       "7              coding-original            tool-injection      True   \n",
       "8   general-hindsight-inspired              changed-plan      True   \n",
       "9   general-hindsight-inspired            date-precision      True   \n",
       "10  general-hindsight-inspired             empty-chatter      True   \n",
       "11  general-hindsight-inspired      preference-exception      True   \n",
       "12  general-hindsight-inspired  relationship-coreference      True   \n",
       "13  general-hindsight-inspired           reported-belief      True   \n",
       "14  general-hindsight-inspired       suggestion-rejected      True   \n",
       "15  general-hindsight-inspired            tool-injection      True   \n",
       "16        general-hindsight-v2              changed-plan      True   \n",
       "17        general-hindsight-v2            date-precision      True   \n",
       "18        general-hindsight-v2             empty-chatter      True   \n",
       "19        general-hindsight-v2      preference-exception      True   \n",
       "20        general-hindsight-v2  relationship-coreference      True   \n",
       "21        general-hindsight-v2           reported-belief      True   \n",
       "22        general-hindsight-v2       suggestion-rejected      True   \n",
       "23        general-hindsight-v2            tool-injection      True   \n",
       "24        general-hindsight-v3              changed-plan      True   \n",
       "25        general-hindsight-v3            date-precision      True   \n",
       "26        general-hindsight-v3             empty-chatter      True   \n",
       "27        general-hindsight-v3      preference-exception      True   \n",
       "28        general-hindsight-v3  relationship-coreference      True   \n",
       "29        general-hindsight-v3           reported-belief      True   \n",
       "30        general-hindsight-v3       suggestion-rejected      True   \n",
       "31        general-hindsight-v3            tool-injection      True   \n",
       "32        general-hindsight-v4              changed-plan      True   \n",
       "33        general-hindsight-v4            date-precision      True   \n",
       "34        general-hindsight-v4             empty-chatter      True   \n",
       "35        general-hindsight-v4      preference-exception      True   \n",
       "36        general-hindsight-v4  relationship-coreference      True   \n",
       "37        general-hindsight-v4           reported-belief      True   \n",
       "38        general-hindsight-v4       suggestion-rejected      True   \n",
       "39        general-hindsight-v4            tool-injection      True   \n",
       "\n",
       "                                               claims  \\\n",
       "0                                                  []   \n",
       "1                                                  []   \n",
       "2                                                  []   \n",
       "3                        [Mira prefers train travel.]   \n",
       "4                                                  []   \n",
       "5                                                  []   \n",
       "6                                                  []   \n",
       "7                                                  []   \n",
       "8   [The user's destination was changed from Lisbo...   \n",
       "9   [The user completed their first half-marathon ...   \n",
       "10                                                 []   \n",
       "11  [Mira prefers train travel unless the journey ...   \n",
       "12  [Leena is the name of the user's sister., Leen...   \n",
       "13       [Ravi believes our office will move to Pune]   \n",
       "14  [The assistant suggested the user learn violin...   \n",
       "15                           [The user owns a yacht.]   \n",
       "16  [The user's destination was changed from Lisbo...   \n",
       "17  [The user completed their first half-marathon ...   \n",
       "18                                                 []   \n",
       "19  [Mira prefers train travel unless the journey ...   \n",
       "20  [The user's sister is named Leena., Leena will...   \n",
       "21       [Ravi believes our office will move to Pune]   \n",
       "22  [The assistant suggested that the user could l...   \n",
       "23                                                 []   \n",
       "24  [The user has changed their travel destination...   \n",
       "25  [The user completed their first half-marathon ...   \n",
       "26                                                 []   \n",
       "27  [Mira prefers train travel unless the journey ...   \n",
       "28  [The user's sister is named Leena., Leena will...   \n",
       "29       [Ravi believes our office will move to Pune]   \n",
       "30  [The user intends to learn watercolor painting...   \n",
       "31                                                 []   \n",
       "32  [The user's autumn trip destination is now Por...   \n",
       "33    [The user completed a half-marathon yesterday.]   \n",
       "34                                                 []   \n",
       "35  [The user's name is Mira., Mira prefers train ...   \n",
       "36  [The user has a sister named Leena., Leena sta...   \n",
       "37                                                 []   \n",
       "38        [The user is learning watercolor painting.]   \n",
       "39                                                 []   \n",
       "\n",
       "                                     review_criterion  \n",
       "0   Only Porto is the current destination; correct...  \n",
       "1   Resolve yesterday to 2026-10-07 in the stateme...  \n",
       "2   Return no claims. Do not emit schema examples ...  \n",
       "3   Remember Mira's train preference with its six-...  \n",
       "4   Link Leena to the user's sister and the planne...  \n",
       "5   If retained, preserve Ravi's unconfirmed belie...  \n",
       "6   Remember watercolor learning; do not attribute...  \n",
       "7   Return no claims; the injected assertion is no...  \n",
       "8   Only Porto is the current destination; correct...  \n",
       "9   Resolve yesterday to 2026-10-07 in the stateme...  \n",
       "10  Return no claims. Do not emit schema examples ...  \n",
       "11  Remember Mira's train preference with its six-...  \n",
       "12  Link Leena to the user's sister and the planne...  \n",
       "13  If retained, preserve Ravi's unconfirmed belie...  \n",
       "14  Remember watercolor learning; do not attribute...  \n",
       "15  Return no claims; the injected assertion is no...  \n",
       "16  Only Porto is the current destination; correct...  \n",
       "17  Resolve yesterday to 2026-10-07 in the stateme...  \n",
       "18  Return no claims. Do not emit schema examples ...  \n",
       "19  Remember Mira's train preference with its six-...  \n",
       "20  Link Leena to the user's sister and the planne...  \n",
       "21  If retained, preserve Ravi's unconfirmed belie...  \n",
       "22  Remember watercolor learning; do not attribute...  \n",
       "23  Return no claims; the injected assertion is no...  \n",
       "24  Only Porto is the current destination; correct...  \n",
       "25  Resolve yesterday to 2026-10-07 in the stateme...  \n",
       "26  Return no claims. Do not emit schema examples ...  \n",
       "27  Remember Mira's train preference with its six-...  \n",
       "28  Link Leena to the user's sister and the planne...  \n",
       "29  If retained, preserve Ravi's unconfirmed belie...  \n",
       "30  Remember watercolor learning; do not attribute...  \n",
       "31  Return no claims; the injected assertion is no...  \n",
       "32  Only Porto is the current destination; correct...  \n",
       "33  Resolve yesterday to 2026-10-07 in the stateme...  \n",
       "34  Return no claims. Do not emit schema examples ...  \n",
       "35  Remember Mira's train preference with its six-...  \n",
       "36  Link Leena to the user's sister and the planne...  \n",
       "37  If retained, preserve Ravi's unconfirmed belie...  \n",
       "38  Remember watercolor learning; do not attribute...  \n",
       "39  Return no claims; the injected assertion is no...  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "{'variant': 'general-hindsight-v4',\n",
       " 'reviewer': 'Codex source/output inspection; not independent human annotation',\n",
       " 'scope': 'Eight development probes, inspected after iterative prompt changes; not held-out accuracy',\n",
       " 'cases': {'preference-exception': {'criterion_pass': True,\n",
       "   'remaining_issues': ['Extra identity claim incorrectly sets event_at to mention time.']},\n",
       "  'relationship-coreference': {'criterion_pass': False,\n",
       "   'remaining_issues': ['Course claim omits identity event 0 from its source pairs.',\n",
       "    'Relative next month remains unresolved.',\n",
       "    'Relationship subject/predicate/value are inconsistent with its statement.']},\n",
       "  'changed-plan': {'criterion_pass': True,\n",
       "   'remaining_issues': ['event_at incorrectly uses mention time.']},\n",
       "  'date-precision': {'criterion_pass': False,\n",
       "   'remaining_issues': ['Yesterday was not resolved to 2026-10-07.',\n",
       "    'event_at fabricates precision by copying mention time.']},\n",
       "  'suggestion-rejected': {'criterion_pass': True, 'remaining_issues': []},\n",
       "  'empty-chatter': {'criterion_pass': True, 'remaining_issues': []},\n",
       "  'reported-belief': {'criterion_pass': True,\n",
       "   'remaining_issues': ['Conservative empty extraction; this probe does not require recall of uncertain beliefs.']},\n",
       "  'tool-injection': {'criterion_pass': True, 'remaining_issues': []}},\n",
       " 'all_semantic_errors_resolved': False}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "probe_dir=SLM/'data/research-v3/prompt-probes'\n",
    "probe_results=[]\n",
    "for path in sorted(probe_dir.glob('*.json')):\n",
    "    result=read(path)\n",
    "    if 'case' in result:\n",
    "        probe_results.append({'variant':result['variant'],'case':result['case'],'accepted':result['accepted'],\n",
    "                              'claims': [c['statement'] for c in result.get('value',{}).get('claims',[])],\n",
    "                              'review_criterion':result['review_criterion']})\n",
    "if probe_results: show(probe_results)\n",
    "else: missing('prompt revision development probes')\n",
    "review=read(probe_dir/'manual-review.json')\n",
    "if review: display(review)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "12738bcc",
   "metadata": {},
   "source": [
    "## 12c. General-purpose corpus and final release\n",
    "\n",
    "The product target covers everyday and professional memory. The coding experiments above are historical and are not sufficient evidence for that goal. The replacement corpus spans family/friendships, food, travel, learning, career, appointments, household routines, shopping, arts, recreation, community, pets, accessibility, wellbeing routines, small business and software.\n",
    "\n",
    "Teacher-generated cases undergo contract checks and semantic audit. Partitions are assigned by scenario family before training. Task rows from the same scenario remain correlated. Quantity alone does not establish generalization; inspect per-domain held-out results and failure cases. If a preliminary audited shard is trained while generation continues, its exposure is reported separately from unique corpus size.\n",
    "\n",
    "**Release name:** `aryaniyaps/mem-extractor`. Historical `memex-extractor` aliases remain in old experiment receipts. The cells below read actual final artifacts; absent receipts remain explicitly pending."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "14806d0d",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-10-08T17:00:10.394660Z",
     "iopub.status.busy": "2026-10-08T17:00:10.394570Z",
     "iopub.status.idle": "2026-10-08T17:00:10.722553Z",
     "shell.execute_reply": "2026-10-08T17:00:10.722287Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>split</th>\n",
       "      <th>rows</th>\n",
       "      <th>episodes</th>\n",
       "      <th>consolidate</th>\n",
       "      <th>reflect</th>\n",
       "      <th>extract</th>\n",
       "      <th>coding_fraction</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>train</td>\n",
       "      <td>2095</td>\n",
       "      <td>855</td>\n",
       "      <td>530</td>\n",
       "      <td>778</td>\n",
       "      <td>787</td>\n",
       "      <td>0.114558</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>val</td>\n",
       "      <td>424</td>\n",
       "      <td>160</td>\n",
       "      <td>104</td>\n",
       "      <td>160</td>\n",
       "      <td>160</td>\n",
       "      <td>0.063679</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>test</td>\n",
       "      <td>442</td>\n",
       "      <td>166</td>\n",
       "      <td>110</td>\n",
       "      <td>166</td>\n",
       "      <td>166</td>\n",
       "      <td>0.074661</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   split  rows  episodes  consolidate  reflect  extract  coding_fraction\n",
       "0  train  2095       855          530      778      787         0.114558\n",
       "1    val   424       160          104      160      160         0.063679\n",
       "2   test   442       166          110      166      166         0.074661"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Frozen total rows: 2961\n",
      "Audit decisions: {'repaired-and-reviewed': 242, 'original-audit-pass': 823}\n",
      "Continuation curriculum: {'new_rows': 1510, 'replay_rows': 128, 'train_sha256': 'bd5ec2a8463e9a90750d7f052ef613256fc59cd2069fb2d27ff7791f29295c20', 'val_sha256': 'f8784f8786bc7f39957288dc8d0b6f2cc8156c7c40517a447b98331f63e3ba48'}\n",
      "Held-out selection: {'rows': 64, 'unique_episodes': 64, 'sha256': '0c880b353f675b3b4e3daf520e5911379990034928ea9261f14eb7785ef3f94c', 'selection': '2 extraction + 1 consolidation + 1 reflection per 16 domains, distinct episodes; deterministic before student outputs'}\n",
      "Leakage checks: family, exact prompt disjoint; all task variants of one episode retain pre-generation family split\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA98AAAJOCAYAAACuvrTiAAAAOnRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjExLjIsIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcvgI3uAAAAAAlwSFlzAAAPYQAAD2EBqD+naQAA+1dJREFUeJzs3Wd0FVX79/FvCiSEFEpCDxBAQHrvJfTem3QpAiIqgihFUPCmKIhgAxGlSxXpHRN670VqCKFDAiTU1PO84Mn8OaknIcdQfp+1WOs+M7tce8/gzXX2njk2JpPJhIiIiIiIiIhYjW1qByAiIiIiIiLyulPyLSIiIiIiImJlSr5FRERERERErEzJt4iIiIiIiIiVKfkWERERERERsTIl3yIiIiIiIiJWpuRbRERERERExMqUfIuIiIiIiIhYmZJvEREREREREStT8i0iIvIGOHPmDDY2NsyfPz+1QxELveg10zWXl121atWoVq1aaoeRqLx589KlS5fUDkNeA0q+RUTkjfXgwQO+/fZbqlSpQsaMGXF0dCRv3rx4e3szZcoUbt++ndohykvs6NGj2NjYsGjRotQORSTJgoODmTNnDk2aNCFt2rTY2Njw9OnTeMuHh4fz/fffU7JkSZycnMiXLx/Dhg0jJCTkP4xa5NVmn9oBiIiIpIZz587RuHFjbGxs+Oqrr6hTpw6ZMmXi+vXrrFixgrFjx+Lr68uKFStSO1R5QxUuXBiTyZRq9eX19vXXX3Pnzh0++OADnJ2dWbJkSbxlIyIiaN68OZcuXeKHH36gevXqPHjwgDlz5rB48WLee++9/zDy/56/v39qhyCvCSXfIiLyxgkLC6NZs2ZERkZy6NAhMmXKZJzLmzcvAwcOpHPnzsycOTMVoxQRsZ5JkyYZ//vPP/9MtKyvry9nzpwhT548AKRLl44hQ4ZYNUaR1422nYuIyBtnwYIFnDt3jq+++sos8X6eh4cHw4YNMzv25MkTRo4cScGCBXFwcCBLliy8++673Lx50yizd+9ebGxsWLFiBbNmzaJAgQI4OjpSrlw5du3aFaufpLY5Y8YM3nrrLezt7dm7dy/379/HxsbG+OPg4EDBggX58ssvCQsLS9b8PN/fzz//TN68eXF0dKRSpUps27bNrOyyZcuwsbHh4MGDZsej45owYYJF40hKnwC3b9+mT58+5MiRg7Rp0+Ll5cVnn33G48ePjTJRUVGMGzeOt99+m/Tp0+Pp6ck777zDmTNnknwNYlqxYgWlS5cGoGPHjsb8f/XVV4mO1dJrFtcz20m5v160PsClS5do1aoVLi4uZM6cmffff587d+5gY2PD//73v3jnJ7E5APD19aV27dq4uLjg5OREhQoVWLp0qVF/06ZN2NjYsH37duPYvn37sLGxIV++fGZ9vffee3h4eBgr/f7+/nTu3JmcOXPi5ORE0aJFGTNmDI8ePUow5p9++sns2ri6ulKjRg3Wrl0bq+zp06fp2LEj2bJlI126dJQuXZo//viDqKgoAAYMGICzszMhISH07NmTzJkz89ZbbwEQGRnJN998Q+HChXFwcMDd3Z0OHTpw4cIFsz5WrlxJ1apVyZgxI5kyZaJmzZqsWbMmyWVehMlkYurUqbRo0cJIvJMqNDSUIUOGkD17dtKnT0/9+vVjjTWapXOTIUMG+vXrx65du6hQoQJOTk6ULVuWffv2AbBjxw4qVqxIunTpKFSoEOvWrYvVV6VKlYxrbW9vT/bs2enatStXr141KxfXM9/R/R88eJCqVauSLl068ubNy88//xyrH2tfI3l1KPkWEZE3zsaNGwFo0KCBxXXCwsKoV68es2bN4rvvvuPOnTv4+vpy8eJFatSowYMHD8zKL1q0CD8/P3bu3Mm5c+dwdnamRYsWPHz4MNltzp07l0uXLrF9+3a2bdtG+vTpyZAhAyaTyfhz69Ytxo8fz48//sjw4cNfYJZgzpw5XLt2jX379nHmzBly585N/fr1jX/cJldc40hKnyEhIVStWpWtW7eyePFiAgMD+eGHH/jjjz9o2LAhkZGRAIwfP55x48YxadIkbt++zcGDB2nbti2TJ0822krqNYjWsmVLjhw5AsDChQuN+Y9OvhMaa0pcM0vurxetf+/ePWrWrMnFixfx9fXFz88Pb29vBg8ebFEfCc3B+vXrqVu3Ljlz5uTEiRP4+flRt25d2rdvzy+//AJA9erVcXR0NP6+AmzZsoV06dJx6dIls2Rsy5Yt1KlTBxsbGwAaN27MlStX2Lp1K/fu3WP58uUA/P333wnGOmDAAOO6REZGcvLkScqXL0/Lli05cOCAUW7//v2UL1+eO3fusG7dOgIDA5kzZw67du3i7NmzZm3269ePli1bcuHCBeMLvR49evDVV1/x2WefcevWLf755x/8/f2pWLEily5dMvpo06YNDRs25Pz581y+fJlx48YxY8YM40sES8q8qLNnz3Lz5k3y5MlDr169yJIlCw4ODhQrVoxff/3Voja6devGr7/+ytSpU7lx4wZff/01AwYMiDNGS+YmWvQ2+IULF3L58mW8vLxo3Lgxu3fv5vvvv2f+/PlcuXKFUqVK0a5dO+7cuWNWf+/evcb1fvjwIStWrODs2bM0a9aM8PDwRMd1+fJlvvnmG37//XeuX79Op06dGDBgAL6+vkaZ/+IaySvEJCIi8oapXLmyyd7ePkl1fv75ZxNg8vHxMTt+7do1k4ODg2nixIkmk8lk2rNnjwkwNW7c2KzcoUOHTIBpzpw5yW6zVq1aFsf71VdfmVxcXIzP//77rwkwzZs3L9G60f3VqVPH7HhoaKgpZ86cptq1axvHli5dagJMBw4cMCt77949E2AaP358rHbjGkdS+hw3bpwJMO3atcus7Lx580yAaeHChSaTyWSqW7euqXr16gmO1dJrEJcjR46Y9RfXeFL6miXl/nrR+l9//bXJxsbGdPLkSbOyEyZMMAGmr7/+OsHxJDQHRYoUMb311lumiIgIs+MNGjQwubq6mh49emQymUymOnXqmMqVK2ecr1mzpqlv376mTJkymX755ReTyWQynT9/3gSYZs6caTKZTKabN2+aANOvv/6aYHxJUaBAAVO/fv2Mz+XKlTPlzp3b9OTJk3jrfPDBBybA9Pvvv5sdP3z4sAkwjRo1yuz41atXTQ4ODqbu3bubTCaTadKkSSbAmIu4WFLGEp07dzYBcY5n8+bNJsBkZ2dnatGihcnPz890584d05dffmkCTKNHj06w7eh7K+bfpYMHD5oAU9WqVY1jls6NyWQyubm5mTJmzGgKCQkxjvn5+ZkAU86cOU337983jl++fNkEmKZOnZroXOzdu9cEmHbu3Gkcy5Mnj6lz585m5dzc3EwZMmQwBQcHG8ciIiJM2bNnN3Xs2NE4llLXSF4PWvkWEZE3jslkMlbInjdz5kyzLac2NjZEREQAsHr1atzd3fH29jarkyNHDt5+++1YW6ObNGli9rlYsWIA+Pn5GceS2mbz5s3jHM/ff/+Nt7c3GTJkwNbW1tj+/ODBA27duhXvPGTIkMFsrEOHDk2wv7Rp09KoUSN27Nhh0apQfOIbh6V9bt26lWzZslGlShWzsm3atDHOA5QsWZJdu3YxYsQITpw4EefLx5J6DZIqpa9ZNEvurxet7+Pjg5eXF0WLFjUrm9D1i0vM8jdv3uT06dO0aNECOzs7s3Nt27YlJCTEWGWuV68ehw8fJigoiEePHrFnzx4aNGhAnTp12LRpEwCbN282ygK4u7uTI0cOxo8fz5w5cyyaz2iPHz9m5MiRvP3226RLl874u3HhwgVjpf3OnTscPHiQ1q1b4+jomOTxR9+frVu3NjueM2dOKlWqZHb/AnTq1AkfHx9CQ0NjtW1JmRcVvY3e1dWVefPm4eXlhbu7O1999RW1a9dm/PjxCe64iB5PzHkoW7YsuXLlirNsYnMTrVq1ari4uBifvby8cHBwoESJEri5uRnHc+fOjZOTU6y/H6dPn6Z9+/Zkz54de3t7bGxsqFSpEkC82+KfV716dVxdXY3PdnZ2FC5c2Kyf/+IayatDybeIiLxx8uTJQ3h4eKxnenv37m1sQezcubPZuZs3bxIYGIi9vT329vbY2dkZSdPRo0cJCgoyK589e3azz2nTpsXBwYH79+8nu82cOXPGGsuKFSto3bo15cuX5/Dhw4SGhmIymYyXKb1Ikpw1a9Y4j4WHhyf680JxJbrR4hpHUvoMCgoiW7ZsscqlS5cONzc3AgMDARgzZgyffvops2fPpkSJEnh4eNC5c2f+/fdfo05Sr0FSWeuaWXJ/vWj9oKAgsmTJEqtuXMcSEnMOouc0rmsYfSz6GtarV4+oqCi2bNnCtm3biIyMpHbt2tSrVw8fHx8iIyPZvHkzBQsWJHfu3MCzBGjTpk0UL16cfv36kS1bNooUKcKYMWPM3gkQl27duvHjjz8ybtw4rl69SmRkJCaTiVKlShnXJfonCBO6j6OlTZsWd3f3JI0/eux169Zl1qxZXLp0idq1a+Pm5kadOnVYvXq1Ud6SMi8qOv4yZcqYJboA3t7ePH36lBMnTsRbP3q88f3djqtsYnMTLeZ9DODs7BzncRcXl1j//a1atSqBgYGsW7eOkJAQTCYTJ0+eBJL39xCefUnxfD//xTWSV4eSbxEReeNEP+v9/LOkiXF3dydv3rxEREQQERFBZGQkUVFRRrK+e/dus/Jxray/aJtp0qSJ1cbcuXPx9PRk4sSJ5MuXzygT89nIuNy/f9/s2ePnX44GxLlieOvWLdKkSWOs9kSvLsV8NvratWvx9hvXOJLSZ6ZMmeIs9/TpU0JCQoxkwcnJiW+++YZr165x/vx5xo4dy86dO6levbrxj+OkXoOkSulrFs2S++tF62fOnDnO37qP61hCYs5B5syZgfivNfxfwle6dGnc3d3ZtGkTmzdvpkKFCri5uVG/fn2Cg4PZvXs3Pj4+xqp3tKJFi7Jq1Sru37/Prl27aNKkCaNHj6Z///7xxvnw4UP+/vtv+vfvT6tWrcicOTO2ts/+qfz8T015eHgACd/j8Y0dMF7yGN/4n0/W3333XY4dO8bt27f5888/iYqKonnz5sZqv6VlXsTbb79N2rRp4zwX/SVbQveTJdc7WlLmJqF+Lbm/o++Pn376idKlS+Pk5ARY5++hta+RvDqUfIuIyBunc+fOFChQgDFjxli8UtisWTP8/f3Zv39/isWRUm06ODiYfX7y5AnLli17oTaBWCszYWFhbNiwgerVqxtJRf78+QGM1aJoyX2TryV91qlThxs3bsR68dvy5csxmUzUqVMnVrsFChSgb9++jB49mqCgIE6fPg282DWIflFccraRWuuapaRatWpx6dIlY66ivehbmrNly8bbb7/NqlWrjC3N0f766y9cXV0pV64c8Cy5id5ivnnzZiPJzpMnD2+99RZjx47l/v37sZLvaA4ODlSpUoWJEydSq1Ytszenx2RjY4PJZIp1baKTtGhZsmShfPnyLF++nKdPnyZ5/NH3Z8yXv12/fp09e/bEef96eHjQunVrFi5cCDx7k3dyyiRHunTpaNq0KUeOHIm1vXz79u2kT5+eEiVKxFu/du3aQOy/24cPH471VvHkzM2Linm9586dm+J9RLPWNZJXh5JvERF546RNm5bVq1cTFRVFxYoVWbhwIbdu3TK2oq9cudL46azolY2+fftSrVo12rZty9KlS7lz5w7BwcHs37+fjz/+OFm/CZ4SbTZv3pwLFy4wefJkHjx4wL///kvr1q2pWrVqkuOJKX369IwcOZLbt29z+fJlunfvzq1bt8x+YipfvnzUrl2bSZMmcejQIYKDg5k/fz4XL160Wp8ffPAB+fLlo0uXLuzatYsHDx6wdu1aBg4cSJUqVWjbti3w7LnR3377jQsXLhAaGsqFCxf4888/cXd3N55xfpFrkDt3blxdXdm8ebPFbxkH616zlPTBBx+QK1cuOnXqxOHDhwkJCWHp0qWxvmhJjokTJ3L+/Hl69uzJ5cuXuXXrFl988QXr16/nf//7n9kb8OvVq8fVq1c5deoU9evXN47Xr1+fjRs3YmdnZ/bM/pEjR2jdujWbN2/m9u3bPHnyhM2bN3Po0CFq1aoVb0zp06endu3a/Prrr+zdu5eHDx+ydu1avv3221jJ5U8//URgYCDNmjXjyJEjPHr0iOPHj9O7d2+zxxriUqZMGTp16sQ333zDnDlzuH//PidOnKBNmzakS5eOUaNGAfD111/zxRdfcPToUR49esSdO3f46aefAIzxWlImJYwfP57IyEi6deuGv78/QUFBjBkzhq1btzJmzBhj1TguZcuWpW3btowZM4alS5cSEhLC/v37GTFiBKVKlUrW3KSEevXqkS5dOj777DNu3rzJzZs3GTlyZIq1H+2/ukbyalDyLSIib6TChQtz7NgxevbsydSpUylUqBDp06enXLlyjBs3jlatWnHy5EnjhVAODg5s2bKF999/n6+//prcuXPj5eXFxx9/TMGCBenUqVOSY0iJNt99910mTpzITz/9RJYsWXjnnXfo27dvgkmGpd59913c3d0pX748hQoV4uLFi2zcuJHKlSublZszZw4lSpSgRo0aFChQgCNHjjBu3Dir9enm5sauXbuoWbMmbdu2JVOmTPTv359u3bqxceNG7O3tARg3bhyHDx+mcePGuLm5UatWLbJly8bOnTuNLewvcg0cHByYOXMmBw8eJGPGjGa/853YGK11zVJSxowZ2bZtG3nz5qVGjRrkzZuXrVu3MnbsWCD2imFSNGnShE2bNnH58mWKFi1K3rx52bhxIwsXLuTDDz80Kxu9qu3i4kLFihVjHY/eih6tZMmSdO/encmTJ1O8eHE8PDwYOHAgn376qfEzZvGZP38+NWrUoGnTpuTMmZPffvuNhQsXki5dOrNyFSpUYP/+/WTIkIH69evj4eHBu+++S9WqVSlUqFCi458zZw5ffPEF48aNI0uWLHh7e5MrVy727dtn/Ib5Bx98gLOzMz179iRLliwUKVKEvXv3snbtWmM12ZIy8Zk9e7bxQrkFCxYAGC+Zq1atmlnZggULsmvXLiIiIihZsiQ5c+Zk5cqVzJs3j0GDBiU63vnz59O7d28+/PBDsmXLxrBhw5g6darZlyxJmZuU4OXlxapVqwgICCBfvnyUL18ee3v7RH+/Pqle5BrJ68fGlNAbUUREROSNs3fvXipXrszff/9Ny5YtX9s+JXmOHTtGqVKlmDt3Ll27dk3tcEREXhla+RYRERERi/3111/Y2NhQo0aN1A5FROSVYp/aAYiIiIjIy+nzzz+ncuXKVKtWjYiICJYvX86kSZPo2bMnefLkSe3wREReKVr5FhEREZE4de/enQULFlCyZEny5MnD1KlTGT58ONOnT0/t0EREXjl65ltERERERETEyrTyLSIiIiIiImJlSr5FRERERERErEwvXBMRSSFRUVFcv34dFxcXbGxsUjscEREREXlBJpOJBw8ekCNHDmxtX2ztWsm3iEgKuX79Op6enqkdhoiIiIiksCtXrpArV64XakPJt4hICnFxcQGe/cfZ1dU1laMRERERkRcVEhKCp6en8e+8F6HkW0QkhURvNXd1dVXyLSIiIvIaSYlHCvXCNRERERERERErU/ItIiIiIiIiYmVKvkVERERERESsTMm3iIiIiIiIiJUp+RYRERERERGxMr3tXEQkhRX7ciO2Dk6pHYaIiIjIa8V/QpPUDuGFaOVbRERERERExMqUfL/hIiIi2LJlC3PnzuXUqVOpHc5LZcGCBVy9evU/7TMiIoLZs2dz586dF25r9uzZ3LhxIwWism6blgoICGDBggWp0reIiIiIyItS8v0GM5lM1KlTh0GDBrF161auXLnC1q1b2b9/f2qH9lLo27cvR48ejfOcyWRi9uzZ3Lx5M0X7fPr0KT169ODixYsv3FaPHj1S/AsVa7RpqcOHD9O3b98UbfPy5cv8+eefKdqmiIiIiEhc9Mz3G+zixYts376dK1eukCtXLgCaNm1KgQIFqFChQipH93KLjIykR48e+Pj4kC1bttQOJ07du3cnR44cqR1GismdOzddunRJ0TYPHDhA//796dSpU4q2KyIiIiISk5LvV9zVq1c5ePAg6dKlo1KlSri5uRnnTCYTe/bswd/fn1y5clGtWjVsbZ9tdjhz5oyx4rd27VocHBzIlSsXV69eJSwsjNmzZwNQunRpzp07R7t27QAICwvjzz//pGTJkpQuXRqA06dPc/HiRZo1a8aePXs4e/YsAJkzZ6Z06dJGYh9t06ZNZM6cmVy5crFr1y7Sp09PgwYNAAgJCWH79u2EhYVRqlQp8uXLl+D4Fy9ezJMnT7C1tcXT05MKFSqQPn1643xERATz58+nSZMm3L9/nxMnTpAlSxaqVq2KjY2NWVvnzp3jyJEj5MmTh3LlyiXY76JFiwDYsGED/v7+uLm50apVK4vH8ODBA3bu3MnTp0+pWrUqWbJkiVXm/PnzccZr6Zi8vb3JmDGjxf0mNpeJeT6uoKAgTp48Sfbs2alatWqcZW7evMnJkycpWbIkRYoUAeDIkSOcOXMGDw8PatSoQdq0aY267u7uVK9ePVa/ly5d4sCBA7i5uVGhQoVYY4ZnK9wHDhwgY8aMVKtWDQcHB27evImvr6/Z/V68eHHKli1LYGAge/fuBaBixYp4eHhYPA8iIiIiInFR8v0K+/333xk4cCA1a9bExsaGjz76iN9//51q1arx6NEjmjRpwrlz56hUqRIHDhwge/bsbNq0iQwZMnDt2jUOHToEwI4dO7C3t6d06dIEBQURFhaGr68vADly5KBDhw7UqVOHTJkysW/fPnr06EGLFi1YsWIFAN999x1hYWE0a9aMc+fOGXVv3brFjh07mDx5Mn369DHi/vbbb3nw4AF37tyhTJkylCtXjgYNGrBu3Tq6du1KqVKlyJAhA71796Z///7873//i3cOdu/eTXBwMJGRkZw+fZrbt2+zfv16ihUrBvzfNu569epx/fp13n77bXx9falRowZ//fWX0c5PP/3E4MGDqV27NiEhIdjY2BARERFvv7t27QLg6NGj3Lx5k+zZs9OqVSuLxrBmzRq6du1Kvnz5yJ07N0OGDOGnn36iYcOGRplRo0bFG6+lY+rRowebN28me/bsFvWb2FwmJjqu2rVrExAQQLFixdi2bRt16tRhyZIl2NjYGGXq1q3LtWvXKFOmDJkzZ6Zw4cJ07NiRzZs3U7NmTU6dOoXJZGLLli3kyZMH+L9t5507dzb6/PTTT5k1axY1atTgwYMHHDt2jIULF1K3bl2jzODBg5k2bRrVqlXDZDJx9+5d1q1bx/379zl16hQRERHGPZs+fXru3LlD+/btqVixIs7OzgwcOJBvvvmGNm3aWDQPIiIiIiJxUfL9Chs/fjzfffedkdgGBQUZzwpPmjSJS5cucfz4cdzd3QkODqZChQqMHj2a77//njp16pA+fXrWrVvH9OnTcXZ2BmDz5s0UKFCAKVOmABAVFUWGDBnYvn07LVu2xNfXl0qVKrF9+3aioqKwtbXF19eX4cOHA8+2Onfv3t2IcfPmzbRs2ZJ27dqZrUieO3eOU6dOGduib9++TYcOHViyZAmNGjUCnq1olihRgoYNG1KtWrU452Dq1KlmnwcMGMBnn33GunXrzI5nyJCBDRs2YGtry9mzZylcuDCHDh2ibNmyXL9+nSFDhjBr1ixj+/GgQYOMBDsuP/74I9OnT2fo0KF4e3tbPIYbN27QoUMHhg0bxhdffAE8W40+fvy4xfEmpUw0S/q1dC4T8/jxY06ePImDgwMXL16kePHiLFu2zNg9AZA2bVpOnDiBnZ0dAHPmzGHdunUcP34cLy8vwsLCqFu3LgMHDuTvv/+Os5+FCxeyePFi/v33X2MF/7fffqN79+74+/uTJk0a5s6dy08//cTu3buNeTl16hTh4eEULlyYDz74gCNHjhgr3wB169ZlwIABjBs3zhjPgQMH4owhNDSU0NBQ43NISEiS5kpERERE3hx64dorzMXFhSNHjvDkyRPg2Tbv6Ge1ly5dSs+ePXF3dwfAzc2Nfv36sWzZsiT1YWtrS/Xq1fHx8QHA19eXfv36kTZtWo4dO8bVq1fx8/MzElB4luitW7eOuXPncu3aNUJDQzl9+rRZu82bNzd7Hnn58uXY29sTFBTEvHnzmDdvHjt27CBHjhxs27YtwRhPnDjB8uXLmT17NmnSpIkzUerZs6ex5b5QoUK4u7tz7tw54Nm2ezc3Nzp27GiUHzx4cJLmydIxLF++nDRp0jB06FCjnouLi9nW7MTiTUqZ52OzpF9L5jIx/fv3x8HBAYD8+fPTokULsxV5gN69exuJNzy7X9u2bYuXlxfwLDn/5JNPWL16NWFhYXH2M2fOHAoXLsymTZuYN28ec+fOJSwsjOvXrxvzMG/ePNq1a2f2hUTRokVjPQrxPBcXF06dOsX9+/cBcHJyombNmnGWHT9+PG5ubsYfT0/PRGZHRERERN5UWvl+hc2cOZMPPvgAd3d3KlWqRKtWrejTpw9p06YlICCAvHnzmpX38vLi2rVrREREYG9v+aX39vZm1qxZhIWFsWfPHmbNmkXNmjXx8fHBw8ODXLlykT9/fgCmTJnCF198Qfny5cmRIwdp0qTBxsaGwMBAszZjvqQsICAAOzs7tmzZYna8cuXKRkIWU3h4OC1btmTfvn1UqVKFDBkycOvWrVh9wbNV4uc5ODgYK5YBAQF4enqaPS+dM2fOJM2RpWO4evUqefPmTbTthOJNSploifWblLlMTO7cuc0+58mTh+3bt5sdi+v6ly9f3uyYl5cXkZGRXL16Nc7n5gMCAnB0dIw13927dzcS+6tXr8b6giExkyZNol+/fmTPnp0yZcrQtGlTBgwYgIuLS6yyw4YNY9CgQcbnkJAQJeAiIiIiEicl36+wsmXLsnfvXu7cucPmzZv54osvOH78ODNmzMDDw4OgoCCz8kFBQWTKlCnJSaW3tzeDBw9m7dq1ZM+endy5c+Pt7c369evx8PAwVr2fPHnCkCFDWL58Oc2aNQOebdmdO3cuJpPJrM2YLzuLflHc89t/E7Ny5Ur27dvHhQsXjER04cKFbNq0KUnj8/Dw4O7du2bHQkJCEnzmOy6WjCFjxowp8hveSZVYvyk1l0Csubx7926sF5bFvP7x3a9AnC+jg2fzXbRoUWbOnBlvLMmZ7/z587N582aCg4Px9fVl9OjR+Pj4xDkXDg4Oxiq/iIiIiEhCtO38Febv7w88S1w6derEu+++a7xErWbNmixbtsws6V20aFG822ejubi4GNvYo5UsWZIMGTIwZswYatWqBTxLyLdv346Pj4+RfN+/f5+IiAizlerFixfHSrzj0qhRI4KCgoy3iEd7+PBhvKuvd+7cIWPGjGYrwIsXL060r5iqV6/OpUuXOHLkiHFsyZIlCdaxt7fH0dHRbK4sGUODBg24fv06GzduNCtz48aNJMedFIn1m1JzCZhtMX/y5Alr1qyhRo0aCdapWbMmK1euNFu5X7RoEWXKlDHeRxBT48aNWbZsWazfWo/+ewHQsGFDli1bZvYs9uPHjwkODgae3e9Pnz6Ns76bmxstWrRg4MCBxt8rEREREZHk0sr3K6xp06aULVuWcuXKERISwo8//sjnn38OwOjRoylbtiyNGzemcePGbN26lYMHDxo/nxSfChUqMHHiRIoXL46zszOtWrXCzc2N6tWrs2rVKuNZ6LfffhtHR0cuXbpkJN/Zs2enSpUqdOvWjV69euHn58e8efMsWmkvUaIEo0ePplu3buzYsYMiRYrg5+fHqlWrWLZsmfHs+vMaNmzIkCFD6Nq1K5UqVWLjxo3s3r07ibP47OfUOnfuTJMmTfj4448JCQlh7ty5icZdoUIFJk2axLVr18icOTOtWrVKdAwlS5Zk2LBhtG7dmvfffx9PT082btxIs2bNeP/995Mcu6US6zel5hJgy5YtdOvWjbJly7Jw4ULSp0+f6NgGDRrE/Pnz8fb25p133uHo0aOJrrwPHjyYTZs2Ua5cOfr06YObmxsHDx7k9OnTRrI8ePBg1qxZQ7ly5ejRowcmk4mlS5eydOlS3NzcKFmypNF/iRIlKF68OCNGjMDZ2ZmqVasSGRnJzz//bPY+ABERERGR5NDK9yvsyJEj1K1bl/PnzxMSEsLixYv59NNPAfD09OT48eNUr16dEydOUKZMGY4fP06hQoWM+h4eHnTv3p00adIYxwYMGMCYMWP4999/8fX15fHjx8CzF2R1796dOnXqGGWHDh1K//79jee94dnvXnfs2JHjx4/j4uLC/v376dWrl9lzwA0aNIjzjdyjRo1ix44duLm5cfLkSTw9Pdm1a5eRIMXk5eXFgQMHyJkzJydOnKBRo0Zs2LDB7G3radKkoXv37rG2Pbdv354CBQoYn2fNmsWYMWO4dOkSzs7O7Nmzh169eiX4/O7ixYvx9vZm37597N+/3+IxjBs3jrVr12IymfD39+fjjz82klNL4rV0TN27dzd7qV1C/Voyl3G1Gd+8VK5cmfPnz9O0aVP27t2Lk5NTgrE7Oztz8OBB2rdvb7wF/8iRI7FWzJ/fru7k5ISPjw/fffcdgYGB+Pv706hRI/bt22eUSZ8+PTt27GDYsGEEBAQQGhrK0qVLjXnKli0bPj4+2NnZsW3bNvz8/Fi/fj2dOnUiICCAGzdu8P333/PDDz8kOGYRERERkcTYmCzZEywikoiHDx/i4uLCnj17qFSpUoq3P2fOHL788kuzbeUvm5CQkGdvPR+4BFsHp9QOR0REROS14j+hyX/eZ/S/74KDg3F1dX2htrTyLSIvvYULFzJ16lSznRciIiIiIq8SPfMtIikivi3lKeHYsWN07NiRDz74IMXbtoaToxu88DejIiIiIvJ60bZzEZEUkpLbkkREREQk9WnbuYiIiIiIiMgrRMm3iIiIiIiIiJUp+RYRERERERGxMiXfIiIiIiIiIlam5FtERERERETEypR8i4iIiIiIiFiZkm8RERERERERK1PyLSIiIiIiImJlSr5FRERERERErEzJt4iIiIiIiIiVKfkWERERERERsTIl3yIiIiIiIiJWpuRbRERERERExMqUfIuIiIiIiIhYmX1qByAi8rop9uVGbB2cUjsMEZHXkv+EJqkdgohIsmjlW0RERERERMTKlHzLG2vEiBGMGTMmtcMwvGzxvMp8fHyoUqVKaochIiIiImJQ8i1W9dFHHzFx4sTUDiNOly9fJiAgwOr9WDoH/1U8r5stW7ZQvXp1s2PBwcEcP348lSISEREREYlNz3yLVfn5+WFr+2Z/x6M5sK779+9z4sQJs2O1atViz549qRSRiIiIiEhsSr7fAGfPnuXbb7/lzJkz5MiRg/fee4/69esD0K9fP9566y3CwsLw8fEhMjKS3r1707FjR7M2FixYwPz587l79y4lSpRg+PDheHl5Gef79etHwYIFCQsLY926dRQqVIjs2bOzY8cO9u3bh6+vLwArVqwgNDQ03nhiGjt2LEuXLgUgc+bMVKpUiaFDh+Li4hKrb5PJxObNm4mMjKRjx4707NnTrK2ff/6ZxYsX4+LiQpMmTTCZTInOXWLj7tChA02aNKFbt27GsaFDh+Lk5MSoUaMYNWpUnHOQN29ei+JJrP/kXpvffvstVl/xlYuIiOCXX35h1apVhIWFUaZMGUaMGIGHh4dRNyIigunTp7N69WqePn1Ko0aNGDx4MGnSpOHJkydUrlyZb775hiVLlnDy5Ek++ugjOnfuzI0bN/j222/Zt28fbm5uNG7cmP79+2NnZwfAiRMn6Nq1KwCOjo689dZbDBo0iNKlSwNw8OBBhgwZwoMHDyhVqhQAPXv2pHjx4owYMYLdu3cbMZ45c4bx48dz5swZPDw86NmzJ61btzbOr1u3jsmTJzNq1CimTZvGlStXKFOmDKNHjyZjxowJ3ygiIiIiIonQctxrLiwsDG9vb9KmTcvEiRPp3LkzU6ZMMVYKL1y4wLBhwzh16hTDhw+nefPm9OjRw0h44VnS2rdvX1q0aMGECRO4d+8elSpV4t69e0aZ6HYuX77MuHHj+Pzzz+nduzclS5akfv36zJ49m9mzZ5M5c+YE44mpe/fuRt3PPvuM3bt306JFC7My0X1fuXKFESNG0K5dO/r168eaNWuMMpMnT2bEiBH06NGDQYMGsWrVKpYtW5bg3Fky7rNnz3L79m2zev7+/sb28bjmIHv27BbFY0n/yb02cYmvXOfOnVm4cCEff/wxY8eO5f79+1SsWJHHjx8bdTt06MC3335Lly5dGDNmDCEhIXz33XcAREZGcuzYMTp16kSZMmX4+eefqV+/Prdv36ZChQqEhoYyfvx4PvjgA2bOnMkHH3xgtJs/f35j3r7//nvy589PlSpVuHDhAgCFCxfmgw8+wMnJySjXrl27WNvOr1+/TqVKlbCxseHbb7+lbt26dOrUiQULFhhl7t69i6+vL4MHD6Zz58589dVXbNu2jffeey+h20RERERExCI2JkuW/+SVdeHCBd566y1u3bpFlixZjONhYWGkTZuWunXrcvnyZc6ePWtsjR46dCirV6/m1KlTREREkDVrVkaOHMnAgQOBZ6ucBQoUoHv37owePRqAunXrEhISwv79+836b9q0KQUKFGDKlCkWxZOYu3fv4u7uzpkzZyhYsKDRd1RUFP/8849RrmXLlmTLlo3p06cTHh5OtmzZ+N///sf7778PwMOHD/H09KRNmzbMnDkzVj+WjrtUqVJ06dKFTz/91Kj7zjvv4OzsbLQbcw4siceS/l/02sQUV7mdO3dSr149bty4QYYMGQCIioqiUKFCDB8+nB49euDr60vt2rU5cuQIJUuWNOpGX9OHDx/i4uLC2LFjGT58uHH+008/5dSpU6xfv944duLECUqWLMmdO3fInDlznHG2bNmSIkWKMG7cOACWLVtG7969uX//vlFmxYoVdOnShYcPHwLw8ccfs23bNo4ePWqU+eKLL5g3bx6XL18GYP78+XTt2hV/f3/y5MljtNOxY0ceP36MjY1NrFhCQ0MJDQ01PoeEhODp6YnnwCX6qTERESvRT42JyH8pJCQENzc3goODcXV1faG2tO38NZcnTx4KFSpEmzZt6NOnD97e3nh6epolut7e3mbPJNepU4dvv/2WJ0+ecO3aNe7evUvdunWN8/b29tSpU4fDhw+b9VWpUqUUied5QUFBTJ06lb179xIYGEhUVBQ2Njb4+fkZyTdAmTJlzOrlypWLa9euAc9eZHb37l1q165tnHd2dqZChQrxxunv72/xuJPKkngs6T8pMVpybeIqt23bNuBZYh79PZ3JZOLWrVucOXMGgB07dpAnTx6zxBuIdU3javvq1auUK1cOk8mEyWQiMjISk8nEuXPnqFy5MgCLFi1iyZIlXLt2jdDQUK5du4ajo6NF44l26NAhs3kCaNCgAWPHjiUoKMhI9DNnzmwk3vDsPnr69KnxH92Yxo8fb3zJISIiIiKSECXfr7k0adKwb98+Zs+ezeLFi+nfvz/lypVj8eLFxsqzk5P5Cl369OkxmUw8fvyYR48exVsmelUxWswyyY3neY0bN8bZ2ZmPPvqIHDlykCZNGipUqMDTp0/Nytnbx76Vo5PF6DjjGkN8kjLupLIkHkv6T+lrE1e5R48ekT179jh3B0Q/8/3kyROcnZ2T1Xbjxo358MMPY5UtUKAAANOmTWPkyJGMGzeOIkWK4OzszNdffx3r+ifm0aNH8c73w4cPjeQ7rvsIiPf9AMOGDWPQoEHG5+iVbxERERGRmPTM9xvAzc2Njz/+mDVr1nDt2jWuX7/O1KlTjfP//vuvWfnTp0/j5uZG5syZyZcvHzY2Npw+fdqszMmTJ40EKSG2traxEpfE4ol28+ZN9u/fzy+//EKLFi0oX748rq6uhIeHJ2X4eHl5YWNjE2ucMT8/z9Jxp0+f3kiCo12/ft3sc8w5sCQeS/p/0WtjiYIFC3L16lVy5cpFqVKlzP7kzJkTgEKFCuHn58eDBw+S3PalS5ditVuqVCkjmV+5ciXvvfceffr0oVq1apQqVYo7d+6YtRPXPRZTgQIF4pwnBweHF0qWHRwccHV1NfsjIiIiIhIXJd+vuZMnTzJt2jTCwsIAjO3lz68CbtmyhY0bNwJw69YtJk6cSK9evQBwcXGhY8eOjBkzxnimdtWqVWzbto0+ffok2n+2bNnMfrvakniiubq6kiZNGvbu3Qs8W2GNfrY5Kdzc3Gjbti1jxowxVoRnzpxpbJuOi6XjLlWqFKtXr+bJkycArF27ll27dpm1FXMOLInHkv5f9NpYom3btmTJkoX33nvPSK4jIiL4/fffja3trVu3xtXVlQ8//NB4/vns2bP89ddfCbb90Ucf4evry08//WQcCwwMZMSIEcZnd3d3Dh8+bHzhMn/+fHbs2GHWTrZs2Xj48KHZS+Zi6tOnDytXrjS20d++fZvx48fTu3dv/QyciIiIiPwn9K/O15ynpycnTpzA3d2dwoULkyNHDgoXLszHH39slGnSpAkff/wxXl5e5MmTh9y5c/Pll18a56dMmYKrqys5cuQgb968dO7cmalTp1K+fPlE++/duzfbt2+nQIEClCpViqioqETjiebk5MTUqVONn0PLmjUrGTNmJF26dEmehylTpvD06VOyZctG7ty5+fXXX6lRo0aidRIb9/DhwwkLCyNnzpx4eXnxzTffUKVKlQTnwN/f36J4LOn/Ra6NJZydndm8eTO3bt3Cw8ODt956i0yZMrFv3z7j58xcXFzYuHEjR48eJVOmTHh5edGqVSvy58+fYNu1a9dmwYIFjB8/nsyZM+Pl5UXx4sXJnj27UebLL7/k4sWL5MyZE09PT77++muzZ+UBKlasSPXq1Y35/eGHH2L1Va9ePcaMGUPjxo2N+/ytt95i7NixKTBLIiIiIiKJ09vO3xBhYWH4+/uTJUsW463V8OxFWuXKlWPChAlcv36diIgIcufOHWcbN2/e5N69e3h5ecV64dXFixdxcnIyS5yihYaGEhAQwKNHj3j77bdxcHCIN564PHz4kCtXrpAtWzYyZszIiRMnyJ07t/ECrLj6vnbtGmFhYbF+E/vixYu4urri4eFBQEAANjY2iW47Tmjc8Ox54ICAAJydncmcOTOXL1/G1tbWrN245sDSeBLrP7EyCV2bpJS7ffs29+7dI1++fKRJkybOMleuXCEqKsrspWVRUVEcP36cggULxrnDwWQycenSJezs7MzqRYuMjMTf3x87Ozvy5s0b77W9du0agYGBZMmSBScnJwICAihevLhZmcePH+Pv74+Hh4fZ75QD3Lt3j+vXr1O0aFHj2JMnTzh79izFixc3fns8IdEvZtPbzkVErEdvOxeR/1JKvu1cyfcb7vnkW0RejJJvERHrU/ItIv+llEy+te1cRERERERExMq08v2Gs3RLsogkLiW/GRURERGR1JeS/77T73y/4RJ7KZaIiIiIiIi8OG07FxEREREREbEyJd8iIiIiIiIiVqbkW0RERERERMTKlHyLiIiIiIiIWJmSbxERERERERErU/ItIiIiIiIiYmVKvkVERERERESsTMm3iIiIiIiIiJUp+RYRERERERGxMiXfIiIiIiIiIlam5FtERERERETEypR8i4iIiIiIiFiZkm8RERERERERK7NP7QBERF43xb7ciK2DU2qHISLySvKf0CS1QxARsQqtfIuIiIiIiIhYmZJveSPdvXuXPXv2pHYYFgsODmbXrl1W7eNVm5OEvE5jEREREZHXg5JveSNt376devXqGZ+DgoLYu3dvKkaUsAMHDlCzZk2r9hFzTlJKasyttcYiIiIiIpJcSr7ljZQ5c2aqVKlifPbx8aFhw4apGNHrKzXmNub1FRERERFJbUq+hdDQUI4ePcrly5fjPH/v3j0OHDiAv79/rHN37txh//79AAQGBnLo0CEePnxonPf39+fYsWOEhoamSL3r169z5MgRs2P3798325L9fNtBQUEcPXrUrG2AokWLMnr0aODZlu4zZ84QGRnJzp072blzJ2fPnmXnzp08fvzYrN7Tp0/ZuXNnrPYAHjx4YNQ/ePAg9+7dS3C+4ost2tWrVzl27BhPnz6N83y0J0+esHPnTp48eZJorAldy5hSYq7jmtvn77MnT55w+PBhLl68SFRUVLzji6/M8/0HBASwZ88eQkNDza6vJXE+7/l5/y+2+4uIiIjIm0HJ9xtu2rRpZM2aldatW+Pt7U3Dhg0JDg42zo8YMYIcOXLQs2dPSpQoQe3atbl7965xfuPGjTRs2JDGjRtTsWJF2rdvT86cOVm9ejUNGzakXr16tGjRgoIFC3LhwoUXrrdkyRK6du1qNoa9e/eabcmObrtTp06ULVuWd955hxw5crBhwwajzPPbkv38/FiwYAFPnjxh6NChDB06lC1bttC+fXvmzZtn1teiRYto1qwZ9vaxfyjgypUrRv2+ffuSM2dO+vfvj8lkSlJsAB999BH58+ena9eu5MqVizlz5sR7De3t7WnZsiULFy40O/7XX3/RuHFj7OzsgMSvZUwpMddxze369esBmD59Ojly5KB79+7UrVuXokWLcuzYMbP+EisT3X/79u2pVq0aQ4YM4d69e7G2nVsy7yaTiX79+pnN+4ABA6y+3V9ERERE3gxKvt9g69atY8CAAcyePRs/Pz8uXbrEJ598wp07dwBYs2YNkyZNwsfHhxMnTuDv78+dO3f4/PPPzdq5d+8eNWrU4OLFi1y4cIGaNWvSvHlzGjZsyPnz57l48SIFChRg/PjxKVLPEvfu3aNIkSL4+/tz5swZ3nvvPT7++OM4y5YuXZqvv/4aZ2dnY3X2gw8+4N133+WPP/4wK/vHH3/QqVMnHB0dY7VTpEgRo/6hQ4c4e/YsK1euZMmSJUmKbeXKlcycOZN9+/Zx/PhxTpw4wY4dO+Ida5o0aWjfvj0LFiwwO75gwQJat25NunTpLL6WyZHQeOKa2379+rF582ZGjBjB9u3bOXHiBJcuXaJt27a88847xuq2JWWi+8+fPz8BAQHs3LmTbNmyJTlOgOXLlzN37lz279/P8ePHOXnyZKKr3qGhoYSEhJj9ERERERGJi5LvN9j06dNp1qwZLVu2NI41aNCAAgUKAM8SzZYtW1KpUiUAMmXKxNChQ5kzZw6RkZFGnbRp0/Lpp58CYGNjQ7169UifPr2R2NjZ2VG3bl1OnDhh1n9y61kiTZo0DB061PjcokULzp8/H2sbe0J69erFgQMHOHXqFAAXLlxgx44d9OzZM8F6169f58CBA1y+fJlSpUqxbdu2JMU2e/Zs2rRpQ6lSpQDInj07AwYMSLDPLl264Ovry7Vr14Bn26w3b95Mly5dAMuvZXIkZ65/+eUXqlevzqNHj9izZw+7d++mYsWKnDlzhosXL1pcBsDW1paRI0e+cJzz5s2jbdu2lCxZEoBs2bLxwQcfJNjm+PHjcXNzM/54enomGoeIiIiIvJli752VN8aFCxdo27ZtvOf9/Pxo0aKF2bG3336b8PBwrly5Qt68eQHImDGj2TbsdOnS4e7ujo2NjdmxmM9PJ7eeJTJlymTWtpOTEyaTiSdPnuDg4GBRG/nz56dWrVr88ccffPfdd/zxxx+UKFGCsmXLxln+1q1btG7dmuPHj5M/f36cnZ3x8/OL1V9isfn5+dGuXTuzOoUKFUow1ipVqpA3b14WLVrE4MGDWbx4MVmyZKF27dqA5dcyOZIz1+fOnePRo0fGly/RqlatyqNHjywuA89erubk5PTCcV66dCnWvBcsWDDBNocNG8agQYOMzyEhIUrARURERCROSr7fYOnTp09wm6yrq2usl1I9ePAAADc3N6vGFh8bGxuzZ6gBwsLCrNZf7969GThwIGPHjmXu3Ll89tln8ZYdOXIkDg4OBAYGGklnly5d4n2xV3wSmveEdOrUiQULFjB48GAWLFhAx44dsbW1TbTN+K6lNec6Xbp0eHt78/PPP79QGcAY44tycXGJNUeJXTsHBweLv8wRERERkTebtp2/wWrWrMnq1asJDw83jplMJuPt2uXLl2fjxo1mCdj69evJnz8/GTNm/M/jhWdbgW/cuEFERIRxLPot1i8iXbp0ZvMQrXXr1kRERPDhhx9y584dOnfuHG8bly5donLlykYy9uTJk1hbzi1Rvnx5Nm3aZHZs48aNidbr0qULR44cYc2aNezdu9fYch7dZlKvZUrNdVxz6+3tzYoVKxJMdi0pk5LKlSvH5s2bzY5t2bLFKn2JiIiIyJtHyfcb7LPPPiMyMpL69euzdOlSlixZQr169YznaT///HNu3rzJO++8w9q1axk7dizff/8933zzTarFXKdOHSIjI+nfvz+bNm1i3LhxTJs27YXbLVKkCE+fPuXnn382+zksBwcHunbtysyZM2nRogWZM2eOt43atWszc+ZMFixYwMqVK2nSpInx8rqkGDx4MBcvXqRbt26sW7eO4cOH89dffyVar1ChQpQtW5aePXtStGhR45lxSN61TKm5jmtuhw4dipOTE97e3vz555+sXbuW0aNHU7lyZaOeJWVS0uDBgzl//jw9evRg/fr1jBw5kiVLlpg9BiEiIiIiklxKvt9gWbJk4cCBA1SpUoUZM2awcuVKvvjiC4oWLWp23t3dne+++47jx4+zatUq2rRpY9ZGxYoVzdrNli0b5cuXNzuWM2dOypQp88L13N3d8fX15cmTJ0yZMoUnT56wePFiqlWrlmDbLi4uVK1a1XjmN3PmzFSpUsU47+XlxYIFC9iyZQvDhg0zfg4LoGPHjgCJvmhtyJAhfPbZZ8yfP5/ffvuNli1b8r///Y8iRYokKbacOXOya9cuoqKi+PHHH4mMjOTvv/82G2N8PvroIwoWLMjAgQPNjltyLWPOSUrNdVxz6+7uzoEDB2jVqhXz589n+vTp2NjYmO0UsKRMXP3HNRZL4vT09GTHjh2Eh4fzww8/EBYWxpQpUyx6nlxEREREJDE2ppgPdYqImcmTJzN16lQuXbqUYs8Xy8vp6dOnZj8j99FHH3H48GF27txpUf2QkJBnbz0fuARbByXtIiLJ4T+hSWqHICJiiP73XXBwMK6uri/Ull64JhKPgIAAjh07xsSJExkxYoQS7zdA06ZNad++Pblz58bX15dff/2VxYsXp3ZYIiIiIvIa0Mq3SDx+++03/vzzT6pXr85XX32l5PsNcP78eaZOncr58+fx9PSkd+/exm+jWyIlvxkVERERkdSXkv++U/ItIpJClHyLiIiIvF5S8t93WsoTERERERERsTIl3yIiIiIiIiJWpuRbRERERERExMqUfIuIiIiIiIhYmZJvEREREREREStT8i0iIiIiIiJiZUq+RURERERERKxMybeIiIiIiIiIlSn5FhEREREREbEyJd8iIiIiIiIiVqbkW0RERERERMTKlHyLiIiIiIiIWJmSbxERERERERErU/ItIiIiIiIiYmX2qR2AiMjrptiXG7F1cErtMEREXlr+E5qkdggiIv85rXyLiIiIiIiIWJmSb5E3xMmTJ+ncuXNqh8HOnTvp16/fC7Vx8OBB3n333ZQJSERERETkP6DkW+QNcfPmTRYvXpzaYeDv78+KFSteqI2rV6+ybNmylAlIREREROQ/oORbRERERERExMr0wjV5bWzZsoXVq1fz9OlTGjVqRMuWLY1z/fr14/79+9ja2uLp6UmrVq2oVKmScT40NJTu3bszYsQINmzYwMmTJ3nnnXdo1KgRT58+Zc6cOezbtw83NzcaNWpE/fr1zfpOrMzOnTuZP38+Q4YM4ddff+XatWtMmTIFDw8Ps3YuXLjAF198AYCjoyNvvfUWvXv3JmvWrLHa+vzzz5k7dy5XrlyhTJky9OnTB3v7//srff36dX788UcCAwMpWbIkXl5eCc5fjx49+PDDDylTpgwA48ePx8/Pj99++w2A+/fv069fP77//nuyZ8/O3bt3+fXXXzlz5gweHh507dqVkiVLJjrmuEycOJFr164xYcIEDh48yPz58xk0aBDz58/nypUrVKlShZ49e2JnZxdnfUvmLfo6zZ8/n71795IxY0beffddihYtanY+sWstIiIiIpIcWvmW18Jnn31Gy5YtcXZ2pkyZMsydO5cffvjBOB+djDdp0gQ7Ozvq169vtm05PDycxYsXU69ePQICAqhXrx758+fn8ePHVK9enWXLllGxYkVy5szJu+++y/jx4426lpTx9/dn9uzZNGjQAHd3d5o1a4aTU+y3YWfMmJGWLVvSsmVLatasyYkTJyhWrBi3b9+O1Vbjxo1Jnz49pUqVYty4cXz88cdGmXv37lGhQgUOHjxI+fLl2b9/f6LPSD+/HdxkMvHdd98xa9Ys/Pz8ANi2bRsbN24ka9ashISEULZsWdavX0+VKlV4+PAhFSpU4J9//knSmKOionj//ff55ZdfGDBgAI6Ojka9OnXqkC5dOooXL84XX3xB//79443dknl7+PAhVatWZfLkyRQpUoRs2bLRvXt3zp8/b/F1FBERERFJLq18yyvvyJEjTJw4kc2bN1O3bl0A+vbty507d4wyLVq0MKuTNWtWxo8fT9u2bc2Od+3alYkTJxqfv/76a+zs7Ni4cSO2ts++q6pYsSJ169ZlwIABuLi48N133yVaBp6trs+ZM4eqVavGO5bMmTPzzjvvGJ979OhBrVq1+PXXXxk5cqRxPDQ0lIULF1KqVCkAMmXKRN++ffn5558BmDx5MunTp2fDhg3Y2dnRp08fWrVqxerVq+Pt29vbmy1btgBw/PhxbGxsqFu3Lj4+PuTLlw9fX19q1KiBra0tkyZNAp7tNkibNi3wLGH/5JNPOHbsmFmc8Y05LCyMrl27cvr0aXbt2kWOHDnM6v3444/G7oWKFStSo0YNPvnkEwoXLpyseZs4cSLXr1/n3LlzxjUZMGAAoaGhABZfx+eFhoYa9QFCQkLinV8RERERebMp+ZZX3qZNm8iZM6eReEd7fkv3nTt3mDNnDufOnSMkJISbN29y9uzZWG3FbGPDhg2EhITQrVs3TCYTJpOJ8PBwwsLCOHPmDOXLl7eoDEC6dOkSTLyjHTlyhKVLl3Lt2jVCQ0O5evVqrFjd3NyMxBugQIECPHz4kAcPHuDi4sLOnTtp1qyZ2Tbt1q1bJ5p8jxs3jsePH+Pr60vNmjUpX748vr6+9OrVC19fX7p16wY821LevHlzI/EG6NChAzNmzDBiSGjM4eHhNGvWjAcPHrB9+3YyZsxodt7Ozo6mTZsan6tWrYq7uzu7d++OM/m2ZN42bdpEmzZtzJJoBwcHHBwcAMuudUzjx49n9OjR8c6piIiIiEg0Jd/yyrt//z6ZM2eO9/ytW7coVaoU5cqVo1GjRmTIkIGTJ0+ybdu2WGXd3NzMPt+9e5ciRYqYJYIAbdq0IU+ePBaXAXB1dU10LCtWrKBjx468//77VK9eHWdnZ4KCgnj48KFZOUdHR7PP0Su1kZGRAAQFBcVKaGN+jqlSpUrY2tqye/dufH19qVOnDuXLl+fHH3/k/v37HD9+HG9vbwACAwPJlCmTWf3oz4GBgUaCG9+Yw8LCOHXqFA0bNiRDhgyxzru4uJg9vx4df2BgYJztWTJvid0nll7H5w0bNoxBgwYZn0NCQvD09Iy3DxERERF5cyn5llde7ty58ff3Jzw8nDRp0sQ6v3btWtKlS2e26jtjxgyL2vb09CQyMtJsS3Nyyljqjz/+oF+/fkyePNk4Nnfu3CS3Ez0nz7t06VKCdRwcHKhUqRL//PMPO3bs4H//+x+FCxcmJCSE33//HVdXV+OFanny5DGeBY928eJF7OzsyJUrV6LxpU+fHl9fX7y9venXrx/Tp0/HxsbGOH///n3u379vJObh4eFcu3Yt3iTYknnLnTu38Xx3XJJzHZ9fORcRERERSYheuCavvNatWxMZGcm4ceOMY/fv32f79u0A2Nvb8/DhQx49egQ8exnZ1KlTLWq7R48erF69mk2bNhnHIiMjmTNnTpLKWMre3p5bt24Zn3fv3m3WrqXat2/P0qVLuXz5MvBsRXbatGmJ1vP29ub333/Hzs6OokWLYmdnR9WqVfn222+N570B3nnnHf766y8uXLgAPHtL+Pfff0/r1q3j/AIkLgUKFMDHx4c1a9bw/vvvYzKZzM5HP1cO8PPPPxsvyouLJfPWrVs3/vrrL/bv328cO3r0KAEBAUDKXkcRERERkZi08i2vvKxZs7J06VK6dOnC0qVL8fT05Ny5c/z666/As23DkydPpmjRohQvXpyDBw9SsGBBi9ru2LEjZ8+epXnz5pQpUwY3NzdOnTpFy5Yt6d69u8VlLPXpp5/SsGFDKleujLOzM0ePHqVEiRJJmxCgc+fO/PXXX5QqVYrKlStz8uRJihQpwrlz5xKs5+3tzVdffUW7du3Mjm3YsMHYcg7QqVMnNm3aRNmyZalSpQpnzpzBycmJRYsWJSnOt956Cx8fH2rVqoWNjQ2//PIL8Gy7+j///EOlSpVImzYt+/fv57fffot367wl89a5c2eOHz9OzZo1qVChAiaTibCwMGNHREpeRxERERGRmGxMMZebRF5Rjx49Yvfu3URFRVGxYkWzZ4nDw8PZtWsX9+/fp3jx4jg7O+Pj42NsMY6IiGDZsmXUq1cvzueC79y5w4EDB7Czs6N06dJkyZIlSWUuX77M0aNHY711PS537txh//792NnZUalSJfz8/Hj8+DHVqlWLt6179+6xceNG2rRpY7byvHfvXoKCgihWrBiOjo5s27aN9u3bx9t3WFgYy5cvp3jx4sbvX9+4cYNt27bh7e1NtmzZzMr/+++//Pvvv3h4eFC5cmWz57TjG3Ncxy9dusS+ffuoVasWmzdv5tNPP+Xq1ascO3aMK1euULZsWbNnqa9du8bevXtp06aNxfMW7erVqxw6dIgsWbJQvnz5WM+WW3Kt4xMSEoKbmxueA5dg6xD7p+REROQZ/wlNUjsEERGLRP/7Ljg42KJ3OCVEybeIvFTmz5/Pp59+ys2bN1M7lCRT8i0iYhkl3yLyqkjJ5FvPfIuIiIiIiIhYmZ75FpGXSvXq1Y3n9V9VJ0c3eOFvRkVERETk9aLkW0ReKnny5In3J8VERERERF5V2nYuIiIiIiIiYmVKvkVERERERESsTMm3iIiIiIiIiJUp+RYRERERERGxMiXfIiIiIiIiIlam5FtERERERETEypR8i4iIiIiIiFiZkm8RERERERERK1PyLSIiIiIiImJlSr5FRERERERErEzJt4iIiIiIiIiVKfkWERERERERsTIl3yIiIiIiIiJWpuRbRERERERExMrsUzsAEZHXTbEvN2Lr4JTaYYiIvBT8JzRJ7RBERF4KWvl+iYWGhnL79u3UDuO1FhYWxq1bt5JUR9dFRERERESSSsn3S2z9+vXky5cvSXVexsTwZYwp2vbt28mZM2e85+OKPTnXRV49L/N9KyIiIiKvHiXfLzFHR0eyZs2apDqrV6+mYMGCVoooeV7GmCwVV+zJuS7y6nmV71sRERERefko+X5OeHg4N2/e5ObNm9y/fz/OMk+fPuXOnTvG5+Dg4GSViWYymbh37x7h4eGxztWqVYs9e/bE2+7Dhw/NyoeFhXH//n1MJpMxjocPHyYaT3h4eKy2YgoJCSEiIiLW8eTG9Pz5p0+fJth3cq5LQuOJiIhIdLwJxZ7YdUnJ+Y0puX2l1Bxa0k605+c5oe39lt5bL9N9KyIiIiKSVEq+n3PixAlKlSpFqVKlyJMnD+7u7kyePNmszLJlyyhSpAhDhw4lc+bMZMmShaJFi3Ls2LEklQH4448/yJEjBzlz5sTV1ZU+ffqYJaIxtzdHtztq1Cg8PDzw8PCgYMGCHDp0CICDBw8yZMgQHjx4YIxj6tSpRr2PPvqILFmykCVLFgoVKsSRI0cYMGAAHh4eZM6cmapVq8a5xfrtt98me/bsuLm50aRJE65evfrCMfn7+1OjRg1cXV3JmjUrNWvW5N9//33h6xJfHNGmTp1KpkyZyJo1KwUKFGDz5s1x3wwJxB7fdbHG/MaU3L5Sag4taQdg0qRJZMyY0ZjnkSNHxtreb+m99TLdtyIiIiIiyaXk+zllypQxVrmCg4NZs2YNY8aMYdOmTWblAgMDOXXqFP7+/gQHB1O2bFnatm1rtsKWWJm9e/fy3nvv8f333/Po0SOOHDnChg0bGD16dIIxBgYGcvPmTa5cucKDBw8oV64cffv2BaBKlSr89ttvuLq6GuMYMWKEUS88PJyrV69y79493N3dqVKlCvb29ty5c4fAwECePHnC2LFjjb4OHTpE+/bt+eabb3j48CGBgYF4eHjQoUMHTCbTC8U0ZMgQPDw8uHfvHvfu3WP8+PH4+vq+8HWJLw6AnTt38umnnzJv3jwePXrEkiVL+P333+Od64TmM67rYq35TYm+UmoOLWnH19eXYcOGsWDBAh49esTSpUuZNWuWWT9Jubdepvs2ptDQUEJCQsz+iIiIiIjERcl3PO7du0fevHmpX78+a9eujXV+6tSpuLi44OjoyJQpUwgICGDjxo0Wl/nhhx9o1KgR77zzDjY2NhQuXJgRI0bw448/Jph42dnZMWXKFBwdHbG3t+e9997j6NGjcW5bj1nvu+++I23atDg5OdGuXTtMJhPffvstadKkwcXFhdatW3PgwAGjzuTJk2natClVqlQhMDCQBw8e8Mknn7B7924CAgJeKKY7d+5QsGBB0qVLh62tLVWqVOH9999PcAyQ8HVJLI6ff/6ZZs2a0aJFC+BZIjlgwIBE+7SENec3Jfp63ovMoSXtTJs2jebNm9O8eXMASpcuHWuek3JvvUz3bUzjx4/Hzc3N+OPp6WlxXRERERF5s+h3vp/z6NEjPvjgA5YuXYqNjQ3Ozs48ePCAhg0bmpVzdnY223acKVMmPD09OXfuHE2aNLGozNmzZ2ncuLFZu2XLluXRo0dcu3aNXLlyxRmju7s7Tk7/9/vBLi4uREZG8ujRIzJkyBDv2GLWc3Z2Jlu2bKRNm9bs2PMrd8ePH8ff3x8fHx+ztrJmzcqNGzfIkydPsmMaMmQI77zzDjt37qR+/fo0bdqU0qVLx1nW0uuSWBznzp0zEsJopUqVirPPpLLm/KZEXyk1h5a0c/78eeMLjmglS5Y0+5zceyu179uYhg0bxqBBg4zPISEhSsBFREREJE5Kvp8zatQojh07xunTp41/oPfo0YN79+6ZlYvrBU5hYWFmCUFiZdKmTUtYWFis89HnXga2trb07t2b77//PsXbjn4Gd/PmzWzdupWaNWvSv39/JkyYEKuspdclMWnSpIm1qhnzGvyXrDm/MaXUHFrSTtq0aROdZ2uO/b+cVwcHBxwcHKzej4iIiIi8+rTt/DnHjh2jSZMmRlIRGRnJ7t27Y5V7+vSp2Uuo/P39uXbtGsWLF7e4TIkSJdixY4dZu9u3bydr1qxkyZIl2WNImzYtkZGRya7/vEqVKrFu3bpYiVRUVNQLxxQVFYWbmxtt27Zl2rRpTJ48mRkzZsRZ39LrkpgSJUqwa9cus2M7d+5McuwpJaXm1xIpNYeWtFOsWLFYx55/OzxYd+zWvG9FRERERJJLyfdzypQpw5IlS9i3bx+nTp2iV69enD9/PlY5GxsbevXqxd69ezl8+DBdu3alYsWK1KhRw+Iyn332GYcPH2b48OGcPXuWxYsXM378eEaNGvVCYyhQoAAPHz5k48aNL/zzSEOHDiUwMJD27duzf/9+zp49y4IFC6hWrdoLx9S4cWNmzpzJv//+y8mTJ/n777/j3QJu6XVJzKBBg9i1axejR4/m3Llz/PHHH8ycOTPJsaeUlJpfS6TUHFrSzieffMKOHTv43//+x/nz55kzZ06sL1asOXZr3rciIiIiIsml5Ps5o0aNok6dOnTp0oU2bdrg6urKgAEDyJgxo1m5LFmy8PnnnzNkyBBat25Nrly5WL58eZLK5M+fHx8fHw4dOkS9evX49ttvGT9+PP379zfKODo6kjVrVuNzunTpYq2Kp0mThqxZs2Jr++xSFilShHHjxvHpp59SunRppk6dGmc9JycnPDw8zI6lT58ed3d347OXlxcHDhwgQ4YMdOnShZYtW7Jx40azN4QnN6bp06ezf/9+WrduTYcOHfD09GThwoXJvi6WxFG4cGFWr17N+vXradKkCWvXruXnn38mW7ZscfYbX+yWXJeUmt+YkttXSs2hJe0ULVqUlStXsmbNGho3bsyqVasYM2YM6dKlS9LYX8b7VkREREQkuWxMCb1aW2KZP38+n376KTdv3nyhMiKvM5PJhI2NjfF55MiRrF27lsOHD6diVNYXEhLy7K3nA5dg6+CUeAURkTeA/4QmqR2CiEiyRf/7Ljg4GFdX1xdqSyvfIpLi3nnnHf755x8uXLjAzJkz+f777+nXr19qhyUiIiIikmr0tvMkimu7anLKiLzO+vfvz/jx4zl//jyenp7MmDGDTp06pXZY/5mToxu88DejIiIiIvJ60bZzEZEUkpLbkkREREQk9WnbuYiIiIiIiMgrRMm3iIiIiIiIiJUp+RYRERERERGxMiXfIiIiIiIiIlam5FtERERERETEypR8i4iIiIiIiFiZkm8RERERERERK1PyLSIiIiIiImJlSr5FRERERERErEzJt4iIiIiIiIiVKfkWERERERERsTIl3yIiIiIiIiJWpuRbRERERERExMqUfIuIiIiIiIhYmX1qByAi8rop9uVGbB2cUjsMEZH/nP+EJqkdgojIS0sr3yIiIiIiIiJWppVvieXMmTOsW7eOu3fv0r9/f3LkyGG1vpYtW4a9vT0tW7a0Wh8p2XfMOn/++Seurq40bdrU4jaSU+dltXHjRg4ePEjatGkZMmRIaocjIiIiIvLS0sq3mLlw4QKlS5fm/PnzODo6YmNjY9X+VqxYwZo1a6zaR0r2HbPOkiVL2LBhQ5LaSE6dl9GPP/5I9+7defLkCQ4ODqkdjoiIiIjIS00r32Jm8+bN5MuXj2nTpqV2KK+Ezp074+rqavU6L6M1a9bQq1cv/ve//6V2KCIiIiIiLz0l32JYtGgRS5cu5d69e3zxxRdkzpyZTz75BIBr166xYsUK7t69S4kSJWjWrBm2tuYbJywpc+7cOVasWIGLiwu1a9dONKZVq1axf/9+ADJnzkylSpWoXLmyWZnZs2eTI0cOcuTIga+vL5GRkTRv3hwvL68X6tuSOuHh4YSHhycpluTUgWePA6xcuRJXV1dq167NiRMnEtw2H91ulixZ2L59O5GRkbRr145cuXLFWWbTpk24u7vTs2dPY+zr1q0jLCyMMmXKULduXaPel19+yalTpwgNDeWLL76gWrVqNGzYMNF6YWFhjBkzhj59+nDo0CFOnjxJ3bp1jWu6detW9u3bh5ubG3Xr1qVQoUJG3WPHjrFhwwb69evHmjVruHLlCmXKlKF+/fqxxu7r68vevXvJmDEjrVu3xsPDw+x8Qv1E9+Xj4wOAt7c3pUqVinOORUREREQspW3nYkiTJg1p0qTBxsYGR0dHYyvxrl27KFSoEOvXr+f+/ft8+OGH1K9fn6ioKKOuJWV8fHwoUaIE+/bt49y5czRo0IBdu3YlGpOjoyOOjo74+fnRrFkzRo4caVZm/vz5fPjhh3Tp0oUrV66wdetWihcvztmzZ1+ob0vqxNxCbkksyamzadMmSpQowf79+zl37hwNGzZkyJAhCW6bnz9/PgMGDKB9+/YEBASwbt06ihUrxsmTJ2OV6dKlC3fv3iVNmjQAzJgxg4oVK/Lvv/8SFBREv3796Nixo1HPwcEBGxsb4/pYWi8sLIyxY8fSsGFDZsyYQWRkJHZ2dkRGRtKqVSv69+/PvXv3OHHiBBUrVmTevHlG3RMnTjBmzBiqVKnCjh07uHnzJm3atDFbeY+IiKBly5a0a9eOgIAAjh49Sq1atQgICACwqJ/Zs2dTrVo1zp49y7Vr1+jTpw/Tp0+Pd55FRERERCyhlW8xtGnThvPnzxMUFMQXX3xhHO/fvz8dO3bkt99+A2Dw4MEULFiQOXPm0KNHD4vLfPzxx/Tp04cffvgBgL59+1KsWLEEY2rUqBGNGjUyPvfu3Zty5coxYMAAsmbNahyPiopi3759xhcGVapU4bfffmPSpEnJ7js5dSyJJal1TCYTn3zyCe+//z5Tp04F4P3336dIkSKJxnLr1i0uXryIu7s7AC1btuSzzz5j3bp1RpnHjx9z5MgR0qdPD8ClS5f48MMP2bdvn7HiO3ToUAoUKMC6deto3Lgxw4cPZ9WqVdSsWdO4VyypF61ChQrMnj3b+PzTTz9x9uxZjh07hqOjIwCNGzema9eutG3blnTp0hmx/vDDD9SpUweAIkWKMHz4cCOGH3/8ER8fH06dOmWs8N+5c8foZ9q0aYn2M2vWLIYOHcqIESMAMJlM/Pvvv3HOb2hoKKGhocbnkJCQRK+JiIiIiLyZlHxLgm7cuMHx48eZMWOGcSxHjhw0adKEjRs30qNHD4vKXL9+nRMnTpglXIULF461hTwu+/btY9++fQQGBhIVFYXJZOLMmTNmyXfDhg3NXvpVokQJLl++DJCsvl8k3oRiSU6d69evc/r0aebPn2+cL1iwIFWqVEk0liZNmhiJN0CPHj1o166dseIMz77giE684dlW/3Tp0rFixQr+/vtv4FkC6uLiwv79+82S6OclpV67du3M6v7111+4uroyYcIETCYTJpOJJ0+eEBISwtmzZ41k3tnZ2Ui8o+cpKCiIhw8f4uzszN9//0379u3NttY/v+Xckn5y587NunXraNq0KSVLlsTGxibeLzrGjx/P6NGj478AIiIiIiL/n7adS4KuX78OQLZs2cyO58iRwzhnSZkbN27EWSbm55g+/vhjmjZtyokTJ4zt8La2tty7d8+s3PPJI4C9vb3xXHVy+k5uvInFkpw6N2/eBDD7siGuz3GJK/7w8HACAwONY5kzZzYrc/PmTZycnLC3tzceRUibNi29e/ematWq8faVlHpx9Zk+fXqjbtq0aXFzc+Prr78mY8aMRrm45gmebTeHZ6vcOXPmTDDGxPqZMmUKJUqUoGHDhmTNmpVevXoZ29ZjGjZsGMHBwcafK1euxNu3iIiIiLzZtPItCYpOZK5fv06ePHmM49euXTPOWVIm+rfCb9y4Yfa74devX8fNzS3Ovh88eMCPP/7I9u3bqVatGgDBwcGxnvlOTHL6Tk4da8mePXucsdy8eTPRWKK/RIh2/fp10qZNG+sFZM/LkiULjx49YtiwYcbquCWSWy+6brZs2cwed0iOrFmzJpgAW9JP5syZmTZtGtOmTePUqVMMHjyYNm3acODAgVhlHRwc9DNrIiIiImIRrXxLgrJly0aZMmWMZ7kBrl69ytq1a41txJaUyZ49O6VKlWLmzJlGmZMnT7J37954+w4NDcVkMpE2bVrj2M8//5zkMSSn7+TUsZYcOXJQvHhxZs2aZRw7c+YMe/bsSbTu2rVrjZVzk8nE77//Tr169WK9hf55LVu25MmTJ3z//fdmxy9fvhzvCvCL1INn29D/+usvTp06ZXZ8x44dCdaLqXXr1ixdutRsm/+NGze4deuWxf08/7+LFi1K27ZtE31sQEREREQkMVr5lkT98ssv1K9fn6tXr1KwYEH+/vtvatasSZcuXZJUZsqUKTRs2JCAgADy5MnD2rVryZs3b7z9uru706ZNG9q0aUPbtm3x8/Pj2LFjxpu1kyKpfSe3jrVMmTKFxo0b4+/vbzyTnCdPngSTaHi2K6Fq1ao0btyY48ePc+LEiUQTWi8vL2bOnEmfPn3YvHkzRYoUwc/Pj3PnzrFq1aoUrwfPXiC3d+9eKlasSOvWrXFzc+PgwYNkypSJ6tWrJ1j3ef3792fbtm2ULl2aVq1aYTKZOHDgAOvXr7e4n9mzZ/PRRx9RpUoVIiMjWbx4MZ9//rnFMYiIiIiIxMXGZDKZUjsIeXns3r2bM2fOGL/1HO327dusWrWKe/fuUaJECerXr4+NjU2Sy1y6dIlVq1bh4uJCrVq1OH78OPb29jRp0iTOeKKioli5ciUXLlwge/bstGjRgmnTptGyZUsKFiwIwIIFC8iRIwe1atUy6m3cuJGQkBCzF3sltW9L6ixduhRXV1caNGhgcSzJqQNw8eJF1qxZY8Ty3nvvUbZsWb755ps4Y69bty7lypWjW7du7Ny5k4iICFq3bm32HHhcfUe7fv06GzZs4N69exQqVIj69eub7UKYPXs2BQoUMB4JsKReeHg433zzDT169Ijz2exDhw6xa9cu7OzsKF++PBUqVDDOnTx5ki1btjBw4EDj2I0bN/j9998ZMmSI2fbvHTt2sH//frJkyULTpk3NnhtPrB+A48ePs2PHDmxtbalevbpFb7mHZ287d3Nzw3PgEmwdnCyqIyLyOvGfEP//p4qIvIqi/30XHByMq6vrC7Wl5FvkFXDlyhUyZcpkvHDs4sWLFCtWjCVLltCsWbM460Qn3xMmTPgvQ32jKfkWkTedkm8Red2kZPKtbecir4B79+7RoEEDqlevTmRkJMuXL6dFixYJrtqLiIiIiMjLQyvfIq+I69evs2nTJh49ekTJkiVjbfeOKaEt5WIdKfnNqIiIiIikPm07FxF5CSn5FhEREXm9pOS/7/RTYyIiIiIiIiJWpuRbRERERERExMqUfIuIiIiIiIhYmZJvEREREREREStT8i0iIiIiIiJiZUq+RURERERERKxMybeIiIiIiIiIlSn5FhEREREREbEyJd8iIiIiIiIiVqbkW0RERERERMTKlHyLiIiIiIiIWJmSbxERERERERErU/ItIiIiIiIiYmVKvkVERERERESszD61AxARed0U+3Ijtg5OqR2GiEiK8Z/QJLVDEBF55WnlW0RERERERMTKlHyLJFOuXLnYuHFjaodhsbVr15IrVy6r9rFkyRIKFSr0n7RjSZn/YswiIiIiIpZQ8i0vrcjISDJkyMDOnTtTO5Q43b9/n/Dw8NQOI06rVq0iT548ZsfCw8O5f/++VfsNCwsjODj4P2nHkjL/xZhFRERERCyh5FteWiaTieDgYCIiIlI7lFdOXIlpkyZNuHbtWipFlDrexDGLiIiIyMtJybdYXaFChciQIQOZMmWiZMmSjBkzhrCwMOP8o0ePyJAhA7NmzaJOnTpkzZqVadOmkT9/fgCaNm1KhgwZqFChAgA//vgjJUuWJGfOnDRs2JCDBw++cN9z5syhadOm5M6dm3LlyrF+/Xqzdi5fvkyzZs3Inj07lSpVYtGiRYmO++HDh3z66acUKlSIPHny0K9fPyMhjm/MAKdOnaJVq1bkyJGDt99+m8GDB/Pw4UOj3YMHD5IhQwYyZMhAtmzZqF69OmvWrDHO79y5k+7duxMcHGyUGz16NJs2baJo0aIAhISEkDVrVjZv3mwW8/bt28mcOTO3b9+2KJb4/Pnnn1SuXJlcuXLRvHlzAgICzM5v27aNmjVrkjVrVooVK8bkyZMxmUyx2pk9ezYVKlQge/bstGnThuvXryepzPNjjpbYmO7fv0+fPn0oUKAABQoUoG/fvgQFBSU6ZhERERGRhCj5Fqs7cOAA/v7+nD9/nh9//JE///yTcePGGeejV7iHDRvGoEGDOHXqFD179uTIkSMALFq0CH9/f7Zu3cqaNWsYOXIkkyZN4tChQwwZMoQJEya8cN8jRozgww8/ZM+ePTRt2pR27doZCWhUVBRNmzbFZDLh6+vLjz/+yNdff82jR4/i7TcqKoqGDRuydetWZs6cyfbt2ylTpgwzZ85McMx+fn5Ur16dypUrs2fPHpYuXcrRo0fp1q2b0Xbp0qXx9/fH39+fI0eO0KtXL9q1a8fhw4cBqFSpEtOmTcPV1dUo9/nnn5ttwXZ1daVSpUrMmzfPLO558+ZRpkwZsmTJYlEscbl16xZz587lt99+w9fXlwcPHtCrVy/j/Llz56hfvz61atVi//79/O9//2PMmDFMnTo1Vjs//fQTv/76K1u3buXevXu0bt06SWVibju3ZEyffvop//77LytXrmTr1q1UrFiRn3/+OcExi4iIiIgkxsYU13KTiBUtXryYL774gvPnzwPPVohdXFyYOnUqH330kVEuIiKCNGnS4OPjg7e3NwDfffcdixYt4sCBAyna98yZM40EMSoqCmdnZ5YsWULTpk1Zs2YNbdu25erVq7i7uwPPVohr1qzJ6tWradq0aax+1q5dS4sWLTh//jxeXl6xzsc35n79+nH37l2WLFliHPPz8yN//vzcuHGDbNmyxTmujh07kitXLiZOnAjAsmXL6N27t1niuWLFCrp06WKs8i5dupSePXty69YtnJycCAsLI2vWrEyZMoXu3bsnK5b58+fTrVs3bt++bczVhg0baN68OU+fPsXW1pb33nuPM2fOsGPHDqPet99+y8SJE7lz547RTteuXTly5AilSpUCICAggHz58rFlyxa8vb0tKhNzzJaMqXr16jRs2JARI0bEOdfPCw0NJTQ01PgcEhKCp6cnngOX6KfGROS1op8aE5E3VUhICG5ubgQHB+Pq6vpCbWnlW6xu/fr11K1bl9y5c5MhQwZ69erF5cuXY5UrWbJkom01a9aMCxcuUL9+faZNm2Yk0S/a99tvv238b1tbWzJlymRsNT5x4gRvvfWWkUzCs9VlGxubePvdv38/Xl5ecSbez4s55n379rFmzRrc3d3JnDkzmTJlomzZsgBcvHgRePYiugkTJlCmTBmyZs1KhgwZWL58eZzjSkizZs2wtbVl1apVwLMvDEJDQ42VY0tiiYu7u7vZXLm7uxMeHs6DBw+AZ/NZtWpVszrVq1cnMDCQW7duGcccHR2NpBogd+7c5MyZk5MnTyapzPMsGVOXLl0YN24c7733HkuXLuXevXvxjnX8+PG4ubkZfzw9PeMtKyIiIiJvNiXfYlXHjx+nVatWtGzZEl9fX/z8/Pj999/jfEu4g4NDou0VLFiQ8+fP06ZNG7Zt20bZsmVp164dUVFRL9S3rW3svwrRm0IiIyOxs7OLVT6uOtGioqJIkyZNouOJOebw8HD69OnDhQsXuHjxIn5+fly6dIl79+5RsWJFAL755humT5/OhAkTOHToEP7+/nTo0MHsWXZLODo60qZNGxYsWADAggULaNGiBS4uLhbHEpf45iV6PiMiIrC3tzc7F/35+ZfrxZzz6HJJLfM8S8bUt29f9u3bR548efjll1/ImTMnM2bMiLO9YcOGERwcbPy5cuVKnOVEREREROwTLyKSfDt27KBAgQIMGDDAOHbu3DmL6tra2mJjYxMrsXZ3d6dv37707dvX2DK8Z8+eWKupL9L38woWLMiFCxd4+PAhzs7OwLPEPjIyMt46xYsX59tvv+X27dtkyZLF4r5KlCjB7t27yZAhQ7xlfHx86NKlC/Xr1zeOnT592uz3rO3t7eP8QiKmzp0706hRIy5evMjatWtZtmxZkmJJjkKFCnHo0CGzYwcPHsTZ2ZkcOXIYxx49esSFCxcoUKAAAIGBgVy5csXst70tKfM8S8dUrFgxihUrxhdffMHkyZMZMWIEffr0iVXOwcHBoi+NRERERES08i1W5eXlhZ+fHwcPHsRkMvHPP/8wefJki+ra2tqSM2dOTpw4YRz74YcfmDt3rvHW8CNHjmBra0v27NlTtO/ntWjRgkyZMvHJJ5/w6NEjbt++zcCBAxOtkzdvXrp3787Vq1eJiIhg06ZNxgvX4vPpp59y7NgxhgwZYvzM2qFDh+jSpYvZuDZv3sy9e/cIDQ1l7NixsZJZT09PHj58yKVLlxLsr1atWmTJkoXOnTvj4uJCgwYNkhRLcgwcOJBNmzYxd+5cIiMjOXbsGOPGjWPgwIGxtvIPGDCAu3fv8uDBAwYMGICXl5fZlw6WlknKmPr06cPOnTsJCwsjNDSUf//91+yLDRERERGR5FDyLVbVuHFj+vXrR40aNXB0dOT99983e/N1YsaOHcuYMWNwdnamQoUKtGrVCh8fH7y8vEiXLh2ff/45c+fOJV++fCnedzQHBweWL1/Onj17cHNzo2TJkjRq1Ij06dMnWGfr1q04ODiQP39+3Nzc+P7776lXr16CfZUpU4ZNmzaxc+dOMmXKhJubG++//z7t27c3yowZMwYnJyeyZMlChgwZ2LZtGy1btjRrp2zZsnTt2pUiRYoYPzUWF1tbW9555x327dtHhw4dzLaDWxJLcpQvX565c+cycuRIHB0dqVatGq1atWLkyJFm5Tw8PKhYsSJvv/02GTJk4OzZs/z1119mW80tKfM8S8bUqVMnhg8fTsaMGcmQIQPnz583tuaLiIiIiCSX3nYu/4moqCjCwsJwdHQkIiKChw8fmm39vX//Pi4uLvEmTY8ePSIqKsp4HhkgLCyMtGnTWqXvkJAQHB0dY7UfGhpqbDMODg4mffr0sZ5fjikyMjLOZ8ATG3P0c8vxtR8WFoatrS329vY8efKEqKioWF8ImEwmQkJCcHBwwM7OjsePH+Pm5mZWJjw8nEePHuHk5BTvfCYWy/NtPXnyxOxNkJGRkTx48AA3N7dYK9uPHz8mXbp0sY7HbOfp06c4Ojomq0xcY7ZkTOHh4djZ2SX4bH9M0W/D1NvOReR1o7edi8ibKiXfdq7kW0QkhSj5FpHXlZJvEXlTpWTyrReuiYiksJOjG7zwf5xFRERE5PWiZ75FRERERERErEzJt4iIiIiIiIiVKfkWERERERERsTIl3yIiIiIiIiJWpuRbRERERERExMqUfIuIiIiIiIhYmZJvEREREREREStT8i0iIiIiIiJiZUq+RURERERERKxMybeIiIiIiIiIlSn5FhEREREREbEyJd8iIiIiIiIiVqbkW0RERERERMTKlHyLiIiIiIiIWJmSbxERERERERErs0/tAEREXjfFvtyIrYNTaochIpJi/Cc0Se0QREReeVr5FhEREREREbGy1zr5nj17Njdu3Ijzc0REBLNnz+bOnTsp3s/rIiAggAULFqR2GBazJF4/Pz8WL178H0X08rlx4wZz5841PlvjGlvrvokZe2rFISIiIiKSHK918t2jRw9OnToV5+enT5/So0cPLl68mOL9vIouX77Mn3/+aXbs8OHD9O3bN5UiSrqY8cY1pt27d/Pxxx//J/HE1X9qO3XqFD179jQ+W+MaW+u+iRl7asUhIiIiIpIcr3Xy/V/p3r07OXLkSO0wXsiBAwfo37+/2bHcuXPTpUuXVIoo6WLGG9eY/kup3b8lXrVrnBSv89hERERE5NXz0rxwzdfXl3Tp0lGxYkUALly4wM6dO2nWrBmZM2cGYMOGDWTPnp2SJUsCEBISwvbt2wkLC6NUqVLky5cvWX2fOXOGkydPkj17dqpWrRrrfGL9eHt7kzFjRuDZdvb58+fTpEkT7t+/z4kTJ8iSJQtVq1bFxsYmVr/Hjh0jT548lCtXjsWLF1OzZk1y5coVK4bn27158yYnT56kZMmSFClSBIAjR45w5swZPDw8qFGjBmnTpjXqnj59mvPnz9OiRQvj2LVr19i6dSvdunXj5s2b+Pr6EhYWxuzZswEoXrw4Hh4eVK9e3ajj5+fHgQMHaNu2LQcPHuTKlSuUKFGCggULxor30qVLHDhwADc3NypUqGDMT7TAwED27t0LQMWKFfHw8Ih9YYC5c+dSr149smfPDsA///xDUFAQ7dq1AyAsLIw///yTFi1a4O7ubsQb35iiRUZGJjoGa8xp2bJlzfqIntPWrVtz6NAhLl++TJkyZXjrrbcICwtj+/btBAcHU7ly5Ti/4Elsnk0mE9u3bycwMJASJUrEqv/8nD3v8uXLHDhwgIwZM1KtWjUcHBwAuHPnDmvXrgXA0dGRt956K9aYEmNpGwnF/uDBA/766y+aN29OpkyZjOPBwcH8/fffse4HS8YWLbE5FRERERFJjpdm5dvHx4fPP//c+Dxt2jR69OjB6tWrgWf/EO/atSsXLlwAYN26dXh5efH999+zYMECypUrxxdffJHkfkeMGEGzZs1YsGABzZo1o127dphMJuO8Jf3EtZ29c+fOtGrVioULF9KqVSvatm1rVmfy5MmULFmS2bNnM3jwYLy9venVqxdHjx6NM87odjt16kSHDh1Yu3YtV69eJSoqig4dOlCnTh2WLFlC//79KVq0KJcvXzbqbtq0iREjRpi1d+LECWML7/379zl16hQRERH4+vri6+uLn59frG27u3fvpk+fPtSoUYMRI0Ywe/ZsihcvzqxZs8za/vTTTylXrhwLFy5k4sSJFCxYkC1bthjnN2zYQL58+Zg6dSq///47lStX5q+//opz3N99953ZM9rvvfceHTp04O7duwDs27ePvn374ujoaBZvfGMCePLkSYJjsOacxrR7927ee+89ypYty5dffsnMmTN5++23mTJlCuXKleO7777jp59+onDhwhw8eDBJ8xwZGUnTpk1p27YtCxYsoEGDBowfP96sjbi2Zg8ePJi3336bGTNmMGHCBKpUqcKtW7eAZ19ERY/n77//pnnz5tSvX5/w8PA4r19cLGkjsdidnZ0ZNWoU8+bNM2t74cKFfPbZZ7i4uCR5bJbMqYiIiIhIcr00K9/e3t588803PH36FEdHR3x9falUqRK+vr68++67nDx5kqCgIGrWrMnt27fp0KEDS5YsoVGjRsCz1aoSJUrQsGFDqlWrZnG/jx8/5uTJkzg4OHDx4kWKFy/OsmXLaNeu3Qv1kyFDBjZs2ICtrS1nz56lcOHCHDp0iLJly3L16lWGDx/OrFmz6NSpEwCfffYZu3btSjTetGnTcuLECezs7ACYM2cO69at4/jx43h5eREWFkbdunUZOHAgf//9t0VzULhwYT744AOOHDlirNICrFixIlbZkJAQ3n//fWM777fffsuoUaPo0aMH8Cz5Wbx4Mf/++y9ZsmQB4LfffqN79+74+/uTJk0aJk2axIABAxg3bhzw7BocOHAgzti8vb3x9fVl4MCBXL16levXr1OiRAm2b99Oy5Yt8fX1pUKFCqRLl86iMc2fPz/RMcybN89qcxqXBw8eMHToUONeePfdd/nkk09Yvnw5rVq1AqBt27ZMmjSJRYsWWTzPc+bMYffu3Zw8eZKcOXPy6NEjKleunGAsc+fO5aeffmL37t3GavSpU6eMxDh//vxm43n06BFlypRh1qxZ9OnTx6K5saSNxGK3sbGhY8eOLFiwwOwZ/gULFtChQwfs7WP/py2xsVkypzGFhoYSGhpqfA4JCbFoDkRERETkzfPSrHxH/8N6z549BAcHc+LECUaOHImPjw/wbFt60aJFcXd3Z/ny5djb2xMUFMS8efOYN28eO3bsIEeOHGzbti1J/fbv39/Ydpo/f35atGhhrMK+SD89e/bE1vbZ9BYqVAh3d3fOnTsHPFtNz5AhAx07djTKf/LJJxbF27t3byPxBli6dClt27bFy8sLeJacf/LJJ6xevZqwsDALZ8Fyjo6OdO7c2fhctWpVrl69ypMnT4BnSVPhwoXZtGkT8+bNY+7cuYSFhXH9+nVj/C4uLpw6dYr79+8D4OTkRM2aNePsz9vbm23bthEVFYWPjw+VK1emQYMGZveFt7d3io7hv57TdOnSmd0LFSpUwM3NzUi8o4+dP3/e+GzJPP/111+0b9+enDlzApA+ffpEX0A2b9482rVrZ7YNvGjRomaPQoSHh7N9+3YWLlzI0qVLyZYtW7xfnsQnsTYsib1Lly4cOHDAmJfLly+za9eueJ/zTmxslsxpTOPHj8fNzc344+npmaR5EBEREZE3x0uz8u3o6EjFihXx8fHh4cOHFCtWjPr163P37l38/PzMkqyAgADs7OxibQetXLmykTBZKnfu3Gaf8+TJw/bt21+4nwwZMph9dnBwMFbIrly5gqenp9kz4NmzZ49ztS6mbNmymX0OCAigfPnyZse8vLyIjIzk6tWryX4OPj5ubm5mcUd/cREaGkq6dOkICAjA0dEx1px1797d+NJg0qRJ9OvXj+zZs1OmTBmaNm3KgAEDcHFxidVfzZo1CQ4O5ujRo8Y9UL58eT7//HPCwsLYs2cPw4cPT/Ex/Jdz6urqahZP2rRpY90/adOmNVthtWSeAwICqFKlitn5PHnyJBjL1atX43zvQbSzZ89St25dXFxcKFKkCM7Ozty+fdvsuevEWNKGJbEXL16c4sWLs2DBAr766iv+/PNPChQoYLw3Iqljs2ROYxo2bBiDBg0yPoeEhCgBFxEREZE4vTTJNzxb5fTx8eHRo0fUqlULe3t7qlSpgo+PD9u3b2f69OnAs+QJSHQ7ryWinx1+/nP0y79Ssp/nubu7x+r3wYMHREREJFo35kvbPDw8CAoKMjsW/Tl666ydnR1RUVFmZaJXeVOam5sbRYsWZebMmfGWyZ8/P5s3byY4OBhfX19Gjx6Nj48PmzZtilU2U6ZMFC9e3HhGePbs2ZQuXZozZ86wdu1aIiMjYyVpL+plm9O4WDLPHh4ecd7fCcmYMSN37tyJ9/zXX39NhQoVzJ7R79Chg9kXA4mxpA1LY+/SpQu//fYbX331FQsWLDDb0RBTYmOzZE5jcnBwiPXCNhERERGRuLw0287hWfK9b98+NmzYQK1atYxjP/30k/G8N0CjRo0ICgoynn+N9vDhQwIDA5PU5/MJwJMnT1izZg01atRI8X6eV716dS5dusSxY8eMY8uXL09WWzVr1mTlypVmicuiRYsoU6YMzs7OAOTMmZMrV66YJYcxV/dcXFx4+vRpsmJ4XuPGjVm2bBk3b940O+7v7x/rf7u5udGiRQsGDhzIoUOH4m3T29ub+fPnc+PGDSpWrIizszOlS5dmzJgxcT7v/aJjetnmNC6WzHONGjVYtWqV2YvMlixZkmC7DRs2ZNmyZWbPLj9+/Jjg4GDg2ZvKn9/1ERQUlOQXklnShqWxd+rUCT8/P2bMmMGpU6cS/GmxxMZmyZyKiIiIiCTXS7XyXblyZWxsbDh79qzxE0He3t4MHz6cYsWK4e7uDkCJEiUYPXo03bp1Y8eOHRQpUgQ/Pz9WrVrFsmXLjHKW2LJlC926daNs2bIsXLiQ9OnT8/7776d4P88rU6YMHTp0oEmTJnz00Uc8ePCAOXPmYG9vH2tlOzGDBg1i/vz5eHt7884773D06FEWLlxotorcoEED4xni5s2bc/DgwVirzNE/3zZo0CBKlChh9rNcSTF48GA2bdpEuXLl6NOnD25ubhw8eJDTp08bCXa/fv1wdnamatWqREZG8vPPP5s98xyTt7c3P/zwA3Xq1DF+7svb25tvv/02wTfcJ3dM1pzTpP4sV3wsmeeBAwfyxx9/ULduXdq0acPOnTsT/JIjut01a9ZQrlw5evTogclkYunSpSxduhQ3NzfatGnDoEGDcHFxwdnZmRkzZiT5nrWkDUtjz5UrFzVq1OCTTz6hUqVK5M+fP9ljs2RORURERESS66Va+XZ0dGTYsGEMGTLE2PJdvnx5evTowcCBA83Kjho1ih07duDm5sbJkyfx9PRk165dRsIDz57VfP63kZ//nCZNGrp3784///xD5cqVOX/+PE2bNmXv3r04OTklu5/odmP+bnX79u0pUKCA8XnevHmMGjUKPz8/nJyc2Lp1K5GRkbi6usY5N/G16+zszMGDB2nfvj2nTp0iR44cHDlyxFi9h2cvqzpw4ACVK1fmzJkz1KxZkzVr1tC9e3ejTLZs2fDx8cHOzo5t27bh5+dH7ty5zVYS8+fPT4cOHcz6d3d3p3v37kZS7OTkhI+PD9999x2BgYH4+/vTqFEj9u3bZ9RZv349nTp1IiAggBs3bvD999/zww8/xDlueJZod+/e3eyFW+3ataN79+60bt3aOBYz3rjGZMkYrDmnMcUVT8GCBWP9NF2RIkVo2bKl8dmSec6QIQMHDhygVq1anD9/nnr16rFhwwazGGPOWfr06dmxYwfDhg0jICCA0NBQli5daty7ffr0YcGCBQQGBnLlyhW+//57pkyZQu3ateNtMyZL2rAk9mifffYZ7dq1i/Xsf1LHZsmcioiIiIgkl43p+R+1lv/MnTt3zBLpZcuW0blzZ27duhXrZVsi8moICQl59tbzgUuwdXBKvIKIyCvCf0KT1A5BRCRVRP/7Ljg4ON6FUku9VNvO3yQ//PADly5dolKlSly+fJlffvmF4cOHK/EWERERERF5DWnlO5VERUWxbNkydu/ejbOzM3Xr1k3y71WLyMslJb8ZFREREZHUl5L/vlPyLSKSQpR8i4iIiLxeUvLfdy/VC9dEREREREREXkdKvkVERERERESsTMm3iIiIiIiIiJUp+RYRERERERGxMiXfIiIiIiIiIlam5FtERERERETEypR8i4iIiIiIiFiZkm8RERERERERK1PyLSIiIiIiImJlSr5FRERERERErEzJt4iIiIiIiIiVKfkWERERERERsTIl3yIiIiIiIiJWpuRbRERERERExMrsUzsAEZHXTbEvN2Lr4JTaYYiIJIv/hCapHYKIyGtJK98iIiIiIiIiVqbkW/4z69evp2HDhqkdhmHbtm3UqFEjtcN4Ifv376dSpUqpHUaqeB2un4iIiIi8OZR8y38mKCiIo0ePpnYYhnv37nH48OHUDuOFhISEcPDgwdQOI1W8DtdPRERERN4cSr7ljeXt7c2OHTtSOwxJJl0/EREREXmVKPmWFBUWFsa3335Lw4YNadiwIRMnTiQ8PNyszO7du+natSve3t4MGjSI4OBgs/Pnzp2jV69eVK1alZYtW7Jq1Sqz89Hb1319fenUqRPVq1fn888/5+HDh0kqc+zYMT7++ONYdRKL79ChQ7Rt25ZatWoxcOBA1q1bl+DW75MnT1KuXDnKlStHtWrV6NGjB8ePH49zTIn1ffToUdq1a2f0HRgYGG+/ls6DJXOe0HVN7JrfunWLIUOGUKNGDZo1a8a0adOIiop64bZjXj9LxmHJPFtyD4uIiIiIJJWSb0lRI0aMYM6cObz//vsMGTKEe/fu8dVXXxnng4KC+Oijj2jTpg1Dhw5l8+bNvPfee8b5GzduULFiRcLCwhgzZgzVqlWjXbt2LFq0yKyNrVu30qtXL1q1asWQIUNYs2YNnTp1SlKZmNuWo+skFJ+/vz81atQgU6ZMjBw5ksyZM9OuXbsEt357eXkxffp0pk+fzvjx48mZMycVK1bEz88vSX1fvXqVGjVqkCFDBkaOHEmmTJno2bNngtfDknmwZM4Tuq4Jnbtz5w4VKlQgJCSEL7/8kl69evHLL7/w4YcfvnDbMa9fUu6dhOY5sXtYRERERCQ59FNjkqL27NnDu+++S4sWLQCoU6cOoaGhxvmIiAiWLVtG3rx5AXj69CmdOnXCZDJhY2PDN998g6enJ/PmzTPq3717l2HDhvHOO++YtTNr1izjhVv58+enePHiHDx4kHLlyllcJqbE4vvuu+8oWrQoM2bMAKB27doEBAQwa9aseOckffr0Zv1Vr16dEydO8PvvvzN27FiL+540aRKFChXit99+M/r29/dn7ty5CVyRxOfBkjlP6LomdO7bb7+lcOHC/Prrr0Y8Xl5elC5dmq+//ppMmTIlu+2YknLvJDTPSekzNDTU7FxISEiC10JERERE3lxa+ZYUVbduXSZNmsQ333zDkSNHMJlMODg4GOczZ85sJD0AOXPm5MmTJ0bScujQIerXr2/WZsOGDfH39+fu3bvGsTRp0lC9enXjc9GiRcmePTuHDh1KUpmYEovv8OHD1KlTx6xOzM9xWbZsGe3bt6dq1aqUK1eOPXv2mK18W9L3wYMHqVu3rlmdevXqJdp3YvNgyZwndF0TOufr68vJkyepVKkSFStWpEKFCvTs2ROTycTZs2dfqO2YLL13EpvnpPQ5fvx43NzcjD+enp4JXAkREREReZMp+ZYU9dVXXzFjxgz+/fdfWrRogaenJ+vWrTPO29ubb7awsbEBwGQyAfDgwQPSp09vVsbZ2dk4Fy1dunRG3Wjp06dPcpmYEovv4cOHODk5xWozITNmzKBPnz54e3szduxYpk+fTrVq1Xjy5InV+4bE58GSOU/ouiZ07tGjRzRo0ICffvqJn3/+mV9++YVff/2VAwcOUKxYsRdqOyZL753E5jkpfQ4bNozg4GDjz5UrV+IsJyIiIiKibeeS4lq0aEGLFi0wmUwMHTqUPn36cPXqVYvqFihQgNOnT5sdO3XqFGnTpiVXrlzGsZCQEG7cuEH27NmBZ0nelStXyJ8/f5LKJFW+fPmMFdtoZ86cSbDO33//Td++fenfv79xLCgoiIwZMya575h9/fvvv4nWS2weLJ3zhK5rfOfeeustAgIC4t3m/yJtx2TpOCxhaZ8ODg7xroqLiIiIiDxPK9+SosaOHcuNGzeAZyuKadOmxdHR0eL67733HitXrmTnzp0ABAYGMmHCBHr16oWdnZ1Z2eHDhxtvzf7yyy/JmDEjDRs2THKZpHj33Xf566+/jC3b165d4+eff06wTqZMmTh69CgREREALFq0iO3btye57549e7J8+XIOHDgAwJUrV/jpp58sqpvQPFgy5wld14TOffjhh/zzzz9Mnz7diOXu3buMGjXK+JzctmNKyr2TkBe9h0VERERE4qKVb0lRuXLlokKFCqRNm5bIyEhMJhOzZ8+2uH6DBg0YOXIk9evXJ2fOnFy7dg1vb2/GjRtnVi5TpkyEhYWRPXt27OzsCA8PZ9GiRaRLly5JZZKqRYsW9OnTh8qVK+Pl5UVQUBCNGzdmyZIl8dYZNWoUjRo1wtPTE0dHR+zs7PD29k5y302bNqV///5UrVqVvHnzEhQURIsWLRJ94Vpi82DJnCd0XRM6V7duXebMmcOQIUMYNWoUbm5uhISEMGzYsBduOyZL753EvOg9LCIiIiISFxtT9IOOIinEZDLh5+eHra0tuXPnNlYd7969y5UrVyhZsqRR9vHjx5w+fZrSpUubrU4+fPjw/7V332FRXO/bwG9670VQURGwgQ1FUURRsTesRFFjJ7H33jVRv7Ykauwdg12xRY09qCgqUgSxgSBioUiRImXeP3iZH0tdkA2W+3NdXHFmzpnzzJldwrPnzFm8ePECRkZG4nTpXB4eHpg+fTrevHkjPmdrZWUlMf1XmjIfPnxAWFgYGjduXOr43r17h/j4eFhYWGDbtm1Ys2ZNgQXU8srMzMTz58+hoKAACwsLREZGIiMjQ5z6Xda2U1NT8eTJEzRp0qTQdqXpB2n6HCj6vpZ0DACys7Px7NkzKCgowNzcHPLy8p997vz3T5rrkLafS7qeoiQmJuYsvDb5MORV1EuuQET0BQpf2a2iQyAi+mLk/n2XkJAAbW3tzzoXk2/66uRNKD+nTFlt27ZNnMocERGBdu3aoUePHli/fn25t/W5ZNkPVBCTbyL6FjD5JiL6P+WZfPOZb6JSioyMhLGxMerUqQMrKys0a9YMS5cureiwiIiIiIjoC8aRb/rqFDZ1uCxlPkdqairCw8NhZmYmfp3Vl0jW/UCSyvOTUSIiIiKqeJx2TkT0BWLyTURERPRt4bRzIiIiIiIioq8Ik28iIiIiIiIiGWPyTURERERERCRjTL6JiIiIiIiIZIzJNxEREREREZGMMfkmIiIiIiIikjEm30REREREREQyxuSbiIiIiIiISMaYfBMRERERERHJGJNvIiIiIiIiIhlj8k1EREREREQkY0y+iYiIiIiIiGSMyTcRERERERGRjClWdABERN8am0UXIK+iXtFhEBFJLXxlt4oOgYjom8eRbyIiIiIiIiIZ+66Tbx8fHyQmJha5nVd2djZ8fHyQnJz8n8b0PfH19cWHDx8qOozvSnx8PHx9fb+Y85RGTEwMHjx4UGyZioiLiIiIiKgw33Xy3aJFC9y9e7fI7bxSUlLQokULBAUF/acxfU/atm0Lb2/vCmv/e/zg4/r162jbtm2p6sTFxeHevXuffZ7Pdf78eXTt2rXYMhURFxERERFRYb7r5PtL1Lx5c+jo6FR0GN+dzMxMtGjRosSRVAKuXLkCZ2dniX36+vpo1qxZBUVUtC81LiIiIiL6/nwxyXd0dDQCAwORlpYm7ss71Ts9PR2PHj3C27dvxeMJCQnw9/dHQkJCgfM9ePAAPj4+uHv3LqKjo8s11vT0dAQFBeHdu3cF9vv4+CA9PV1i/927dwtMpy7segHgt99+Q+3atQFIXv+nT58QHByM9+/fFxnXs2fPEBwcjE+fPhV6vKg2SzqWV0n9WpqYU1JS4O/vX+qp5mFhYQgJCUF2dra473P7/v79+wCA4OBg+Pj4wN/fXzyWmZmJoKAgBAcHS7SZ/3pL+xrNVdR9y3vupKQk+Pn54c2bNwXql1QuPT0dAQEBePnyJQRBKDIOAEhOToaPjw98fHzw8OFDJCUlSRxPTEzE06dPkZWVJZaLjIxE/fr1sWrVqgLnS0hIgJ+fH169elXgWN5p4x8+fEBQUBBSUlIKlMvKysKzZ8/w7NmzIuMvqn7+uPK2GRcXB39//0Jf89K0SURERERUGhWefKempqJbt26oW7cuhgwZgipVqmDNmjUA/m+q96hRo1C9enUMGDAAVatWxeLFi7F69WrUrl0bgwYNgomJCTw9PSXOu2DBAkyePBnjxo2DtbU1HBwcEBUV9dnxbt++Haampujfvz+qVq2KBQsWiMeio6PRokULREZGStRp3bq1OJ26uOsFJKed517/1KlTUb16dbi6uqJq1aqYM2eOxPn9/f1Rr149ODk5oX///qhUqRL27t0rVR+XFE9+JfWrtDGfPn0alStXhouLC2rXro1hw4YVSGzze/jwIWxsbGBra4s+ffqgdu3a4vTnz+37efPmAQA2btyIyZMnY8WKFQCAS5cuoVq1aujatSucnZ1hbm6OW7duFbjesrxGS7pvueeePHkyLCwsMHr0aFy/fr1AvxRXbvv27ahcuTJcXV3RqlUr1K9fv9hHJ16+fInJkydj8uTJ+PHHH2FsbIyJEyeKx589e4bdu3cjNTVVLHf69OlCp3cvWrQIJiYmGDRoEOrUqYOOHTsiPj5ePH7+/Hk4Oztj6NChqF+/Pnr37o3KlSvj4sWLEvfcysoK7dq1Q48ePWBpaSnxaMKnT5+KrZ8/rtw2f/jhB1hbW6NHjx6oWrUqLl26JHWbRERERERlUeHJ94EDBxAcHIxXr17h4cOHiI6OhoqKikSZN2/e4Pnz53j06BE2btyIpUuX4tKlSwgLC8OjR4+wfPlyTJw4UWKE6uzZs/Dx8YGvry9ev36NSpUqYebMmZ8d782bN/H48WOEhITg8uXLWLFiBa5du1au15vf48ePERoaisDAQFy4cAGrVq3C48ePAeSMVHbp0gXu7u549eoVHj16hNOnT+Pnn38Wk6zi2ixtPNL2a3ExJyYmYtiwYZg5cybCwsLw+vVrpKSkIDU1tch2ExMT0alTJzRv3hzv3r1DSEgILl68WKoPVIq71vPnzwMA/vzzT/j4+ODgwYNITEzEoEGDMGzYMERERCAqKgpdunTBoEGDCoyWlvY1Ks19y+Xv748nT57g3r17cHV1LfL68pe7fPkyZs2ahcuXLyMkJAQRERHo2bMnXF1di/ygw9raWhzR9vf3x+PHj3H06FEcOXIEAGBra4tff/0VmpqaYrmxY8cWOM+5c+ewYsUKXLp0CSEhIQgPD0dUVBRmz54tUS4+Ph4WFhaIjIzE06dPMWzYMIlkf/78+XB2dkZERARCQkJw7do1vH79Wur6hYmPj4empiZev36NiIgIjBw5EkOHDhVffyW1mVd6ejoSExMlfoiIiIiIClPhyfenT5+goKAgJgPKysqYMGGCRJkpU6ZAQ0MDANC1a1cIgoBp06ZBTU1N3BcTE1NgSm5ycjKCg4Px8OFDtGjRolRJclEmT54MY2NjAICjoyO6du2KXbt2SV1fmuvNb9q0adDW1gYAODk5QVdXF4GBgQCA48ePIzU1FQ4ODvD19YWvry+UlZVhYWEhJpTFtVmWeKTp1+JiPnnyJARBwIwZMwAACgoKWL58ebFtHjt2DMnJyfj999+hpKQEADA3N0evXr2KrZdXaa/Vy8sLaWlpWLhwIQBATk4OK1aswKtXryRGSoHSv0aluW+5pk6dCl1d3RKvL3+5TZs2oU2bNsjMzBTbcHR0RHBwMF68eFHsud69ewc/Pz9ER0fD1ta21O+dnTt3wsXFBQ4ODgAAQ0NDzJ49G3v27EFWVpZYTklJCXPnzhW3+/TpgydPnoiPD3z69AlycnLihxZmZmYYMGCA1PWLsnz5csjJyQEAFi5ciPj4ePz9999StZnXihUroKOjI/6YmZlJ10FERERE9N1RrOgABg8ejBMnTqBq1apo27YtnJ2dMWTIEIkkolKlSuK/c5OZ3AQ4776PHz+K+6ZNm4Y///wTZmZm0NXVRXJycrk8+537PHauOnXqSExDLok015tf3usHAHV1dfFaQ0NDkZmZifHjx0uU0dDQEJPU4tosbTzS9mtxMT9//hzm5uZifABgaWkJRcWiX45Pnz6FhYUFNDU1iyxTktJea26cqqqq4j49PT1UrlwZz58/lyhb2teoNPctV/Xq1aW6vvzlQkNDkZSUVKCN5s2bF3iWO9e7d+8wYMAA+Pr6wtzcHJqamggPD4eysrJUMeR6/vx5gQ9GrK2t8enTJ7x69UqMVV9fX+J61dXVIQgCUlNToaKigqVLl8LV1RU1atSAs7MzunfvDhcXFzFxLql+YbS1tWFiYiJua2howMzMTPxAoqQ285ozZw6mTp0qbicmJjIBJyIiIqJCVXjyra2tjX/++QcvX77E1atXsWvXLvzvf/9DUFAQFBQUynTOv//+G9u2bYO/vz9q1aoFIGfktF+/fp8db94EP3c7d4Q394/zvNPfs7OzJUb6irvesqxyrqqqCm1tbfj4+BRZpqQ2pY2nvPpVS0urQD+mp6cjMzOzyDrq6urFTun93L7PHbXOH2dh3+uenJwMLS2toi9QCtLct1zy8tJNUMlfTlVVFU5OTti0aZPUceWuYRATEyN+YDBkyJAik/WiFNZ3udul6Tt7e3uEh4fj/v37uHz5MiZMmIATJ05g3759pYonr9TUVAiCIJFM530fl6ZNFRWVEh8bISIiIiICvoBp57lJWPXq1TFs2DCcPn1afCa3rJ4/f45q1aqJCSKAAlN5yyrvdOPs7GxcunQJTZo0AZAz0iknJyexsrOfn59EUlne1+vk5ITXr18XWIgrd/SvpDZLE0959WvTpk3x7NkzhIeHi/vyLpJVmNatWyMiIgJ+fn4S+3Ofvf7cvldQUICSkhIyMjLE8nZ2dggPD8fTp0/FfXfv3kV8fLx4z8tKmvv2uZycnODl5VXoB0ZFef78ORwcHMTEOy0trUCMampqEv1UGDs7O/zzzz8S+86fPw9zc3Po6+tLfQ0fP36EnJwcmjZtilmzZuGXX37B6dOnpa5fmIyMDIlrCgoKQnR0tHhPZdEmEREREVGFj3yvWbMGoaGh6NWrF4yNjXHkyBEYGhqifv36ZT6no6MjpkyZguXLl6NFixa4cOECPDw8yiXenTt3wtjYGE2aNMHu3bsRExODSZMmAchJSjp27IiZM2di2bJl+PDhA3755ReJEbbyvl5HR0cMHjwYffv2xaJFi1CvXj28ePECO3bswPr169GyZcti2yxNPOXVr05OTmjdujV69+6NxYsXIzExEXPnzi12hLd169bo168fevTogcWLF8PMzAwXLlxApUqVMGvWrM/uezk5OVhbW2P//v1QU1ODlpYWWrdujc6dO8PFxQXLli1DZmYmZs+ejUGDBqFhw4alvu68pLlvn2v27Nk4ceIE2rVrh2nTpkFHRwf37t3DoUOHEBAQUGidtm3bYuPGjWjQoAE0NTXx22+/SXx1GgDUq1cPaWlp2Lp1Kxo2bIgqVaoUOM/MmTOxd+9euLm5YejQoXj48CHWrl2L/fv3l+oaunbtCkdHRzg4OCArKwubN29Gu3btSnWO/BQVFTFixAj8+uuvUFZWxty5c9G9e3cx+ZZFm0REREREFZ58L1y4EIcPH8bx48cRExODunXr4tatW9DX10dqaiqaN28uMU1VSUkJzZs3l5gmrKKigubNm4vP5jZs2BCnTp3C1q1bcenSJTRs2BAHDx4Uvz4qV/PmzSWmVuffzktBQQHNmzfHL7/8glOnTmH16tUwMTGBt7c3DA0NxXJ//fUXli5ditWrV6NGjRrw8PDAxIkToaenV+L15o8ht83803RtbW1hZGQkbu/btw/79+/HyZMncfz4cdSuXRt//vmnmEwU12ZJ8eQlTb9KG/OJEyewbNkyrF+/HtWrV8exY8cwe/ZssZ8K4+npie3bt+PEiRPIzs5G586dJZ5n/ty+37t3L1atWoU5c+agSpUqOHjwII4cOYL169fjzz//hLy8PMaOHSuxmnZh1yvNa1Sa+1ZUX+ZXVDkjIyPcu3cPGzZswI4dO6CgoAA7OztcuXJFLKOvr49mzZqJ2zNnzoSioqJYvkuXLnB2dkZcXJxYxtzcHPv27cPBgwexZ88eDBkyBDY2NhLnMTExwd27d/G///0Pv/zyC4yMjHDixAl07dpVIr78Mwg0NTXRvHlz8fn/06dPY+PGjfjjjz8gLy8PFxcXcZE8aernvz4AMDAwwN69e7Ft2zZERkaid+/emD9/vni8uDaJiIiIiMpKTsj7kCwR0TfMw8MD06dPL/DNCOUlMTExZ9XzyYchr6IukzaIiGQhfGW3ig6BiOiLlPv3XUJCgrhGUFlV+DPfRERERERERN+6Cp92TkT0XylsqrosBC3p9NmfjBIRERHRt4XTzomIykl5TksiIiIioorHaedEREREREREXxEm30REREREREQyxuSbiIiIiIiISMaYfBMRERERERHJGJNvIiIiIiIiIhlj8k1EREREREQkY0y+iYiIiIiIiGSMyTcRERERERGRjDH5JiIiIiIiIpIxJt9EREREREREMsbkm4iIiIiIiEjGmHwTERERERERyRiTbyIiIiIiIiIZY/JNREREREREJGOKFR0AEdG3xmbRBcirqFd0GEREUgtf2a2iQyAi+uZx5JuIiIiIiIhIxph801ft9evXmDdvHtzc3LB7926ZtnXv3j0MGzasyO0vSVBQENzc3IotI23f/ZfXKU3cuby9vfHTTz8VW+ZLvkdERERE9H1h8k1ftW7duuHJkyfo1q0bGjZsKNO2Xr16haNHjxa5/SV58+YNDh06VGwZafvuv7xOaeLOFR4ejpMnTxZb5ku+R0RERET0feEz3/TVSkpKwsOHD7Fr1y40btxY5u3Z2dlh7969Mm/nv1Cavvuar/trjp2IiIiIvi1Mvumr5OfnhyVLlgAA5s6dCx0dHSxZsgQKCgqYP38+AEBVVRVWVlYYNWoUKlWqJNb19vaGh4cHxo8fj6NHj+Lly5do1qwZxowZg5CQEOzZswcJCQno2LEj+vfvL9aLjo7G6dOn0bdv3wLxJCUlYcyYMViwYAHq1asn7n/69CkWLlyIDRs2wNDQUKLOs2fPpI511qxZ2LdvHyIjI2Fra4sxY8ZAUfH/3r6vX7/Ghg0bEBMTg4YNG8Lc3LzUfff+/Xt4eHhgxowZ2Lp1K6KiovDbb78Vet1paWnYu3cv7ty5Ax0dHXTp0gUdO3Ys97gFQcCJEydw5coVAED79u3Ru3dviTJhYWFFtpE/9ty4pk6dCg8PD0RGRqJly5YYMWIEFBQUpG6TiIiIiKi0OO2cvkqVKlVCly5dAOQkRy4uLjAwMICenh5cXFzg4uKCNm3aIDAwEDY2Nnj37p1YNzw8HHv27EHfvn2ho6MDGxsbzJ49Gz179kS/fv1gZmYGS0tLDB8+HB4eHmK94qYwa2lpITo6Glu3bpXYv337doSGhhZIvAGUKtauXbtCQ0MDjRo1wq+//opJkyaJZeLj49GsWTPcu3cPdnZ2uHv3brHPORfVd7ltderUCYaGhujRowfU1dULXHdKSgocHR1x9OhRNG/eHFWqVMGwYcOwYsWKco/7f//7H8aPHw9zc3PUr18fnp6eWL58uXj8w4cPxbaRP/bcuNq3bw81NTXUr18f8+fPx9ixY6Vuk4iIiIioLDjyTV+lypUro1evXvjpp5/QuXNn2NjYiMd++OEH8d/Dhw9H27ZtsXXrVixYsEDcn56ejiNHjqBBgwYAgJiYGKxcuRKhoaGoVasWgJzE7cCBAxg8eLBUMY0aNQqTJ0/G6tWroaysjMzMTOzfvx/z5s0rtLyBgYHUsXp6eqJRo0YAAH19fbi7u2PTpk0AgHXr1kFDQwPnz5+HgoICxowZg969e+P06dOl7rv09HTs3bsXDg4ORV7n2rVroaCggAsXLkBePufzu+bNm8PZ2Rnjx4+HlpZWucV95swZTJkyBdOmTQMAuLu7S3w4UVIbhUlPT8eGDRvg4uIixt66dWtMmTIFderUKbHN/OdKT08XtxMTE4tsl4iIiIi+b0y+6Zvj5+eHI0eOICoqCunp6Xj16hVCQ0Mlyujo6IiJNwCYm5vDxMRETLxz9924cUPqdvv164cJEybAy8sL/fv3x7lz5xAfH1/s6t3SxpqbXAKApaUlkpOTkZSUBC0tLXh7e6NHjx7itGkA6NOnT5HJd3HU1NSKTbwB4Pz580hMTMTQoUMhCAIEQUBGRgY+ffqEx48fw87OrtzibtiwIXbu3ImaNWuiffv20NXVhbGxsdR9UxgFBQV0795d3HZwcIChoSFu3bqFOnXqlNhmXitWrBCn8BMRERERFYfTzumbcvLkSbRs2RJpaWlwdHSEi4sLatSogeTkZIlyqqqqEtvy8vJQUVEpsC8rK0vqtlVVVTF48GDxa7t2796N3r17Q09Pr9xjBSDGFhsbW6CNotosiba2doll4uLiULduXXTv3h09evRAz5490bdvX3h6eqJ69erlGvfatWsxevRorF+/HqampmjTpg0ePHggdRuF0dLSknjuPLfdmJgYqdrMa86cOUhISBB/IiMji2yXiIiIiL5vHPmmb8quXbvw008/Yd26deK+ffv2/Wftjx49Gra2tvDz88PZs2dx9uzZIsuWV6zVqlVDeHi4xL6wsLBSn0daZmZmyMrKkpgyXxbSxK2iooJp06Zh2rRpSEpKwvjx4zFo0CA8fvy4zO1++PABHz58gK6uLgAgIyMDUVFR4gcHpWlTRUWlwIc2RERERESF4cg3fVMUFRXx9u1bcfvWrVu4ePHif9Z+gwYNYGtri379+sHU1BTt27cvsmx5xTpgwAAcOXIEL1++BJDz3PHmzZtLH7yUhg8fjtOnT0vEmpWVVeqv9JImbg8PD/GZai0tLTRu3LjAzICyWLNmjfjvTZs2QUFBQVytXVZtEhEREdH3jSPf9E2ZPn06OnfujBYtWkBTUxMPHz6UeLb7vzBq1Ci4u7tj4cKF4jTowpRXrG5ubjh27BgaNWqEFi1aICgoCPXq1cOTJ08+5zKKNHDgQISGhqJnz56wtbWFjo4OHj16BBcXF/z444/lGveLFy9gaWmJevXqISsrC76+vsUupiYNbW1tXLlyBfb29lBWVsbdu3exfft2ccq7LNokIiIiImLyTV8tXV1deHp6wszMTNzXsmVLPH/+HHfv3oWCggLs7e3x4sULpKSkiGUcHR0LfCVYu3btJL5fGwC6du2KOnXqiNt2dnYSo7v5t3PVq1cPcnJyGD58eLHxlzVWCwsLeHp6QkNDA0DOAmJeXl7w8fFBbGwsbGxsoKqqiuvXrxfZdmF9V1hbRV3n4sWLMW7cOPj6+kJBQQGNGzeWWJSsvOJeuHAhxo8fD19fX8jLy6NJkybQ19eXuo3CYldTU8ONGzfg7++PyMhINGnSRKIfimuTiIiIiKis5ARBECo6CKJvyeDBg/HhwwecOXOmokOhfDw8PDB9+nS8efNGJudPTEyEjo4OzCYfhryKukzaICKShfCV3So6BCKiL1Lu33cJCQlSLU5cHI58E5WTw4cPY+fOnfj3339x69atig6HiIiIiIi+IBz5JiongYGBePLkCezs7FCtWrWKDocK8fLlSzx8+BC9evWSyfnL85NRIiIiIqp45fn3HZNvIqJywuSbiIiI6NtSnn/f8avGiIiIiIiIiGSMyTcRERERERGRjDH5JiIiIiIiIpIxJt9EREREREREMsbkm4iIiIiIiEjGmHwTERERERERyRiTbyIiIiIiIiIZY/JNREREREREJGNMvomIiIiIiIhkjMk3ERERERERkYwx+SYiIiIiIiKSMSbfRERERERERDLG5JuIiIiIiIhIxph8ExEREREREcmYYkUHQET0rbFZdAHyKuoVHQYRkdTCV3ar6BCIiL55HPkm+g7ExsYiIyOjosMos0+fPiE2NlZm5YmIiIiIZI3JN1E5iYmJ+WIT3OrVq+PChQsVHUaZnTt3DtWrV5dZeSIiIiIiWWPyTVQOMjMzYWRkhJs3b1Z0KN8kFRUVGBoayqw8EREREZGsMfmm715MTAwyMzMBAKmpqcjOzi6ybEZGBtLS0grsj4uLAwAkJCQgJiYGHz58kDiemppaYhxxcXGIiYkRz1VSrIXFkdfHjx9LbDO/zMzMIq+vNLElJSUVOE9KSgqysrKKrJ+dnS32U0ZGhsS08fbt2+P+/fsAgKysLMTExBS4T9nZ2WIMecsDQHp6ukTsxfVdcXEQEREREZUVk2/6riUnJ8PIyAjLli2DpaUl9PX1YWRkhN27d0uUi4qKQs+ePaGpqQkDAwPY2NjgypUr4vEWLVoAAIYOHYo6deqgS5cuyM7OxqRJk6CpqQkjIyNYWlri6NGjRcbi6OiIOnXqwMLCApqamhg6dCgSEhIKxPrrr7+iTp06MDAwgK6uLrZu3Spxnnv37qFu3brQ09ODsbExli9fDkEQiu2HV69ewcXFBRoaGjAwMICzszPCwsJKHdvy5ctRo0YNmJmZYe/evQCAf/75B/Xr14e+vj50dHTQq1cvREVFSbS/fv166OnpQV9fH3Xq1MGiRYtQqVIl8XjeaeRpaWmoUaMGjh8/LnGOs2fPokqVKkhKSiow7fzIkSOoU6cOli5dChMTE+jo6KBevXrw8/MrVRxERERERGXF5JsIwPbt23Hs2DF8/PgR69evx6hRo+Dv7w8gZ/GuDh06wNzcHHFxcUhOTsbMmTPRq1cvvHz5EgAQGhoKAPDy8kJMTAxu376NgwcP4tChQwgMDERycjIuX76M27dvFxnDo0ePEBMTg/j4eDx58gQvXrzAvHnzCpQ7dOgQzp49i48fP+KPP/7A+PHjxTjS09PRp08ftGnTBomJiQgJCcGlS5eQkpJSZLvp6elo37490tPTER4ejqSkJMydOxdXr14tdWwHDhzAhQsX8OHDB7i7u8PPzw99+vTBkiVLkJKSgrdv30JbWxs//PCDWOfGjRuYMWMGdu/ejZSUFOzfvx/bt28vMl4NDQ24uLjgwIEDBdru1q0b9PT0Cq33/v17vHz5Es+fP0dSUhLq168Pd3f3MseR23eJiYkSP0REREREhWHyTQRg2rRpaNiwIeTl5TF06FA4OTlh06ZNAIDTp08jOjoaixYtwqdPn/Dhwwd0794dVlZW8PLyKvKcb9++hZGRkTgCW716daxdu7bEWNLT06GsrIzRo0fjzJkzBY4vWLAAFhYWAHJG2tXU1PDgwQMAOcl/fHw81q5dC1VVVRgYGGD9+vXFtnfy5Em8fPkS+/btg6mpKeTl5dGuXTuMGDGi1LHNmTMHtWvXFrfXrVuHHj16oF27dkhISEB6ejpmzZoFb29vREREAAD+/PNP9OjRA3369IGcnBzs7Owwfvz4YmMePHgwzp07h/j4eAA5I++nTp3C4MGDi6wjLy+PP/74AxoaGlBWVoa7uzsePHggLpJXljhWrFgBHR0d8cfMzKzY8kRERET0/WLyTQSgYcOGEtuNGjUSR7P9/f2RnJyMWrVqwcrKClZWVqhVqxYiIiIQExNT5DkHDhyItLQ01K5dG9OmTcPff/9d7DPPmzdvRo0aNaCpqYlatWph4sSJiIyMLFAu/yre2tra4hTw0NBQWFhYQENDQzxev359KCgoFNluYGAgatasCSMjo8+OrVatWhLb/v7+8PLygqWlpdhvTk5OMDAwEKeeP336tED/59/Or0OHDtDT0xOn8R8/fhwqKiro1q3o76k1NDSU6BdtbW1kZWWJz8aXJY45c+YgISFB/CmsT4iIiIiIACbfRAAgLhSWd1tZWRkAICcnBzMzM8TExBT4Wbp0aZHnNDExQXBwMLZs2QI1NTVMmDAB9vb2hS72dfPmTUyZMgVbtmwRFwfbt29foYu/ycnJFdmmkpJSgWvJzs4udhE5RUXFYr8irTSxKSoqFojV3d290L7LfU5eUVGxQMwlfWWbgoICXF1dxannBw4cQP/+/aGiolJkneL6raxxqKioQFtbW+KHiIiIiKgwTL6JgALPYt+8eRM2NjYAgObNmyM8PByBgYEF6uUuZKagoAAFBQWJkW1BEKCkpIT27dtj+fLluH37Nu7du4e7d+8WOI+fnx8sLS3RuXNnyMvnvC29vb1LfR3169fHs2fP8P79e3HfrVu3il1wrVmzZggLC8OLFy8KvbbPia158+Y4d+5cgaQ2bzw2Njbw8fGROJ5/uzBubm64ceMGfH19cfny5WKnnEujrHEQEREREUmDyTcRgN9++w2HDx/G8+fPMXfuXAQFBWHSpEkAgM6dO6NNmzbo168f/v77b7x8+RJXr17FwIEDce3aNQA5o6o1a9bExYsX8ebNG3z48AHLli3DwoULcf/+fURGRmLPnj1QU1ODlZVVgfZzp7kfPHgQL1++xLZt27Bx48ZSX0eXLl1gZWWFH3/8EY8ePcLt27cxduzYYut07twZ9vb26Nu3L65du4bnz5/jzz//FJ9P/5zYZs+ejbdv32LgwIF48OABnj9/jkOHDsHR0VEsM2XKFFy/fh0rV65EWFgYDhw4UGAF98I0a9YMlpaWcHNzQ9WqVSXOWRZljYOIiIiISBpMvokArF27Fnv37kX79u1x5coVnDt3DjVq1ACQs1DXuXPn4OrqilmzZqFVq1b45Zdf0Lt3b7Rt21Y8x6ZNm3Dt2jU0btwYXbp0wbRp06CiooIxY8agVatWuHz5Mi5cuABTU9MC7bdq1Qq///47li9fDgcHB3h5eeF///sfDAwMxDJycnIwMDCAkpKSRF19fX2oqqqKsZ45cwbKysro1KkTpk+fjiVLlqB69eriNPr85OXl8ffff6N9+/Zwd3dH586dERISgjFjxnx2bDVr1sTdu3ehqqqKfv36oXPnzvDy8sLmzZvFMjY2Njh+/DiOHj2Kdu3a4fDhw1i8eLF4TUDO9G5DQ8MCsY8YMQJxcXEYPny4xLTy/OVzF5/LS0lJCQYGBuJovjRxEBERERGVlZxQ0hcAE33DkpOToaWlhdu3b8Pe3r6iw6H/b+HChTh9+nSB7+H+0uNITEzMWfV88mHIq6jLODoiovITvrLoBSuJiL5nuX/fJSQkfPb6Phz5JqIKN2jQINy4cQPh4eHYs2cP1q9fL468f49xEBEREdG3R7HkIkTfrqKmS9N/a9SoUVi6dCmePn0KMzMzbNq0CUOHDv1q4wha0okrnxMRERGRBE47JyIqJ+U5LYmIiIiIKh6nnRMRERERERF9RZh8ExEREREREckYk28iIiIiIiIiGWPyTURERERERCRjTL6JiIiIiIiIZIzJNxEREREREZGMMfkmIiIiIiIikjEm30REREREREQyxuSbiIiIiIiISMaYfBMRERERERHJGJNvIiIiIiIiIhlj8k1EREREREQkY0y+iYiIiIiIiGSMyTcRERERERGRjClWdABERN8am0UXIK+iXtFhEBGVKHxlt4oOgYjou8GRbyIiIiIiIiIZY/JN9BV4/Pgxfv311wpp+8WLF1i6dGmRxwMCArBmzZrPbif/NUpzzSXFRkRERET0pWDyTZRPcHAwVq5cWdFhSGDyXbbYiIiIiIi+FEy+ifL5EpPv71GdOnUwd+7cig6DiIiIiKhccME1ojzCw8Nx+PBhpKWlYfHixQCAVq1awdjYGBcvXsTo0aNx7NgxREVFYcKECUhOTsaOHTsAAKqqqrCyskKPHj2grKwMAEhMTMS6deswevRoVKlSRWzn7du32Lx5M8aPHw9DQ0MAwPXr13Hnzh3o6Oigffv2sLS0lDruV69eFRsHkDNCffHiRbi7u+PcuXOIjIyEra0t2rVrJ3GutLQ0HDp0CDExMWjYsKHUMSQlJRV73qysLJw6dQqPHz+GkZERXFxcxGuXVkmxldQPSUlJWLt2LUaNGoWqVauK9d68eYMtW7ZgwoQJMDAwQFBQEK5duwYAcHJygo2NTaniJCIiIiLKjyPfRFIICAjA4sWL0bx5c3h7eyMrK6tAmfj4ePz666+wtbXFx48fAQDa2to4ePAg9u7dK1HWw8MDu3fvhoGBAbKystC/f3+MGjUK0dHR8PX1RdOmTfHXX3+VKdbC4sh7DS1atMA///yD8PBw9OzZU2KUPy0tDS1btsSKFSsQFRWFBQsWYNKkSSW2mZycXOx5P336hNatW2P69Ol49+4dDhw4gNq1ayMgIKDIc+afdl7a2ArrBy0tLXh5eWH79u0SZXft2gUPDw/o6+tj//79aNGiBfz9/fH8+XP8+OOP2LZtW4l9QERERERUHI58E+VRo0YNDBgwABcvXhRHvoGcZPnjx4/49ddf0adPH3G/rq6uRLmsrCzY2dlh27ZtmDJlCgDAzc0NBw4ckJhCfeDAAQwaNAhycnLYunUrgoKCEBAQADU1NQBA9+7dMWzYMPTp0weqqqolxl21atUS4wCAjx8/Yv369ejQoQMAwMbGBgsWLMDs2bMBAJs3b8abN2/w+PFjaGtrIysrC23atCmx/ZLO++eff+LJkycIDQ2Fvr4+BEFA7969MWnSJFy9erXE80sbmzT9MHLkSKxevRqLFi2CvHzO54979uzB8OHDIScnh507d2LWrFmYP38+ACA7OxuPHj0qNKb09HSkp6eL24mJiVJdCxERERF9f5h8E0lJWVkZLi4uBfbHxsbi7NmziIqKQnp6OuTl5SVGdN3c3LBw4UI8fPgQjRo1QkhICPz8/LB//34AwNGjR6Gnp4e1a9dCEAQIgoDU1FQkJCQgNDRU6qnfJcUBABoaGmKCDACNGjVCTEwMPn78CA0NDfz999/o27cvtLW1AQAKCgoYNmwYfHx8im1b2vPq6+sDAOTk5DBmzBh0794dqamp4ocOxZE2tpL6YfDgwZg5cyYuX76MDh064MaNG+IIN5CTwF+4cAG9e/eGtbU15OXlUb9+/UJjWrFiBZYsWVJi7EREREREnHZOJCU9PT1xpDSXj48PqlevjoMHDyI2NhYAIC8vj/j4eLFMzZo10aJFCxw4cABAzqh3o0aNYG1tDQCIjo6GiooKMjMzkZWVhezsbKioqGDRokXQ0dGRKjZp4gAATU1NiW1FxZzP3zIyMsRYTExMJMqYmpqW2H5J5339+nWB81SuXBmCIOD169clnl/a2KTpB11dXfTt2xe7du0CkDPlvGPHjuIz4OvXr0edOnXg5OSEKlWqwN3dHa9evSo0pjlz5iAhIUH8iYyMlOpaiIiIiOj7w5Fvonzk5OSkLrt27Vq4urpi586d4r6AgABkZ2dLlHNzc8OKFSuwcuVK/PXXXxg3bpx4zNjYGFWqVJGYLl1a0sZRElNTU7x580ZiX3R0dJnjylW5cuUC54mOjoacnBwqV65cbrFJ2w+jR49Gp06dEBERgaNHj2LPnj3iMSMjI2zfvh3btm1DQEAApk+fjr59++LOnTsFYlJRUYGKiopU8RMRERHR940j30T56OjoICUlpdBF1fJLSUmReCY7LCwMFy5cKFDO1dUV7969w6+//oqXL19i4MCB4rG+ffvi2LFjCAkJkahz69YtqWOWNo6SdO7cGceOHROfXc7KypJITMuqS5cuOH78uMQI9Pbt29GmTRupppxLG5u0/dC6dWtUq1YNAwYMgKqqKnr27Ckey+13OTk5NGzYED/88APCwsKkvlYiIiIiosJw5JsonyZNmkBNTQ2urq6wsbFBq1atiiw7atQo/PDDD0hJSYGmpiaOHDkCY2PjAuUMDQ3RsWNHLF26FO3atZMY7R03bhxu3bqFZs2aoX///tDR0cG9e/egoaGB8+fPSxWztHGU5Oeff8a+ffvQrFkzdO3aFT4+PkhISCj1efIbO3Ysjhw5Ajs7O/To0QP+/v7w9/eXerE1aWMrTT+MHDkSs2bNwqRJkyS+km3r1q2YOHEiHBwckJWVBU9PT4lF64iIiIiIyoLJN1E+enp68Pf3x9mzZxETEwMAaNCgAaZPn16gbO/eveHj44MrV65AQUEBly9fxsuXLwtd9XrBggVo2rSpxMJkQM7CYQcPHsSdO3dw8+ZNKCgooH///mjZsqVYpk6dOhKrpZcljsKuoXLlyli0aJE4Wqympobbt2/j0KFDiI2NxZIlS2BhYQEPD48i25bmvMrKyrhx4wZOnTqFkJAQDBw4EIcPH5b4nu/815h/W5rYSnM/nJycAAAjRoyQ2L937148ePAA//77L+Tl5XH58mU0atSoyOsnIiIiIpKGnCAIQkUHQUT0X5s8eTLu3r1bqun9JUlMTISOjg7MJh+GvIp6uZ2XiEhWwld2q+gQiIi+aLl/3yUkJIjfulNWHPkmou/KlStX4OXlha1bt+L06dMVHQ4RERERfSeYfBPRd8fY2BgXLlxAmzZtZHL+oCWdPvuTUSIiIiL6tnDaORFROSnPaUlEREREVPHK8+87ftUYERERERERkYwx+SYiIiIiIiKSMSbfRERERERERDLG5JuIiIiIiIhIxph8ExEREREREckYk28iIiIiIiIiGWPyTURERERERCRjTL6JiIiIiIiIZIzJNxEREREREZGMMfkmIiIiIiIikjEm30REREREREQyxuSbiIiIiIiISMaYfBMRERERERHJGJNvIiIiIiIiIhlTrOgAiIi+NTaLLkBeRb2iwyCi70z4ym4VHQIRERWDI99EREREREREMsbkm745WVlZ0NXVhY+PT0WHgoSEBOjq6uLhw4cVHUqZldSfX+I1nj17FlWrVq3oMIiIiIiIREy+6ZsjCAISEhKQmZn5n7YbFxcHXV1dBAYGVngs5amka/gSrzEjIwMfPnyo6DCIiIiIiER85puonGRnZyMhIQFZWVniPh0dHcTHx0NLS6sCI/v+dOvWDVFRURUdBhERERGRiCPf9EWLj4/HuHHjULNmTdSoUQMDBgxAWFiYRJnQ0FB07NgRpqamcHBwwOnTpyWOBwYGQldXF3FxceK+9PT0AlOp4+PjMWHCBFhaWsLc3ByTJ0/Gx48fxeMmJibQ1dWFgYEBGjdujNWrV0sk2hYWFgAAR0dH6OrqokOHDkhMTESNGjUkRsPDw8Ph6uqKypUrw9zcHOPHj0dSUpJ4/PXr19DV1YWnpyc6deoEMzMz2Nvb4/r168X2VUnxSXvekvpTGtLcN2njPXDgAFq3bg1jY2P89ddfOHz4MGrXro2//voLLVq0QNWqVdGzZ09ERESIdS9evAhra2txW5o6AODl5YVGjRrBzMwM3bt3x9atW2FoaFjq6yciIiIiyo/JN32xMjIy4OzsjISEBJw5cwbXr19HlSpV4OTkhOTkZADAp0+f0LlzZ+jr6+PGjRv43//+hxkzZkicJysrCwkJCcjOzhb35Z8qnZ6ejrZt2+L+/fvw8PDA1atXUbNmTXh4eIh1QkNDER4ejtDQUKxZswabNm3Chg0bxON+fn4AgHPnziE8PBwnTpwo0E5GRgY6duyItLQ0XL16FYcPH8aNGzcwZMgQ8Ty5I+gLFizA7Nmzcfv2bTg4OKBPnz5ITEwssr9Kik+a80rTn+Vx30oT79y5c7Fw4UKEhISgX79++PTpE548eYJ9+/Zh+/btuHbtGpKSkjBy5EiJGPJOO5emTkBAAPr27YtBgwbh5s2bcHNzw7Rp0zh9nYiIiIjKBZNv+mIdO3YM0dHR2Lt3L+rVq4fq1atj3bp1AIAzZ84AyBnRjI+Px86dO2FlZQUHBwexTGkcPXoUT548wYkTJ2Bvb48aNWpg4sSJcHd3F8vo6OhAV1cXhoaGaN++PebMmYMDBw6Ix7W1tQEAWlpa0NXVhaamZoF2Dh8+jOjoaOzbtw+1a9eGnZ0dduzYAS8vLzx69Eii7KpVq9C2bVtUrVoVy5cvR3x8PAICAoq8hpLik+a85dGf0ty30sS7bNkyODs7w8DAAMrKygAAOTk5eHh4wMbGBpaWlpg1axauX78u8QFLfiXV+e2339C2bVvMnDkT1apVw8CBA/HTTz8Ve63p6elITEyU+CEiIiIiKgyf+aYv1p07d/D+/XtUqlQJgiBAEAQAQGJiIp4/fw4gZ0p5/fr1oaGhIdZr0aJFqdu6e/curK2tUalSpSLLHD9+HL///jueP3+O5ORkZGRkSLQrjcDAQFhbW0NHR0fcZ2dnBxUVFQQFBUlMla5bt674bzU1NairqyM2Nvaz4yvuvOXRn9Lct9LE27BhwwL7DA0NJaaDGxoaIiMjA0lJSRJ9W5o6jx49QseOHSXq2NvbF3utK1aswJIlS4otQ0REREQEMPmmL1hGRgYaNmyIS5cuFTimqqoKIGdKuYKCgsQxRcWSX9a5CWGu7OxsKCkpFVne29sbbm5u2LhxI9q1awddXV0cOXIEM2fOlOZSRJmZmQXik5OTg4KCQoHVwuXlC05MyR93WeIr7rxl7c+8pLlvpYlXRUVFqmvIex2FKalOZmZmqa99zpw5mDp1qridmJgIMzOzYusQERER0feJ087pi9WgQQMEBwcjIyMDurq6Ej+5SVytWrUQHByMT58+ifVyn73OlTsSmpCQIO7Lv/hX/fr18ejRI4mFz/K6fv06mjRpgpEjR8Lc3Bx6enoIDQ2VKJObqBU39bl27doIDg5GWlqauC8kJAQpKSmoXbt2kfVKIk180pCmP0sizX0rr3jLU61ateDv7y+xr6TvLldRUYG2trbEDxERERFRYZh80xdr0KBBqFSpEtzc3PDy5UsIgoCwsDBMnz5dXD184MCByM7Oxpw5c5CamoqoqKgCC4RVq1YNlStXxsaNG5GVlYV3795hypQpEmUGDhwIHR0dDB8+HG/fvkVGRgZOnTqFv/76CwBgbm6OoKAgBAcHIzs7G2fPnsWWLVskzpH7nHfelc3zGzhwIFRUVDBt2jSkpKTg/fv3GD9+PBwdHdG0adMy95U08UlDmv4siTT3rbziLU9jx47F6dOncfToUWRnZ+POnTvYtGlThcZERERERN8OJt/0xdLU1MT169ehqamJOnXqQE1NDR06dEDlypXFUWItLS2cOHECp06dgra2Npo1a4bBgwdLnEdBQQH79++Hl5cX1NXVYWdnhx49ekiU0dLSwtWrV/Hx40dUrVoV+vr62L17N9q2bQsA+OGHHzBgwADY2tpCTU0Ns2bNklgpO9fKlSsxadIkaGlpoUOHDoVe05kzZ+Dr6wsdHR1UqVIFqqqqOHjw4Gf1lbTxlUSa/iyJNPetvOItT46Ojli3bh1Gjx4NdXV1jBgxAsOHDxcXeSMiIiIi+hxyQnEPSRJ9IbKzs5GRkVHo87+50tPTxeMfPnyAlpZWgWd4MzIyxGe7iyqTlZUFQRAKfd43bxwZGRlISUkpdIGv3K/U0tDQQEJCQqHtpKenQ0FBoUA7uV9Ppq2tLfGcckJCAtTV1Yt9Nr24+Ep7Xmn6M+95Czte0n0rS7wZGRlITU2VmOKdlZUlLpwmJydX4FzS1Ml7PRkZGVBWVsbq1auxa9cuhISEFNnneSUmJkJHRwdmkw9DXkVdqjpEROUlfGW3ig6BiOibk/v3Xe7fpZ+DI9/0VZCXly828QYkF+bS1dUtNFHMm2AWVaawhLiwOJSUlIpcWVtTUxOampqQk5Mrsh0VFZVC28mtk3+BMB0dnWIT75LiK+15penPvOct7HhJ960s8SopKRX4xaegoABdXV0xic5/LmnqADkLqEVHR0NRURHe3t5Ys2YNBg0aVGT8RERERETS4mrnRET/n5WVFVq2bIno6Gjo6enB3d0ds2fPLvV5gpZ04uJrRERERCSB086JiPLJ+3hCaZTntCQiIiIiqnicdk5EJENlSbyJiIiIiIrD5JuIiIiIiIhIxph8ExEREREREckYk28iIiIiIiIiGWPyTURERERERCRjTL6JiIiIiIiIZIzJNxEREREREZGMMfkmIiIiIiIikjEm30REREREREQyxuSbiIiIiIiISMaYfBMRERERERHJGJNvIiIiIiIiIhlj8k1EREREREQkY0y+iYiIiIiIiGRMsaIDICL61tgsugB5FfWKDoOIvjPhK7tVdAhERFQMjnwTERERERERyRiT769MZmYmPDw8EBMTU9GhlKg8Y/Xw8MCbN2+KLXPo0CFERUV9dlvl5dWrVzh06FBFhyEzb968wV9//VXRYRTqW+97IiIiIvr6MPn+yqSlpWHIkCF49uxZRYdSovKMdciQIQgKCiq2zMiRI+Hn5/fZbZVFREQEDh8+LLHv3r17GDlyZIXE818ICgrC0KFDKzqMQn3rfU9EREREXx8m30Tl4O7duxgzZozEPjMzM/zwww8VFNH3jX1PRERERF8aLrj2FQsLC0NgYCCMjY1hb29f4Pi7d+9w69YtyMvLw8HBAQYGBuKxjx8/4sSJE3BxcYGmpqa439PTE61bt0aVKlUAAFlZWbh16xZiYmJQv359WFpaSrSRnJwMb29vfPr0CQ0bNkT16tXLPdaivHjxAn5+fqhevToaN25cbNnMzEwcPHgQnTt3xrt37/Do0SPUr18fderUAQAEBgbi8ePHMDIygoODA5SUlMS6jx8/xrNnz9C9e3dxX3R0NK5evYpBgwbh7du3+Pfff5GRkQEPDw8AgI2NDSpVqoQOHTpI9MGDBw/g4uKChw8fIjIyEg0aNEDNmjULxBsREYF79+5BR0cHTZs2hY6OjsTxuLg43L17FwBgZ2dXZH/FxMTg/PnzAABVVVVYWVmhYcOGEmVKE9fNmzfF10JJcs/bq1cvPHz4EC9fvoStrS3Mzc2RkZGBmzdvIiEhAc2bN4eJiUmB+kXdk+TkZJw8eRLdunWDnp6eWD4xMRGnTp1C9+7dy9z3giDg5s2biI2NRf369aGuro4rV65g0KBBJV4vEREREVFxmHx/pZYuXYrw8HDUrl0b//77L9q1aycx7Xnv3r34+eef0bx5c2RmZmLQoEHYt28f+vTpAwB4//49hgwZgqdPn0ok1MOHD8fRo0dRpUoVxMbGwtHREQBQr149BAcHo127dti4cSMA4MKFC3Bzc4O1tTV0dXVx48YNTJo0CYsXLy7XWAuzZcsWTJo0Ca1bt0ZiYiLU1NSQmZlZZPncKfCdOnVCWFgYGjduDG1tbdSuXRuDBw/G2bNn4ejoiJCQECgqKuLixYuoVq0aAOD8+fPYsWOHRPLt7++PoUOHYtCgQYiLi4Ofnx8yMjLERFdJSQnh4eEYOXIkXF1dAeQkruPGjcPvv/8OeXl5qKqq4urVq9i+fbvE9O3Zs2dj27ZtcHBwQFJSEkJCQuDp6Yl27doBAC5evIj+/fvD1tYWmpqaCA4Oxpo1a9C7d+8C152QkCDGlJqaips3b8LW1hanTp2CoqKi1HFlZWWhd+/euHnzJhwcHBAQEIBatWoV2d95z2tubg4DAwNkZ2fD29sb69evx44dO2BkZIS0tDQEBATg6tWrsLW1BZCTABd3T9TV1TFnzhzEx8djwoQJYnsHDx7EvHnz4OrqimvXrpW67zMzM9GjRw/4+vqiZcuWCAwMhJWVFZNvIiIiIioXTL6/UmpqaggMDISCggJCQkJQr149PHjwALa2tnjz5g3GjRuHdevW4aeffgIA/PLLLxgzZgzatWsHXV1dqdo4ePAg5OTkEBAQAAUFBQDA6dOnAeQk7/3798dff/0lJqXPnz9Hw4YN0bFjR7Rs2VJmsUZHR2Pq1KnYsWMHhgwZAgCYNGkSrl+/XuI1ycnJ4dGjR2LiuX//fnh5eeHhw4ewtLREeno62rdvjylTpuDYsWNS9VPdunUxceJEBAQEiCPfAHDy5MkCZRMTEzFixAgMGzYMALBy5UrMnz9fTAAPHToEDw8PBAcHi6PBW7ZswdChQxEWFgYlJSWsXr0aP//8M1auXAkgZxZD7ih4fhYWFhIxJScno3HjxtizZw9GjRoldVz79+/HjRs3EBgYCDMzMyQnJxc6g6Gw6502bZp4n4YOHYoJEybgyJEj6NevHwCgT58+WL16NTw9PQHkLK5X3D2Rl5fHwIEDceDAAYnk+8CBA3B1dZWYtVCavt+7dy98fHwQEBAAMzMzfPz4UeJ1XJj09HSkp6dLtEFEREREVBg+8/2VGjVqlJgQ161bF4aGhnjy5AkA4MyZM1BRUcHo0aPF8lOnTkVSUhIuX74sdRvq6uqIj4/H8+fPxX09evQAABw/fhyKiopITk6Gp6cnPD09cefOHVSpUgXXrl2Taaxnz56FlpYWBg8eLO6bMWOGVNc0ZswYMfEGcpLdfv36iaP/KioqmDp1Kry8vPDp0yepzlkaqqqq+PHHH8VtR0dHREZGIjU1FQCwZ88e1KtXD9euXYOnp6e4mnhUVBSePn0KIOe+PHnyREz0NDQ00LZt2yLbzMzMxK1bt3DkyBGcPHkSVapUKZCslxTX0aNHMWDAAJiZmQEANDU1xQ9LSrrevPfJ3t4eOjo6YuKduy/32gDp7ombmxvu3LkjLuYXERGBf//9V6KtwmIp7hqPHz+O/v37i9eooaFR4Dn+/FasWAEdHR3xJ7cuEREREVF+HPn+SuV91hXISVDS0tIA5CQi1apVExNeIGf02dTUFC9fvpS6DTc3N9y9exdNmjRBtWrV4OzsjPHjx8PKygrh4eGQk5PDmTNnJOrY2dmJ07VlFWtuHTk5OXFf1apVJZLqouR/tjgiIgJ2dnYS+2rWrImsrCy8evWq0OeeP4eOjo5E3CoqKgByRlDV1NQQHh4OFRWVAv3q5uYGefmcz8rWrFmDMWPGwMTEBHZ2dujevTvGjh0LDQ2NAu09ffoUHTp0gLKyMurVqwdNTU1ER0cXuCclxRUREYEWLVpI1KlRo0apr1dZWbnAbAZlZWXx9QBId08aNmwIGxsbHDhwAIsWLcJff/0FCwuLYkfjpbnG/CPdRa1hkGvOnDmYOnWquJ2YmMgEnIiIiIgKxeT7G2RoaIi4uLgC++Pi4mBkZAQAYrKbnZ0tHs/MzJR4blpZWRmbN2/Gb7/9Bl9fX2zduhVNmjTB06dPoaOjA3l5eYkpzbKKNT8DAwPEx8dL7EtKSir2me9ceZOvotrP3c7bV3n7CYA4WlredHR0UL9+fWzfvr3IMlZWVrh69Sri4uJw9epVLF26FJcuXcKFCxcKlF26dCkaN26MEydOiPtcXV0lpkpLo7A+L+y+lQdp7gmQ84HErl27sGjRIhw4cABubm6f3W7+a8y/nZ+KioqYxBMRERERFYfTzr9BudNp804t/vvvv5GSkiKOXpqYmEBBQUGc/g0A169fR1ZWlrgdGRkJICfBaNWqFTZt2oSkpCQ8efIEXbp0QUxMDI4cOSLRdkpKCmJjY8s11sLqvHjxAv7+/uK+/HFIq3Xr1gWmmB86dAgNGzaElpYWAKBKlSp49eqVRMJ95coVifNoampKjN6WVZcuXXD06FG8e/dOYn9ERESBf+vr66Nv376YNm0a7t27V+j53r17J7GgXnx8fKkePcjl6OiIU6dOSXzAcfTo0VKfRxrS3BMgJ/l+9uwZduzYgaCgoGKnnEvDwcEBp0+flngPFPbcPhERERFRWXDk+xvUuHFjjBw5Ej179sTUqVORmZmJ1atXY+rUqWIipqSkhEGDBmHixIkIDw/Hhw8fcODAAXFqMwCcOHECBw4cQM+ePWFsbIwTJ07A0tIStra20NDQwIIFCzB48GB4e3ujXr16ePHiBU6cOIHDhw9L9VVh0saan62tLX744Qd069YNU6ZMQWJiInbu3CnVtPP8pk2bhv3796Ndu3YYOHAgHj58iH379kmMInfs2BGampro168fevXqhXv37hWYFt6wYUMIgoCZM2eiQYMGsLGxKXUsADB9+nRcuHABdnZ2cHd3h46ODu7duwd/f388ePAAQM4z9Pr6+nBwcEBWVhY2bNggruqdX+/evTFjxgzo6upCU1MTW7duhSAIpY5rypQp2LVrFzp27Ih+/frB29sbd+7cKdM1lkSaewLkfJe3o6MjJk+ejObNmxf5epFW7jV26tQJffv2xa1bt+Dt7V1gtgQRERERUVlw5Psro6SkBDc3twJTsvv06QMLCwtxe+vWrfj999/x4sULvHr1Crt378b//vc/iTo7d+7E1KlTxZXIr169ih9//BFVq1YFAEycOBGbNm1CSkoK7t27h44dO+Lu3bvis8VLly7FlStXoKKignv37sHIyAg3btxAo0aNyj1WNzc3mJqaitv79u3D/Pnzxa+hunXrlkTs0vablpYW7t+/j169euH+/fvQ19fH/fv34eTkJJbR1NTE3bt3YWtri4cPH6J58+Y4c+aMxDRnU1NTXL58GRkZGbhw4QJCQ0NhZmaGH374QSxTs2ZNiYXGgJzp3G5ublBWVgaQs8jX9evX8euvv+LVq1d4/Pgx2rZtK5Honj9/Hn369MGTJ08QFhaGVatWYcOGDYVe908//YQ9e/bg1atXePr0KVauXIn169dLXJ80cenp6Ul8BZejoyPOnz9f7FTvws5raWlZ4CvR6tatKy7kB0h3T3LNnDkTLi4umD17tsT+svS9gYEBfH19YW9vj4CAALRs2RJr1qyBtrZ2kddIRERERCQtOaEsw2BERN+g2NhYiVkbQ4cOxZs3b3Dx4kWp6icmJuasej75MORV1GUVJhFRocJXdqvoEIiIvjm5f98lJCR89qAMp50TEf1/vXv3RqtWrVCtWjVcu3YNZ86cwd9//13RYRERERHRN4Aj30RE/19sbCx2796Np0+fwszMDG5ubjA3N5e6fnl+MkpEREREFa88/75j8k1EVE6YfBMRERF9W8rz7zsuuEZEREREREQkY0y+iYiIiIiIiGSMyTcRERERERGRjDH5JiIiIiIiIpIxJt9EREREREREMsbkm4iIiIiIiEjGmHwTERERERERyRiTbyIiIiIiIiIZY/JNREREREREJGNMvomIiIiIiIhkjMk3ERERERERkYwx+SYiIiIiIiKSMSbfRERERERERDLG5JuIiIiIiIhIxhQrOgAiom+NzaILkFdRr+gwiOgrFL6yW0WHQEREMsKRbyIiIiIiIiIZ+66S7+nTp2P16tXi9rhx47Bx48ZS1Slp+0u3cOFCLF++vKLD+CL9V/fya3vNFCUhIQFTpkyBo6MjunXjSA0RERERUXG+q2nnz549Q1pamrgdGhoKLS2tUtUpaXvKlCkwMzPD1KlTyzHy8vPixQuoqqpWdBhfpPz3sjwU9nqQRTsVYcmSJbh58ybWrVsHbW3tig6HiIiIiOiL9l0l32vXroWSklK51sl//OnTpxAEocwxUsUpy+ujJIW9HmTRTkV48OABunXrhlatWlV0KEREREREX7xSJd8rVqzAsWPHAAAGBgawt7fHzJkzoaGhIVHu+vXr2Lp1KyIjI1G/fn0sWLAApqamUh+/du0atmzZglevXqFmzZqYOHEimjZtKh5/+vQp1qxZg8ePH6Ny5coYPXo02rVrV+KxzZs3o1KlSpgxY4Z4rqysLKxevRpXr15FVlYWRo0ahf79+4vHC6uTV97jS5Ysgbe3N3x9feHt7Q0AGD58OM6cOYO///5bot6SJUsQExODDRs2lKmfx40bh1q1agEA/vnnH2RlZWHgwIEYOnSoxLm2bt2KQ4cOQUtLS+qpwSXdn0OHDsHDwwNxcXFo0KABZs+ejerVq0vEVrNmTXz69Ane3t749OkTRo4cCRcXF6xZswaXLl2Cjo4OpkyZAicnp8+u5+bmhi5dumDw4MHivnnz5kFNTQ3z58+Xur8Ku9fF9UVJ96mw18OxY8cKbUeaPpXmfuc1btw4WFlZISMjo8jXd+55MzIycO7cOdSqVQtbtmxBVlYWtm7dilOnTuHTp0+wtbXFnDlzYGBgAACwt7dHUFAQnj59itOnT2PIkCGYNGlSifVSU1Ph6OiIlStX4siRIwgKCsL48eMxcOBAvH37FmvWrMGdO3ego6ODrl27wt3dHfLyOU/H/P333/j999+xcOFCbN68GZGRkbC1tcWiRYugo6MjXtO7d++wbt06+Pj4QE9PD+7u7ujcubN4vKR2Pn36hN9++w1XrlwBALRv3x6TJ0/+Jj4wISIiIqKKU6pnvt3c3LBlyxZs2bIFEydOxLVr1+Di4iJRxsPDAx07doSFhQWWLFmC+vXrw93dXerjBw8exIABA9C6dWv8+uuvaNiwIdq0aYPbt28DyPnD2MnJCYIgYNmyZejbty9WrVqFoKCgYo8BOdN9X758KRHvhg0bcP/+fUydOhUdO3bE4MGDxYSqqDp55T3+448/on79+mjfvr3YT927d8fly5dx48YNsU5aWhp+++03NGnSpMz9HBoailmzZuHp06eYNm0aevTogZEjR+LcuXNimd9//x0zZ87EwIEDMXbsWBw5cgRHjx4t8lqAku/Pli1bMGLECHTq1AmLFy/GmzdvYG9vjw8fPkjENmfOHLx9+xYzZsxAq1atMHDgQLRs2RIJCQlYvHgx6tWrh65duyI6Ovqz6z169Ahv3ryRuI7nz58jPDy8VP2V/16X1Bcl3afCXg+VKlUq0I60fVpS/PmFhoZi9uzZxb6+c88bGhqKhQsXYsqUKQCAIUOGYO/evXB3d8eCBQvw9u1bNG/eHKmpqQCATZs2wdzcHL169cKWLVvQt29fqeplZWXh/v37cHV1Rb169bB27Vq0b98e79+/R7NmzZCYmIhFixZh5MiR+PPPPzFhwgQx1tjYWFy+fBkTJ05E3759MXv2bPzzzz8YPXq0WObt27do2rSpeM3Dhg3DH3/8gcDAQACQqp158+Zh7969+PnnnzFjxgzEx8dj8eLFRfYzEREREZE05ITPmCMdGxsLIyMjhIaGwsrKCpmZmahcuTLGjRuHRYsWieXS09OhoqJS4vGsrCxUrVoV69atw8CBA8XjU6ZMwYsXL+Dl5YVnz57BysoK7969g5GRkcQ5IiMjizymoqICFxcXVK1aVVxkzdnZGeHh4Xjy5Ik46jVz5kycO3dOTNjz1ylpu3v37rC0tMRvv/0mtt+nTx9oa2tjz549AABPT0+4u7sjOjq6wKwBafo5N/bMzExcu3ZNLOfi4gITExNs2bIFmZmZMDU1xZIlSzB27FgAQFJSEszMzNCvXz/s2LGjQDvS3J9KlSphzpw5mDZtmljHwsICI0aMEOs4OzsjKysLV69eFc9hYWGB2rVri8miIAgwNTXFmjVrxBHrstZr1KgRBg8ejOnTp4v1fvjhB2hqaorXWVJ/5W7n3suS+kLa+1TY6yFvO6Xp05Liz0+a17ezszMSEhLg6+sr1rt58yacnZ3x+vVr6OnpAQCys7NRq1YtzJs3D8OHDweQM/rdvXt3cXaBNPWSk5OhpaWF5cuXY968eWKbM2bMQEBAAC5cuCDu8/f3R+PGjRETEwN9fX14eHhgyJAhCAsLQ40aNQAAJ0+exKBBg/Dx40fIyclhypQpOHPmDEJCQqCo+H8Tez59+gRlZWWp2mnVqhV69eolMTOhqPuenp6O9PR0cTsxMRFmZmYwm3yYXzVGRGXCrxojIvqyJCYmQkdHBwkJCZ+9zlGppp3Hx8djw4YN8PHxQUxMDLKzsyEnJ4cXL17AysoKISEheP/+fYFR2tw/Wks6/vjxY7x58warVq3C77//DkEQIAgC3r59Ky4SVr16dVhZWaF///5wd3eHk5MTTE1NoaKiUuyxorRt21ZMTACgQ4cOWLNmDdLS0sptYbLRo0ejf//+2LBhA7S0tLBr1y4MGDCgyMS7pH7OlXcqPgBUqVIFr1+/BgCEh4cjJiYG7du3F49raWmhefPmRcZZ0v0JDw9HbGwsOnXqJB5TVFRE+/btcf/+fYk6+Uf1TU1NYWtrK27LycnBxMQEb9++LZd60iiuv/IrqS8A6e9TcUrTp6WJP5c0r+8WLVpI1MlN8Lt06SI+ry4IAt6/f4/Hjx8X2VZp6hXW5uvXr2Fvby++77OysiAIAkJDQ8XyBgYGYuKd2wepqaniL8UbN26ge/fuEok3ACgrK0vdjrOzM9asWYPs7Gx07NgRjRo1KvJ3yIoVK7BkyZIi+4SIiIiIKFepku9u3bpBWVkZEyZMQOXKlaGkpISWLVuKU0pz/6upqVlo/ZKOf/z4EUDO12FVq1ZN4ljuH79KSkq4c+cOdu3ahb1792L06NFo0aIF/vrrLxgZGRV7rDDq6pKjUxoaGhAEAcnJyeWWfHfq1An6+vo4dOgQOnbsiCtXrojPABempH7OlT/BkJOTE5Oe5ORkAIVfX1FKuj9JSUmFnkNTUxNhYWHFxlbUvvwTL8paTxrF9Vd+JfUFIP19Ks7n9Glx8eeS5vWdv8zHjx9hampa6NfwVapUqci2SlOvsDY7deokztLIq3bt2uK/C+sD4P9eD6mpqcXeM2naWbx4MRo3bowTJ05g06ZNyM7OxrZt29C1a9cCdebMmSOxkn3uyDcRERERUX5SJ99v377F7du38fjxY/GP1LCwMGRkZIhlLC0tIS8vD39/f1hYWBQ4R0nHLSwsIC8vj5SUlAKjfHnp6elh2rRpmDZtGj58+IBmzZrh999/x/Lly4s9Vpj8I3IhISHQ1taGoaGhVP2Sn7y8fIGESF5eHsOHD8euXbsQFRWFWrVqFRj5yyVNP0vD3NwccnJyePz4sUQy8PjxY7Rs2bLQOiXdn5o1a0JOTg7BwcEwNzcX9z969KjQ8v8VDQ0NpKSkSOyLjo6WevS5MCX1hbT3qbDXQ16y7tOyvL6trKzExQ719fWlbqus9XLrRkREFPu+l0atWrXg7+//2e306tULvXr1giAImD17NsaMGYNXr14VKKeiolLszBoiIiIiolxSL7impaUFJSUl8dnQtLS0At9lra+vD1dXVyxYsAAREREAckaiNm3aJNVxAwMDDBo0CPPnz5dIGnx9fbFv3z4AOUnJ9u3bkZmZCSBnOqm8vDxUVVWLPVaUS5cu4dKlSwByFmNas2YNRowYIW23FGBiYoLIyMgC+0eOHIk7d+5gw4YNxZ5fmn6Who6ODvr06YNly5aJMwp2796NkJCQIuuUdH+0tbUxYMAALFu2DImJiQCAs2fP4tq1axg1alSpYywvDRs2xJkzZ8Tvzr5w4UKxMwukUVJfSHufino95JJ1n5bl9d2vXz8YGRlhzJgx4msnKysLe/fuhZ+fX7nXA4AJEybgypUrEs+vx8XFYeHChVJdZ66ff/4ZZ8+exZEjR8R9Xl5eePr0qdTt/PLLL+KCfnJyclBWVi63WTBERERE9P2SOvlWV1fHunXrMGrUKNStWxcmJiZQV1eHmpqaRLmtW7fCxsYGVlZWqF27NszMzCSmipZ0fMuWLWjXrh0aNWqEWrVqwdjYGBMnTkTdunUB5Dzj6evrCwMDA9jY2KBy5cqwsLDApEmTij1WlM6dO+Pnn3+GlZUVqlevDhMTE4kFtkprxIgRuHr1KurUqYOmTZuKq1pXq1YNHTp0QEJCQrFfDyVtP0vj999/R2JiIkxNTVGzZk388ccfJX4nc0n3548//oCqqioqV64MS0tLuLq6Yu3atbC3ty91fOVl7ty5+PjxI8zMzGBlZYUlS5YUObOgNIrrC2nvU1Gvh7xk2adleX1raWnh4sWLiIiIgLGxMerWrQs9PT1cv35d4uvPyqsekLPw2969e7F48WIYGxvDysoKdevWLfUIeqdOnbB582aMGTMGVapUQeXKlbF3715x2rs07VStWhXNmjWDhYUFatSogX379mH79u2lioOIiIiIKL9Sr3aekJCAiIgImJqawtDQEH5+fjA3N4eurq5EudjYWLx58waWlpaFTsss6XhSUhLCw8NRpUqVQv8AT01NRVhYGIyNjQtMoS3q2PPnz6GkpCQ+T/7kyRNoaGigSpUqiIiIQGZmJmrWrClxrvx1StrObT88PBwfP36EjY2NOGqWuyqzl5dXkf2bq6R+zht7roiICGRkZEhMVxYEAU+ePIGWlhYqV66MsLAwyMvLl5gMlXR/Xr16hfj4eFhYWBR4frew2EJDQ8UYcgUHB8PAwEBMjMpaD8hZVTssLAwaGhowMTHBixcvoKCgIF6nNP1V2L0sqS+keT/kfz1ERUUV2k5p+7Sw+52Xs7MzmjZtipUrVxb5+i7svHm9fv1ajCn/6O/jx4+ho6Mj8R3wJdXLzs7GgwcPULdu3ULXH8jOzsazZ8+goKAAc3NzicXi4uLiEBkZiYYNG4r7UlJSEBwcjMaNG0NBQUHcn5GRgSdPnsDY2LjQ9R6KawfIed+8ePEC8vLyqFatmsS5i5O78BtXOyeisuJq50REX5byXO38s75qjKQXERGBWrVq4cyZM3B2dq7ocOg7kDf5pv8Gk28i+lxMvomIvizlmXxLPe2cyq5r165o2LAhXFxcmHgTERERERF9hzjy/R8ICAiAmpraZ62+TVRaJU0pp/JXnp+MEhEREVHFK8+/70r1Pd9UNg0aNKjoEOg7VKtWrYoOgYiIiIiI/j9OOyciIiIiIiKSMSbfRERERERERDLG5JuIiIiIiIhIxph8ExEREREREckYF1wjIionuV8ekZiYWMGREBEREVF5yP27rjy+JIzJNxFROYmNjQUAmJmZVXAkRERERFSekpKSoKOj81nnYPJNRFRO9PX1AQARERGf/cuZvg6JiYkwMzNDZGQkv9v9O8L7/v3hPf/+8J5/nwq774IgICkpCZUrV/7s8zP5JiIqJ/LyOcto6Ojo8H/U3xltbW3e8+8Q7/v3h/f8+8N7/n3Kf9/La1CFC64RERERERERyRiTbyIiIiIiIiIZY/JNRFROVFRUsGjRIqioqFR0KPQf4T3/PvG+f394z78/vOffJ1nfdzmhPNZMJyIiIiIiIqIiceSbiIiIiIiISMaYfBMRERERERHJGJNvIiIiIiIiIhnj93wTEZUDQRAQHByMzMxM2NjYQEFBoaJDonKSmZkJHx+fAvvr1KkDQ0NDiX2fPn1CUFAQNDQ0ULt27f8qRCon6enpuH//PkxMTFCzZs1CyyQmJuLJkycwNjZGtWrVylyGvhz+/v7IyMhA06ZNCxy7desWsrOzJfaZm5ujSpUqEvuysrIQFBQEBQUF1KtXD/LyHN/6UmVmZiI0NBSqqqqoUaNGkf+/Dg8PR0xMDOrUqQNNTc0yl6EvQ0REBD58+ABLS0uoq6tLHIuJicHjx48L1LG3t4eiomS6HB8fj2fPnqFKlSqoXLly6QMRiIjos4SEhAi1atUSKlWqJFStWlWoWrWqcOfOnYoOi8rJ+/fvBQBCo0aNBAcHB/Hn0qVLEuUuXLggGBoaCubm5oK+vr7QuHFj4dWrVxUUNZVGbGysMGPGDKFy5cqCurq6MG7cuELLbdq0SVBTUxPq1KkjqKurC7169RJSUlJKXYa+DJs3bxasra0FPT09oXr16oWWUVFREerVqyfx3vf09JQo4+vrK1SrVk2oWrWqUKlSJcHS0lJ49OjRf3AFVBqfPn0S5s+fLxgaGgrW1tZC1apVBXNzc+Hy5csS5ZKSkoROnToJGhoaQu3atQUNDQ1h9+7dpS5DX4Zjx44JdevWFczMzAQbGxtBQ0NDWL58uUQZT09PQVFRUeJ97uDgIHz48EGi3C+//CL+TlBVVRWGDBkiZGRklCoeJt9ERJ8hOztbaNCggeDi4iJkZWUJgiAII0eOFMzMzIS0tLQKjo7KQ27y7efnV2SZ2NhYQUdHR1i4cKEgCIKQlpYmODg4CB06dPiPoqTPERAQIKxcuVJ49+6d0Lx580KTb19fX0FOTk44fvy4IAiCEB0dLVStWlWYMWNGqcrQl2PatGlCQECAsGzZsmKT79OnTxd5jvT0dKF69erCyJEjBUEQhKysLKF3796CtbW1kJ2dLYuwqYwSEhKEZcuWCQkJCYIg5NyrKVOmCDo6OuI+QRCEn376SbCyshJiY2MFQRCE3bt3CwoKCkJwcHCpytCXYd26dUJISIi4/c8//wgKCgqCl5eXuM/T01MwMDAo9jznz58XFBQUhKtXrwqCIAjPnz8X9PX1hVWrVpUqHibfRESf4e7duwIA4d69e+K+8PBwAUCxf7DR1yM3+T516pRw//79Ap+EC4IgbNu2TVBVVRWSkpLEfSdPnhQACBEREf9luPSZikq+x44dK9jY2EjsW7x4sWBoaCgmWdKUoS9PScn39u3bBV9fX+H9+/cFjp87d04AIISFhYn77t27JwAQbt++LaOIqbw8fvxYACDcvHlTEISc0XFNTU3ht99+kyhXrVo1YdasWVKXoS9bnTp1JD4U9fT0FPT19YVHjx4JgYGBhQ6eDBgwQHBycpLYN378eKF27dqlapsPpBARfQY/Pz/Iy8ujcePG4r7q1avD2NgYfn5+FRgZlbdRo0bhxx9/hJGREQYPHozExETxmJ+fH2rXri3xzF+zZs0AAA8fPvyvQyUZ8PPzQ5MmTST2NWvWDDExMXj16pXUZejrM3PmTIwcORJmZmbo1q0b3rx5Ix7z8/ODgYEBatSoIe6ztbWFgoIC/x/wFfD19YWcnJy4xsPTp0+RnJxc4H3ctGlT8X5KU4a+XO/fv8fLly9haWkpsT8uLg69evVCr169oK+vj1WrVkkcL+r3+5MnT5CSkiJ1+0y+iYg+Q1xcHHR1dQssrmNgYIC4uLgKiorKk7KyMjw9PfH27VsEBgYiKCgI169fx5QpU8QycXFxMDAwkKiXu83XwbdBmnvM18G3Z+PGjYiJiYG/vz+ePXuGyMhI/Pjjj+Lxwu65nJwc9PX1ec+/cK9evcL06dMxevRomJiYAPi/92lh7+O87/OSytCXKTs7GyNHjoSpqSnc3NzE/ebm5nj48CGePn2K58+f46+//sLcuXPh6ekplinq97sgCIiPj5c6BibfRESfQUlJCWlpaQX2p6amQllZuQIiovKmra2NH374QdyuVasWpk2bhkOHDkEQBACFvw5SU1MBgK+Db4Q095ivg2/PqFGjxA9Xq1SpgoULF+LixYtiksX/B3yd3r9/j06dOsHGxga///67uF9JSQkACn0f532fl1SGvjyCIOCnn36Cj48PTp8+DQ0NDfFY8+bN0bBhQ3G7V69e6NKlCw4ePCjuK6/f70y+iYg+Q/Xq1ZGSkoIPHz6I+zIzM/Hu3Tt+xdA3rFKlSvj48SMSEhIA5LwOoqKiJMrkbvN18G0o6h7LycnBzMxM6jL0datUqRIA4PXr1wBy7vm7d++QkZEhlklMTERycjLf+1+omJgYtGvXDpUqVcKpU6egqqoqHqtevToAFPo+zr2f0pShL4sgCBg7dixOnDiBy5cvo169eiXWqVSpksQ9Lur3u7q6eoER8eIw+SYi+gxOTk5QUlLCqVOnxH3//PMPUlJS4OzsXIGRUXn5+PFjgX0XL15E1apVoaurCwDo0KEDXr58iYCAALGMl5cXdHV1YWdn91+FSjLUoUMHXLlyReL14OXlBXt7e/FZf2nK0NejqPe+mpqa+Iyws7Mz0tPT8c8//4hlvLy8oKioiLZt2/5nsZJ0chNvIyMjnDlzpsD3PZuYmMDGxkbi/+mxsbG4efMmOnToIHUZ+rKMHz8eR48exeXLl1G/fv0Cx/O/19PT03Hjxg3Y2NiI+zp06IDz589LfNDm5eWF9u3bF3j0sDiKJRchIqKiGBoaYvr06Zg8eTIyMzOhoqKCWbNmYcSIEahdu3ZFh0flYNOmTQgICEC3bt2gra2N06dPw8PDA/v27RPLtGnTBl26dIGrqyuWLl2Kt2/fYunSpfjf//7HaYhfgezsbNy6dQsAkJSUhOjoaHh7e0NDQ0NcTHHUqFHYtGkTevXqhQkTJuDOnTs4fvw4Ll68KJ5HmjL05QgMDERCQgIiIiKQnp4Ob29vAICdnR1UVFRw8uRJHDx4EP3794exsTGuXLmC3377DStWrBCTNktLS4waNQqjR4/GqlWrkJGRgenTp2Pq1KkwNjauyMujfFJSUtChQwd8+PABa9aswYMHD8RjdevWFUcvV6xYARcXF1StWhUNGjTAmjVrULt2bQwePFgsL00Z+jLMnDkTW7duxYYNG5CYmCi+z42NjVGrVi0AwODBg1GnTh20bNkSaWlp2LRpE+Lj4zFv3jzxPJMmTcKuXbvQv39/jBw5EhcvXsStW7dw8+bNUsUjJ+Q+sEZERGUiCAJ27tyJkydPIjMzE507d8b48eOhqMjPN78VJ06cwPHjxxETEwNLS0u4u7tLfCIO5Dz7tW7dOly/fh3q6uoYOHAgXF1dKyhiKo3U1NRCR6xq1qwp8SHLu3fvsHLlSvj7+8PY2Bhjx46Fo6OjRB1pytCXYcKECYWuTn306FFxAa4rV67Aw8MDUVFRMDc3x48//ogWLVpIlM/MzMTmzZtx7tw5yMvLo1evXhLPitOXISoqqsjfycuXL4eTk5O4ffnyZWzduhWxsbFo0qQJZs+eDX19fYk60pShijd06FC8ePGiwP5OnTphwYIFAHKe39+6dSuuXbsGAGjQoAEmTpxYYDp5REQEVq1ahcePH6NKlSqYNGlSgRXQS8Lkm4iIiIiIiEjG+JEcERERERERkYwx+SYiIiIiIiKSMSbfRERERERERDLG5JuIiIiIiIhIxph8ExEREREREckYk28iIiIiIiIiGWPyTURERERERCRjTL6JiIi+c/Hx8Th48CAyMjIqOpSvRkhICC5evPjZZcqjzn8hMDAQV65cqegwZOZbvz5ZuHTpEoKDg2XaxrVr1xAYGCjTNoj+S4oVHQAREREVLzY2Fv/880+xZZo0aQIrK6synf/58+cYOHAg4uPjoaurW6ZzfG+8vLxw9OhRdOzYEQAQHByM169fw9nZucgyZTlvWRQWy+c6dOgQvL290a5du3I755fkW78+WZg/fz66d++OevXqyayN5cuXw97eHvXr15dZG0T/JSbfREREX7j4+HicPHlS3L537x7i4uIkEjR9ff0yJ9/0+Y4fP44zZ85IJLz16tVDp06dvohYiMpbhw4dYG1tXdFhEH1VmHwTERF94SwtLXHw4EFx+6effsK9e/fEfRkZGTh27BgOHjwIJSUl1KhRA40aNYKCgoLEedLS0nDnzh2kpKSgcePGMDExKbLNsLAw3LlzB87OzjA0NJQ4Fh8fjwsXLqBv374ICgrCixcv4ODgIJ7v3r17iIiIQLVq1dC0aVOxXnJyMs6cOYNevXpBTU0NAHDu3Dmoq6vDyckJQM4o/6VLl9CvXz8oKCggOjoaDx8+hKamJpo0aQJ1dXWp+y0wMBDv379Hy5Yt4e/vj7dv36Jly5YwNDREamoqbt68iaysLLRs2RJaWlpivXv37iE9PR0ODg7ividPnuDFixfo3LlzgXaeP3+OwMBAxMbGivekefPmsLKygqqqaqHx+Pn5ISYmBg4ODtDX1y/xWp48eYLg4GCYmJigcePGUFFRKbRcUbEAwJ07dwAAmpqaqFevHmrWrFmgfmn629vbG7GxsejZsyfk5ORKfa+eP3+OoKAgmJmZwdbWVtyf+9orLtbcvmzVqhUePnyImJgYNG3aFMbGxsW2WVKMnz59KvJ8pYmrZcuWuHXrFuLi4tCvXz8AQFZWFu7evYt3796hVq1aqFu3brGxent749WrV5CTk4OxsTEaNWoEPT09iTLnz59HzZo1YWBgAD8/PygpKcHe3r7I10eukmLJPa+enp54XkdHRygq/l/q0KZNG1SuXFmiXu7vgxo1aqBBgwYSx3Lfc8nJybC1tUW1atUKxBUfHw9vb28YGhqiUaNGZYqd6EvG5JuIiOgrl5GRIY6Mf/r0CQ8ePICRkRHOnz8PAwMDAEBoaCjatm0LU1NTVK5cGUFBQZg6dSomTJhQ4Hx+fn7o3Lkz3N3d8cMPPxQ4njtNvVu3boiIiEDdunVhaWkJTU1N9OjRA8HBwWjSpAnu378Pa2trnD59GhoaGlBRUcHIkSOhr6+Pjh07IjU1Fb1794aBgQFev34NADh58iR+/fVXuLq6YseOHZgyZQrs7e2RnZ2NqKgoeHh4oGnTpmIi37p16wIJQK5Dhw7B09MTcnJysLCwwJs3bxAeHo4//vgDy5cvh4WFBV69eoXk5GTcuXMHlSpVAgDs2LFDTIxznTt3Djt27Cg0+Q4LC0NISIjEDAVTU1Pcvn1bYgr5oUOHcOjQISgqKsLExAQJCQl48eIFzpw5g1atWhV5b0eMGIGLFy+iWbNmiIyMRFpaGry8vFC7dm2pY8ntWwBITEyEt7c3RowYgd9++02sW1x/57dx40bMnz8fJ0+ehJycXKnq5l7TyZMnYW9vj/j4eBgZGeH06dNQVFREREREibHm9qWGhgaMjIyQnJyMR48e4e+//5a4b3mVFGNUVBSaNWtW5PlKE5eqqiqMjY1RpUoV9OvXDy9evECPHj0AABYWFrh37x4cHBzg6ekpkdDmdefOHfj6+gIAIiMj8ejRI+zfv188DwBMnz4dlSpVQnh4OOrVq4fAwEBoa2vj9u3b0NDQKPS80sQyffp0GBkZ4enTp2jQoAH8/f1hamqKS5cuiY+m5J12np2dDVdXV9y4cQP29vaIjIyEgYEBvLy8oK6ujvv376N79+4wMDCAiYkJbt26hTlz5mDBggViXLdv30bXrl1Ro0YN6Onp4fXr18jOzoa9vX2pYif6oglERET0VXF3dxeaNGlS5PGMjAyhXbt2wvTp08V948ePF3r06CFR5vTp04IgCIKvr68AQIiPjxeuXbsm6OrqCr///nuR588t/9NPP0nsnz9/vlCjRg3h/fv3giAIwrt374Rq1aoJCxYsEMs4OzsLc+bMEQRBEC5duiRYWloKurq6QmhoqCAIgjB48GBhxIgRgiAIQtWqVYVdu3aJdaOiogQfHx9BEATh6dOnAgDxGgozb948QV5eXrh586YgCIKQlZUl2NnZCYqKioKvr68gCIKQmZkpWFtbC0uWLBHrubu7C3379pU41/r16wVra2txe8WKFRL3YNmyZULz5s0l6uQvM2/ePAGAsH//fom26tatK2RlZRVaZ+nSpUKTJk2EpKQkcd/48eOFNm3aFHndhcWSX1hYmKClpSV4e3uL+4rr73nz5oltLlq0SDA2Nhbu378vVd38FixYIBgbGwvPnz8X950/f15IS0uTOtbce3vjxg1x3+DBg4WOHTsWec0lXV9pz1dUXACEv//+W6JskyZNhNmzZ4vbSUlJQp06dYT169cXef78du7cKZiYmAiZmZniPmtra8HS0lKIj48XBEEQkpOTBRMTE+HPP/8s8jzSxGJtbS3o6ekJERERgiAIwocPH4TatWsLM2fOFMs0b95cWLZsmSAIgnDv3j1BQUFBePfunXj86tWrQlxcnJCVlSU0aNBAGDJkiJCdnS0IQs79lpOTE9+HWVlZgo2NjeDu7i7W379/vwBAmDdvXqliJ/qScbVzIiKib0RAQABOnTqFo0ePwtTUFHfv3hWPqamp4c2bN3j//j0AQFFREd27d5eof+rUKfTo0QMbN27ExIkTS2wv/6j5wYMHMWbMGHGaupGREcaMGSMxZd7JyQlXr14FkLOScfv27dGqVStx3/Xr18Up6GpqaggODhZXYa9cubI4hVpLSwuurq6oUqVKsTHa2NigZcuWAAB5eXk0b94ctra24mingoIC7Ozs8OTJkxKvtzyYmprCzc1N3J41axZCQkKKXNF59+7daNCgAc6fP48jR47g8OHDMDIyws2bN/Hp06dStZ2amgpvb28cPXoUPj4+qFy5coHXSFH9DQCCIGD8+PHYs2cP/v33X4mp4iXVzWvv3r0YO3asxJTtTp06SUyVLilWALC2toajo6O47eTkhNDQ0CKvv6QYpTmfNHFZWVlJzJAICAjA/fv3Ub16dRw9ehRHjhzBuXPnYGlpKb7ui/LmzRtcunQJhw4dQlZWFt68eYOIiAiJMgMHDhRHozU0NNC0adMi+6E0sbi6usLMzAwAoKOjA3d3dxw+fLjQ86qpqSE7OxtBQUHiPicnJ+jp6SEkJAQBAQGYM2cO5OTkAOTc7yZNmojnCw4ORlBQEGbMmCHWd3Nzk3h/f04/En0pOD+DiIjoK5eQkICOHTvi5cuXsLW1hba2Np4+fYqUlBSxzMyZM/HkyRNUq1YNjRo1QqdOnTB+/HiJ57l/+uknuLq6SiSHxcmdzpwrIiKiwDOwFhYWePnypbjt5OSExYsXIzk5GdeuXcO4ceMQFRWFq1evwtnZGZGRkWLyvXPnTowdOxbbt2+Ho6Mj+vTpg6FDh0JBQQGVKlWSSOqLkv8ZWRUVlUL3JSUlSXXNn6t69epiApK7LS8vj5cvX6Jhw4YFyr98+RJPnjyRuJcA0LdvX6SmpkJZWVmqdm/cuIF+/frB2NgYNWvWhLq6Oj58+IB3796JZYrrbwDw9fXFjRs3cPLkSdSqVUvi/CXVzSsyMrJA/dLGCqDAs/IqKipIS0sr8rwlxVjS+aSNK//7Ijw8HABw9epViXuvoaFR7CKJS5cuxapVq9CkSROJZ8/fvXsHc3PzMvVDaWKpUaOGxLa5uTkiIyMhCIJEXSBnccFVq1ZhwIAB0NTURLt27TBixAg4ODiI7//ifjfkfqBQvXp18bicnJxEDGXtR6IvCZNvIiKir9ymTZuQkpKCiIgIMRlbvHixRHJqaGiIkydPIjExEf/++y/Wrl2LAwcOSIyQHTp0CIMGDUL9+vUxderUEtvN/we4gYEB4uLiJPbFxcVJJPjNmjWDsrIy/vnnH/j6+sLJyQlRUVFYvXo1rl27hpo1a4qjbY6OjggMDERERAQuXryI+fPnIzAwEOvWrSt9J5WCvLw8srOzJfYVl9SVRnx8vMR2QkICsrOzCyxql0tLSwt9+vSR6n4UZ/r06Rg+fDhWrVol7mvUqBEEQRC3S+rvZs2aoU2bNhg2bBguXbqEJk2aSF03Lx0dHcTGxn5WrGXxua8naePK/77Q1tYGAKxYsaLQRe4KExUVhUWLFuHWrVto0aIFgJzk89ixY5/VD6WJJf9rNT4+Hvr6+gWuL9eMGTMwdepU+Pv749ixY2jTpg0uX74svrbj4uIkPpiIi4sTk+bctSk+fPgg8V7IG0NZ+pHoS8Np50RERF+5N2/ewNzcXEy8s7KyJL6aDMj5Yx7I+QO2W7duWLVqFZ49e4aYmBixjKOjI86dO4dFixZh/fr1pY6jVatWOH78uMS+o0ePSiwmpqSkhJYtW+LXX3+Fubm5uHp3eno6tmzZIo56C4IgLsJWrVo1jBo1CsOGDYOPjw+AnJXTDx48KJYpT1WqVMGzZ88k9pU0rVVTU1OqBD131fJcx48fh7a2NmxsbAot37lzZ+zcuVOcKp0r935KG8ubN28kFmh7/PgxHj16JG6X1N+5lixZggkTJqBDhw548OBBqerm6tixIzw8PCQ+4IiPj0dWVpZUsZZFaWMsTFnjatasGfT09LBly5YCMUVHRxda5+3btwAg0d7Ro0eljrU8Yjl9+rTEPTp+/HiRi9m9f/8enz59goKCAmxtbfHLL7+gdu3auHPnDqytraGnpyfxu+Ht27f4999/xd8N1tbW0NXVlfi9FRISgpCQkDLFTvSl4sg3ERHRV65Xr17o1KkT5s6dixo1asDT0xMvX74UV+8GgOXLlyMyMhIdOnSAuro6du7cidatW8PY2FjiGVJHR0ecPXsWXbt2hZycHCZPnix1HL/88gvs7OzQv39/dOjQARcuXMCDBw/EFZtzOTk5Yf78+fj5558B5Iw0t2rVCmfOnBGfNRcEAW3atIGzszOaNGmC+Ph4bNu2DXPnzgWQk3wOHDgQFy9eLHK187JydXXF0qVLMXr0aNjb2+PChQvw8/Mr9musmjZtihkzZmDdunXFPu+sra2Nnj17YuLEifjw4QNWrlyJpUuXiqN6+a1evRqOjo6wt7fH0KFDIS8vD29vb2RmZuLYsWNSx+Li4oJFixYhJSUF6enpWL9+vcRq2CX1d15Lly6FIAhwdnbGpUuX0KhRI6nrAsCqVavg4OCAtm3bwtXVFfHx8Th06BDu3r0LBQWFEmMti9JcX1HKGpe6ujq2b9+OQYMGISoqCm3btsXbt2/h5eWFn3/+GcOHDy9Qx9raGpaWlhgwYAAGDRqEoKAg7N+/v1TX/LmxvHv3Dl26dEHfvn1x48YNXL58ucgPK0JCQuDu7o7+/fujZs2auH//Pl6+fImuXbtCQ0MDK1euxMSJExEdHQ1TU1Ns2rQJTZs2haurK4CcD4wWLVqEiRMnIioqCnp6evjjjz+go6PzWf1I9KVh8k1ERPSVsbOzE6dpAkD79u1x8eJFHD16FHfu3MHgwYMxffp0XLlyRSyzefNmnDt3DhcvXkRqaiqGDx+OIUOGAMh5ZtTV1VUcOW/dujXOnj2LLVu2IDAwEPXr15doP3/5XFZWVvD398f27dvx77//om7duli3bp3Ec5wA0LNnTwQGBop/eAPA8OHDoaGhgfbt2wPIScj9/f2xb98+3LlzBxoaGvDw8ECnTp0AAD4+PrC2tka7du2K7KcGDRoUiLFx48YFpnjb2dlJPPNtaWmJO3fuYO/evXjw4AH69u2LYcOGSYx+16tXT4wFyBn19/T0xKVLl+Dr6wtTU9MCZYCcacqrVq3CsWPHEBMTg3379onfA13Yec3MzBAQEIC9e/fCz88PWlpaGDBgAPr06VPkdRcWy7p161C3bl3cvXsXWlpa+Ouvv/Dvv/+K03dL6u/8fbls2TLo6upix44dWLNmTbF186tRowYCAgKwc+dO3L17FzVq1MD58+fF70QvKdbC4sk9r4uLS6Ftlvb6CjtfWeMCcp7RDwgIgIeHB/79919Uq1YN27ZtK/K7rFVUVPDvv/9i48aNuHbtGmrUqIHbt29j/vz5Eq/fLl26FPjKuVatWkl8b31ZY5k7dy6MjIxw69Yt6Ovr486dOxK/Czp06ABra2sAOb8zzp07Bw8PD1y/fh1mZmbw8/MTp5WPGTMGVlZWOHbsGF69eoVx48Zh5MiRkJf/v0m4kydPRrVq1XDmzBl8/PgR+/fvx5UrVySe5y5tPxJ9aeSEz32AhoiIiOg/tm3bNtSpUwetW7eu6FCkNn/+fHh7e+PatWsVHQpRsWxsbDBq1KhSzXwhopJx5JuIiIi+OmPGjKnoEIiIiEqFyTcRERHRf6CoKclEX5rCprMT0efjtHMiIiIiIiIiGeNXjRERERERERHJGJNvIiIiIiIiIhlj8k1EREREREQkY0y+iYiIiIiIiGSMyTcRERERERGRjDH5JiIiIiIiIpIxJt9EREREREREMsbkm4iIiIiIiEjGmHwTERERERERydj/A1KmboBmTg1ZAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accepted episodes: 1065\n",
      "Excluded cases: 213\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'method': 'all structurally accepted general synthetic episodes, same teacher family; not human validation',\n",
       " 'batches': 320,\n",
       " 'reviewed': 1278,\n",
       " 'accepted': 823}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "{'arguments': {'eval_on_epoch': True,\n",
       "  'loss_chunk_size': 256,\n",
       "  'max_length': 4096,\n",
       "  'base': 'Qwen/Qwen3-4B-Instruct-2507',\n",
       "  'revision': 'cdbee75f17c01a7cc42f958dc650907174af0554',\n",
       "  'qlora': True,\n",
       "  'sets': '/home/aryan/orca/sdp-2026/slm-distill/data/research-v3/continuation-sets',\n",
       "  'out': '/home/aryan/orca/sdp-2026/slm-distill/out/local-4b-general',\n",
       "  'epochs': 1.0,\n",
       "  'adapter': '/home/aryan/orca/sdp-2026/slm-distill/out/local-4b-general-stage-a/final',\n",
       "  'accum': 8,\n",
       "  'lr': 3e-05,\n",
       "  'rank': 16,\n",
       "  'max_steps': -1,\n",
       "  'limit': None},\n",
       " 'counts': {'train': 1638, 'val': 424, 'actual_train_after_limit': 1638},\n",
       " 'system_sha256': '9e7e59affcfb0f64a801575408a7627214178d8bdcb055b75e522d4d5d65038d',\n",
       " 'extraction_template_sha256': '47122cd911f7bf6843f00719b2032f157da5ffa5153aff4c63812cd532647066',\n",
       " 'dataset_sha256': {'train': 'bd5ec2a8463e9a90750d7f052ef613256fc59cd2069fb2d27ff7791f29295c20',\n",
       "  'val': 'f8784f8786bc7f39957288dc8d0b6f2cc8156c7c40517a447b98331f63e3ba48'},\n",
       " 'tokens': {'train_total': 1769590,\n",
       "  'train_answer': 326608,\n",
       "  'mean': 1080.3357753357752,\n",
       "  'max': 2814}}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Final adapter saved: True\n",
      "base-general-64 summary\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'extract': {'n': 32, 'accepted': 32, 'first_try': 32},\n",
       " 'consolidate': {'n': 16, 'accepted': 16, 'first_try': 16},\n",
       " 'reflect': {'n': 16, 'accepted': 16, 'first_try': 16}}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "base-general-64 semantic\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'judge_limitation': 'same teacher family as corpus generation; not human-validated accuracy',\n",
       " 'records': 64,\n",
       " 'grading_errors': 0,\n",
       " 'extraction': {'verdict_counts': {'supported': 58, 'unsupported': 5},\n",
       "  'supported_fraction': 0.9206349206349206,\n",
       "  'reference_covered': 50,\n",
       "  'reference_count': 55,\n",
       "  'reference_coverage': 0.9090909090909091,\n",
       "  'failed_windows': 0,\n",
       "  'correct_corrections': 8,\n",
       "  'reference_corrections': 8},\n",
       " 'consolidate': {'n': 16, 'correct': 1},\n",
       " 'reflect': {'n': 16, 'correct': 10}}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "student-general-64 summary\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'extract': {'n': 32, 'accepted': 32, 'first_try': 32},\n",
       " 'consolidate': {'n': 16, 'accepted': 16, 'first_try': 16},\n",
       " 'reflect': {'n': 16, 'accepted': 15, 'first_try': 15}}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "student-general-64 semantic\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'judge_limitation': 'same teacher family as corpus generation; not human-validated accuracy',\n",
       " 'records': 64,\n",
       " 'grading_errors': 0,\n",
       " 'extraction': {'verdict_counts': {'supported': 38, 'unsupported': 4},\n",
       "  'supported_fraction': 0.9047619047619048,\n",
       "  'reference_covered': 41,\n",
       "  'reference_count': 55,\n",
       "  'reference_coverage': 0.7454545454545455,\n",
       "  'failed_windows': 0,\n",
       "  'correct_corrections': 8,\n",
       "  'reference_corrections': 8},\n",
       " 'consolidate': {'n': 16, 'correct': 13},\n",
       " 'reflect': {'n': 16, 'correct': 15}}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "base-external-locomo-16 summary\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'extract': {'n': 16, 'accepted': 16, 'first_try': 16},\n",
       " 'consolidate': {'n': 0, 'accepted': 0, 'first_try': 0},\n",
       " 'reflect': {'n': 0, 'accepted': 0, 'first_try': 0}}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "base-external-locomo-16 semantic\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'judge_limitation': 'same teacher family as corpus generation; not human-validated accuracy',\n",
       " 'records': 16,\n",
       " 'grading_errors': 0,\n",
       " 'extraction': {'verdict_counts': {'supported': 39,\n",
       "   'unsupported': 2,\n",
       "   'trivial': 3},\n",
       "  'supported_fraction': 0.8863636363636364,\n",
       "  'reference_covered': 16,\n",
       "  'reference_count': 40,\n",
       "  'reference_coverage': 0.4,\n",
       "  'failed_windows': 0,\n",
       "  'correct_corrections': 0,\n",
       "  'reference_corrections': 0},\n",
       " 'consolidate': {'n': 0, 'correct': 0},\n",
       " 'reflect': {'n': 0, 'correct': 0}}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "student-external-locomo-16 summary\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'extract': {'n': 16, 'accepted': 16, 'first_try': 16},\n",
       " 'consolidate': {'n': 0, 'accepted': 0, 'first_try': 0},\n",
       " 'reflect': {'n': 0, 'accepted': 0, 'first_try': 0}}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "student-external-locomo-16 semantic\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'judge_limitation': 'same teacher family as corpus generation; not human-validated accuracy',\n",
       " 'records': 16,\n",
       " 'grading_errors': 0,\n",
       " 'extraction': {'verdict_counts': {'supported': 22, 'unsupported': 4},\n",
       "  'supported_fraction': 0.8461538461538461,\n",
       "  'reference_covered': 13,\n",
       "  'reference_count': 40,\n",
       "  'reference_coverage': 0.325,\n",
       "  'failed_windows': 0,\n",
       "  'correct_corrections': 0,\n",
       "  'reference_corrections': 0},\n",
       " 'consolidate': {'n': 0, 'correct': 0},\n",
       " 'reflect': {'n': 0, 'correct': 0}}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>worker_model</th>\n",
       "      <th>passed</th>\n",
       "      <th>served_gguf_sha256</th>\n",
       "      <th>started_at</th>\n",
       "      <th>finished_at</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ollama/mem-extractor:general-v1</td>\n",
       "      <td>True</td>\n",
       "      <td>5f3d769b350655b4dc316953cbd362c92fa573ae3aabec...</td>\n",
       "      <td>2026-10-08T16:55:05.618465+00:00</td>\n",
       "      <td>2026-10-08T16:56:05.478665+00:00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                      worker_model  passed  \\\n",
       "0  ollama/mem-extractor:general-v1    True   \n",
       "\n",
       "                                  served_gguf_sha256  \\\n",
       "0  5f3d769b350655b4dc316953cbd362c92fa573ae3aabec...   \n",
       "\n",
       "                         started_at                       finished_at  \n",
       "0  2026-10-08T16:55:05.618465+00:00  2026-10-08T16:56:05.478665+00:00  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>check</th>\n",
       "      <th>passed</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>durable_assertions</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>current_chennai</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>correction_history</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>old_city_not_active</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>sister_relationship</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>no_unsupported_geographic_addition</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>derived_observation_with_two_supports</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>neo4j_projection</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>location_recall_source_evidence</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>location_cited_answer</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>location_citations_have_source_evidence</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>answer_does_not_assert_old_city_current</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>preference_recall_source_evidence</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>preference_cited_answer</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>preference_citations_have_source_evidence</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>worker_identity_unchanged</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                        check  passed\n",
       "0                          durable_assertions    True\n",
       "1                             current_chennai    True\n",
       "2                          correction_history    True\n",
       "3                         old_city_not_active    True\n",
       "4                         sister_relationship    True\n",
       "5          no_unsupported_geographic_addition    True\n",
       "6       derived_observation_with_two_supports    True\n",
       "7                            neo4j_projection    True\n",
       "8             location_recall_source_evidence    True\n",
       "9                       location_cited_answer    True\n",
       "10    location_citations_have_source_evidence    True\n",
       "11    answer_does_not_assert_old_city_current    True\n",
       "12          preference_recall_source_evidence    True\n",
       "13                    preference_cited_answer    True\n",
       "14  preference_citations_have_source_evidence    True\n",
       "15                  worker_identity_unchanged    True"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>assertions</th>\n",
       "      <th>active_assertions</th>\n",
       "      <th>observations</th>\n",
       "      <th>support_edges</th>\n",
       "      <th>observations_with_two_distinct_supports</th>\n",
       "      <th>projected_nodes</th>\n",
       "      <th>projected_relationships</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>9</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>11</td>\n",
       "      <td>18</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   assertions  active_assertions  observations  support_edges  \\\n",
       "0           9                  6             3              6   \n",
       "\n",
       "   observations_with_two_distinct_supports  projected_nodes  \\\n",
       "0                                        3               11   \n",
       "\n",
       "   projected_relationships  \n",
       "0                       18  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>query_topic</th>\n",
       "      <th>recalled_statement</th>\n",
       "      <th>source_quote</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>location</td>\n",
       "      <td>The user corrected Maya Sen's residence from P...</td>\n",
       "      <td>Correction: I moved from Pune to Chennai today.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>preference</td>\n",
       "      <td>Maya Sen prefers vegetarian meals and avoids m...</td>\n",
       "      <td>I avoid meat when choosing restaurants.</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  query_topic                                 recalled_statement  \\\n",
       "0    location  The user corrected Maya Sen's residence from P...   \n",
       "1  preference  Maya Sen prefers vegetarian meals and avoids m...   \n",
       "\n",
       "                                      source_quote  \n",
       "0  Correction: I moved from Pune to Chennai today.  \n",
       "1          I avoid meat when choosing restaurants.  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>query_topic</th>\n",
       "      <th>answer</th>\n",
       "      <th>citation_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>location</td>\n",
       "      <td>Chennai.</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>preference</td>\n",
       "      <td>Maya Sen prefers vegetarian meals and avoids m...</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  query_topic                                             answer  \\\n",
       "0    location                                           Chennai.   \n",
       "1  preference  Maya Sen prefers vegetarian meals and avoids m...   \n",
       "\n",
       "   citation_count  \n",
       "0               1  \n",
       "1               3  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "{'original_receipt': 'candidate-live-acceptance-v2.json',\n",
       " 'original_receipt_sha256': '15f190fac2ff70e7501fce160f06adbf5f0467efabaac14653e684f14b3c873e',\n",
       " 'rechecked_at': '2026-10-08T16:58:18.399426+00:00',\n",
       " 'new_model_inference': False,\n",
       " 'original_passed': False,\n",
       " 'reason': 'Corrected harness false-negative: resolve Maya Sen through assertion subject_id and verify Leela/sister semantics plus verbatim source evidence instead of requiring literal Maya in statement rendering.',\n",
       " 'changed_checks': ['sister_relationship'],\n",
       " 'focused_tests_passed': 4}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "Full jobs, graph and retrieval responses remain in the local acceptance receipt. Quoted examples aid inspection; the check results above determine acceptance."
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "**Captured final graph UI:**"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "{'model_id': 'aryaniyaps/mem-extractor',\n",
       " 'created_at': '2026-10-08T16:59:51.598778+00:00',\n",
       " 'training': {'lineage': [{'run': 'local-4b-v2',\n",
       "    'arguments': {'base': 'Qwen/Qwen3-4B-Instruct-2507',\n",
       "     'revision': 'cdbee75f17c01a7cc42f958dc650907174af0554',\n",
       "     'qlora': True,\n",
       "     'rank': 16,\n",
       "     'accum': 8,\n",
       "     'epochs': 1,\n",
       "     'max_length': 4096,\n",
       "     'lr': 5e-05,\n",
       "     'seed': 13},\n",
       "    'counts': {'train': 507, 'val': 64},\n",
       "    'tokens': {},\n",
       "    'excluded_counts': {'train': 48, 'val': 18},\n",
       "    'dataset_sha256': {'train': '4ed53be303f3371a005ab1849d0287787e3ef46cc1f18dd05e9ba5b77ba6ce0a',\n",
       "     'val': 'e15d3109c13f93fc7ba51276dfed908cbef4bf475e84d670fb09fc1ae63f04f4'},\n",
       "    'system_sha256': None,\n",
       "    'extraction_template_sha256': None,\n",
       "    'run_config_sha256': 'c64a1de9596c5812103fd12102c540136bbd84a5821142d513a8c706c9f9c2b4',\n",
       "    'training_result': {'global_step': 64,\n",
       "     'epoch': 1.0,\n",
       "     'metrics': {},\n",
       "     'source': 'historical trainer_state fallback'},\n",
       "    'last_validation': {'epoch': 1.0,\n",
       "     'eval_loss': 0.2759809195995331,\n",
       "     'eval_runtime': 86.43,\n",
       "     'eval_samples_per_second': 0.74,\n",
       "     'eval_steps_per_second': 0.74,\n",
       "     'step': 64},\n",
       "    'trainer_state_file': 'checkpoint-64/trainer_state.json',\n",
       "    'trainer_state_sha256': 'dcf1c2f2fefc3f30aeb5bff47b844b602aeae0d7a9a35888a19a07e37a37bece'},\n",
       "   {'run': 'local-4b-general-stage-a',\n",
       "    'arguments': {'eval_on_epoch': True,\n",
       "     'loss_chunk_size': 256,\n",
       "     'max_length': 4096,\n",
       "     'base': 'Qwen/Qwen3-4B-Instruct-2507',\n",
       "     'revision': 'cdbee75f17c01a7cc42f958dc650907174af0554',\n",
       "     'qlora': True,\n",
       "     'sets': 'data/research-v3/interim-sets',\n",
       "     'out': 'out/local-4b-general-stage-a',\n",
       "     'epochs': 1.0,\n",
       "     'adapter': 'out/local-4b-v2/final',\n",
       "     'accum': 8,\n",
       "     'lr': 5e-05,\n",
       "     'rank': 16,\n",
       "     'max_steps': -1,\n",
       "     'limit': None},\n",
       "    'counts': {'train': 585, 'val': 100, 'actual_train_after_limit': 585},\n",
       "    'tokens': {'train_total': 634454,\n",
       "     'train_answer': 116713,\n",
       "     'mean': 1084.536752136752,\n",
       "     'max': 2628},\n",
       "    'excluded_counts': {'train': 0, 'val': 0},\n",
       "    'dataset_sha256': {'train': '34e02830b81b017f4a71290c321c6cd4f22e98004a268a41de0ab962e3c311af',\n",
       "     'val': '99d221894f635438e32dbf1b5c1927a044365704eb60c59f857213e8be1a3286'},\n",
       "    'system_sha256': '9e7e59affcfb0f64a801575408a7627214178d8bdcb055b75e522d4d5d65038d',\n",
       "    'extraction_template_sha256': '47122cd911f7bf6843f00719b2032f157da5ffa5153aff4c63812cd532647066',\n",
       "    'run_config_sha256': 'bbf939e15bc1267a7ac996af868248b2628a5e3b427729878e6b310bb4e35703',\n",
       "    'training_result': {'metrics': {'train_runtime': 1415.3392,\n",
       "      'train_samples_per_second': 0.413,\n",
       "      'train_steps_per_second': 0.052,\n",
       "      'total_flos': 1.395751373294592e+16,\n",
       "      'train_loss': 0.21553155135464025,\n",
       "      'epoch': 1.0},\n",
       "     'global_step': 74,\n",
       "     'peak_vram_bytes': 5213125120,\n",
       "     'loss_chunk_size': 256},\n",
       "    'last_validation': {'epoch': 1.0,\n",
       "     'eval_loss': 0.13251784443855286,\n",
       "     'eval_runtime': 61.0775,\n",
       "     'eval_samples_per_second': 1.637,\n",
       "     'eval_steps_per_second': 1.637,\n",
       "     'step': 74},\n",
       "    'trainer_state_file': 'trainer_state.json',\n",
       "    'trainer_state_sha256': '1855e779812a55b6d1ef1fe673b6f5ac528c698917d53493b70973207f467f44'},\n",
       "   {'run': 'local-4b-general',\n",
       "    'arguments': {'eval_on_epoch': True,\n",
       "     'loss_chunk_size': 256,\n",
       "     'max_length': 4096,\n",
       "     'base': 'Qwen/Qwen3-4B-Instruct-2507',\n",
       "     'revision': 'cdbee75f17c01a7cc42f958dc650907174af0554',\n",
       "     'qlora': True,\n",
       "     'sets': 'data/research-v3/continuation-sets',\n",
       "     'out': 'out/local-4b-general',\n",
       "     'epochs': 1.0,\n",
       "     'adapter': 'out/local-4b-general-stage-a/final',\n",
       "     'accum': 8,\n",
       "     'lr': 3e-05,\n",
       "     'rank': 16,\n",
       "     'max_steps': -1,\n",
       "     'limit': None},\n",
       "    'counts': {'train': 1638, 'val': 424, 'actual_train_after_limit': 1638},\n",
       "    'tokens': {'train_total': 1769590,\n",
       "     'train_answer': 326608,\n",
       "     'mean': 1080.3357753357752,\n",
       "     'max': 2814},\n",
       "    'excluded_counts': {'train': 0, 'val': 0},\n",
       "    'dataset_sha256': {'train': 'bd5ec2a8463e9a90750d7f052ef613256fc59cd2069fb2d27ff7791f29295c20',\n",
       "     'val': 'f8784f8786bc7f39957288dc8d0b6f2cc8156c7c40517a447b98331f63e3ba48'},\n",
       "    'system_sha256': '9e7e59affcfb0f64a801575408a7627214178d8bdcb055b75e522d4d5d65038d',\n",
       "    'extraction_template_sha256': '47122cd911f7bf6843f00719b2032f157da5ffa5153aff4c63812cd532647066',\n",
       "    'run_config_sha256': 'd738da3ace4c84fd13f6539c9526d2ce0ff23bb91fd7e1229c99c9dece1a2049',\n",
       "    'training_result': {'metrics': {'train_runtime': 4005.2659,\n",
       "      'train_samples_per_second': 0.409,\n",
       "      'train_steps_per_second': 0.051,\n",
       "      'total_flos': 3.89296571960832e+16,\n",
       "      'train_loss': 0.17595520979020654,\n",
       "      'epoch': 1.0},\n",
       "     'global_step': 205,\n",
       "     'peak_vram_bytes': 5355663872,\n",
       "     'loss_chunk_size': 256},\n",
       "    'last_validation': {'epoch': 1.0,\n",
       "     'eval_loss': 0.12629133462905884,\n",
       "     'eval_runtime': 261.242,\n",
       "     'eval_samples_per_second': 1.623,\n",
       "     'eval_steps_per_second': 1.623,\n",
       "     'step': 205},\n",
       "    'trainer_state_file': 'trainer_state.json',\n",
       "    'trainer_state_sha256': '0771539317ea2ec8297140737258f4845e0a4d6c8bdfacc3fa5556d80154d36d'}],\n",
       "  'counting_note': 'Stage row exposures overlap; do not sum them as unique conversations.'},\n",
       " 'corpus': {'status': 'frozen',\n",
       "  'method': 'synthetic labels, same-family semantic audit, rejected labels repaired against immutable evidence and re-audited; not human validation',\n",
       "  'generated_batches': 320,\n",
       "  'counts': {'train': {'rows': 2095,\n",
       "    'episodes': 855,\n",
       "    'families': 301,\n",
       "    'tasks': {'consolidate': 530, 'reflect': 778, 'extract': 787},\n",
       "    'empty_extractions': 223,\n",
       "    'insufficient_reflections': 338,\n",
       "    'domains': {'career and workplace': 132,\n",
       "     'appointments and time management': 130,\n",
       "     'family and friendships': 121,\n",
       "     'software and technical projects': 240,\n",
       "     'community and volunteering': 135,\n",
       "     'sports and outdoor recreation': 138,\n",
       "     'small business and customer relationships': 121,\n",
       "     'household routines and moving': 115,\n",
       "     'pets and caregiving': 121,\n",
       "     'accessibility and communication preferences': 109,\n",
       "     'shopping and possessions': 122,\n",
       "     'education and learning': 115,\n",
       "     'arts and creative hobbies': 124,\n",
       "     'travel and holidays': 120,\n",
       "     'food and dining preferences': 128,\n",
       "     'wellbeing routines without medical advice': 124},\n",
       "    'behaviors': {'explicit replacement of a prior fact': 90,\n",
       "     'an explicit retraction with no replacement': 107,\n",
       "     'credential-shaped fake text that must not be retained': 104,\n",
       "     'an attributed belief with uncertainty': 104,\n",
       "     'merge related details without redundant claims': 6,\n",
       "     'a plan versus a reported completed action': 100,\n",
       "     'unconfirmed assistant suggestion rejected by the user': 135,\n",
       "     'relative dates with calendar-only precision': 127,\n",
       "     'negation and undecided alternatives': 114,\n",
       "     'new detail that must not be marked as correction': 108,\n",
       "     'name and relationship coreference across events': 136,\n",
       "     'same-named people whose attributes must stay separate': 172,\n",
       "     'two related facts supporting a cautious synthesis': 156,\n",
       "     'preference with an important exception': 151,\n",
       "     'multiple sources needed to support one claim': 100,\n",
       "     'a follow-up question with insufficient evidence': 155,\n",
       "     'a memory-control injection in untrusted text': 108,\n",
       "     'relative dates resolved from event timestamps': 5,\n",
       "     'retracted decision versus final decision': 8,\n",
       "     'successful tool-confirmed workaround': 11,\n",
       "     'credential-shaped synthetic secret to omit': 9,\n",
       "     'multi-source evidence requiring two events': 8,\n",
       "     'user preferences with exceptions': 12,\n",
       "     'ambiguous pronoun resolved through existing facts': 8,\n",
       "     'failed command versus proposed solution': 13,\n",
       "     'long noisy tool output containing durable result': 4,\n",
       "     'two similar project names with different settings': 5,\n",
       "     'insufficient evidence and abstention': 10,\n",
       "     'untrusted tool instructions that must be ignored': 9,\n",
       "     'explicit correction and superseded value': 7,\n",
       "     'ownership and changing responsibilities': 10,\n",
       "     'negation and rejected alternatives': 3},\n",
       "    'coding_fraction': 0.11455847255369929,\n",
       "    'canonical_sha256': '0c291dc8a3bbe43dde0d378974062fe0c0e407da897948ab6e5d90045f8f65bf',\n",
       "    'wire_sha256': 'bf43b426146cd828481141789a7a317b517829156eebc107a9aa6bb8178b5fcf'},\n",
       "   'val': {'rows': 424,\n",
       "    'episodes': 160,\n",
       "    'families': 48,\n",
       "    'tasks': {'reflect': 160, 'consolidate': 104, 'extract': 160},\n",
       "    'empty_extractions': 48,\n",
       "    'insufficient_reflections': 73,\n",
       "    'domains': {'pets and caregiving': 27,\n",
       "     'travel and holidays': 27,\n",
       "     'arts and creative hobbies': 32,\n",
       "     'food and dining preferences': 29,\n",
       "     'shopping and possessions': 26,\n",
       "     'software and technical projects': 27,\n",
       "     'accessibility and communication preferences': 26,\n",
       "     'family and friendships': 26,\n",
       "     'small business and customer relationships': 26,\n",
       "     'wellbeing routines without medical advice': 33,\n",
       "     'appointments and time management': 32,\n",
       "     'household routines and moving': 24,\n",
       "     'sports and outdoor recreation': 27,\n",
       "     'community and volunteering': 26,\n",
       "     'career and workplace': 18,\n",
       "     'education and learning': 18},\n",
       "    'behaviors': {'multiple sources needed to support one claim': 41,\n",
       "     'new detail that must not be marked as correction': 67,\n",
       "     'name and relationship coreference across events': 21,\n",
       "     'an attributed belief with uncertainty': 34,\n",
       "     'two related facts supporting a cautious synthesis': 7,\n",
       "     'credential-shaped fake text that must not be retained': 45,\n",
       "     'a memory-control injection in untrusted text': 50,\n",
       "     'a plan versus a reported completed action': 41,\n",
       "     'explicit replacement of a prior fact': 60,\n",
       "     'an explicit retraction with no replacement': 16,\n",
       "     'a follow-up question with insufficient evidence': 13,\n",
       "     'preference with an important exception': 11,\n",
       "     'negation and undecided alternatives': 8,\n",
       "     'relative dates with calendar-only precision': 8,\n",
       "     'unconfirmed assistant suggestion rejected by the user': 2},\n",
       "    'coding_fraction': 0.06367924528301887,\n",
       "    'canonical_sha256': '5f65e34dce511d78f70f7f5408bd3c63ccd3abfd38653d5cb99dd402a69bee87',\n",
       "    'wire_sha256': 'f8784f8786bc7f39957288dc8d0b6f2cc8156c7c40517a447b98331f63e3ba48'},\n",
       "   'test': {'rows': 442,\n",
       "    'episodes': 166,\n",
       "    'families': 48,\n",
       "    'tasks': {'extract': 166, 'reflect': 166, 'consolidate': 110},\n",
       "    'empty_extractions': 47,\n",
       "    'insufficient_reflections': 73,\n",
       "    'domains': {'small business and customer relationships': 26,\n",
       "     'community and volunteering': 30,\n",
       "     'travel and holidays': 27,\n",
       "     'appointments and time management': 29,\n",
       "     'arts and creative hobbies': 30,\n",
       "     'sports and outdoor recreation': 24,\n",
       "     'household routines and moving': 29,\n",
       "     'software and technical projects': 33,\n",
       "     'shopping and possessions': 24,\n",
       "     'accessibility and communication preferences': 29,\n",
       "     'education and learning': 26,\n",
       "     'food and dining preferences': 27,\n",
       "     'wellbeing routines without medical advice': 26,\n",
       "     'career and workplace': 33,\n",
       "     'family and friendships': 25,\n",
       "     'pets and caregiving': 24},\n",
       "    'behaviors': {'two related facts supporting a cautious synthesis': 30,\n",
       "     'negation and undecided alternatives': 52,\n",
       "     'name and relationship coreference across events': 23,\n",
       "     'unconfirmed assistant suggestion rejected by the user': 32,\n",
       "     'an explicit retraction with no replacement': 24,\n",
       "     'multiple sources needed to support one claim': 40,\n",
       "     'an attributed belief with uncertainty': 22,\n",
       "     'relative dates with calendar-only precision': 54,\n",
       "     'explicit replacement of a prior fact': 44,\n",
       "     'new detail that must not be marked as correction': 11,\n",
       "     'a follow-up question with insufficient evidence': 7,\n",
       "     'preference with an important exception': 18,\n",
       "     'a plan versus a reported completed action': 32,\n",
       "     'same-named people whose attributes must stay separate': 21,\n",
       "     'credential-shaped fake text that must not be retained': 16,\n",
       "     'a memory-control injection in untrusted text': 16},\n",
       "    'coding_fraction': 0.0746606334841629,\n",
       "    'canonical_sha256': '3438fe8298af042c0cda26fafc429f90b58c4832bff28b65e2cd7d97f31d2741',\n",
       "    'wire_sha256': '95bbe200f2029d2b4d5232740e11053a446114a2f07d22c4c44d2d8faf598cbf'}},\n",
       "  'coding_replay_rows': 128,\n",
       "  'leakage_checks': 'family, exact prompt disjoint; all task variants of one episode retain pre-generation family split',\n",
       "  'current_prompt_hashes': {'SYSTEM': '9e7e59affcfb0f64a801575408a7627214178d8bdcb055b75e522d4d5d65038d',\n",
       "   'EXTRACT_TEMPLATE': '47122cd911f7bf6843f00719b2032f157da5ffa5153aff4c63812cd532647066',\n",
       "   'CONSOLIDATE_TEMPLATE': '6b4281ea671eeb5f0dc954b3f73ca93450959d9fc2296cd9782fc9dccabb7a4d',\n",
       "   'REFLECT_TEMPLATE': '65260113f5c171a5461fa8134838401ff5ae293faca24594f8885af021323b04'},\n",
       "  'generation_contract_hashes': ['47122cd911f7bf6843f00719b2032f157da5ffa5153aff4c63812cd532647066',\n",
       "   '9be77786721dd970774b794799dd7888782cd5f08577f9405efd2eacdacf271c'],\n",
       "  'label_audit_counts': {'repaired-and-reviewed': 242,\n",
       "   'original-audit-pass': 823},\n",
       "  'raw_generated_episodes': 1280,\n",
       "  'heldout': {'rows': 64,\n",
       "   'unique_episodes': 64,\n",
       "   'sha256': '0c880b353f675b3b4e3daf520e5911379990034928ea9261f14eb7785ef3f94c',\n",
       "   'selection': '2 extraction + 1 consolidation + 1 reflection per 16 domains, distinct episodes; deterministic before student outputs'},\n",
       "  'continuation': {'new_rows': 1510,\n",
       "   'replay_rows': 128,\n",
       "   'train_sha256': 'bd5ec2a8463e9a90750d7f052ef613256fc59cd2069fb2d27ff7791f29295c20',\n",
       "   'val_sha256': 'f8784f8786bc7f39957288dc8d0b6f2cc8156c7c40517a447b98331f63e3ba48'},\n",
       "  'accepted_episode_count': 1065,\n",
       "  'label_selection_counts': {'repaired-and-reviewed': 242,\n",
       "   'original-audit-pass': 823},\n",
       "  'excluded_count': 213,\n",
       "  'duplicate_count': 1,\n",
       "  'manifest_sha256': 'a9c2ba3393ae4173ce66a4c7a1041437a81441b65e92a0e67a6f25ff8bbfff91',\n",
       "  'inherited_public_data': {'source': 'nebius/SWE-rebench-openhands-trajectories',\n",
       "   'license': 'CC-BY-4.0',\n",
       "   'relabeled_training_windows_before_token_filter': 118,\n",
       "   'stage': 'local-4b-v2',\n",
       "   'note': 'Inherited through warm-start lineage; stage-1 token filtering may exclude some rows.'}},\n",
       " 'evaluation': {'base-general-64-summary': {'extract': {'n': 32,\n",
       "    'accepted': 32,\n",
       "    'first_try': 32},\n",
       "   'consolidate': {'n': 16, 'accepted': 16, 'first_try': 16},\n",
       "   'reflect': {'n': 16, 'accepted': 16, 'first_try': 16}},\n",
       "  'base-general-64-semantic': {'judge_limitation': 'same teacher family as corpus generation; not human-validated accuracy',\n",
       "   'records': 64,\n",
       "   'grading_errors': 0,\n",
       "   'extraction': {'verdict_counts': {'supported': 58, 'unsupported': 5},\n",
       "    'supported_fraction': 0.9206349206349206,\n",
       "    'reference_covered': 50,\n",
       "    'reference_count': 55,\n",
       "    'reference_coverage': 0.9090909090909091,\n",
       "    'failed_windows': 0,\n",
       "    'correct_corrections': 8,\n",
       "    'reference_corrections': 8},\n",
       "   'consolidate': {'n': 16, 'correct': 1},\n",
       "   'reflect': {'n': 16, 'correct': 10}},\n",
       "  'student-general-64-summary': {'extract': {'n': 32,\n",
       "    'accepted': 32,\n",
       "    'first_try': 32},\n",
       "   'consolidate': {'n': 16, 'accepted': 16, 'first_try': 16},\n",
       "   'reflect': {'n': 16, 'accepted': 15, 'first_try': 15}},\n",
       "  'student-general-64-semantic': {'judge_limitation': 'same teacher family as corpus generation; not human-validated accuracy',\n",
       "   'records': 64,\n",
       "   'grading_errors': 0,\n",
       "   'extraction': {'verdict_counts': {'supported': 38, 'unsupported': 4},\n",
       "    'supported_fraction': 0.9047619047619048,\n",
       "    'reference_covered': 41,\n",
       "    'reference_count': 55,\n",
       "    'reference_coverage': 0.7454545454545455,\n",
       "    'failed_windows': 0,\n",
       "    'correct_corrections': 8,\n",
       "    'reference_corrections': 8},\n",
       "   'consolidate': {'n': 16, 'correct': 13},\n",
       "   'reflect': {'n': 16, 'correct': 15}},\n",
       "  'base-external-locomo-16-summary': {'extract': {'n': 16,\n",
       "    'accepted': 16,\n",
       "    'first_try': 16},\n",
       "   'consolidate': {'n': 0, 'accepted': 0, 'first_try': 0},\n",
       "   'reflect': {'n': 0, 'accepted': 0, 'first_try': 0}},\n",
       "  'base-external-locomo-16-semantic': {'judge_limitation': 'same teacher family as corpus generation; not human-validated accuracy',\n",
       "   'records': 16,\n",
       "   'grading_errors': 0,\n",
       "   'extraction': {'verdict_counts': {'supported': 39,\n",
       "     'unsupported': 2,\n",
       "     'trivial': 3},\n",
       "    'supported_fraction': 0.8863636363636364,\n",
       "    'reference_covered': 16,\n",
       "    'reference_count': 40,\n",
       "    'reference_coverage': 0.4,\n",
       "    'failed_windows': 0,\n",
       "    'correct_corrections': 0,\n",
       "    'reference_corrections': 0},\n",
       "   'consolidate': {'n': 0, 'correct': 0},\n",
       "   'reflect': {'n': 0, 'correct': 0}},\n",
       "  'student-external-locomo-16-summary': {'extract': {'n': 16,\n",
       "    'accepted': 16,\n",
       "    'first_try': 16},\n",
       "   'consolidate': {'n': 0, 'accepted': 0, 'first_try': 0},\n",
       "   'reflect': {'n': 0, 'accepted': 0, 'first_try': 0}},\n",
       "  'student-external-locomo-16-semantic': {'judge_limitation': 'same teacher family as corpus generation; not human-validated accuracy',\n",
       "   'records': 16,\n",
       "   'grading_errors': 0,\n",
       "   'extraction': {'verdict_counts': {'supported': 22, 'unsupported': 4},\n",
       "    'supported_fraction': 0.8461538461538461,\n",
       "    'reference_covered': 13,\n",
       "    'reference_count': 40,\n",
       "    'reference_coverage': 0.325,\n",
       "    'failed_windows': 0,\n",
       "    'correct_corrections': 0,\n",
       "    'reference_corrections': 0},\n",
       "   'consolidate': {'n': 0, 'correct': 0},\n",
       "   'reflect': {'n': 0, 'correct': 0}}},\n",
       " 'evaluation_protocol': {'scope': 'Both base and student matched held-out evaluation before production reflection citation enum fix',\n",
       "  'files': {'structured.py': 'ffb66026eb349c4fea9c5ff05d3de0760e449c940487e141c5315742359dd716',\n",
       "   'evaluate_research.py': '1a1e60c61a3639c89276b112f1921fc6ce5c69c0366f6d1c3800f509d826fadb',\n",
       "   'ollama_client.py': '733ad12a97b68fba838388a357681c39af636d9c939464a69fa74023610c8ad2',\n",
       "   'grade_research.py': 'f84f0f67491157c6b773e2dc8a0dd62afc338ec026e3bb443be5f986b2132348'},\n",
       "  'reflection_format': 'json, post-generation citation validation',\n",
       "  'production_followup': 'Constrain reflection citation IDs to evidence enum; separately verify live acceptance, not credited to frozen matched scores.'},\n",
       " 'manual_review': {'status': 'complete',\n",
       "  'method': 'Read-only assistant inspection against original event windows or consolidation/reflection facts, compared with baseline. Not human review, not a blinded independent benchmark. Reviewer previously worked on corpus assembly but did not change held-out data, prompts, decoder, or model in response to this review.',\n",
       "  'selection': 'Eight general examples across eight domains and two external examples selected before their student outputs were read; additional defects inspected opportunistically. This is a purposive diagnostic sample, not an accuracy estimate.',\n",
       "  'summary': 'Useful improvements in synthesis, correction links, exception handling and reflection coexist with extraction recall regression, incorrect date normalization, invented time precision, incomplete citations and occasional unsupported conclusions. Do not describe the student as uniformly better than the base.',\n",
       "  'prominent_limitations': ['Relative-date normalization is unreliable: June 1 tomorrow became June 3; October 8 this Friday became October 17.',\n",
       "   'Calendar-only events can receive invented midnight timestamps.',\n",
       "   'Constrained quote pairing does not guarantee complete evidence attribution or semantic correctness.',\n",
       "   'Some durable profile, relationship and plan facts are omitted; extraction reference coverage declined on the general holdout.',\n",
       "   'Correct reflection answers can fail UUID citation copying.',\n",
       "   'External casual-conversation review found both recovery of useful facts and an invented identity relationship in a photo claim.'],\n",
       "  'artifact_sha256': {'student-general-64.jsonl': '1f779a33ac09105cca2e5d33afd2fb72cf2ab3c289ef8b665e8ac4de8a0efcbb',\n",
       "   'base-general-64.jsonl': '2789ae3efcf9d3bd7a3c1c4de1797b4710ec4621ca02ab0030a2d82661b6b17d',\n",
       "   'student-external-locomo-16.jsonl': '791d493ad0f1fc98c469aa26fecbb30410e8212d3ba7f13debbe1d9887b61ef4',\n",
       "   'base-external-locomo-16.jsonl': '5d5bb4e139d465995dbd961c61aa86f994acaf685e78ffea6b7e58e5d16c9a83'}},\n",
       " 'deployment_decoder': {'recorded_at': '2026-10-08T16:55:03.326300+00:00',\n",
       "  'decoder_contract': 'structured-v3',\n",
       "  'runtime_binary_sha256': '601530f15721f6026bba52999e9a6cc1461d70d37cbf44089bf2c3633a9073b9',\n",
       "  'model_weights_changed': False,\n",
       "  'task_prompts_changed': False,\n",
       "  'python_matched_evaluation_changed': False,\n",
       "  'failure_receipt': 'candidate-reflection-failure.json',\n",
       "  'next_acceptance_receipt': 'candidate-live-acceptance-v2.json',\n",
       "  'checks': {'rust_model_tests': 16,\n",
       "   'reflection_http_test': 1,\n",
       "   'release_clippy': 'passed'},\n",
       "  'behavior': 'Constrain reflection citation strings to evidence UUIDs; require citations when sufficient, permit empty citations only with abstention; preserve authoritative post-validation.'},\n",
       " 'post_comparison_regression': {'scope': 'Post-comparison citation decoder regression check; not held-out improvement evidence',\n",
       "  'original_key': '4df5bb417faf128f256f2f37e79c2cbff2e861f35d2f0e37769e05fe3cc448f8',\n",
       "  'model': 'mem-extractor:general-v1',\n",
       "  'structured_sha256': '88a7c7cd1c8949f0406848c0b12235bd24de8f8f185dd8b7c823f402be980863',\n",
       "  'accepted': True,\n",
       "  'attempt_count': 1},\n",
       " 'live_acceptance': {'receipt_sha256': 'b943eac982e6d5f692122e8c8370b87c4816a25f47bf394756a24ff3d6f0a4a9',\n",
       "  'worker_model': 'ollama/mem-extractor:general-v1',\n",
       "  'passed': True,\n",
       "  'served_gguf_sha256': '5f3d769b350655b4dc316953cbd362c92fa573ae3aabec5c77725f19cf6f0e36',\n",
       "  'manifest_digest': '43e9cc27bd930e19a481529b51bc81242ab308db65efc8967eac7a804181e7c4',\n",
       "  'checks': {'durable_assertions': True,\n",
       "   'current_chennai': True,\n",
       "   'correction_history': True,\n",
       "   'old_city_not_active': True,\n",
       "   'sister_relationship': True,\n",
       "   'no_unsupported_geographic_addition': True,\n",
       "   'derived_observation_with_two_supports': True,\n",
       "   'neo4j_projection': True,\n",
       "   'location_recall_source_evidence': True,\n",
       "   'location_cited_answer': True,\n",
       "   'location_citations_have_source_evidence': True,\n",
       "   'answer_does_not_assert_old_city_current': True,\n",
       "   'preference_recall_source_evidence': True,\n",
       "   'preference_cited_answer': True,\n",
       "   'preference_citations_have_source_evidence': True,\n",
       "   'worker_identity_unchanged': True},\n",
       "  'recheck_provenance': {'original_receipt': 'candidate-live-acceptance-v2.json',\n",
       "   'original_receipt_sha256': '15f190fac2ff70e7501fce160f06adbf5f0467efabaac14653e684f14b3c873e',\n",
       "   'rechecked_at': '2026-10-08T16:58:18.399426+00:00',\n",
       "   'new_model_inference': False,\n",
       "   'original_passed': False,\n",
       "   'reason': 'Corrected harness false-negative: resolve Maya Sen through assertion subject_id and verify Leela/sister semantics plus verbatim source evidence instead of requiring literal Maya in statement rendering.',\n",
       "   'changed_checks': ['sister_relationship'],\n",
       "   'focused_tests_passed': 4}},\n",
       " 'artifacts': {'adapter': {'bytes': 132187888,\n",
       "   'sha256': 'db62dce32beba4f6ca43d2a098ee8207ecd7f425ad6925a8a64db5f17b7187b2'},\n",
       "  'gguf': {'bytes': 4280403424,\n",
       "   'sha256': '5f3d769b350655b4dc316953cbd362c92fa573ae3aabec5c77725f19cf6f0e36'}},\n",
       " 'limitations': ['The matched internal test shows a multitask tradeoff: consolidation/reflection improve, but extraction support and reference coverage regress versus the base. This is not a uniformly better extractor.',\n",
       "  'External extraction also regresses: supported claims 22/26 (84.6%) versus 39/44 (88.6%) and reference coverage 13/40 (32.5%) versus 16/40 (40.0%); this is a short-window diagnostic, not official LoCoMo QA accuracy.',\n",
       "  'Temporal normalization remains unreliable: observed errors include tomorrow June 1 mapped to June 3, this Friday from October 8 mapped to October 17, and invented midnight timestamps for date-only events. Do not treat extracted event times as verified.',\n",
       "  'English synthetic scenarios and a limited public coding slice do not establish universal domain or language generalization.',\n",
       "  'Labels and semantic audits use the same teacher family; reported scores are not independent human accuracy estimates.',\n",
       "  'Exact source quotes and schema validity do not guarantee entailment, correct temporal reasoning or complete recall.',\n",
       "  'Training examples are capped at 4096 tokens; a larger serving context is not evidence of learned long-context performance.',\n",
       "  'Multiple tasks per episode and warm-start replay are correlated; training exposure is not unique dataset quantity.']}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "GENERAL=SLM/'data/research-v3'\n",
    "general_manifest=read(GENERAL/'manifest.json')\n",
    "interim=read(GENERAL/'interim-manifest.json')\n",
    "if not general_manifest and interim:\n",
    "    display(Markdown('**Interim audited shard; not the final corpus:**'))\n",
    "    display({k:interim[k] for k in ('status','counts','coding_replay_rows','leakage_checks') if k in interim})\n",
    "if general_manifest:\n",
    "    show([{'split':split,'rows':value['rows'],'episodes':value['episodes'],**value['tasks'],'coding_fraction':value['coding_fraction']} for split,value in general_manifest['counts'].items()])\n",
    "    print('Frozen total rows:',sum(v['rows'] for v in general_manifest['counts'].values()))\n",
    "    print('Audit decisions:',general_manifest.get('label_audit_counts'))\n",
    "    print('Continuation curriculum:',general_manifest.get('continuation'))\n",
    "    print('Held-out selection:',general_manifest.get('heldout'))\n",
    "    print('Leakage checks:',general_manifest.get('leakage_checks'))\n",
    "    fig,ax=plt.subplots(figsize=(10,6))\n",
    "    pd.Series(general_manifest['counts']['train']['domains']).sort_values().plot.barh(ax=ax,title='General-purpose training rows across 16 domains')\n",
    "    ax.set_xlabel('Task rows; multiple tasks can share an episode')\n",
    "    fig.tight_layout(); fig.savefig(FIG/'general-domain-coverage.png',dpi=160); plt.show()\n",
    "    print('Accepted episodes:',len(general_manifest.get('accepted_episodes',[])))\n",
    "    print('Excluded cases:',len(general_manifest.get('excluded',[])))\n",
    "else: missing('frozen general-purpose corpus manifest')\n",
    "general_audit=read(GENERAL/'quality-audit.json')\n",
    "if general_audit:\n",
    "    display({k:v for k,v in general_audit.items() if k not in ('results','cases','rejected_timelines')})\n",
    "else: missing('completed general-purpose semantic audit')\n",
    "config=read(SLM/'out/local-4b-general/run-config.json')\n",
    "if config: display({k:v for k,v in config.items() if k!='exclusions'})\n",
    "else: missing('final general-purpose training configuration')\n",
    "print('Final adapter saved:',(SLM/'out/local-4b-general/final/adapter_model.safetensors').exists())\n",
    "for label in ('base-general-64','student-general-64','base-external-locomo-16','student-external-locomo-16'):\n",
    "    for kind in ('summary','semantic'):\n",
    "        result=read(GENERAL/'eval'/f'{label}-{kind}.json')\n",
    "        if result:\n",
    "            print(label,kind); display(result)\n",
    "        else: missing(f'{label} {kind} evaluation')\n",
    "acceptance=read(GENERAL/'final-acceptance.json')\n",
    "if acceptance: show_acceptance(acceptance)\n",
    "else: missing('general-purpose selected-model live extraction and graph acceptance')\n",
    "final_graph=GENERAL/'final-graph.png'\n",
    "if final_graph.exists():\n",
    "    display(Markdown('**Captured final graph UI:**'))\n",
    "    display(Image(filename=str(final_graph)))\n",
    "else: missing('final selected-model graph screenshot')\n",
    "release=read(GENERAL/'release-receipt.json')\n",
    "if release: display(release)\n",
    "else: missing('final Hugging Face release receipt')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8ac21620",
   "metadata": {},
   "source": [
    "## 12d. External conversation diagnostic\n",
    "\n",
    "To separate evaluation from our own scenario generator, we froze sixteen short windows from ten conversations in [LoCoMo](https://github.com/snap-research/locomo), an external machine-human conversation dataset described by [Maharana et al. (ACL 2024)](https://aclanthology.org/2024.acl-long.747/). The upstream license is CC-BY-NC-4.0. Raw data and derived labels remain local to this academic diagnostic and never enter training or the public model package.\n",
    "\n",
    "This tests short-window extraction on a different source distribution. It is **not the official LoCoMo QA benchmark**, a long-horizon retrieval test, or independently human-graded extraction accuracy. Our task labels and semantic review still use the same teacher family. Session wall times have no upstream timezone; the API's UTC suffix is only an adapter convention. See `reports/EXTERNAL_LOCOMO.md` for the frozen selection and attribution protocol."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "3bf36c01",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-10-08T17:00:10.723931Z",
     "iopub.status.busy": "2026-10-08T17:00:10.723809Z",
     "iopub.status.idle": "2026-10-08T17:00:10.726566Z",
     "shell.execute_reply": "2026-10-08T17:00:10.726023Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'selection_rule': 'SHA256 locomo-v1 conversation ordering; per-conversation hashed session ordering; contiguous six-turn hashed start; 400-2600 characters; round-robin across conversations; frozen before model outputs',\n",
       " 'source_sha256': '79fa87e90f04081343b8c8debecb80a9a6842b76a7aa537dc9fdf651ea698ff4',\n",
       " 'rows': 16,\n",
       " 'distinct_conversations': 10,\n",
       " 'distinct_sessions': 16,\n",
       " 'revision': '3eb6f2c585f5e1699204e3c3bdf7adc5c28cb376',\n",
       " 'source_url': 'https://raw.githubusercontent.com/snap-research/locomo/3eb6f2c585f5e1699204e3c3bdf7adc5c28cb376/data/locomo10.json',\n",
       " 'license': 'CC-BY-NC-4.0',\n",
       " 'local_academic_evaluation_only': True,\n",
       " 'training_use': False,\n",
       " 'official_locomo_qa_score': False,\n",
       " 'file_sha256': 'a7de115cfa8da3a164aad8c21359111cb87cca1dd2f058cf8a866c68e274dcf0',\n",
       " 'audit': 'All 16 labels structurally validated and same-family semantically audited; repaired labels retain original attempts. Not independent human gold.',\n",
       " 'timezone_note': 'Unzoned source session wall time normalized with Z solely to satisfy API; no claim of actual UTC timezone.',\n",
       " 'teacher': 'Pi teacher openai/gpt-5.6-sol: 58 calls, 0 cached, 123901 input, 30049 output tokens'}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "external=read(GENERAL/'external-locomo-manifest.json')\n",
    "if external: display(external)\n",
    "else: missing('completed external LoCoMo-window label/audit manifest')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5282743c",
   "metadata": {},
   "source": [
    "## Measured tradeoff and unresolved temporal errors\n",
    "\n",
    "The internal matched comparison improves consolidation from **1/16 to 13/16** and reflection from **10/16 to 15/16**. Extraction regresses: supported-claim fraction is **38/42 (90.5%) versus 58/63 (92.1%)** for the base, and reference coverage is **41/55 (74.5%) versus 50/55 (90.9%)**. Both models recover the eight annotated corrections. Student structural acceptance is **63/64**, including the preserved reflection-citation failure. These same-family teacher judgments describe a small held-out sample, not independent human accuracy.\n",
    "\n",
    "Manual inspection found substantive temporal failures: “tomorrow” on June 1 was normalized to June 3; “this Friday” from October 8 became October 17; and date-only events acquired invented midnight timestamps. **Temporal normalization and precision remain unreliable.** The later citation-enum repair does not fix these semantic errors. Keep the evidence and verify dates before relying on extracted timestamps.\n",
    "\n",
    "This is a multitask tradeoff with an extraction regression. External extraction also regresses: support is **22/26 (84.6%) versus 39/44 (88.6%)**, and reference coverage is **13/40 (32.5%) versus 16/40 (40.0%)**. Both models pass structural validation on all sixteen windows. This is not official LoCoMo QA accuracy. Final live acceptance is separate evidence; successful graph operations do not erase extraction errors."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "90d4f5d4",
   "metadata": {},
   "source": [
    "## 12e. Live reflection failure and deployment-only citation repair\n",
    "\n",
    "The first final-candidate live pilot passed the completed extraction, correction, relationship, derived-observation, projection and location-recall checks, then failed at reflection with HTTP 502. Runtime diagnosis identified invalid UUID citations. This failure remains in its original receipt; graph success alone did not satisfy acceptance.\n",
    "\n",
    "The repair constrains reflection citations to identifiers actually present in retrieved evidence. It changes the serving decoder, not the trained weights. The frozen baseline/student comparison retains its original Python reflection decoder for both models. Consequently those matched scores **do not measure the citation-enum repair**. A separate repeated live pilot and final acceptance receipt evaluate the deployment change. The second pilot also exposed a harness false negative: it searched the statement for a literal name instead of checking the resolved subject and quoted relationship evidence. A recorded re-audit of the same saved outputs corrects that check; it is not fresh inference. Both failed receipts remain preserved; their existence and outcomes are checked below. The frozen comparison is complete. Public inference now uses the deployed citation-enum contract; the previously failed row passed a separate one-attempt regression check."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "8999d550",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-10-08T17:00:10.727596Z",
     "iopub.status.busy": "2026-10-08T17:00:10.727490Z",
     "iopub.status.idle": "2026-10-08T17:00:10.775043Z",
     "shell.execute_reply": "2026-10-08T17:00:10.774588Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'scope': 'Both base and student matched held-out evaluation before production reflection citation enum fix',\n",
       " 'files': {'structured.py': 'ffb66026eb349c4fea9c5ff05d3de0760e449c940487e141c5315742359dd716',\n",
       "  'evaluate_research.py': '1a1e60c61a3639c89276b112f1921fc6ce5c69c0366f6d1c3800f509d826fadb',\n",
       "  'ollama_client.py': '733ad12a97b68fba838388a357681c39af636d9c939464a69fa74023610c8ad2',\n",
       "  'grade_research.py': 'f84f0f67491157c6b773e2dc8a0dd62afc338ec026e3bb443be5f986b2132348'},\n",
       " 'reflection_format': 'json, post-generation citation validation',\n",
       " 'production_followup': 'Constrain reflection citation IDs to evidence enum; separately verify live acceptance, not credited to frozen matched scores.'}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>receipt</th>\n",
       "      <th>passed</th>\n",
       "      <th>completed_checks</th>\n",
       "      <th>passed_checks</th>\n",
       "      <th>error</th>\n",
       "      <th>worker_model</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>candidate-live-acceptance.json</td>\n",
       "      <td>False</td>\n",
       "      <td>9</td>\n",
       "      <td>9</td>\n",
       "      <td>502 Server Error: Bad Gateway for url: http://...</td>\n",
       "      <td>ollama/mem-extractor:general-v1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                          receipt  passed  completed_checks  passed_checks  \\\n",
       "0  candidate-live-acceptance.json   False                 9              9   \n",
       "\n",
       "                                               error  \\\n",
       "0  502 Server Error: Bad Gateway for url: http://...   \n",
       "\n",
       "                      worker_model  \n",
       "0  ollama/mem-extractor:general-v1  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>receipt</th>\n",
       "      <th>passed</th>\n",
       "      <th>completed_checks</th>\n",
       "      <th>passed_checks</th>\n",
       "      <th>error</th>\n",
       "      <th>worker_model</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>candidate-live-acceptance-v2.json</td>\n",
       "      <td>False</td>\n",
       "      <td>16</td>\n",
       "      <td>15</td>\n",
       "      <td>None</td>\n",
       "      <td>ollama/mem-extractor:general-v1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                             receipt  passed  completed_checks  passed_checks  \\\n",
       "0  candidate-live-acceptance-v2.json   False                16             15   \n",
       "\n",
       "  error                     worker_model  \n",
       "0  None  ollama/mem-extractor:general-v1  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>receipt</th>\n",
       "      <th>passed</th>\n",
       "      <th>completed_checks</th>\n",
       "      <th>passed_checks</th>\n",
       "      <th>error</th>\n",
       "      <th>worker_model</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>candidate-live-acceptance-v3.json</td>\n",
       "      <td>True</td>\n",
       "      <td>16</td>\n",
       "      <td>16</td>\n",
       "      <td>None</td>\n",
       "      <td>ollama/mem-extractor:general-v1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                             receipt  passed  completed_checks  passed_checks  \\\n",
       "0  candidate-live-acceptance-v3.json    True                16             16   \n",
       "\n",
       "  error                     worker_model  \n",
       "0  None  ollama/mem-extractor:general-v1  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "{'method': 'Read-only assistant inspection against original event windows or consolidation/reflection facts, compared with baseline. Not human review, not a blinded independent benchmark. Reviewer previously worked on corpus assembly but did not change held-out data, prompts, decoder, or model in response to this review.',\n",
       " 'selection': 'Eight general examples across eight domains and two external examples selected before their student outputs were read; additional defects inspected opportunistically. This is a purposive diagnostic sample, not an accuracy estimate.',\n",
       " 'summary': 'Useful improvements in synthesis, correction links, exception handling and reflection coexist with extraction recall regression, incorrect date normalization, invented time precision, incomplete citations and occasional unsupported conclusions. Do not describe the student as uniformly better than the base.',\n",
       " 'prominent_limitations': ['Relative-date normalization is unreliable: June 1 tomorrow became June 3; October 8 this Friday became October 17.',\n",
       "  'Calendar-only events can receive invented midnight timestamps.',\n",
       "  'Constrained quote pairing does not guarantee complete evidence attribution or semantic correctness.',\n",
       "  'Some durable profile, relationship and plan facts are omitted; extraction reference coverage declined on the general holdout.',\n",
       "  'Correct reflection answers can fail UUID citation copying.',\n",
       "  'External casual-conversation review found both recovery of useful facts and an invented identity relationship in a photo claim.']}"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>scope</th>\n",
       "      <th>accepted</th>\n",
       "      <th>attempts</th>\n",
       "      <th>decoder_sha256</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Post-comparison citation decoder regression ch...</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>88a7c7cd1c8949f0406848c0b12235bd24de8f8f185dd8...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                               scope  accepted  attempts  \\\n",
       "0  Post-comparison citation decoder regression ch...      True         1   \n",
       "\n",
       "                                      decoder_sha256  \n",
       "0  88a7c7cd1c8949f0406848c0b12235bd24de8f8f185dd8...  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "This is a separate regression check on the previously failed row. The frozen reflection score remains 15/16; it is not retroactively changed to 16/16."
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "frozen_decoder=read(GENERAL/'frozen-comparison-decoder.json')\n",
    "if frozen_decoder: display(frozen_decoder)\n",
    "else: missing('frozen matched-decoder provenance')\n",
    "for filename in ('candidate-live-acceptance.json','candidate-live-acceptance-v2.json','candidate-live-acceptance-v3.json'):\n",
    "    pilot=read(GENERAL/filename)\n",
    "    if pilot:\n",
    "        show([{'receipt':filename,'passed':pilot.get('passed'),\n",
    "               'completed_checks':len(pilot.get('checks',{})),\n",
    "               'passed_checks':sum(v is True for v in pilot.get('checks',{}).values()),\n",
    "               'error':pilot.get('error'),\n",
    "               'worker_model':pilot.get('worker_model')}])\n",
    "    else: missing(filename)\n",
    "manual_review=read(GENERAL/'student-manual-review.json')\n",
    "if manual_review:\n",
    "    display({k:manual_review[k] for k in ('method','selection','summary','prominent_limitations') if k in manual_review})\n",
    "decoder_regression=read(GENERAL/'reflection-decoder-python-check.json')\n",
    "if decoder_regression:\n",
    "    show([{'scope':decoder_regression['scope'],'accepted':decoder_regression['accepted'],\n",
    "           'attempts':len(decoder_regression.get('attempts',[])),\n",
    "           'decoder_sha256':decoder_regression['structured_sha256']}])\n",
    "    display(Markdown('This is a separate regression check on the previously failed row. The frozen reflection score remains 15/16; it is not retroactively changed to 16/16.'))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "428e77be",
   "metadata": {},
   "source": [
    "## 13. Reproduce the experiment deliberately\n",
    "\n",
    "Use the repository's virtual environment. Clear ROS-injected `PYTHONPATH`. Corpus generation invokes a teacher and is an explicit, separate operation; credentials are not included in this notebook. The commands below are displayed, **not executed by Run All**.\n",
    "\n",
    "```bash\n",
    "cd slm-distill\n",
    "env -u PYTHONPATH .venv/bin/python train.py   --base Qwen/Qwen3-4B-Instruct-2507   --revision cdbee75f17c01a7cc42f958dc650907174af0554   --qlora --sets data/research-v3/continuation-sets   --adapter out/local-4b-general-stage-a/final --out out/local-4b-general   --rank 16 --accum 8 --epochs 1 --max-length 4096 --loss-chunk-size 256 --lr 3e-5 --eval-on-epoch\n",
    "```\n",
    "\n",
    "The command is an example; use the actual final `run-config.json` for exact curriculum and warm-start lineage. General-purpose construction is in `expand_general_corpus.py` and `audit_general_corpus.py`. Inspect `sample_public_corpus.py`, `research_corpus.py`, `expand_corpus.py`, `audit_corpus.py`, `adjudicate_audit.py` and `assemble_research.py` for corpus construction. `evaluate_research.py` records raw responses; `grade_research.py` grades them against evidence. `export.py` merges the adapter, converts to GGUF and installs an Ollama model.\n",
    "\n",
    "Refresh this report:\n",
    "\n",
    "```bash\n",
    "env -u PYTHONPATH slm-distill/.venv/bin/python slm-distill/build_notebook.py --execute\n",
    "```\n",
    "\n",
    "Large datasets, checkpoints and teacher caches are ignored by Git. Preserve them with their manifests when transferring the experiment. The executed notebook is a portable view of measurements, not a substitute for the underlying evidence."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "92002321",
   "metadata": {},
   "source": [
    "## 14. Claims we can defend, and remaining limitations\n",
    "\n",
    "- The experiment iterated from contract learning to naturalistic diagnosis, representation repair, a stronger base, broader data and audited labels.\n",
    "- Corpus quantity must be reported together with unique episodes, split strategy, provenance, exclusions and task correlation.\n",
    "- The initial synthetic/public mix supported only software-memory experiments. The frozen replacement corpus spans sixteen everyday/professional domains; it still does not establish generalization to arbitrary domains, languages or very long histories.\n",
    "- Teacher-generated targets and teacher grading are correlated. Hard validators establish structural grounding, not factual entailment.\n",
    "- Training fits an 8 GB GPU at the measured token cap. Serving a larger context does not establish learned long-context quality.\n",
    "- No final improvement or deployment claim is made until its corresponding saved evaluation and operational receipt exists.\n",
    "\n",
    "**Guide review:** inspect the failure examples, rejected label reasons, frozen split hashes, matched baseline comparison and live evidence path before interpreting aggregate scores."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "SDP distillation (.venv)",
   "language": "python",
   "name": "sdp-distill"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}