{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "84c36161",
   "metadata": {},
   "source": [
    "# AAAI Lab: Developing AI Agents for IT Automation Tasks with ITBench\n",
    "## Evaluating Agents"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7bababea",
   "metadata": {},
   "source": [
    "January 20, 2026\n",
    "\n",
    "Authors: Paulina Toro Isaza, Saurabh Jha"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "87f091bf",
   "metadata": {},
   "source": [
    "# Preliminaries"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "416f0c50",
   "metadata": {},
   "source": [
    "## Installation and Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "uhbkrofupp",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[2mAudited \u001b[1m7 packages\u001b[0m \u001b[2min 226ms\u001b[0m\u001b[0m\n"
     ]
    }
   ],
   "source": [
    "!cd /data/ITBench-SRE-Agent && /home/user/.local/bin/uv pip install matplotlib seaborn numpy pandas plotly tqdm pyyaml"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "e2ffea69",
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "import sys\n",
    "from pathlib import Path\n",
    "from dataclasses import dataclass, field\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "b0fc9634",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Add project root to path\n",
    "PROJECT_ROOT = Path(\"/data/ITBench-SRE-Agent\")\n",
    "sys.path.insert(0, str(PROJECT_ROOT))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7ef28e4a",
   "metadata": {},
   "source": [
    "## Paths and Global Variables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "2b300412",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Paths\n",
    "LEADERBOARD_DIR = PROJECT_ROOT / \"ITBench-Trajectories\" / \"ReAct-Agent-Trajectories\"\n",
    "OUTPUT_BASE_DIR = PROJECT_ROOT / \"ITBench-Trajectories\" / \"output\"\n",
    "\n",
    "# Minimum runs per scenario required for inclusion\n",
    "MIN_RUNS_PER_SCENARIO = 2\n",
    "\n",
    "# Minimum scenarios needed after filtering\n",
    "MIN_QUALIFYING_SCENARIOS = 20\n",
    "\n",
    "# Success threshold for binary classification\n",
    "SUCCESS_THRESHOLD = 0.5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "pz42i6nppa9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "✓ Created output directories at: /data/ITBench-SRE-Agent/ITBench-Trajectories/output\n"
     ]
    }
   ],
   "source": [
    "# Create all output directories upfront\n",
    "OUTPUT_BASE_DIR.mkdir(parents=True, exist_ok=True)\n",
    "(OUTPUT_BASE_DIR / \"consistency\").mkdir(parents=True, exist_ok=True)\n",
    "(OUTPUT_BASE_DIR / \"inferences\").mkdir(parents=True, exist_ok=True)\n",
    "(OUTPUT_BASE_DIR / \"tool_failures\").mkdir(parents=True, exist_ok=True)\n",
    "(OUTPUT_BASE_DIR / \"discovery\").mkdir(parents=True, exist_ok=True)\n",
    "\n",
    "print(f\"✓ Created output directories at: {OUTPUT_BASE_DIR}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1134e25a",
   "metadata": {},
   "source": [
    "## Data"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b2cb4841",
   "metadata": {},
   "source": [
    "To evaluate our agents, we will be using a variety of files:\n",
    "- The ground truth file for each scenario that includes the root cause entity as well as aliases\n",
    "- The individual trajectories for each trial that contain each turn of the agent, including input and output of each turn\n",
    "- The end-to-end evaluation done by an LLM-as-a-Judge"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7gpq7ct50cg",
   "metadata": {},
   "source": [
    "## Download Ground Truth Data\n",
    "\n",
    "The ground truth files contain the root cause entity information and aliases for each scenario."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "y3ffif24x",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of files in the repo is unreliable. Using `list_repo_tree` to ensure all files are listed.\n",
      "Downloading (incomplete total...): 0.00B [00:00, ?B/s]\n",
      "Fetching ... files: 35it [00:00, 7422.80it/s]\n",
      "Download complete: : 0.00B [00:00, ?B/s]              /data/ITBench-SRE-Agent/ITBench-Lite\n",
      "Download complete: : 0.00B [00:00, ?B/s]\n"
     ]
    }
   ],
   "source": [
    "!source /data/ITBench-SRE-Agent/.venv/bin/activate && hf download \\\n",
    "  ibm-research/ITBench-Lite \\\n",
    "  --repo-type dataset \\\n",
    "  --include \"snapshots/sre/v0.2-*/Scenario-*/ground_truth.yaml\" \\\n",
    "  --local-dir /data/ITBench-SRE-Agent/ITBench-Lite"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "nxza58xw7v",
   "metadata": {},
   "source": [
    "### Check Downloaded Ground Truth Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "lg601pti47f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "total 16K\n",
      "drwxr-sr-x. 37 user user 16K Jan 19 11:19 v0.2-B96DF826-4BB2-4B62-97AB-6D84254C53D7\n"
     ]
    }
   ],
   "source": [
    "!ls -lh /data/ITBench-SRE-Agent/ITBench-Lite/snapshots/sre/ | head -5"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8b47a303",
   "metadata": {},
   "source": [
    "# End-to-End Evaluations"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b8d516c1",
   "metadata": {},
   "source": [
    "## Accuracy"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "879b5e4e-71c6-4093-9c19-65908b4cbff5",
   "metadata": {},
   "source": [
    "**Problem**: As we are dealing with a system environment in which entities (pods, services, etc.) can have multiple aliases or different names at run time, there is no simple heuristic that can determine if the agent picked the right root cause(s). "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7e6b4af2",
   "metadata": {},
   "source": [
    "<details>\n",
    "  <summary><b>Solution:</b></summary>\n",
    "    Use an LLM-as-a-Judge to determine if the predicted root cause entity from the agent output matches any of the known aliases of the ground truth root cause entity. \n",
    "</details>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "50af6f88",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-block alert-info\">\n",
    "<b>Warning:</b> Never blindly trust that an LLM-as-a-Judge makes the right judgment! This is especially true for non-generic domains such as the IT domain. \n",
    "</div>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8343cba7-b455-43eb-9dc9-494ab7120c7e",
   "metadata": {},
   "source": [
    "Our LAAJ has been manually verified on a few dozen examples by subject matter experts for this lab, but a systemic approach with statistically significant sample sizes is always recommended."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "63a13ce6-3d7b-478c-a4cc-1bf6d9924ab2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2026-01-19 12:09:25,809 - itbench_evaluations - INFO - Loading ground truths from: ./ITBench-Lite/snapshots/sre/v0.2-B96DF826-4BB2-4B62-97AB-6D84254C53D7\n",
      "2026-01-19 12:09:25,809 - itbench_evaluations.loader - INFO - Scanning directory for scenario ground truths: ITBench-Lite/snapshots/sre/v0.2-B96DF826-4BB2-4B62-97AB-6D84254C53D7\n",
      "2026-01-19 12:09:26,242 - itbench_evaluations.loader - INFO - Loaded 35 ground truths from directory\n",
      "2026-01-19 12:09:26,242 - itbench_evaluations - INFO - Loaded 35 ground truth scenarios\n",
      "2026-01-19 12:09:26,242 - itbench_evaluations - INFO - Loading agent outputs from: ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,243 - itbench_evaluations.loader - INFO - Loaded 1 valid trial outputs for incident 1, found 0 bad runs\n",
      "2026-01-19 12:09:26,244 - itbench_evaluations.loader - WARNING - No directory found for incident 102 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,245 - itbench_evaluations.loader - INFO - Loaded 2 valid trial outputs for incident 105, found 0 bad runs\n",
      "2026-01-19 12:09:26,246 - itbench_evaluations.loader - WARNING - No directory found for incident 11 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,246 - itbench_evaluations.loader - WARNING - No directory found for incident 12 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,246 - itbench_evaluations.loader - WARNING - No directory found for incident 13 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,246 - itbench_evaluations.loader - WARNING - No directory found for incident 14 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,246 - itbench_evaluations.loader - WARNING - No directory found for incident 15 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,246 - itbench_evaluations.loader - WARNING - No directory found for incident 16 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,246 - itbench_evaluations.loader - WARNING - No directory found for incident 17 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,246 - itbench_evaluations.loader - WARNING - No directory found for incident 18 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,246 - itbench_evaluations.loader - WARNING - No directory found for incident 19 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,247 - itbench_evaluations.loader - WARNING - No directory found for incident 2 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,247 - itbench_evaluations.loader - WARNING - No directory found for incident 20 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,247 - itbench_evaluations.loader - WARNING - No directory found for incident 21 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,247 - itbench_evaluations.loader - WARNING - No directory found for incident 22 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,247 - itbench_evaluations.loader - WARNING - No directory found for incident 23 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,247 - itbench_evaluations.loader - WARNING - No directory found for incident 24 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,247 - itbench_evaluations.loader - WARNING - No directory found for incident 25 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,247 - itbench_evaluations.loader - WARNING - No directory found for incident 29 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,247 - itbench_evaluations.loader - WARNING - No directory found for incident 31 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,247 - itbench_evaluations.loader - WARNING - No directory found for incident 33 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,247 - itbench_evaluations.loader - WARNING - No directory found for incident 34 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,248 - itbench_evaluations.loader - WARNING - No directory found for incident 35 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,248 - itbench_evaluations.loader - WARNING - No directory found for incident 38 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,248 - itbench_evaluations.loader - WARNING - No directory found for incident 4 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,248 - itbench_evaluations.loader - WARNING - No directory found for incident 5 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,248 - itbench_evaluations.loader - WARNING - No directory found for incident 6 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,248 - itbench_evaluations.loader - WARNING - No directory found for incident 7 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,248 - itbench_evaluations.loader - WARNING - No directory found for incident 8 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,248 - itbench_evaluations.loader - WARNING - No directory found for incident 80 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,248 - itbench_evaluations.loader - WARNING - No directory found for incident 81 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,248 - itbench_evaluations.loader - WARNING - No directory found for incident 83 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,248 - itbench_evaluations.loader - WARNING - No directory found for incident 9 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,249 - itbench_evaluations.loader - WARNING - No directory found for incident 91 under ./outputs/agent_outputs\n",
      "2026-01-19 12:09:26,249 - itbench_evaluations - INFO - Loaded 3 trials across 2 incidents\n",
      "2026-01-19 12:09:26,249 - itbench_evaluations - INFO - Starting batch evaluation...\n",
      "2026-01-19 12:09:26,305 - itbench_evaluations.agent - INFO - LAAJ Agent initialized with model: google/gemini-2.5-pro\n",
      "2026-01-19 12:09:26,305 - itbench_evaluations.agent - INFO - Starting batch evaluation of 3 trials across 35 incidents (max_concurrent=5)\n",
      "2026-01-19 12:09:26,306 - itbench_evaluations.agent - INFO - Starting evaluation for incident 1, trial 1\n",
      "2026-01-19 12:09:26,308 - itbench_evaluations.agent - INFO - Starting evaluation for incident 105, trial 1\n",
      "2026-01-19 12:09:26,313 - itbench_evaluations.agent - INFO - Starting evaluation for incident 105, trial 2\n",
      "2026-01-19 12:09:29,183 - httpx - INFO - HTTP Request: POST https://openrouter.ai/api/v1/chat/completions \"HTTP/1.1 200 OK\"\n",
      "2026-01-19 12:09:29,285 - httpx - INFO - HTTP Request: POST https://openrouter.ai/api/v1/chat/completions \"HTTP/1.1 200 OK\"\n",
      "2026-01-19 12:09:29,399 - httpx - INFO - HTTP Request: POST https://openrouter.ai/api/v1/chat/completions \"HTTP/1.1 200 OK\"\n",
      "2026-01-19 12:10:09,728 - itbench_evaluations.agent - INFO - Computed entity@k metrics for k=[1, 2, 3, 4, 5]\n",
      "2026-01-19 12:10:09,728 - itbench_evaluations.agent - INFO - Successfully evaluated incident 1, trial 1\n",
      "2026-01-19 12:10:18,303 - itbench_evaluations.agent - INFO - Computed entity@k metrics for k=[1, 2, 3, 4, 5]\n",
      "2026-01-19 12:10:18,303 - itbench_evaluations.agent - INFO - Successfully evaluated incident 105, trial 1\n",
      "2026-01-19 12:10:52,503 - itbench_evaluations.agent - INFO - Computed entity@k metrics for k=[1, 2, 3, 4, 5]\n",
      "2026-01-19 12:10:52,503 - itbench_evaluations.agent - INFO - Successfully evaluated incident 105, trial 2\n",
      "2026-01-19 12:10:52,504 - itbench_evaluations.agent - INFO - Batch evaluation complete: 3 results\n",
      "2026-01-19 12:10:52,504 - itbench_evaluations - INFO - Calculating statistics...\n",
      "2026-01-19 12:10:52,539 - itbench_evaluations - INFO - Results written to: outputs/evaluation_results.json\n",
      "\n",
      "============================================================\n",
      "EVALUATION SUMMARY\n",
      "============================================================\n",
      "Incidents evaluated: 2\n",
      "Total trials: 3\n",
      "Bad runs: 0\n",
      "\n",
      "Overall Statistics:\n",
      "  root_cause_entity_precision: mean=0.1250\n",
      "  root_cause_entity_recall: mean=0.2500\n",
      "  root_cause_entity_f1: mean=0.1667\n",
      "  root_cause_entity_k_precision: mean=0.1250\n",
      "  root_cause_entity_k_recall: mean=0.2500\n",
      "  root_cause_entity_k_f1: mean=0.1667\n",
      "  root_cause_reasoning: mean=0.1250\n",
      "  propagation_chain: mean=0.3393\n",
      "  fault_localization_component_identification: mean=1.0000\n",
      "  root_cause_reasoning_partial: mean=0.5000\n",
      "  root_cause_proximity_no_fp_precision: mean=0.1458\n",
      "  root_cause_proximity_no_fp_recall: mean=0.3750\n",
      "  root_cause_proximity_no_fp_f1: mean=0.1842\n",
      "  root_cause_proximity_with_fp_precision: mean=0.2054\n",
      "  root_cause_proximity_with_fp_recall: mean=0.4500\n",
      "  root_cause_proximity_with_fp_f1: mean=0.2255\n",
      "============================================================\n"
     ]
    }
   ],
   "source": [
    "!cd ITBench-SRE-Agent/ && export JUDGE_BASE_URL=\"https://openrouter.ai/api/v1\" && export JUDGE_MODEL=\"google/gemini-2.5-pro\"  && export JUDGE_API_KEY=\"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\" && /home/user/.local/bin/uv run itbench-eval \\\n",
    "  --ground-truth ./ITBench-Lite/snapshots/sre/v0.2-B96DF826-4BB2-4B62-97AB-6D84254C53D7 \\\n",
    "  --outputs ./outputs/agent_outputs \\\n",
    "  --result-file ./outputs/evaluation_results.json"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "09i1edpm6gp8",
   "metadata": {},
   "source": [
    "## Download Trajectory Data\n",
    "\n",
    "Before running deeper evaluation, we need to download the trajectory data from previous agent runs from HuggingFace."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "46uhrcvf9un",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading (incomplete total...): 0.00B [00:00, ?B/s]\n",
      "Fetching 389 files:   0%|                               | 0/389 [00:00<?, ?it/s]\u001b[A\n",
      "Fetching 389 files:  16%|███▍                 | 63/389 [00:00<00:00, 613.16it/s]\u001b[A\n",
      "Fetching 389 files:  43%|████████▌           | 166/389 [00:00<00:00, 854.60it/s]\u001b[A\n",
      "Fetching 389 files:  68%|█████████████▋      | 266/389 [00:00<00:00, 913.55it/s]\u001b[A\n",
      "Fetching 389 files: 100%|████████████████████| 389/389 [00:00<00:00, 831.85it/s]\u001b[A\n",
      "Download complete: : 0.00B [00:00, ?B/s]              /data/ITBench-SRE-Agent/ITBench-Trajectories\n",
      "Download complete: : 0.00B [00:00, ?B/s]\n"
     ]
    }
   ],
   "source": [
    "!source /data/ITBench-SRE-Agent/.venv/bin/activate && hf download \\\n",
    "  ibm-research/ITBench-Trajectories \\\n",
    "  --repo-type dataset \\\n",
    "  --local-dir /data/ITBench-SRE-Agent/ITBench-Trajectories"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "osdoat6de7d",
   "metadata": {},
   "source": [
    "### Check Downloaded Trajectory Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "0228zciecll1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "total 8.0K\n",
      "-rw-r--r--. 1 user user 8.0K Jan 19 07:39 README.md\n",
      "drwxr-sr-x. 3 user user   33 Jan 19 07:39 ReAct-Agent-Trajectories\n",
      "drwxr-sr-x. 6 user user   81 Jan 19 07:39 output\n"
     ]
    }
   ],
   "source": [
    "!ls -lh /data/ITBench-SRE-Agent/ITBench-Trajectories/"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8f447261",
   "metadata": {},
   "source": [
    "#### Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "05fc8b32",
   "metadata": {},
   "outputs": [],
   "source": [
    "from analysis_src.extract_consistency_data import extract_all_data as extract_accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "facee07d",
   "metadata": {},
   "outputs": [],
   "source": [
    "OUTPUT_DIR = OUTPUT_BASE_DIR / \"consistency\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "20094953",
   "metadata": {},
   "source": [
    "#### By Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "a6023526",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 1 'react with code' agent directories:\n",
      "  - OpenAI-GPT-OSS-120B\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Reading agent data:   0%|          | 0/1 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Reading: OpenAI-GPT-OSS-120B\n",
      "  Processing: OpenAI-GPT-OSS-120B (21 scenarios with 3+ runs)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "  OpenAI-GPT-OSS-120B metrics:   0%|          | 0/6 [00:00<?, ?it/s]\u001b[A\n",
      "Reading agent data: 100%|██████████| 1/1 [00:00<00:00,  2.60it/s]   \u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "✓ Included 1 models: ['OpenAI-GPT-OSS-120B']\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "_, perf_df, scenario_df = extract_accuracy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "1b88ab3f",
   "metadata": {},
   "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>metric</th>\n",
       "      <th>metric_raw</th>\n",
       "      <th>performance</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>RC Entity F1</td>\n",
       "      <td>root_cause_entity_f1</td>\n",
       "      <td>0.164021</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>RC Entity Prec</td>\n",
       "      <td>root_cause_entity_precision</td>\n",
       "      <td>0.158730</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>RC Entity Rec</td>\n",
       "      <td>root_cause_entity_recall</td>\n",
       "      <td>0.174603</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>RC Proximity F1</td>\n",
       "      <td>root_cause_proximity_with_fp_f1</td>\n",
       "      <td>0.462661</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Prop. Chain</td>\n",
       "      <td>propagation_chain</td>\n",
       "      <td>0.397140</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Fault Loc.</td>\n",
       "      <td>fault_localization_component_identification</td>\n",
       "      <td>0.571429</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 model           metric  \\\n",
       "0  OpenAI-GPT-OSS-120B     RC Entity F1   \n",
       "1  OpenAI-GPT-OSS-120B   RC Entity Prec   \n",
       "2  OpenAI-GPT-OSS-120B    RC Entity Rec   \n",
       "3  OpenAI-GPT-OSS-120B  RC Proximity F1   \n",
       "4  OpenAI-GPT-OSS-120B      Prop. Chain   \n",
       "5  OpenAI-GPT-OSS-120B       Fault Loc.   \n",
       "\n",
       "                                    metric_raw  performance  \n",
       "0                         root_cause_entity_f1     0.164021  \n",
       "1                  root_cause_entity_precision     0.158730  \n",
       "2                     root_cause_entity_recall     0.174603  \n",
       "3              root_cause_proximity_with_fp_f1     0.462661  \n",
       "4                            propagation_chain     0.397140  \n",
       "5  fault_localization_component_identification     0.571429  "
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "perf_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "e6ada97d",
   "metadata": {},
   "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>performance</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>0.164021</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 model  performance\n",
       "0  OpenAI-GPT-OSS-120B     0.164021"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "perf_df[perf_df['metric'] == 'RC Entity F1'][['model', 'performance']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "65ab5b3c",
   "metadata": {},
   "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>performance</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>0.174603</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 model  performance\n",
       "2  OpenAI-GPT-OSS-120B     0.174603"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "perf_df[perf_df['metric'] == 'RC Entity Rec'][['model', 'performance']]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b22c20cc",
   "metadata": {},
   "source": [
    "#### By Scenario and Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "f36da129",
   "metadata": {},
   "outputs": [],
   "source": [
    "def add_scenario_num_column(df):\n",
    "    \"\"\"\n",
    "    Extract the numeric part from Scenario-X format and create a scenario_num column.\n",
    "    \n",
    "    Args:\n",
    "        df: pandas DataFrame with a 'scenario' column containing values like \"Scenario-1\"\n",
    "        \n",
    "    Returns:\n",
    "        DataFrame with a new 'scenario_num' column containing the extracted numbers\n",
    "    \"\"\"\n",
    "    df['scenario_num'] = df['scenario'].str.extract(r'Scenario-(\\d+)')[0].astype(int)\n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "cca2ecf7",
   "metadata": {},
   "outputs": [],
   "source": [
    "scenario_df = add_scenario_num_column(scenario_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "16a82c37",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1600x400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(16, 4))\n",
    "sns.barplot(scenario_df, x=\"scenario_num\", y=\"mean\", hue=\"model\")\n",
    "plt.xticks(rotation=45)\n",
    "plt.title(\"Scenario Performance by Model\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "566253ea",
   "metadata": {},
   "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>scenario</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>trials</th>\n",
       "      <th>is_flaky</th>\n",
       "      <th>scenario_num</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-1</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>[0.0, 0.0, 0.0]</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-102</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.577350</td>\n",
       "      <td>[0.0, 0.0, 1.0]</td>\n",
       "      <td>True</td>\n",
       "      <td>102</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-105</td>\n",
       "      <td>0.555556</td>\n",
       "      <td>0.509175</td>\n",
       "      <td>[1.0, 0.6666666666666666, 0.0]</td>\n",
       "      <td>True</td>\n",
       "      <td>105</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-11</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>[0.0, 0.0, 0.0]</td>\n",
       "      <td>False</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-12</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>[0.0, 0.0, 0.0]</td>\n",
       "      <td>False</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-13</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>[0.0, 0.0, 0.0]</td>\n",
       "      <td>False</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-15</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>[0.0, 0.0, 0.0]</td>\n",
       "      <td>False</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-17</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>[0.0, 0.0, 0.0]</td>\n",
       "      <td>False</td>\n",
       "      <td>17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-19</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>[0.0, 0.0, 0.0]</td>\n",
       "      <td>False</td>\n",
       "      <td>19</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-2</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>[0.0, 0.0, 0.0]</td>\n",
       "      <td>False</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-20</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.577350</td>\n",
       "      <td>[0.0, 0.0, 1.0]</td>\n",
       "      <td>True</td>\n",
       "      <td>20</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-21</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>[0.0, 0.0, 0.0]</td>\n",
       "      <td>False</td>\n",
       "      <td>21</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-22</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.577350</td>\n",
       "      <td>[0.0, 1.0, 0.0]</td>\n",
       "      <td>True</td>\n",
       "      <td>22</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-23</td>\n",
       "      <td>0.888889</td>\n",
       "      <td>0.192450</td>\n",
       "      <td>[1.0, 0.6666666666666666, 1.0]</td>\n",
       "      <td>False</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-24</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.577350</td>\n",
       "      <td>[0.0, 0.0, 1.0]</td>\n",
       "      <td>True</td>\n",
       "      <td>24</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-29</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.577350</td>\n",
       "      <td>[0.0, 0.0, 1.0]</td>\n",
       "      <td>True</td>\n",
       "      <td>29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-34</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>[0.0, 0.0, 0.0]</td>\n",
       "      <td>False</td>\n",
       "      <td>34</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-5</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>[0.0, 0.0, 0.0]</td>\n",
       "      <td>False</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-6</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>[0.0, 0.0, 0.0]</td>\n",
       "      <td>False</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-81</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.577350</td>\n",
       "      <td>[0.0, 0.0, 1.0]</td>\n",
       "      <td>True</td>\n",
       "      <td>81</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-83</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>[0.0, 0.0, 0.0]</td>\n",
       "      <td>False</td>\n",
       "      <td>83</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  model      scenario      mean       std  \\\n",
       "0   OpenAI-GPT-OSS-120B    Scenario-1  0.000000  0.000000   \n",
       "1   OpenAI-GPT-OSS-120B  Scenario-102  0.333333  0.577350   \n",
       "2   OpenAI-GPT-OSS-120B  Scenario-105  0.555556  0.509175   \n",
       "3   OpenAI-GPT-OSS-120B   Scenario-11  0.000000  0.000000   \n",
       "4   OpenAI-GPT-OSS-120B   Scenario-12  0.000000  0.000000   \n",
       "5   OpenAI-GPT-OSS-120B   Scenario-13  0.000000  0.000000   \n",
       "6   OpenAI-GPT-OSS-120B   Scenario-15  0.000000  0.000000   \n",
       "7   OpenAI-GPT-OSS-120B   Scenario-17  0.000000  0.000000   \n",
       "8   OpenAI-GPT-OSS-120B   Scenario-19  0.000000  0.000000   \n",
       "9   OpenAI-GPT-OSS-120B    Scenario-2  0.000000  0.000000   \n",
       "10  OpenAI-GPT-OSS-120B   Scenario-20  0.333333  0.577350   \n",
       "11  OpenAI-GPT-OSS-120B   Scenario-21  0.000000  0.000000   \n",
       "12  OpenAI-GPT-OSS-120B   Scenario-22  0.333333  0.577350   \n",
       "13  OpenAI-GPT-OSS-120B   Scenario-23  0.888889  0.192450   \n",
       "14  OpenAI-GPT-OSS-120B   Scenario-24  0.333333  0.577350   \n",
       "15  OpenAI-GPT-OSS-120B   Scenario-29  0.333333  0.577350   \n",
       "16  OpenAI-GPT-OSS-120B   Scenario-34  0.000000  0.000000   \n",
       "17  OpenAI-GPT-OSS-120B    Scenario-5  0.000000  0.000000   \n",
       "18  OpenAI-GPT-OSS-120B    Scenario-6  0.000000  0.000000   \n",
       "19  OpenAI-GPT-OSS-120B   Scenario-81  0.333333  0.577350   \n",
       "20  OpenAI-GPT-OSS-120B   Scenario-83  0.000000  0.000000   \n",
       "\n",
       "                            trials  is_flaky  scenario_num  \n",
       "0                  [0.0, 0.0, 0.0]     False             1  \n",
       "1                  [0.0, 0.0, 1.0]      True           102  \n",
       "2   [1.0, 0.6666666666666666, 0.0]      True           105  \n",
       "3                  [0.0, 0.0, 0.0]     False            11  \n",
       "4                  [0.0, 0.0, 0.0]     False            12  \n",
       "5                  [0.0, 0.0, 0.0]     False            13  \n",
       "6                  [0.0, 0.0, 0.0]     False            15  \n",
       "7                  [0.0, 0.0, 0.0]     False            17  \n",
       "8                  [0.0, 0.0, 0.0]     False            19  \n",
       "9                  [0.0, 0.0, 0.0]     False             2  \n",
       "10                 [0.0, 0.0, 1.0]      True            20  \n",
       "11                 [0.0, 0.0, 0.0]     False            21  \n",
       "12                 [0.0, 1.0, 0.0]      True            22  \n",
       "13  [1.0, 0.6666666666666666, 1.0]     False            23  \n",
       "14                 [0.0, 0.0, 1.0]      True            24  \n",
       "15                 [0.0, 0.0, 1.0]      True            29  \n",
       "16                 [0.0, 0.0, 0.0]     False            34  \n",
       "17                 [0.0, 0.0, 0.0]     False             5  \n",
       "18                 [0.0, 0.0, 0.0]     False             6  \n",
       "19                 [0.0, 0.0, 1.0]      True            81  \n",
       "20                 [0.0, 0.0, 0.0]     False            83  "
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scenario_df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4dab00f2",
   "metadata": {},
   "source": [
    "## Consistency"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4600b694",
   "metadata": {},
   "source": [
    "**Problem**: Because agents are backed by probabilistic LLMs, we cannot guarantee that agent behavior is consistent. "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3b388b5c",
   "metadata": {},
   "source": [
    "<details>\n",
    "  <summary><b>Solution:</b></summary>\n",
    "    Run multiple trials for each given scenario and determine how *consistent* the results are, not just how correct. Use metrics that reflect the distribution of trial performance, not just an average. \n",
    "</details>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "207ba269",
   "metadata": {},
   "source": [
    "### Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "fe76b9f6",
   "metadata": {},
   "outputs": [],
   "source": [
    "from analysis_src.extract_majority_vote_data import extract_all_data as extract_majority_vote, plot_pass_vs_majority"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "1843aaa0",
   "metadata": {},
   "outputs": [],
   "source": [
    "OUTPUT_DIR = OUTPUT_BASE_DIR / \"consistency\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e7497064",
   "metadata": {},
   "source": [
    "### Calculation"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c5358c6e",
   "metadata": {},
   "source": [
    "Our metrics for consistency will be Pass@$k$ and Majority@$k$.\n",
    "\n",
    "`Pass@$k$`: Given $k$ trials over $n$ scenarios, what proportion of scenarios had a least one successful trial whose F1 score was higher than the success threshold?\n",
    "\n",
    "`Majority@$k$`: Given $k$ trials over $n$ scenarios, what proportion of scenarios had a $majority$ of trials whose F1 scores were higher than the success threshold?\n",
    "\n",
    "We set `success_threshold` at `F1 = 0.5`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "e044c4ba",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 1 'react with code' agent directories:\n",
      "  - OpenAI-GPT-OSS-120B\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Reading agent data: 100%|██████████| 1/1 [00:00<00:00, 28.30it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Reading: OpenAI-GPT-OSS-120B\n",
      "  Processing: OpenAI-GPT-OSS-120B (34 scenarios)\n",
      "\n",
      "✓ Included 1 models: ['OpenAI-GPT-OSS-120B']\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Processing metrics:   0%|          | 0/3 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "--- Extracting for metric: F1 (root_cause_entity_f1) ---\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "  F1:   0%|          | 0/1 [00:00<?, ?it/s]\u001b[A\n",
      "                                           \u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "--- Extracting for metric: Precision (root_cause_entity_precision) ---\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "  Precision:   0%|          | 0/1 [00:00<?, ?it/s]\u001b[A\n",
      "                                                  \u001b[A"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "--- Extracting for metric: Recall (root_cause_entity_recall) ---\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "  Recall:   0%|          | 0/1 [00:00<?, ?it/s]\u001b[A\n",
      "Processing metrics: 100%|██████████| 3/3 [00:00<00:00, 41.48it/s]\n"
     ]
    }
   ],
   "source": [
    "majority_vote_results = extract_majority_vote()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "672ea55e",
   "metadata": {},
   "source": [
    "#### Overall Aggregate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "d6e00265",
   "metadata": {},
   "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>metric</th>\n",
       "      <th>n_scenarios</th>\n",
       "      <th>n_trials</th>\n",
       "      <th>pass_at_k</th>\n",
       "      <th>majority_at_k</th>\n",
       "      <th>all_at_k</th>\n",
       "      <th>consistent_correct</th>\n",
       "      <th>consistent_wrong</th>\n",
       "      <th>inconsistent</th>\n",
       "      <th>overall_mean</th>\n",
       "      <th>overall_std</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>34</td>\n",
       "      <td>2</td>\n",
       "      <td>0.147059</td>\n",
       "      <td>0.058824</td>\n",
       "      <td>0.058824</td>\n",
       "      <td>0.058824</td>\n",
       "      <td>0.852941</td>\n",
       "      <td>0.088235</td>\n",
       "      <td>0.088235</td>\n",
       "      <td>0.265793</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 model metric  n_scenarios  n_trials  pass_at_k  \\\n",
       "0  OpenAI-GPT-OSS-120B     F1           34         2   0.147059   \n",
       "\n",
       "   majority_at_k  all_at_k  consistent_correct  consistent_wrong  \\\n",
       "0       0.058824  0.058824            0.058824          0.852941   \n",
       "\n",
       "   inconsistent  overall_mean  overall_std  \n",
       "0      0.088235      0.088235     0.265793  "
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "majority_vote_results['F1'][0]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a50f0c4a",
   "metadata": {},
   "source": [
    "#### By Scenario"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "e48e34bc",
   "metadata": {},
   "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>metric</th>\n",
       "      <th>scenario</th>\n",
       "      <th>n_success</th>\n",
       "      <th>n_trials</th>\n",
       "      <th>majority_correct</th>\n",
       "      <th>consistency_type</th>\n",
       "      <th>mean_score</th>\n",
       "      <th>std_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-102</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-105</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>True</td>\n",
       "      <td>correct</td>\n",
       "      <td>0.833333</td>\n",
       "      <td>0.166667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-11</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-12</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-13</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-14</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-15</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-16</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-17</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-18</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-19</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-2</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-20</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-21</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-22</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>inconsistent</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-23</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>True</td>\n",
       "      <td>correct</td>\n",
       "      <td>0.833333</td>\n",
       "      <td>0.166667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-24</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-25</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>inconsistent</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-29</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-31</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-33</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>inconsistent</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-34</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-35</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-4</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-5</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-6</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-7</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-8</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-80</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-81</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-83</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-9</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>F1</td>\n",
       "      <td>Scenario-91</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>False</td>\n",
       "      <td>wrong</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  model metric      scenario  n_success  n_trials  \\\n",
       "0   OpenAI-GPT-OSS-120B     F1    Scenario-1          0         2   \n",
       "1   OpenAI-GPT-OSS-120B     F1  Scenario-102          0         2   \n",
       "2   OpenAI-GPT-OSS-120B     F1  Scenario-105          2         2   \n",
       "3   OpenAI-GPT-OSS-120B     F1   Scenario-11          0         2   \n",
       "4   OpenAI-GPT-OSS-120B     F1   Scenario-12          0         2   \n",
       "5   OpenAI-GPT-OSS-120B     F1   Scenario-13          0         2   \n",
       "6   OpenAI-GPT-OSS-120B     F1   Scenario-14          0         2   \n",
       "7   OpenAI-GPT-OSS-120B     F1   Scenario-15          0         2   \n",
       "8   OpenAI-GPT-OSS-120B     F1   Scenario-16          0         2   \n",
       "9   OpenAI-GPT-OSS-120B     F1   Scenario-17          0         2   \n",
       "10  OpenAI-GPT-OSS-120B     F1   Scenario-18          0         2   \n",
       "11  OpenAI-GPT-OSS-120B     F1   Scenario-19          0         2   \n",
       "12  OpenAI-GPT-OSS-120B     F1    Scenario-2          0         2   \n",
       "13  OpenAI-GPT-OSS-120B     F1   Scenario-20          0         2   \n",
       "14  OpenAI-GPT-OSS-120B     F1   Scenario-21          0         2   \n",
       "15  OpenAI-GPT-OSS-120B     F1   Scenario-22          1         2   \n",
       "16  OpenAI-GPT-OSS-120B     F1   Scenario-23          2         2   \n",
       "17  OpenAI-GPT-OSS-120B     F1   Scenario-24          0         2   \n",
       "18  OpenAI-GPT-OSS-120B     F1   Scenario-25          1         2   \n",
       "19  OpenAI-GPT-OSS-120B     F1   Scenario-29          0         2   \n",
       "20  OpenAI-GPT-OSS-120B     F1   Scenario-31          0         2   \n",
       "21  OpenAI-GPT-OSS-120B     F1   Scenario-33          1         2   \n",
       "22  OpenAI-GPT-OSS-120B     F1   Scenario-34          0         2   \n",
       "23  OpenAI-GPT-OSS-120B     F1   Scenario-35          0         2   \n",
       "24  OpenAI-GPT-OSS-120B     F1    Scenario-4          0         2   \n",
       "25  OpenAI-GPT-OSS-120B     F1    Scenario-5          0         2   \n",
       "26  OpenAI-GPT-OSS-120B     F1    Scenario-6          0         2   \n",
       "27  OpenAI-GPT-OSS-120B     F1    Scenario-7          0         2   \n",
       "28  OpenAI-GPT-OSS-120B     F1    Scenario-8          0         2   \n",
       "29  OpenAI-GPT-OSS-120B     F1   Scenario-80          0         2   \n",
       "30  OpenAI-GPT-OSS-120B     F1   Scenario-81          0         2   \n",
       "31  OpenAI-GPT-OSS-120B     F1   Scenario-83          0         2   \n",
       "32  OpenAI-GPT-OSS-120B     F1    Scenario-9          0         2   \n",
       "33  OpenAI-GPT-OSS-120B     F1   Scenario-91          0         2   \n",
       "\n",
       "    majority_correct consistency_type  mean_score  std_score  \n",
       "0              False            wrong    0.000000   0.000000  \n",
       "1              False            wrong    0.000000   0.000000  \n",
       "2               True          correct    0.833333   0.166667  \n",
       "3              False            wrong    0.000000   0.000000  \n",
       "4              False            wrong    0.000000   0.000000  \n",
       "5              False            wrong    0.000000   0.000000  \n",
       "6              False            wrong    0.000000   0.000000  \n",
       "7              False            wrong    0.000000   0.000000  \n",
       "8              False            wrong    0.000000   0.000000  \n",
       "9              False            wrong    0.000000   0.000000  \n",
       "10             False            wrong    0.000000   0.000000  \n",
       "11             False            wrong    0.000000   0.000000  \n",
       "12             False            wrong    0.000000   0.000000  \n",
       "13             False            wrong    0.000000   0.000000  \n",
       "14             False            wrong    0.000000   0.000000  \n",
       "15             False     inconsistent    0.500000   0.500000  \n",
       "16              True          correct    0.833333   0.166667  \n",
       "17             False            wrong    0.000000   0.000000  \n",
       "18             False     inconsistent    0.500000   0.500000  \n",
       "19             False            wrong    0.000000   0.000000  \n",
       "20             False            wrong    0.000000   0.000000  \n",
       "21             False     inconsistent    0.333333   0.333333  \n",
       "22             False            wrong    0.000000   0.000000  \n",
       "23             False            wrong    0.000000   0.000000  \n",
       "24             False            wrong    0.000000   0.000000  \n",
       "25             False            wrong    0.000000   0.000000  \n",
       "26             False            wrong    0.000000   0.000000  \n",
       "27             False            wrong    0.000000   0.000000  \n",
       "28             False            wrong    0.000000   0.000000  \n",
       "29             False            wrong    0.000000   0.000000  \n",
       "30             False            wrong    0.000000   0.000000  \n",
       "31             False            wrong    0.000000   0.000000  \n",
       "32             False            wrong    0.000000   0.000000  \n",
       "33             False            wrong    0.000000   0.000000  "
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "majority_vote_results['F1'][1]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "88514245",
   "metadata": {},
   "source": [
    "#### Pass@k vs. Majority@k"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "9022aa67",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 325x325 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saved: fig_pass_vs_majority.png\n"
     ]
    }
   ],
   "source": [
    "plot_pass_vs_majority(majority_vote_results['F1'][0])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6cb24da9",
   "metadata": {},
   "source": [
    "# Trajectory Level Evaluations"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "213fb8c8",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-block alert-info\">\n",
    "<b>The first step in evaluating agents is parsing trajectory output into structured data for your downstream analysis. \n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7314a1ec",
   "metadata": {},
   "source": [
    "How you parse the trajectories will depend on the kind of analyses you want to run. The analyses below focus on different aspects of the trajectories."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a3553717",
   "metadata": {},
   "source": [
    "## Inferences "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f59a95cd",
   "metadata": {},
   "source": [
    "**Problem:** One of the most basic ways of analyzing agents is by looking at the number and distribution of inferences they make. However, it is not clear by just these numbers how this behavior aligns with performance."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4c0de3ac",
   "metadata": {},
   "source": [
    "<details>\n",
    "  <summary><b>Solution:</b></summary>\n",
    "    Analyze inference (turn) numbers in the context of end-to-end performance with a simple scatter-plot."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4108256f",
   "metadata": {},
   "source": [
    "### Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "1c9b1600",
   "metadata": {},
   "outputs": [],
   "source": [
    "from analysis_src.extract_inference_data import extract_all_data as extract_inference_data, plot_inference_vs_performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "4167b853",
   "metadata": {},
   "outputs": [],
   "source": [
    "OUTPUT_DIR = OUTPUT_BASE_DIR / \"inferences\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "51cc4085",
   "metadata": {},
   "source": [
    "### Parsing Trajectories"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "id": "7299959a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 1 'react with code' agent directories\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Processing agents:   0%|          | 0/1 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Reading: OpenAI-GPT-OSS-120B\n",
      "  Processing: OpenAI-GPT-OSS-120B (35 scenarios)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "  OpenAI-GPT-OSS-120B scenarios:   0%|          | 0/35 [00:00<?, ?it/s]\u001b[A\n",
      "Processing agents: 100%|██████████| 1/1 [00:01<00:00,  1.07s/it]       \u001b[A\n"
     ]
    }
   ],
   "source": [
    "inference_summary_df, inference_details_df = extract_inference_data()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "853782be",
   "metadata": {},
   "source": [
    "##### Aggregate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "id": "d76939e2",
   "metadata": {},
   "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_scenarios</th>\n",
       "      <th>n_trials</th>\n",
       "      <th>avg_inference_count</th>\n",
       "      <th>std_inference_count</th>\n",
       "      <th>avg_input_tokens</th>\n",
       "      <th>avg_cached_tokens</th>\n",
       "      <th>avg_output_tokens</th>\n",
       "      <th>avg_total_tokens</th>\n",
       "      <th>total_inference_count</th>\n",
       "      <th>total_tokens</th>\n",
       "      <th>avg_tool_call_count</th>\n",
       "      <th>total_tool_calls</th>\n",
       "      <th>avg_code_execution_count</th>\n",
       "      <th>total_code_executions</th>\n",
       "      <th>top_tools</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>35</td>\n",
       "      <td>105</td>\n",
       "      <td>95.133333</td>\n",
       "      <td>46.630067</td>\n",
       "      <td>26763.333333</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3209.580952</td>\n",
       "      <td>29972.914286</td>\n",
       "      <td>9989</td>\n",
       "      <td>3147156</td>\n",
       "      <td>27.504762</td>\n",
       "      <td>2888</td>\n",
       "      <td>25.07619</td>\n",
       "      <td>2633</td>\n",
       "      <td>{'shell': 2376, 'apply_patch': 257, 'mcp__sre_...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 model  n_scenarios  n_trials  avg_inference_count  \\\n",
       "0  OpenAI-GPT-OSS-120B           35       105            95.133333   \n",
       "\n",
       "   std_inference_count  avg_input_tokens  avg_cached_tokens  \\\n",
       "0            46.630067      26763.333333                0.0   \n",
       "\n",
       "   avg_output_tokens  avg_total_tokens  total_inference_count  total_tokens  \\\n",
       "0        3209.580952      29972.914286                   9989       3147156   \n",
       "\n",
       "   avg_tool_call_count  total_tool_calls  avg_code_execution_count  \\\n",
       "0            27.504762              2888                  25.07619   \n",
       "\n",
       "   total_code_executions                                          top_tools  \n",
       "0                   2633  {'shell': 2376, 'apply_patch': 257, 'mcp__sre_...  "
      ]
     },
     "execution_count": 71,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "inference_summary_df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c8963f15",
   "metadata": {},
   "source": [
    "##### By Scenario"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "id": "5cf9bfdc",
   "metadata": {},
   "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>scenario</th>\n",
       "      <th>inference_count</th>\n",
       "      <th>input_tokens</th>\n",
       "      <th>cached_input_tokens</th>\n",
       "      <th>output_tokens</th>\n",
       "      <th>total_tokens</th>\n",
       "      <th>tool_call_count</th>\n",
       "      <th>code_execution_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-1</td>\n",
       "      <td>58</td>\n",
       "      <td>26284</td>\n",
       "      <td>0</td>\n",
       "      <td>3003</td>\n",
       "      <td>29287</td>\n",
       "      <td>21</td>\n",
       "      <td>20</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-1</td>\n",
       "      <td>126</td>\n",
       "      <td>27059</td>\n",
       "      <td>0</td>\n",
       "      <td>4495</td>\n",
       "      <td>31554</td>\n",
       "      <td>37</td>\n",
       "      <td>30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-1</td>\n",
       "      <td>86</td>\n",
       "      <td>23518</td>\n",
       "      <td>0</td>\n",
       "      <td>2203</td>\n",
       "      <td>25721</td>\n",
       "      <td>24</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-102</td>\n",
       "      <td>110</td>\n",
       "      <td>17624</td>\n",
       "      <td>0</td>\n",
       "      <td>1424</td>\n",
       "      <td>19048</td>\n",
       "      <td>23</td>\n",
       "      <td>19</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-102</td>\n",
       "      <td>116</td>\n",
       "      <td>26039</td>\n",
       "      <td>0</td>\n",
       "      <td>958</td>\n",
       "      <td>26997</td>\n",
       "      <td>37</td>\n",
       "      <td>25</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-9</td>\n",
       "      <td>92</td>\n",
       "      <td>22289</td>\n",
       "      <td>0</td>\n",
       "      <td>2384</td>\n",
       "      <td>24673</td>\n",
       "      <td>27</td>\n",
       "      <td>27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>101</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-9</td>\n",
       "      <td>32</td>\n",
       "      <td>15850</td>\n",
       "      <td>0</td>\n",
       "      <td>2550</td>\n",
       "      <td>18400</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>102</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-91</td>\n",
       "      <td>105</td>\n",
       "      <td>42046</td>\n",
       "      <td>0</td>\n",
       "      <td>3199</td>\n",
       "      <td>45245</td>\n",
       "      <td>28</td>\n",
       "      <td>21</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>103</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-91</td>\n",
       "      <td>141</td>\n",
       "      <td>36222</td>\n",
       "      <td>0</td>\n",
       "      <td>4935</td>\n",
       "      <td>41157</td>\n",
       "      <td>43</td>\n",
       "      <td>42</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>104</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-91</td>\n",
       "      <td>21</td>\n",
       "      <td>14960</td>\n",
       "      <td>0</td>\n",
       "      <td>15</td>\n",
       "      <td>14975</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>105 rows × 9 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                   model      scenario  inference_count  input_tokens  \\\n",
       "0    OpenAI-GPT-OSS-120B    Scenario-1               58         26284   \n",
       "1    OpenAI-GPT-OSS-120B    Scenario-1              126         27059   \n",
       "2    OpenAI-GPT-OSS-120B    Scenario-1               86         23518   \n",
       "3    OpenAI-GPT-OSS-120B  Scenario-102              110         17624   \n",
       "4    OpenAI-GPT-OSS-120B  Scenario-102              116         26039   \n",
       "..                   ...           ...              ...           ...   \n",
       "100  OpenAI-GPT-OSS-120B    Scenario-9               92         22289   \n",
       "101  OpenAI-GPT-OSS-120B    Scenario-9               32         15850   \n",
       "102  OpenAI-GPT-OSS-120B   Scenario-91              105         42046   \n",
       "103  OpenAI-GPT-OSS-120B   Scenario-91              141         36222   \n",
       "104  OpenAI-GPT-OSS-120B   Scenario-91               21         14960   \n",
       "\n",
       "     cached_input_tokens  output_tokens  total_tokens  tool_call_count  \\\n",
       "0                      0           3003         29287               21   \n",
       "1                      0           4495         31554               37   \n",
       "2                      0           2203         25721               24   \n",
       "3                      0           1424         19048               23   \n",
       "4                      0            958         26997               37   \n",
       "..                   ...            ...           ...              ...   \n",
       "100                    0           2384         24673               27   \n",
       "101                    0           2550         18400                2   \n",
       "102                    0           3199         45245               28   \n",
       "103                    0           4935         41157               43   \n",
       "104                    0             15         14975                0   \n",
       "\n",
       "     code_execution_count  \n",
       "0                      20  \n",
       "1                      30  \n",
       "2                      23  \n",
       "3                      19  \n",
       "4                      25  \n",
       "..                    ...  \n",
       "100                    27  \n",
       "101                     1  \n",
       "102                    21  \n",
       "103                    42  \n",
       "104                     0  \n",
       "\n",
       "[105 rows x 9 columns]"
      ]
     },
     "execution_count": 72,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "inference_details_df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "860432a8",
   "metadata": {},
   "source": [
    "### Inference Requests vs. Performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "f835f595",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Skipping inference vs performance: no performance data\n"
     ]
    }
   ],
   "source": [
    "plot_inference_vs_performance(inference_summary_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d27f1c99",
   "metadata": {},
   "source": [
    "## Tool Calling"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "06314bec",
   "metadata": {},
   "source": [
    "**Problem**: Agent behaviors of planning and tool usage are tightly coupled, making it difficult to know if low performance comes from low planning abilities and/or low tool usage abilities."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "44d2024f",
   "metadata": {},
   "source": [
    "<details>\n",
    "  <summary><b>Solution:</b></summary>\n",
    "    For now, we will simply consider tool failures: calls to tools (typically generated by an LLM) that resulted in some execution error. We can think of this as a minimum capability of the agent. If it can't use tools correctly, it will fail no matter how well it plans and even if it picks the right tool to use.<p>\n",
    "    This type of analysis is generic: some form can applied to any agentic system that uses tool calling. However, the evaluation for each individual tool will be specific to that tool as we will see below.\n",
    "</details>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7e39512b",
   "metadata": {},
   "source": [
    "### Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "f54cca55",
   "metadata": {},
   "outputs": [],
   "source": [
    "from analysis_src.extract_inference_data import plot_tool_breakdown_heatmap\n",
    "from analysis_src.extract_tool_failures import extract_all_data as extract_tool_failures, plot_failure_rate_by_model, plot_failure_categories_stacked"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "id": "030b36e1",
   "metadata": {},
   "outputs": [],
   "source": [
    "OUTPUT_DIR = OUTPUT_BASE_DIR / \"tool_failures\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "730af422",
   "metadata": {},
   "source": [
    "### Tool Call Distribution"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5169c83c",
   "metadata": {},
   "source": [
    "\n",
    "| Tool | LLM / Deterministic Input | General / Domain |\n",
    "|------|---------------------------|------------------|\n",
    "| update plan | LLM | General |\n",
    "| apply patch | LLM | General |\n",
    "| shell - bash | LLM | General |\n",
    "| shell - python | LLM | General |\n",
    "| metric analysis | Deterministic | SRE |\n",
    "| log analysis | Deterministic | SRE |\n",
    "| alert analysis | Deterministic | SRE |\n",
    "| event analysis | Determinstic | SRE |\n",
    "| alert summary | Determinstic | SRE |"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "890c50a4",
   "metadata": {},
   "source": [
    "Let's look at the distribution of tool calls per model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "id": "0c11333d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 975x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saved: fig_tool_usage_heatmap.png\n"
     ]
    }
   ],
   "source": [
    "plot_tool_breakdown_heatmap(inference_summary_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7a60e809",
   "metadata": {},
   "source": [
    "### Tool Failures"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "id": "cab5bb61",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 1 'react with code' agent directories\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Processing agents:   0%|          | 0/1 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Processing: OpenAI-GPT-OSS-120B\n",
      "  Processing: OpenAI-GPT-OSS-120B (35 scenarios)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "  OpenAI-GPT-OSS-120B scenarios:   0%|          | 0/35 [00:00<?, ?it/s]\u001b[A\n",
      "Processing agents: 100%|██████████| 1/1 [00:00<00:00,  1.42it/s]       \u001b[A\n"
     ]
    }
   ],
   "source": [
    "tool_summary_df, tool_detail_df, tool_failures_df = extract_tool_failures()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c2fc4b25",
   "metadata": {},
   "source": [
    "#### Overall Aggregate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "id": "7c54555d",
   "metadata": {},
   "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_scenarios</th>\n",
       "      <th>n_trials</th>\n",
       "      <th>total_tool_calls</th>\n",
       "      <th>total_failed_calls</th>\n",
       "      <th>avg_tool_calls_per_trial</th>\n",
       "      <th>avg_failed_calls_per_trial</th>\n",
       "      <th>avg_failure_rate_pct</th>\n",
       "      <th>std_failure_rate_pct</th>\n",
       "      <th>failure_categories</th>\n",
       "      <th>tool_failure_rates</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>35</td>\n",
       "      <td>105</td>\n",
       "      <td>2888</td>\n",
       "      <td>1257</td>\n",
       "      <td>27.504762</td>\n",
       "      <td>11.971429</td>\n",
       "      <td>35.263406</td>\n",
       "      <td>27.182595</td>\n",
       "      <td>{'timeout': 680, 'other': 327, 'file_not_found...</td>\n",
       "      <td>{'shell': {'total': 2376, 'failures': 1210, 'r...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 model  n_scenarios  n_trials  total_tool_calls  \\\n",
       "0  OpenAI-GPT-OSS-120B           35       105              2888   \n",
       "\n",
       "   total_failed_calls  avg_tool_calls_per_trial  avg_failed_calls_per_trial  \\\n",
       "0                1257                 27.504762                   11.971429   \n",
       "\n",
       "   avg_failure_rate_pct  std_failure_rate_pct  \\\n",
       "0             35.263406             27.182595   \n",
       "\n",
       "                                  failure_categories  \\\n",
       "0  {'timeout': 680, 'other': 327, 'file_not_found...   \n",
       "\n",
       "                                  tool_failure_rates  \n",
       "0  {'shell': {'total': 2376, 'failures': 1210, 'r...  "
      ]
     },
     "execution_count": 78,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tool_summary_df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cc5e5e05",
   "metadata": {},
   "source": [
    "#### By Scenario"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "id": "addf1471",
   "metadata": {},
   "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>scenario</th>\n",
       "      <th>trial</th>\n",
       "      <th>total_tool_calls</th>\n",
       "      <th>failed_tool_calls</th>\n",
       "      <th>failure_rate_pct</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-1</td>\n",
       "      <td>0</td>\n",
       "      <td>21</td>\n",
       "      <td>7</td>\n",
       "      <td>33.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-1</td>\n",
       "      <td>1</td>\n",
       "      <td>37</td>\n",
       "      <td>15</td>\n",
       "      <td>40.540541</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-1</td>\n",
       "      <td>2</td>\n",
       "      <td>24</td>\n",
       "      <td>15</td>\n",
       "      <td>62.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-102</td>\n",
       "      <td>0</td>\n",
       "      <td>23</td>\n",
       "      <td>18</td>\n",
       "      <td>78.260870</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-102</td>\n",
       "      <td>1</td>\n",
       "      <td>37</td>\n",
       "      <td>9</td>\n",
       "      <td>24.324324</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-9</td>\n",
       "      <td>1</td>\n",
       "      <td>27</td>\n",
       "      <td>16</td>\n",
       "      <td>59.259259</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>101</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-9</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>102</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-91</td>\n",
       "      <td>0</td>\n",
       "      <td>28</td>\n",
       "      <td>4</td>\n",
       "      <td>14.285714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>103</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-91</td>\n",
       "      <td>1</td>\n",
       "      <td>43</td>\n",
       "      <td>21</td>\n",
       "      <td>48.837209</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>104</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-91</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>105 rows × 6 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                   model      scenario  trial  total_tool_calls  \\\n",
       "0    OpenAI-GPT-OSS-120B    Scenario-1      0                21   \n",
       "1    OpenAI-GPT-OSS-120B    Scenario-1      1                37   \n",
       "2    OpenAI-GPT-OSS-120B    Scenario-1      2                24   \n",
       "3    OpenAI-GPT-OSS-120B  Scenario-102      0                23   \n",
       "4    OpenAI-GPT-OSS-120B  Scenario-102      1                37   \n",
       "..                   ...           ...    ...               ...   \n",
       "100  OpenAI-GPT-OSS-120B    Scenario-9      1                27   \n",
       "101  OpenAI-GPT-OSS-120B    Scenario-9      2                 2   \n",
       "102  OpenAI-GPT-OSS-120B   Scenario-91      0                28   \n",
       "103  OpenAI-GPT-OSS-120B   Scenario-91      1                43   \n",
       "104  OpenAI-GPT-OSS-120B   Scenario-91      2                 0   \n",
       "\n",
       "     failed_tool_calls  failure_rate_pct  \n",
       "0                    7         33.333333  \n",
       "1                   15         40.540541  \n",
       "2                   15         62.500000  \n",
       "3                   18         78.260870  \n",
       "4                    9         24.324324  \n",
       "..                 ...               ...  \n",
       "100                 16         59.259259  \n",
       "101                  0          0.000000  \n",
       "102                  4         14.285714  \n",
       "103                 21         48.837209  \n",
       "104                  0          0.000000  \n",
       "\n",
       "[105 rows x 6 columns]"
      ]
     },
     "execution_count": 79,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tool_detail_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "id": "4835b16f",
   "metadata": {},
   "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>scenario</th>\n",
       "      <th>trial</th>\n",
       "      <th>tool_name</th>\n",
       "      <th>category</th>\n",
       "      <th>error</th>\n",
       "      <th>exit_code</th>\n",
       "      <th>output_snippet</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-1</td>\n",
       "      <td>0</td>\n",
       "      <td>shell</td>\n",
       "      <td>timeout</td>\n",
       "      <td>timed out</td>\n",
       "      <td>124.0</td>\n",
       "      <td>command timed out after 10152 milliseconds\\n</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-1</td>\n",
       "      <td>0</td>\n",
       "      <td>shell</td>\n",
       "      <td>other</td>\n",
       "      <td>Unknown Error</td>\n",
       "      <td>NaN</td>\n",
       "      <td>failed to parse function arguments: Error(\"EOF...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-1</td>\n",
       "      <td>0</td>\n",
       "      <td>shell</td>\n",
       "      <td>other</td>\n",
       "      <td>Unknown Error</td>\n",
       "      <td>NaN</td>\n",
       "      <td>failed to parse function arguments: Error(\"EOF...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-1</td>\n",
       "      <td>0</td>\n",
       "      <td>shell</td>\n",
       "      <td>other</td>\n",
       "      <td>Unknown Error</td>\n",
       "      <td>NaN</td>\n",
       "      <td>failed to parse function arguments: Error(\"EOF...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>Scenario-1</td>\n",
       "      <td>0</td>\n",
       "      <td>shell</td>\n",
       "      <td>file_not_found</td>\n",
       "      <td>No such file or directory</td>\n",
       "      <td>1.0</td>\n",
       "      <td>cat: alerts/alert-2023-10-26T09-00-00Z.json: N...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 model    scenario  trial tool_name        category  \\\n",
       "0  OpenAI-GPT-OSS-120B  Scenario-1      0     shell         timeout   \n",
       "1  OpenAI-GPT-OSS-120B  Scenario-1      0     shell           other   \n",
       "2  OpenAI-GPT-OSS-120B  Scenario-1      0     shell           other   \n",
       "3  OpenAI-GPT-OSS-120B  Scenario-1      0     shell           other   \n",
       "4  OpenAI-GPT-OSS-120B  Scenario-1      0     shell  file_not_found   \n",
       "\n",
       "                       error  exit_code  \\\n",
       "0                  timed out      124.0   \n",
       "1              Unknown Error        NaN   \n",
       "2              Unknown Error        NaN   \n",
       "3              Unknown Error        NaN   \n",
       "4  No such file or directory        1.0   \n",
       "\n",
       "                                      output_snippet  \n",
       "0       command timed out after 10152 milliseconds\\n  \n",
       "1  failed to parse function arguments: Error(\"EOF...  \n",
       "2  failed to parse function arguments: Error(\"EOF...  \n",
       "3  failed to parse function arguments: Error(\"EOF...  \n",
       "4  cat: alerts/alert-2023-10-26T09-00-00Z.json: N...  "
      ]
     },
     "execution_count": 81,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tool_failures_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1ed0f040",
   "metadata": {},
   "source": [
    "#### Average Failure Rate by Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "id": "6eec6bad",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1012.5x450 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saved: fig_failure_rate_by_model.png\n"
     ]
    }
   ],
   "source": [
    "plot_failure_rate_by_model(tool_summary_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9a7b43c8",
   "metadata": {},
   "source": [
    "#### Tool Failure Distribution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "id": "8dbb8bfb",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0    {'timeout': 680, 'other': 327, 'file_not_found...\n",
      "Name: failure_categories, dtype: object\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1012.5x450 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saved: fig_failure_categories_stacked.png\n"
     ]
    }
   ],
   "source": [
    "plot_failure_categories_stacked(tool_summary_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a4d869e7",
   "metadata": {},
   "source": [
    "## Planning"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7e2679eb",
   "metadata": {},
   "source": [
    "**Problem**: How can we properly evaluate the planning abilities of LLMs in this task if there are no ground truth labels for planning turns in the trajectory?"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fe2ad88b",
   "metadata": {},
   "source": [
    "<details>\n",
    "  <summary><b>Solution:</b></summary>\n",
    "    We can leverage the ground truth we do have: the root cause entity. Since we know that the agent is ultimately exploring over a list of entities, we can use the ground truth entity to determine how \"close\" the agent gets in its planning.<p>\n",
    "    Such a specific evaluation metric is not immediately obvious. It helps to think: What needs to happen in order for any of the next turns to get closer to the final answer?<p>\n",
    "    Alternatively, you can also work your way backwards. What had to have happened for this particular turn to have led to the right next step?<p>\n",
    "</details>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1602ef84",
   "metadata": {},
   "source": [
    "### Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "id": "cdbfb063",
   "metadata": {},
   "outputs": [],
   "source": [
    "from analysis_src.extract_discovery_trajectory import extract_all_data as extract_discovery_data, plot_pipeline_funnel"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "4126cf3d",
   "metadata": {},
   "outputs": [],
   "source": [
    "OUTPUT_DIR = OUTPUT_BASE_DIR / \"discovery\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8d73417c",
   "metadata": {},
   "source": [
    "### Discovery"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "46d54569",
   "metadata": {},
   "source": [
    "Using some simple heuristics matching over entity aliases, we can search the individual turns of each trial for the appearance of the root cause (RC) entity. \n",
    "\n",
    "There are only a few possiblities:\n",
    "- RC never seen: The root cause entity was never explored or mentioned. This is a severe planning error, although we need to dig deeper to consider the severity of the error.\n",
    "- RC seen but not queried: The root cause entity was seen in the data (such as logs), but the agent never decided to investigate further.\n",
    "- RC queried, not asserted: The root cause entity was investigated but for some reason the agent decided it was not a contributing factor.\n",
    "- RC asserted, not in output: The root cause entity was investigated and determined to be a contributing factor, but for some reason this information did not make it to the final answer.\n",
    "- RC asserted, in output: The root cause entity was investigated, determined to be a contributing factor in a previous step, and this was reflected in the agent's final answer. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "id": "db405330",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Loading ground truth data...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Loading ground truths: 100%|██████████| 35/35 [00:00<00:00, 74.31it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loaded 31 ground truth files\n",
      "Found 1 agent models\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Analyzing models:   0%|          | 0/1 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Analyzing OpenAI-GPT-OSS-120B...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "  OpenAI-GPT-OSS-120B scenarios:   0%|          | 0/35 [00:00<?, ?it/s]\u001b[A\n",
      "\n",
      "    Scenario-1 trials: 100%|██████████| 3/3 [00:00<00:00, 39.59it/s]\n",
      "\n",
      "\n",
      "    Scenario-102 trials: 100%|██████████| 3/3 [00:00<00:00, 48.80it/s]\n",
      "\n",
      "  OpenAI-GPT-OSS-120B scenarios:   6%|▌         | 2/35 [00:00<00:02, 13.05it/s]\u001b[A\n",
      "\n",
      "    Scenario-11 trials: 100%|██████████| 3/3 [00:00<00:00, 53.14it/s]\n",
      "\n",
      "\n",
      "    Scenario-12 trials: 100%|██████████| 3/3 [00:00<00:00, 58.81it/s]\n",
      "\n",
      "  OpenAI-GPT-OSS-120B scenarios:  14%|█▍        | 5/35 [00:00<00:01, 18.77it/s]\u001b[A\n",
      "\n",
      "    Scenario-13 trials: 100%|██████████| 3/3 [00:00<00:00, 31.70it/s]\n",
      "\n",
      "\n",
      "    Scenario-14 trials: 100%|██████████| 3/3 [00:00<00:00, 187.45it/s]\n",
      "\n",
      "  OpenAI-GPT-OSS-120B scenarios:  20%|██        | 7/35 [00:00<00:01, 17.36it/s]\u001b[A\n",
      "\n",
      "    Scenario-15 trials: 100%|██████████| 3/3 [00:00<00:00, 56.86it/s]\n",
      "\n",
      "\n",
      "    Scenario-16 trials: 100%|██████████| 3/3 [00:00<00:00, 92.02it/s]\n",
      "\n",
      "  OpenAI-GPT-OSS-120B scenarios:  26%|██▌       | 9/35 [00:00<00:01, 17.97it/s]\u001b[A\n",
      "\n",
      "    Scenario-17 trials: 100%|██████████| 3/3 [00:00<00:00, 41.35it/s]\n",
      "\n",
      "\n",
      "    Scenario-18 trials: 100%|██████████| 3/3 [00:00<00:00, 66.31it/s]\n",
      "\n",
      "  OpenAI-GPT-OSS-120B scenarios:  31%|███▏      | 11/35 [00:00<00:01, 16.89it/s]\u001b[A\n",
      "\n",
      "    Scenario-19 trials: 100%|██████████| 3/3 [00:00<00:00, 45.24it/s]\n",
      "\n",
      "\n",
      "    Scenario-2 trials:   0%|          | 0/3 [00:00<?, ?it/s]\u001b[A\u001b[A\n",
      "\n",
      "    Scenario-2 trials: 100%|██████████| 3/3 [00:00<00:00, 24.27it/s]\u001b[A\u001b[A\n",
      "\n",
      "  OpenAI-GPT-OSS-120B scenarios:  37%|███▋      | 13/35 [00:00<00:01, 13.52it/s]\u001b[A\n",
      "\n",
      "    Scenario-21 trials: 100%|██████████| 3/3 [00:00<00:00, 35.44it/s]\n",
      "\n",
      "\n",
      "    Scenario-22 trials: 100%|██████████| 3/3 [00:00<00:00, 57.42it/s]\n",
      "\n",
      "  OpenAI-GPT-OSS-120B scenarios:  46%|████▌     | 16/35 [00:01<00:01, 15.49it/s]\u001b[A\n",
      "\n",
      "    Scenario-24 trials: 100%|██████████| 3/3 [00:00<00:00, 59.10it/s]\n",
      "\n",
      "\n",
      "    Scenario-25 trials: 100%|██████████| 3/3 [00:00<00:00, 58.47it/s]\n",
      "\n",
      "  OpenAI-GPT-OSS-120B scenarios:  54%|█████▍    | 19/35 [00:01<00:00, 18.17it/s]\u001b[A\n",
      "\n",
      "    Scenario-29 trials: 100%|██████████| 3/3 [00:00<00:00, 52.44it/s]\n",
      "\n",
      "\n",
      "    Scenario-33 trials: 100%|██████████| 3/3 [00:00<00:00, 81.13it/s]\n",
      "\n",
      "  OpenAI-GPT-OSS-120B scenarios:  63%|██████▎   | 22/35 [00:01<00:00, 20.56it/s]\u001b[A\n",
      "\n",
      "    Scenario-34 trials: 100%|██████████| 3/3 [00:00<00:00, 60.99it/s]\n",
      "\n",
      "\n",
      "    Scenario-35 trials: 100%|██████████| 3/3 [00:00<00:00, 69.64it/s]\n",
      "\n",
      "\n",
      "    Scenario-38 trials: 100%|██████████| 3/3 [00:00<00:00, 148.09it/s]\n",
      "\n",
      "  OpenAI-GPT-OSS-120B scenarios:  71%|███████▏  | 25/35 [00:01<00:00, 21.15it/s]\u001b[A\n",
      "\n",
      "    Scenario-4 trials: 100%|██████████| 3/3 [00:00<00:00, 47.82it/s]\n",
      "\n",
      "\n",
      "    Scenario-5 trials: 100%|██████████| 3/3 [00:00<00:00, 60.89it/s]\n",
      "\n",
      "\n",
      "    Scenario-6 trials: 100%|██████████| 3/3 [00:00<00:00, 33.99it/s]\n",
      "\n",
      "  OpenAI-GPT-OSS-120B scenarios:  80%|████████  | 28/35 [00:01<00:00, 17.71it/s]\u001b[A\n",
      "\n",
      "    Scenario-7 trials: 100%|██████████| 3/3 [00:00<00:00, 102.14it/s]\n",
      "\n",
      "\n",
      "    Scenario-8 trials: 100%|██████████| 3/3 [00:00<00:00, 47.80it/s]\n",
      "\n",
      "  OpenAI-GPT-OSS-120B scenarios:  86%|████████▌ | 30/35 [00:01<00:00, 17.95it/s]\u001b[A\n",
      "\n",
      "    Scenario-80 trials: 100%|██████████| 3/3 [00:00<00:00, 122.31it/s]\n",
      "\n",
      "\n",
      "    Scenario-81 trials: 100%|██████████| 3/3 [00:00<00:00, 42.94it/s]\n",
      "\n",
      "  OpenAI-GPT-OSS-120B scenarios:  91%|█████████▏| 32/35 [00:01<00:00, 17.87it/s]\u001b[A\n",
      "\n",
      "    Scenario-83 trials:   0%|          | 0/3 [00:00<?, ?it/s]\u001b[A\u001b[A\n",
      "\n",
      "    Scenario-83 trials: 100%|██████████| 3/3 [00:00<00:00, 15.96it/s]\u001b[A\u001b[A\n",
      "\n",
      "\n",
      "    Scenario-9 trials: 100%|██████████| 3/3 [00:00<00:00, 122.23it/s]\n",
      "\n",
      "  OpenAI-GPT-OSS-120B scenarios:  97%|█████████▋| 34/35 [00:02<00:00, 13.93it/s]\u001b[A\n",
      "\n",
      "    Scenario-91 trials: 100%|██████████| 3/3 [00:00<00:00, 34.80it/s]\n",
      "  OpenAI-GPT-OSS-120B scenarios: 100%|██████████| 35/35 [00:02<00:00, 16.26it/s]\n",
      "Analyzing models: 100%|██████████| 1/1 [00:02<00:00,  2.16s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  Trials: 93\n",
      "  Encounters: 55 (59.1%)\n",
      "  Assertions: 4 (4.3%)\n",
      "  Recoveries: 0 (0.0%)\n",
      "\n",
      "============================================================\n",
      "Generating output files...\n",
      "Saved: /data/ITBench-SRE-Agent/ITBench-Trajectories/output/discovery/discovery_trials.csv\n",
      "Saved: /data/ITBench-SRE-Agent/ITBench-Trajectories/output/discovery/discovery_summary.csv\n"
     ]
    }
   ],
   "source": [
    "discovery_summary_df, discovery_trial_df, discovery_trials_n = extract_discovery_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "id": "77a4a7c0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABAMAAAGzCAYAAABaTN78AAAAOnRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjEwLjgsIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcvwVt1zgAAAAlwSFlzAAAXEgAAFxIBZ5/SUgAAphFJREFUeJzs3XdUFFcbBvBnl7L0IioCIigKKNgblggWjF2T2BvYjb2XxAgaTezip8YuaDR2Y9TYFewNK4oaFbAXRJDe7/cH2ZV1l6Yomn1+5+yJmVvmzuzdYeadO3ckQggBIiIiIiIiItIY0qJuABERERERERF9WgwGEBEREREREWkYBgOIiIiIiIiINAyDAUREREREREQahsEAIiIiIiIiIg3DYAARERERERGRhmEwgIiIiIiIiEjDMBhAREREREREpGEYDCAiIiIiIiLSMAwGEBEREREREWkYBgOIiIiIiIiINAyDAUREREREREQahsEAIiIiIiIiIg3DYAARERERERGRhmEwgIiICoW3tzckEkmuH6lUCjMzM9SvXx9+fn5ISUkp6mZ/EpmZmdixYwd69+4NR0dHmJqaQldXFyVKlICbmxsGDhyIjRs3IjIysqib+lkzMzPLs4+p+3h4eBRZm2/dugVbW1s0aNAAycnJRdYOdWJiYnLdbzKZDKVKlUL16tUxePBg7NixAxkZGXnWO3LkSJiammLz5s2fYCuosL3Pb0wikcDe3r6om17kgoKCVPZLREREUTeLKEcSIYQo6kYQEdGX78mTJ4iOjsZff/2FKVOmAAAOHjwIa2trAIAQAo8fP8bx48excOFCpKamokaNGjh27BhMTU2LsulKfH19MW3aNLi7uyMoKOiD67t06RK8vLxw69YtdO7cGR06dEDZsmUhlUrx5MkTHDt2DP7+/oiLi4OWlhYaN26Mw4cPf/iG/AfdunULGRkZmDJlCv766y9YW1vj4MGDuZbp06cPDA0NC+W7zC4oKAiNGzcGkNW3czJ//nyMGzcOABAcHIyaNWsqpUdERKBs2bIAgPDw8E96QZWZmYnQ0FAAWfspODgYtWrVgr+/PwAgKSkJL1++xIULF7Bt2zbcunULZcqUwfz589GxY8cc6zU2NkZ8fDzatGmDPXv2fJJtocJz48YNAOr7RE6mTJmCq1evavyFb0JCAsLDw/H06VN8/fXXAD7975qoILSLugFERPTfYGNjAxsbGwQHByuWOTo6Kp0EVa5cGS1btkTTpk3RvHlzXL58GZMmTcKyZcuKoMUf399//41OnTpBR0cHx48fR8OGDZXSa9Wqhfbt22P8+PH4+uuvERoaiiNHjhRRaz9/FStWBJA1QgAAdHR04OrqmmsZQ0PDj92sXHXr1g0HDhyAra0tqlWrVqRteZdUKlXsP/l+MjQ0VNmnrVu3ho+PD9asWYPhw4ejU6dOGD16NBYsWKC23l9++QUbNmxQBEHoy5KfPvEu+W9S08n3lZGRUVE3hShf+JgAERF9cp6ennBxcQEAbNq0KV9Dj780t27dQrdu3ZCUlIT169erBAKyK126NHbu3AltbcboC1ulSpVQqVKlIlu/tbU1Dh8+jLVr10JLS6vI2vGhpFIpBgwYgO3bt0MikWDhwoWYO3eu2rzDhw/H+fPn4e7u/olbSUXFwcEB1atXL+pmEFEBMRhARERFQn6B9ubNG7x69aqIW1P4xowZg7i4ONSrVw/t27fPM7+TkxPatWv3CVqmWX777Tf89ttvRd2M/4w2bdrA29sbADB16lQ8fvy4aBtEn4WffvoJf/75Z1E3g4gKiMEAIiIqEtnvguvq6qrNk5ycDD8/P9SvXx/m5ubQ09ODvb09vLy8cPXq1VzrL2hZX19fSCQSTJs2DQBw/PhxlYmg8is0NBQHDhwAkDVMPL8mTpyIX3/9VWV5REQEZs2aBU9PT1hZWUFHRwcmJiaoWbMmpk+fjtjYWJUyBw4cyLX9z58/z9dEV+np6di4cSMaN24MW1tbxcSHHh4e8PX1VTxzrq7c6tWr4eHhgWLFikEmk8Ha2hrt27fHX3/9le998j4iIiIgkUgQEBCgkpbTBF/Hjx9H8+bNYW5uDn19fdSoUQPr16/Psbx8vgBAdcI1+fwE7y7P3h55G+XzBQBA2bJlVfJ7eHionawt+xwI8r6b07oK25gxYwBk/cYWL16cYztyek76/v37GDp0KFxdXWFiYgJ9fX04OTmhd+/e2LJlC5KSknIt5+joCAMDA+jr68PBwQGdOnWCv78/4uPj1Za7d+8ehgwZggoVKkBfX1/x2/n555/x5s0bpbw5TYSafX8GBASopPv6+irVEx0dDR8fH1StWhVGRkYwMDBA+fLl0bdvX1y7dk2lje9+z97e3khKSsK0adPg4uICQ0NDpf1aGG0sDOrqXb58ea794Ny5c7kem+S/jXf7+7Vr1/Dtt9+iRIkS0NPTQ6VKlTB//ny1c3ao2zcRERHo3bs3rKysIJPJ4ODggB9//BGpqam5bmNwcDB69OgBW1tbyGQymJubo27duvjll19U+g/RF0cQEREVIn9/fwFAABDh4eE55qtbt64AICpWrKg2/cGDB8LFxUUAED169BCHDx8WFy9eFMuXLxdWVlZCKpWK+fPnF1rZFy9eiJCQEPH9998LAKJWrVoiJCRE6ZNf8+fPV+yDc+fO5btcTtzd3QUA4e7uLnbs2CEuXrwodu/eLXr06CEkEolwcHAQz549UyoTFxcnQkJCxNq1axVtyS4tLU2EhISIgwcP5vh9ZWRkiK+//lqxH/fv3y+Cg4PFrl27RMuWLdXWK4QQr1+/Fg0bNhQARNu2bcWff/4pzp49K9asWSMcHBwU9WVkZLzX/vDy8hIAhJ2dndr08PBwAUD4+/urpMXHx6ts96xZs0SjRo3Enj17xLlz54Sfn58wMDAQAMTy5cvVls++X9/tJ/Hx8UIIofh/a2trlfakpqaqtOPgwYNK9URHR4uwsDBx9OhRIZFIBAAxd+5cpXUI8bbvurm5idq1ayvKFkT2PpYfpUqVEgBE5cqVVdoxY8aMHL+fo0ePCplMJmxsbMSyZcvEmTNnxMmTJ8WcOXNEsWLFBADh6+urUm7jxo1CJpMJExMTMWvWLHHu3DkRGBgoZs2apShXq1atHMuZmpqKuXPnivPnz4ugoCAxbNgwIZVKha2trbh+/boi/+PHj8WlS5eEoaGhACBGjx6tsj+jo6NFSEiI6Nixo7C2thYhISHixYsXivSrV68Ka2trIZVKxejRo8WhQ4dEUFCQ+PXXX4WxsbHQ0tISS5cuVWpnWFiYCAkJEe3btxcARJcuXYSbm5sYOXKkOHHihDh16pRo1aqVACBmzJghLl26JIyMjAQAMWrUqBzb+O233wobGxuVNuZHfvoEAOHj46O07PXr17n2g8TExFyPTfLfRkhIiCL9559/FjVr1hRbt24VFy5cEGvWrBHFixcXAMSkSZNU2vXub2/y5MmiSpUqIiAgQFy8eFFs3rxZlCtXTgAQXbt2zXH7Zs+eLSQSibCxsRErVqwQZ8+eFbt27RJdu3YVAIS9vb24ffu2Sjn5MSivv4NERY3BACIiKlT5CQbcvHlTSKVSoaWlJfbu3auSnpycLKpUqSIAiCFDhqik3717V3GxtnXr1kIrK4QQPj4+BbooUqd3796KfRAZGfne9ci5u7uLmjVritTUVJW0yZMnCwDi22+/VVs2MDAwx4t2IXI/af3rr78EAFG/fn2VcpmZmaJZs2Zq623RooUAILp3766SFhsbqzhB//XXX9W2KS8fEgx4N4/8gjYpKUkp/bfffhMAhK2trdryee3X7Ozs7HJsT34vGuRBmebNm6tNj4iIEFKpVKxbty7P9qhT0GCAp6enACCkUqlKUEd+DFD3/VSvXl0AEIcOHVJJO3HihNqLy5MnTwptbW0hlUrFmTNnVMpduHBB6OjoiKpVq+ZY7uzZsyrlfvnlF8V3HBUVpZTWv39/AUC4uLio3f7ExERhZmYmfvrpJ6Xlr169EjY2NgKAWLlypdq2SiQSIZVKxalTp1TS5X1bS0tLzJ07VyntyZMnSv1o0KBBAoBwdnZW28b4+HhhYmIipk2bpjY9L+8bDJDLrR8Ikb/fkDy9VKlSKsfSffv2CQBCX19fxMbGqi0v/+0ZGBiIO3fuKKWFhoYq6r97965K2c2bNwsAwtzcXDx//lwlfdiwYQKAqFSpkkhPT1dKYzCAvhR8TICIiD6Z58+fY8uWLWjZsiUaNWqEY8eOoXXr1ir5/P39cf36dWhpaakd2lq+fHn06dMHADBu3Dikp6cXStnCEhkZqfi3iYnJB9fn7e2N+fPnQ0dHRyVt0KBBAIC//vpL7eMCH0L+CIC6mbElEgmGDh2q8v0dPHhQ8YjErFmzVMoZGxtj6NChAIC5c+d+0P5/8OABtLW1VT7ly5cvUD2DBg2Cnp6e0jJPT08AwKNHjz6L16UNGTIEAHD48GHcu3dPJX358uUoVqwYunTp8knaU6xYMQBZryeMiorKd7nc+tRXX32FTp06wdHRUWm5/Hfapk0b1KtXT6Vc7dq14eHhobI8ezk3NzeV9LFjx6JYsWJ49OgR5s2bp5Q2ePBgAMDNmzdx+vRplbJbt25FbGws+vfvr7R83rx5ePLkCcqXL6+SJm9rs2bNkJmZidmzZ6uky5mYmGDEiBFKy6ytrREeHq54raO8T9y+fRtHjx5VqWPDhg1ISkrCwIEDc1xPfhw/flzt7+xTTnjas2dPFC9eXGlZkyZNIJVKkZSUhEuXLuVavlWrVir9qmLFirCxsQEAnDp1SiktPT0d48ePBwCMHj0alpaWKnVOnjwZQFaf/tiPPhF9LAwGEBHRR1O+fHmlE0crKyt07doVderUwebNm9GoUSO15bZu3QoAqFq1KkqUKKE2T/PmzQEADx8+xLlz5wqlbGEpyPwC+eHt7Z3jzOx2dnYAgIyMDNy9e7dQ1ys/eT506JDauQk6dOiAvXv3Ki2T7/9y5crB1tZWbb3Ozs4AgNevX+PKlSvv3T5ra2tcvXpV5bNv374C1VO7dm2VZfKLBCAriFXU2rRpgzJlykAIofIqztTUVKxZswZ9+/aFTCb7JO0R2Z7TLkh/l/epQYMG4ezZsyrpW7duRffu3RX///DhQ5w/fx4A0KxZsxzrXbJkidJEkfkpp6urq/hdyfutXM2aNVGrVi0AwIoVK1TKrlixAi1btkSZMmVU2g8AjRo1ynG/yPt/UFAQMjMz1eapU6eO2rlU7O3tFYGUKlWqoEGDBgCgdpLMZcuW4ZtvvkGpUqXUriO/atWqpfZ3lte8LYVJ3W9UJpMpAgR5/UbVlQfe/s7fLX/27Fk8evQIANQGmoCs44882KsuGEP0JWAwgIiIPpp9+/YpThqPHTuGMWPGQCqVYvv27fDw8EBCQoLactevXweQdUGZk+wTr2WfkOtDyhaW7EGIwrhbn5mZiQ0bNqB169YoXbo09PX11d6dy2kCtffVoUMHxRsOfHx8YGlpiXbt2mH58uV48uSJ2jLy/RkWFpbj3cROnTop8j98+PC926ejowNXV1eVz7t3APNiYWGhskxfX1/x7+Tk5PduY2GRSqWKUSD+/v5Kk+xt27YNUVFR+P777z9Ze16/fq1ol3yUQH4sWrQIxsbGCAkJQf369eHo6IixY8fi2LFjal8xKv89A7n/ph0dHVG/fv0Cl5MfC+7fv69yPJKPDti2bRuio6MVy2/cuIGzZ88qvg+5+Ph4hIWFAcj6jnLq/0uWLAEAxMXFKdWbXU6BzHfJRwfs3r1b6Td5+vRpXLt2TTEK50MYGhqq/Z25urp+cN35pe43Crz9neb1Gy1o+ex/Fxo3bpzjdyk/vn/IcYyoKPGFxkRE9NE4OjoqzSTdqFEjlCxZEpMmTcLt27cxZ84cxez92clnaM5+QfYuAwMDlfwfWrawVKlSRfHv+/fvqwxvLYi0tDS0bt0ahw8fhrW1NSZMmIDq1asrXYBVrlwZANTOqv0hpFIpdu3ahW3btmHlypUIDAzEnj17sGfPHgwZMgStWrXCggULlC6+5fuzRo0aWLduXZ7rKF26dKG2Gci6e1qQfaGlpVXobfgY+vfvj2nTpiE6OhqbNm1C3759AWTdFW7ZsmWOs/d/DPKLpcqVK0Mqzf+9pcaNG+PGjRtYvHgxNmzYgLt372LBggVYsGABrKysMHHiRIwYMUJxVz377zO33/S78lvu3WOBoaGh4v+7deuGsWPH4s2bN1i3bh1GjRoFIGtUgK2tLVq1apXjOgcOHIhhw4bl2c6cHiPKb5/s2LEjRo8ejZcvX2LFihWYPn06gKw+4erqmuPoq8JS2MecnHzob7Sg5bN/l3v27MlxlJOcusdeiL4EDAYQEdEnNW7cOGzcuBEhISHw8/PDqFGjYG5urpTH1NQUUVFRSExMzLGe7GmmpqaFUrawtGzZEmPHjgUAXLhwAXXr1n3vun777TccPnwY2traOHToEFxcXAqrmQCQ5zP7EokEnTt3RufOnfHy5Uvs2rULmzdvRmBgIP7++2+cPXsWN27cgJWVFYC3+zMjI+OT3jnUBCVLlsR3332HTZs2YdmyZejbty+uX7+OM2fOFPjRiA9x5coVxbwY714Q50eZMmUwd+5czJ49G6dPn8aOHTuwceNGPHv2DKNGjcLz588Vr9jM/vvM6ZWD6mQv977HAgMDA/Tq1QtLlizBypUrMWrUKCQlJWHDhg0YPXq0ygVm9vJ6enqfpP/r6uqiX79++PXXX7F69Wr89NNPiImJwfbt2+Hn5/fR1/8hPsZ8LYUl+3dpZWXFYxn9Z/ExASIi+qS0tLTwyy+/AMgaQq/uhFV+Z10+5Fad7GlVq1YtlLKFpWLFimjbti0AYPPmzfkqk5ycjMqVK6N06dI4ePCgYrn8WVRHR8f3CgRkf4Y8JSVFJf3Vq1f5rqtkyZIYOHAgjh07hqCgIOjr6+P169dYvXq1Io98f967d0/tsG+5Y8eOYfXq1YX+aMO7oqOjP4tn/guLfFh4cHAwLly4gN9++w0ODg5o0aLFJ2vDggULAGTdcc/P3e+cSKVSfPXVV/Dz88ODBw8UE+MtWLBAcaGY/feZ2286NTUV8fHximfws4/Oyc+xwMHBQWlUgJz8UYBbt27hxIkT2LJlC+Li4tCvXz+VvEZGRopHEm7fvp3jOgFg48aN2LJlS6558mvQoEGQSqV49uwZ/vzzT6xZswZ6enro1atXodT/vuTHHnXHHaBgx55PLXu/y+27fPToEVavXo3Lly9/imYRFToGA4iI6JNr06aNYuKrRYsWISYmRildPiP69evX8fLlS7V1HDp0CEDWXcbsM4V/SFkAimfwsw9/FUJg8+bNCA4Oztf2AcD8+fNhamqKM2fOqEyyp87y5ctx48YNmJiYoGnTporl8oubnIbj5jXTffbJw54+faqSfuHChRzLzps3L8eZ+d3d3fH1118DAJ49e6ZYLt//CQkJCAwMVFs2PT0d3bp1w7Rp09RegBWm0aNHq51J/kNln6sh+3dz9OjRAt2lz6mec+fOYdu2bSr5GzZsqLjQ/fXXX7Fx40YMHjy40CetzMnOnTuxYcMGAMDMmTNhbW1doPKlSpVSmQARyLoLP2XKFABZF/byOQlsbW0VbxA4fPhwjvU2adIElpaWiuBS9t92TuVSU1Nx/PhxAMjxLQyurq6KY9WKFSuwcuVKtGnTRmmCyezk9Zw4cSLH+UJCQkLQs2dP7NmzJ8ftKQg7OzvFWz0WL16MFStWoFevXkU+dF1+7Hn16hXS0tJU0nM79hS1evXqKSaHzO347efnhwEDBij6K9GXhsEAIiIqEvLXzr158waLFi1SSvP29kaVKlWQkZGBn376SaXsvXv3EBAQACDrgjX7BdWHlAXensBmP7l78uQJunXrlq+LerkKFSpg586d0NfXR69evdTOnC63a9cuTJgwAQYGBli/fr1Sm7766isAwJ07d3Dx4kWVskuXLs21Hfb29ooLl2PHjimlJSUlqZ0pXS4+Ph7379/HkSNHVNKyv72gTp06iuXNmzdXDB3/4Ycf1E7sNX36dLx8+RI//PDDJ7uILWzZgyzZ+8qgQYMwZ86cfNdTvHhxxXDz7PVMmTJF8Wqzd8knCty1axcyMjIUcwd8TJmZmVi2bBm6du0KIOtxn9GjRxe4nhcvXmDTpk1qZ9GX34G1t7dXmkBP/juVP5byroMHD+L06dPo1auX0jP4c+fOhZaWFv7++2+1rwecP38+oqOjYWtrq3isRx35RIJbt25VO3FgduPGjUPp0qWRlJSEH3/8USU9PT0dI0eOhI6ODiZMmJBjPQUlHzFy6tQpREREKP6/KNWsWRN6enpIT0/HiRMnlNIiIyOxcePGImpZ3rS1tTF37lwAwJYtWxRvpsguJCQEK1asQN26dXN90wXRZ00QEREVgsePH4uQkBAxY8YMAUAAEAcPHhQhISEiLCxMbZnWrVsLAMLc3FxcunRJhISEiOjoaCGEEBEREaJSpUoCgOjevbs4cuSICA4OFitWrBBWVlZCKpWK+fPnq633Q8o+fPhQyGQyIZVKxdq1a8W5c+dEp06dhJaWlggODi7wfrl69apwdXUVUqlUdO/eXWzdulWcP39enDhxQqxbt040b95cABB2dnbizJkzKuXj4+NF1apVBQBRvHhxsWDBAnHmzBlx4MAB4e3tLUxNTRX7e+3atSIkJETEx8cr1bF06VIBQBgZGYnFixeL8+fPiz179oivvvpKzJkzR+X7kpf39fVVfD8///yzOHr0qAgODhZ//vmn+PrrrwUA8fXXX4v09HSl9UVHRwsPDw8BQNSoUUP88ccfIjg4WOzZs0d07dpVABB9+vQRmZmZBdqXQUFB4vDhw4p9ZmlpKQ4fPpzrp3nz5sLOzk5RR2pqqggJCREHDx5U2e7U1FQhhBAhISEiJCQkz/0q72OjRo0SFy5cUPR9Pz8/pXqsra0FADFjxgwREhIiXrx4oVRPy5YtBQDRtWtXcf78ebFixQqhpaUlRo0apXY/xMXFCRMTEwFAeHt7F2gfZpeRkaFoY61atQQAUatWLcWyixcvir///ltMnTpVODk5CQCiTJkyYvv27Wrre/HihdIxwNraWlGXnHyfNmvWTGzdulVcuHBBHD9+XMyaNUuYmZkJPT09cejQIZW6N23aJGQymTA1NRVz5swR58+fF4GBgWLq1KnC0NBQ1K1bVyQkJKiU27Bhg9DV1RWmpqZi7ty5ivUNGzZMSKVSUbp0aXH9+vVc91NycrKwsLBQ/E4zMjJyzX/9+nVha2srAIiOHTuKffv2iYsXL4pNmzaJ2rVrCy0tLbFmzRqlMvLjZ/v27QUA0b59e8W+e7ffqZOZmSkcHBwEAOHh4ZFn/tzk1idyOpbnZPz48Yrfqr+/v7h48aLYtm2bqFmzptLfCnn9+f0N3rlzR+1vS/73I6/0sLAwpW38/vvv1f4258+fL6RSqTA2NhYzZswQZ86cESdOnBC//PKLMDMzE2XLlhXh4eGK/PHx8TkeX/LzPRJ9agwGEBFRofDy8lKc/Lz7cXd3V1vm2rVrQiqVKuX19/dXpCcmJooFCxYINzc3YWJiInR1dUWZMmVE7969xeXLl3Ntz4eU/fvvv0Xt2rWFnp6eMDY2FrVq1RI7d+4s6C5RyMjIENu3bxfdunUTZcuWFQYGBkJXV1fY2NiIVq1aiZUrV4qkpKQcy8fFxYmffvpJVKxYUchkMiGTyUSFChXE0KFDRUREhMr+DgwMVKnD399fVK9eXejp6QlTU1PRrFkzERQUJMLDw3Msn5ycLDZv3iy6d+8uHB0dhYGBgdDS0hIWFhaicePGYs2aNSqBALn09HSxbt060bRpU1GsWDGhpaUlihcvLlq0aCF27NjxXvsxe+CjIJ/swQB12yv/yE/qc0p/d7/evn1btGrVSpiamgqZTCYcHR3FL7/8otgnOdXj4+OjVM+zZ89Ely5dhIWFhdDR0RFly5YV48aNU3txKzdkyBABQFy8ePG99qUQWUGb3Pabjo6OKFmypKhataoYOHCg2LFjR47ftxBC+Pj45FhX9n02depU8dVXX4kSJUoIbW1toaenJxwdHcXgwYPFP//8k2P99+7dE0OGDBHly5cXenp6Ql9fX1StWlXMnDkz13119+5dMXjwYOHg4CBkMpkwMjIS1apVE9OmTRMxMTH52ldjx45VXFTmx5s3b8SMGTNEjRo1hLGxsdDR0RG2traiZ8+eao8/uR0/1f2e1ZEH9rZt25av/DnJrU/kdCzPSUZGhpg7d65wdnYWurq6wsLCQnTo0EFcu3ZNBAYGvvdv0M7OTm26/O9HXunu7u75+m0KkRXQ9fb2FnZ2dkJXV1cYGBiIqlWrCl9fX/HmzRulvOq2qaDfI9GnJBHiE70ThIiIiIgKRceOHfHw4cPP+rlr+rTGjRuHTZs24cGDByqPPxERqcM5A4iIiIi+IC9fvsTu3bsxdOjQom4KfSZSU1Oxbt06DBw4kIEAIso3BgOIiIiIPmMDBw5UzHoPZE0aWaxYsRxnwKf/vilTpii9bWLjxo2Ij4/HwIEDi7BVRPSlYeiQiIiI6DN27Ngx3Lp1C3PnzsX9+/cxZ84cLFy4EHp6ekXdNCoi586dw86dO2FpaYk3b95g4sSJGDlyJKysrIq6aUT0BWEwgIiIiOgzVrduXRw8eBCNGjWCtbU1fH19Fa+7I81Uo0YN3LhxA82aNUPx4sXRu3dvzJw5s6ibRURfGE4gSERERERERKRhOGcAERERERERkYZhMICIiIiIiIhIwzAYQERERERERKRhGAwgIiIiIiIi0jAMBhARERERERFpGAYDiOi9tWvXDu3atSvqZhARERERUQFpF3UDiOjLdf/+/aJuAhERERERvQeODCAiIiIiIiLSMAwGEBEREREREWkYBgOIiIiIiIiINAyDAUREREREREQahsEAIiIiIiIiIg3DYAARERERERGRhmEwgIiIiIiIiEjDMBhAREREREREpGEYDCAiIiIiIiLSMAwGEBEREREREWkYBgOIiIiIiIiINAyDAUREREREREQahsEAIiIiIiIiIg3DYAARERERERGRhmEwgIiIiIiIiEjDMBhAREREREREpGEYDCAiIiIiIiLSMAwGEBEREREREWkYBgOIiIiIiIiINAyDAUREREREREQahsEAIiIiIiIiIg3DYAARERERERGRhmEwgIiIiIiIiEjDMBhAREREREREpGEYDCAiIiIiIiLSMAwGEBEREREREWkYBgOIiIiIiIiINAyDAUREREREREQahsEAIiIiIiIiIg3DYAARERERERGRhmEwgIiIiIiIiEjDMBhAREREREREpGEYDCAiIiIiIiLSMNpF3QAi+rKJjHQk37lV1M0goi+MjoUppJmpRd0M+q8zMQak6UXdCqL/rHSpLjLwcY7lUUlAdEriR6m7KLkWL1vUTVBgMICIPkxGBiK6flPUrSCiL0yF3X8hc9u0om4G/cdJB80GXmwp6mYQ/WdlWHfDw4RNH6XulLR2aLFz4kepuyg9GfRnUTdBgY8JEBEREREREWkYBgOIiIiIiIiINAyDAUREREREREQahsEAIiIiIiIiIg3DYAARERERERGRhmEwgIiIiIiIiEjDMBhAREREREREpGEYDCAiIiIiIiLSMAwGEBEREREREWkYBgOIiIiIiIiINAyDAUREREREREQahsEAIiIiIiIiIg3DYAARERERERGRhmEwgIiIiIiIiEjDMBhAREREREREpGEYDCAiIiIiIiLSMAwGEBEREREREWkYBgOIiIiIiIiINAyDAUREREREREQahsEAIiIiIiIiIg3DYAARERERERGRhmEwgIiIiIiIiEjDMBhAREREREREpGEYDCAiIiIiIiLSMAwGEBEREREREWkYBgOIiIiIiIiINAyDAUREREREREQahsEAIiIiIiIiIg3DYAARERERERGRhmEwgIiIiIiIiEjDMBhAREREREREpGEYDCAiIiIiIiLSMAwGEBEREREREWkYBgOIiIiIiIiINAyDAUREREREREQahsEAIiIiIiIiIg3DYAARERERERGRhmEwgIiIiIiIiEjDMBhAREREREREpGEYDCAiIiIiIiLSMAwGEBEREREREWkYBgOIiIiIiIiINAyDAUREREREREQahsEAIiIiIiIiIg3DYAARERERERGRhmEwgIiIiIiIiEjDMBhAREREREREpGEYDCAiIiIiIiLSMAwGEBEREREREWkYBgOIiIiIiIiINAyDAUREREREREQahsEAIiIiIiIiIg3DYAARERERERGRhmEwgIiIiIiIiEjDMBhAREREREREpGEYDCAiIiIiIiLSMAwGEBEREREREWkYBgOIiIiIiIiINAyDAUREREREREQahsEAIiIiIiIiIg3DYAARERERERGRhmEwgIiIiIiIiEjDMBhAREREREREpGEYDCAiIiIiIiLSMAwGEBEREREREWkYBgOIiIiIiIiINAyDAUREREREREQahsEAIiIiIiIiIg3DYAARERERERGRhmEwgIiIiIiIiEjDMBhAREREREREpGEKNRiQmJiIxYsXo2nTpihZsiR0dXVRsmRJeHh4YP78+YiPjy/M1RWJnTt3QiKRoG7durnm8/X1hUQiUfn4+voWeJ3h4eGYMGECatWqBQsLC+jo6MDMzAzVqlWDt7c3AgICEBkZqVTG29tb7folEgmkUiksLCzQrFkzbN++Pc825/XJj8uXL2PgwIFwdHSEoaEhDA0NUb58efTt2xfnzp3LtWxSUhLmz5+PevXqwdzcHLq6urCyskKNGjXg7e2NlStX4tGjRyrlYmJi4OPjgxo1asDExAQymQylS5dGnTp1MHjwYKxfvx6vXr3KV/vV2b9/P+zt7fPcB6mpqdiwYQO++eYblC5dGrq6ujA1NYWbmxsWLVqEtLS0HMsmJydj+vTpcHJygp6eHqysrNCrVy+Eh4er5PXz88v1e9LV1UWZMmXQvXt3hISEvPd2ExERERHRl6/QggEXLlyAs7MzFixYgF69euHq1auIj4/H1atX0adPHyxZsgTly5fHyZMnC2uVRWLNmjUAsrb3xo0bOebz9fWFEAJeXl4AAB8fHwghChQMEEJg5syZcHR0RGhoKH799Vfcu3cPcXFxuHjxIgYOHIijR4+iT58+KFWqFF6/fq0oGxAQACEE3N3dAQD+/v4QQkAIgaioKAQEBOD+/fvo1KkTRo0apSjn7u6uyCf/yOsIDAxUWh4YGJjnNqSlpWH48OGoVasWMjIysGHDBrx48QIvXrzApk2bIJVKUb9+ffTp0wcpKSkq5ePj49GgQQP8+OOP6Ny5M65du4a4uDhcunQJkydPxtmzZzFo0CDF9yL35MkTVK9eHUuWLMHQoUNx+/ZtxMbG4syZMxg8eDB2794NLy8v7N27N9/fh9zLly/RvXt39OzZEw8ePMgzf48ePdCrVy9IpVLs378fMTExuHDhAhwdHTFq1Ch4eHggKSlJpVxSUhKaNm2KuXPnYsaMGXj9+jUOHDiAkJAQVK9eHdeuXVPKP2rUKAgh4OPjAwDw8vJSfFdJSUm4du0avv76a2zatAlubm64fv16gbedNM+WVzHwuvcQHW5H4OvQMNQNuYsOtyOw4kUUUjIzAQApmZlY+SIK39yOQOOb9/F1aBi+Dg3DL49f4GVauqKuwzFxaH0rHM1Cw/DdnQhcjk9UWtfOqDdofSscqZnik24j0ac05q/LqLvwIGrM2w8rn51wmLEb3dafxpXHb/+Grzl3H82WHUXlOX/DYcZu2Pj+iaa/HcWuEOXA983nb+Cx9AjKz9wNx1/2YPW5e0rpd17GwvyHbbgbGfdJto0IAFZsPIsmXZejWosFcPSYjWJVpqJaiwX4ZelRJCe/vQFy5cYTtOmzBta1p8PWbQbqtFuErXuvqdRVvtGvKNfwF3h0XobQuy+U0n0WHESz7is+yXbRlykzIRWxf97ES98jePHDQbyYcggvfY8gdncoMpOy+mPCiXC8mHIIL348hFfzTyLtaaxSHbF7buHVwlNF0fyPrlCCAVevXkXTpk2hr6+PixcvwtvbG9bW1tDV1YW1tTW8vLxw8eJFmJqawtPTE+fPny+M1X5yT548wcGDB2FhYQEAKheghW3ChAmYMmUKRowYgb1798LT0xPm5ubQ09NDhQoVMGTIEJw6dQqWlpbIzMxE5r8n5nkxNzdH27ZtsXHjRgDAokWLcPny5Y+yDf369cOSJUuwaNEirFmzBnXq1IGRkRGMjIxQu3ZtrF69Gn5+fggICEDXrl0hhPJFgJ+fH65cuYIpU6Zg9OjRKFOmDGQyGaytrdGpUyccPHgQenp6Kuv18fFBREQEFi1ahH79+sHa2hoymQxlypRB3759sXPnzvfansTERFSsWBFCCNy6dStfZZKSklC5cmVs2bIFlStXhoGBAZycnLBu3Tq4ubnhzJkz+N///qdSztfXF2fOnMGsWbPQqVMnGBgYoGrVqtixYwfi4+PRo0cPZGRk5KsNenp6qFixIlauXAlXV1fFKB6ivATGxqOCngw7nexwsFI5TLYpiTvJKfB79gq/PnkJAPj1yUssfPYKt5NTsLJcaeyvWBam2lr4/VUM+tx7hEwh8DItHeMfPIOBVIoDFcviZVo6RkY8VQQU3qRnYP6zSPxUuiR0pfkbcUT0JVpx5i6+q2qLy+Na4sFP7eFqZYpt1x6iwf8O43R41ii/HdcfwkSmg+AxLXB/SjsM/8oRx++/RMeAU5gX+PZvT88Np3Eu4hUujWmB6jbm+H77RaWgwvCdwRjRyAkVShh/8u0kzbX36C24OJXC5X2j8E/QRPj5tMP1288wZe4BjJq+GwBwLfQpGnX+DcfPh+HC7hEIOzkZ6emZ6DpsA9ZsuQAAuBTyGEOm/IkarqVx+e/ROHM5Aj1GbFSs517EK/itPYkl078pku2kL0P02mDEH7yL9FeJKPFDY5T0bYbMpHTE7/sH0SsvIPVBDN5sugadMmYo8aMHUu+/RvTaYEX59JfxSDh6H6bdqhbhVnw8HxwMSE9PR9euXREfH48lS5agePHiavMVL14cixcvRkpKCrp3757r0OjPVUBAACpUqIApU6YAADZs2IDU1NSPsq7AwEDMmzcP5cqVw6xZs3LMZ2dnh/Hjx7/XOurXrw8jIyMAQFBQENq1a1egupycnLBq1aoc0zds2IDff/8d9evXx/Dhw3PMN2LECNSrVw+7du3C2rVrldLOnDkDAKhevbrasvb29mjVqpViO/Jbzs3NDbVr14a+vn6O7VJHIpFgw4YN2LRpE0qWLJmvMhUqVMCAAQOgra2tUlfbtm0BAAcPHlRKS0xMxNKlS6Grq6sYXSLn4OCAZs2a4ebNm9i3b1+B2+/s7AwAePz4cYHKkmbyKmGOoaUsIP33cRhP07cXFbeTskbz3Ep6O6rHXk8XUokEdjIdAEBYSirCUlJxPSEJKULAXk8H2hIJbGU6eJWegfCUrGPowmeRqGdkADdjw0+1aURForKVGUa7Zx2HdbW1MNajIgAgNSMTy07fBQDYmhlicjMXyLS1AABjPSpCVyvrlG3RiTsAgJikVIQ8e4OSRnow1ddFhRLGEAI4GZYVUNh65QEiXidgclOXT7p9RKP6fQWfkZ6QSrP67LctKivSrt96CgCYNOtvJCSmooarDUpbmUFHRwstG2f9LsbP3Ivk5DScvBAOIQQqlC0OM1N9lLQwwrVbz/AmNms05UjfvzC0dwM4l8/f+RhpprSHbwAAUkNdSI10IdGWQsvCAACQcisSKaEvAAFolzSC1EAXUmMZ0h/HKkYNvNkSAkOPctAp9d8Mqn5wMGDbtm24c+cOypQpg2bNmuWa19PTE7a2tggLC8O2bdsQEBCg9Ezz4cOH8euvv8LZ2Rl6enooWbIk+vTpg2fPnqmtLyIiAv3794etrS1kMhmsrKzQrVs33Lx5UynfrFmzlNYTFBSEJUuWwMnJSXG3+Oeff1a5K52dEAJr166Ft7c3evToAR0dHbx69Qp//fVXwXdaPvz6668AgJ49e0JHRyfXvD179oSPjw8MDAw+aJ01atRA69at853fysoK/fv3zzH9559/BgD07ds3z7r69esHAJg9e7bSchMTEwDA6dOncyy7Y8cOjBs3rsDlLly4gC5duuTZtuz09fXRsmXLApVZuHBhjsEQU1NTtcuPHTuGhIQEVK1aVSXQAQANGzYEAOzevbtAbRFCIDQ0FABQpUqVApUlzVTP2BDm/wayMoXAplfRAAAJgG+LZfXfDsVMIL+XfycpBWlC4H5y1kW+jkSC4tra0P43mCA/zMr/qy2R4GZiMg7ExGGiDU/o6L/v3KivoaP19vSrhJFM8e/4lKzHalZ0roPaZSwUy3W0pDDT1/k3T9YJqva/I2jkZy4iW974lDSM33MFft/UhJ6O1kfaEiL1mjaogOLFsgK7mZmZ+O33rBs0EokE3p1qIyExFYdPZQW+Slq8PcexLJ51sRUTm4Sgc/eho5P1O8n+d0MikUBbWwt/HgjBrXsvMGV400+1WfSFMqhfBgCQGZ+CjJgkZCanI+NVAgBAaqwLqb6a6ywJAKkESVeeIv15HIxaOX7CFn9aHxwM+PvvvwEAdevWzXMitewT7+3fvx/e3t5Kz9WPHDkSz58/R2BgIF6/fg0/Pz9s2bIF9erVw/Pnz5XqCg4ORo0aNRAYGIhNmzYhLi4Ohw4dwoMHD1CnTh2cOHFCkXfSpElK65k9ezaioqJw4sQJhIWFoV69epg6dWquw6aDgoLw4MED9OrVCyVKlFBcNH+MRwUSExNx/PhxAFl3sPNiaWkJX1/fAgcDzpw5o5jU0cPDo8DtzM29e/fwzz//AADq1auXZ375dt69exf3799XLJe3a+7cuZg8eTKePn2ar/XLy40ZMwZz5sxRmk/hc3L79m0AqvtfPh9A2bJl1ZYrV66cUr68JCcnIzQ0FP369UNoaCjq16+PH3744T1bTZpoZ9Qb1A25h/nPXsFUS4qZtqXQubgZAKBbcXPMsbOCiZYU3e8+gPuN+7iVlAJTLSl+trWEmbYWqhnqw0RLinvJqUjOzMSDlDTYyXRQRlcX0x+/wNBSxVFCRzv3RhD9B9179XZy5aaOlmrzxCanITIh5d88pQAARjIdfFWuBF7EJeNlXDJuPIuBrpYUnk6lMO3gDdQuY4FWFa0//gYQ5cB/60UUq+qDSbP2wdxUH2vmdMLA7m64Gx6JzH/nhtHXe3shZpDtouz2/Uh4NnSEjo4Wbtx5hpev4vHiVRw83MpBIgFG/7wbfj7tYaCv+8m3i74sJt+4wKSjKyCV4OXUI3gx+QAyY1MgNdWDWZ9akFUsCWhJkPbkDTJiU5AZmwxdx+KABIjdFgLTzpUh1f3vnp98cDDgzp2s4WplypTJV35bW1sAby+CsjM2NsaiRYtgZWUFAwMDdO/eHTNmzMCDBw8wduxYRb60tDR07twZ0dHRWLt2LRo2bAhdXV3Fc9mpqano3bs30tPTVdYBAFKpFD4+PrC0tISNjY3iee2AgIAc271mzRp8/fXXsLbO+sPap08fAMDhw4fx8OHDfG17foWFhSkeP5Dvr8IUExODvXv3okePHgCygjA1atQo1HXI+wWQv76RfTuz943+/fujcePGSE9Px6xZs2Bra4uGDRti5syZuHLlSo71TZ48Ga6urkhISMDEiRNRqlQpNG/eHAsWLFBqW1FKSUnB1q1bYWFhgaFDhyqlyYNfxYoVU1vW3NxcKZ8669atU4yG0dfXh4uLC7Zt24ZJkybhwIEDOY5KIFLnWwtTnK1cHmOsiuNNRiZ+ePQcq15EAQB2vX6DSQ+eITlTYK9zWZxwdUB9YwNUMdBHOb2sEzUzbS2sKlcaVrra6HX3EeoaG2B5udLY9foN0oRAJwtT/Pb8FbzuPUTvuw+x9PkrTiRIGuG3U1mB8yYVLDGoXnn1eU7/AyEAWzMDzGv39vG3P3o1QO9a9mi/9gRS0jOxq18jpKRnwv/CfSxsXwNbrjyA57JjaPLbUQzaegEv4lQnqyX6WPp0ro1XV3zx68RWiH6ThL7jt2LWb8cQn/j2EVutbHPEaEnfXpbEJ6bAsVwJ/LXKG8kp6WjXby28OtbCxkXdMWPxUVStaI3G9RwwdsYeNO6yDE26LsfigFO5jvIlzRS37w5it9+A1EAXJWc2h+WsFtC2MYFueQtoFzeAtqURig1xg0jLxOvfzsGgXhmY962F+H3/QLu0KXSdiuPN9hC8mn8SrxacQnzg/f9UP/vgYEBMTAwA5PvZa0PDrGFDb968UUnr1auXyjL5BevWrVsRF5c1G+6ePXsQHh4OJycnxUz3cra2tnB3d8eDBw9w6NAhtW345hvliUYsLS1RrFgxxZ3sd8XExGDHjh3w9vZWLGvVqhVKliyJzMxM+Pv757C17yf7vinoM+056dOnj+LCsFixYujVqxfKlSuHrVu3ws/Pr1DWkZ28XwD52wZ5vwCUt19bWxuHDx/GmjVrUKtWLWRmZuL06dOYMmUKatSogYoVKyomQszO1NQUFy9exIIFC1CxYkWkpaXh8OHDGDt2LJydnVGnTh2V5/Q/NR8fH0RGRmLt2rWKSSnlEhOzZlnX1VUf8ZbJZEr51Mn+NoGMjAw8ePAAfn5+WL16NVxcXJRGz+TFxcVF7Sf7KA7679OWSDDA0gJOeln9b/HzKESnZ2DOk0hkAKikL0MZmS60JRJ4mhrjZFwCet99hCf/DmuuYqiPZeVKY5uTHRbaW8NUSwt+z17Bp7Ql1rx8jcXPo9CjuDl6ljDHkudRWPUyqgi3lujj+3HfNRz+5zmGNKiAvf3doautOqR/z80nmH7wBho5lMSZkc1hX+ztsGorE32s7uqGsyOb48CgxmjuZIXhO4MxsUklhL1OQI8NZ1DJ0gSbezWA/4UwdP93uDbRp6KtrYWJ3zdGFWcrAICv3yGlEQAZ2YK+GdkmwjYyyPo708LDGYc2DMS5v0ZgzZzOeBOXjOUbzmKRT3uM9P0LC1efwLwpbdG6aUWM9P0L63dc+kRbRl+CzPhUxO3LugkocywOLWMZpDJt6LlaIvnSE7yadxKZSWnQc7FE8VENUGKSO8x614BISkPCiXCYdq6MN1tCkHDkPkw6VoZeZUvEbglB0jnVV5p/qT44GGBmZgYAal+Npk5CQoJSuezs7OxUlllaWsLc3Bzp6emKV6HJ30uf0+Rw8jvRFy5cUJtuY2OjsszIyEjRtndt3LgRBgYGaNeunWKZtrY2evbsCeDta/sKS/Y7tjld7G3evFnlPfLZgxXvyv5qwczMTERHR+Po0aPo1KlTobU7u+zfb376RvZ9/27f0NLSQt++fXHx4kU8fPgQK1asQNu2bSGTyXD79m307NlTMcdCdnp6ehg9ejRCQ0Nx584d+Pn5oVmzZtDS0sLFixfRsmVL/PHHH++9jR9i586dmDdvHhYuXKjUr+Tkj3zkNEGl/DWM+X00RCqVokyZMujXrx+WL1+OR48eoW3btoiMjHzPLSBN8SRVdbLXsv/e7U8TAg9TUhH971stjLXeXsgY//tMdLIQCIyNV6kDABY8jUQzMyNUNdRH0JusY0AFPRkq/BtsOB6r/phM9KWLT0lDzw1n8MelCPw9wAP/+7YWACD0+dtguBAC8wNvoffGM/ildVUcGdwEVib6CH3+Bqnp6t8k88elCEQlpGCUuzP+Dn0CAHC1MkNJYz2UNJLh+P2XSEhRP2qSqLA8eBytsszZIWtOmNTUDMQnpEL674iApGyvGkxMevtvp3Il1NY9wncXxg1yh71tMew9lvVmjcpOpeD67+Mz+wLz96Yn0gzpkfFAelaQSaL/dqi/5N+AVGZsClLvvlIp92bLdRh5lod2cUOkhGSNwtWxMYG2ddacZMkhOY/M/dJ8cDDAyckJAPI9VP7Ro6xISsWKFVXSjI3Vz9Iov2ssv9scHZ11kFF3QSyRSBR36l+8eKG2PnV3qnOb72DNmjXo1q2b4m6snPxRgQcPHuDIkSM5llfH19dXpd3y58bLlSunuCOc03vs5a/hy/5e+U/N29s7x4CEvF8A+esb8n4BqO8bcra2thg4cCB2796NJ0+eKCYnnD59umLkiDqOjo4YOXIkDh8+jPDwcLRp0wZCCEyYMEEpn4eHh8o2+fr65tn+gjh48CC6d++OWbNmYeTIkWrzlCqV9Uctp7kO5L8BKyurAq+/Xbt20NLSQmxsLPbs2ZOvMjdv3lT7cXBwKPD66cvy3Z0IxL3zCsuXaW8vJoy0pND59/iZkO2uTkLG23+rC5VeS0hCYGw8xlhlnfBl/Jsr+6E4/T80DI9I7sLDKNRZeBAGOlq4Nr4Vvv73junT2CS0XZ01X9Dz2CS0WX0cf914jAujW2CUu7Pi4qnt6uN4GqsaZI9NTsPEvVex+Nta0NGSIj1T+TclP89Jz+driIneV43WCxUz/ss9efE20FXG2gzNGlQAALyMehsslv/b1FgPHvVUzy+27r2GR0/fYNyArFHB6fKLPIlE0c/T0vL3ymXSDNJsE7SK5HS1/373JCUp+AkyopNg5JnVR4V89IoEbw+oGf+d85MPDgbIJ9I7f/58nnfHhRA4f/48ACheqZZdThdz744mkD8v3bdvX8UFsbrPsmXL3mubsrty5QquXLmCpUuXqlwkVq789lUpBZ1I0NfXV6W9QUFBALLu9soDA6dOnfrgbfhYAgICVLZBPu9C+fLl4eiYNfPm2bNn86xLPtrDxcVFadK827dv49Yt9VFeCwsLrFixAtbW1khOTlZ6zOPq1asIDw9XW87W1hYbN26EtrY2njx5gqiot0ORg4KCVLapMIMBu3fvRocOHbBw4UKVNyBkV61aNQDIcRvCwsIAvN8bAXR0dBSvAM1tzgEiAMgUgN+zV8j49/h+MCYOlxOyTvJamhnDQU+Gbv9OJHgrKRmR/wYKTvx7V99ES4rmpkbv1Ckw/fELjLYqAbN/h0XXNsoa5XI/OQVh/458kS8j+q+IS05Do8WHce9VPP668RgVZu6G5U87YPnTDtRe8PbRtW/9T+Lg7We49SIWDf93SJHH8qcdeBSjfsSg74HraFrBEo3+vQPr8e9/Q1/EIiohBS/jk1HN2gymnHCNPrLMTIEf5x1Axr9B4e37ruN0cAQAoEubqrC3LYZfJ7aCgb4Ortx8iifP3yA9PQMHgrLmjJo9ubXSxIIAEJ+QgnEz92DJz99A99/J3DzcsiZTDr37ArfuvQQAuLvxJgW9pV3CEHpVswKuqfeikJmcDpGRiZTQrP6iZWEAXee3o1Ayk9PxZscNmHatCol21mWyrELWOXP6szikP8u6VtV1LP4pN+Oj+uCpETt16oRp06bhzp07OHLkCDw9PXPMe/jwYTx69AguLi749ttvVdLV3QV//vw5oqOjoa2trbjwkc88n9OFUlpaGo4ePQoXF5cPnoBvzZo1qFevnuK99e9av349vLy8sGvXLrx+/TrHCd8KavLkyTh06BA2bNiAqVOnfvBrA4vCTz/9hF69esHf31/x6sCcrF27VlEmu1mzZuHBgwcIDAxUW05bWxuWlpZ4+vSp0siNUaNGoWzZsjnO52BiYgJjY2NER0fn+Fx+Ydu2bRt69eqF5cuXKz3S8ejRI2zZskUpONC4cWMYGhri+vXrSEhIUJpTAXj7ykR1jxjkJS0tTfF4QMmSfJUb5a5nCXNciE9E+9sRSMzMxJuMDLjoy9DG3AQ9SmQFZidal4CzngxbomLQ9Z8H0Po3ct7JwhR9SxaDpa7ySd2mVzHQk0rxbTETxbJhpSwQk56BXx6/hADQ1twEw0opz6VB9KXLEEJxxz4qUfUxMJN/L4BS/n0MIDpJ/aNi77r+NBp/XH6A6+NbKZa1cy2NaS0qw/9CGHbfeIyGZUtgyXe1PnQTiPI0vE9DBJ29jypfz0dCUiqiohNRs3JpdGtfDcO9sl6NXN3VBse3DsHU+QdRq40ftLW1UKqEMf74Xw90bVdNpc5piw7jq9pl0aT+20k2F0//BpmZAh0GBEBAYLh3QwzpVf9TbSZ9Icz710LCiXAknXuEyJ+PAQAk2lIYNnOAkWcFSGVvL4fj/r4NWXkLyLIFCEy7VgGEwOvfsm5cGjYuB0MP9W/7+hJ9cDBAW1sbmzdvxldffYVhw4bh9OnTiruO2b169QrDhw+HsbEx/vjjD0ilqoMSNm7ciGHDhiktkz/T3aVLF8VjBG3btkXZsmVx4sQJ3Lt3D+XLK8++u379egwcOBB37979oG1LTk7Gxo0bsXTp0hzzdOnSBePGjUNkZCQ2bNiAESNGfNA65Tw8PDB+/HjMnTsXw4cPx+rVq3N8lCEj4/McEtWzZ08cPHgQGzZswOLFizF8+HC1+RYvXowzZ87Ay8sLXbp0UUm/cOECwsLCFK/Ty+7hw4e4efMmbGxsUKlSJaW0Q4cOITo6WjGSJLvz588jOjoa9erVy/HxlMIk75MBAQHo2rWrUtr9+/cxfvx4pWCAgYEBhg0bhtmzZ2PdunUYMmSIIi0sLAxHjhyBi4sLWrVqhYL6888/kZmZCR0dnVyDd0QAMMIq7+i3VCLBNxam+MYif2+o6FHCXBFIkDPW0sJsu4I/9kL0JTHT10X6/G555rs0tmWB6q1ibY7n01Vvsvzo6YofPV0LVBfRh5o+5ut85atZuTT+Dsj9ZpHc3B/aqCyzKWWKXav7FKhtpHkkOlowaloeRk3Vv60lO9PvVI+XWub6KDYk71e9f6k++DEBIGtI89GjR5GcnIzatWtj3bp1ePbsGdLS0vDs2TOsW7cOderUQXJyMo4cOZLj0GYjIyOMHj0az58/R1JSEjZt2oSffvoJZcqUwbx58xT5dHR0sHXrVpiYmKBly5Y4ePAgYmJiEBkZieXLl2P48OGYMWOG2ovHgti5cydkMhk6duyYYx6ZTIb+/fsDKPijAnmZPXs2pk2bhnXr1qFJkybYt28fXr9+jfT0dLx48QJ///03evXqhVmzZkFfXx+1an1+Ef+1a9di2LBhGDVqFPr374/g4GAkJCQgISEBwcHBGDBgAEaNGoVhw4Zh1apVautITExEixYtsGnTJjx69Aipqal48uQJNm/ejGbNmkEikWD16tUqAaanT5/C09MTe/bswdOnT5GSkoKHDx9i1apV6NChA8zMzHIN9BSWVatWwdvbG1ZWVti1axe6du2q9Jk+fbracj4+PqhXrx4mTZqEHTt2ICkpCdevX8d3330HAwMDbNy4EVpaqjNPq5OZmYlHjx5h9erVisDCr7/+qnbSTiIiIiIi+u+TiEKcBj8pKQlr1qzBzp07ERISgpiYGJiZmaFy5cr49ttv0a9fP7WT93l7e2PdunUIDAzErVu3sHTpUty7dw/GxsZo3bo1fvnlF1hbW6uUe/DgAX799Vfs378fz549g4WFBSpXrozhw4crzUkQEBCgmOxPzs7ODhEREfD19cW0adOU0ry8vBAUFKT02IK/v7/a2frV3a2XT+j3br3ytII+g37//n0sW7YMR44cQUREBOLj42FoaAhra2u4urqiefPm+Oabb5RGZMj36bvc3d0VcxPkxcPDA8ePH1dZHhgYqJjTIL+uXLmCZcuWISgoCI8fPwaQ9VYHDw8PfP/996hRo4baci9evMChQ4dw5MgRXL9+Hc+fP8erV68gk8lgZ2cHDw8PjBw5UjE/gdyDBw9w8OBBHD16FLdu3cLz58/x+vVrGBgYwMHBAZ6enhg1apTafpWXnPYtoH7/VqtWDdeuXcuzXnU/xeTkZMyePRsbN27Ew4cPYWpqCk9PT0yfPl0l2OXn54fRo0fnWL++vj5Kly4NNzc3DBo0CA0aNMizTXlxcXGBSE3BTqMPHmRERBqmwu6/gG2qfyeJCpN00Gzg1ZaibgbRf1aKdTc8TNj0cepOa4cWOyd+lLqL0pNBfxZ1ExQKNRjwvrIHAwp6kUlERYfBACJ6XwwG0KfAYADRx8VgQMF9TsGAQnlMgIiIiIiIiIi+HAwGEBEREREREWmYIg0GBAQEQCKRKJ6/bty4MSQSCSIiIoqyWURERERERET/aUUaDPD29oYQQuVjb29flM0iIiIiIiIi+k/jYwJEREREREREGobBACIiIiIiIiINw2AAERERERERkYZhMICIiIiIiIhIwzAYQERERERERKRhGAwgIiIiIiIi0jAMBhARERERERFpGAYDiIiIiIiIiDQMgwFEREREREREGobBACIiIiIiIiINw2AAERERERERkYZhMICIiIiIiIhIwzAYQERERERERKRhGAwgIiIiIiIi0jAMBhARERERERFpGAYDiIiIiIiIiDQMgwFEREREREREGobBACIiIiIiIiINw2AAERERERERkYZhMICIiIiIiIhIwzAYQERERERERKRhGAwgIiIiIiIi0jAMBhARERERERFpGAYDiIiIiIiIiDQMgwFEREREREREGobBACIiIiIiIiINw2AAERERERERkYZhMICIiIiIiIhIwzAYQERERERERKRhGAwgIiIiIiIi0jAMBhARERERERFpGAYDiIiIiIiIiDQMgwFEREREREREGobBACIiIiIiIiINw2AAERERERERkYZhMICIiIiIiIhIwzAYQERERERERKRhGAwgIiIiIiIi0jAMBhARERERERFpGAYDiIiIiIiIiDQMgwFEREREREREGobBACIiIiIiIiINw2AAERERERERkYZhMICIiIiIiIhIwzAYQERERERERKRhGAwgIiIiIiIi0jAMBhARERERERFpGAYDiIiIiIiIiDQMgwFEREREREREGobBACIiIiIiIiINw2AAERERERERkYZhMICIiIiIiIhIwzAYQERERERERKRhGAwgIiIiIiIi0jAMBhARERERERFpGAYDiIiIiIiIiDQMgwFEREREREREGobBACIiIiIiIiINw2AAERERERERkYZhMICIiIiIiIhIwzAYQERERERERKRhGAwgIiIiIiIi0jAMBhARERERERFpGAYDiIiIiIiIiDQMgwFEREREREREGobBACIiIiIiIiINw2AAERERERERkYZhMICIiIiIiIhIw0iEEKKoG0FEXyYXFxeIjHRc/mtXUTeFiL4wOhamkGamFnUz6L/OxBiQphd1K4j+s9KlusjAxzmWRyUB0SmJH6XuouRavGxRN0FBu6gbQERfNomWNvScKhZ1M4iIiIjoE9P59/Mx2Bhnfejj4WMCRERERERERBqGwQAiIiIiIiIiDcNgABEREREREZGGYTCAiIiIiIiISMMwGEBERERERESkYRgMICIiIiIiItIwDAYQERERERERaRgGA4iIiIiIiIg0DIMBRERERERERBqGwQAiIiIiIiIiDcNgABEREREREZGGYTCAiIiIiIiISMMwGEBERERERESkYRgMICIiIiIiItIwDAYQERERERERaRgGA4iIiIiIiIg0DIMBRERERERERBqGwQAiIiIiIiIiDcNgABEREREREZGGYTCAiIiIiIiISMMwGEBERERERESkYRgMICIiIiIiItIwDAYQERERERERaRgGA4iIiIiIiIg0DIMBRERERERERBqGwQAiIiIiIiIiDcNgABEREREREZGGYTCAiIiIiIiISMMwGEBERERERESkYRgMICIiIiIiItIwDAYQERERERERaRiJEEIUdSOI6MtkbGyMtLQ0ODg4FHVTiIiIiIg+ew4ODti9e3dRNwMAoF3UDSCiL1dSUhIYT6T3df/+fQBgMIneC/sPfQj2H3pf7Dv0Ie7fv4+HDx8WdTMUGAwgovfm5OQEALh582YRt4S+RC4uLgDYf+j9sP/Qh2D/offFvkMfQt5/PhecM4CIiIiIiIhIwzAYQERERERERKRhGAwgIiIiIiIi0jAMBhARERERERFpGAYDiIiIiIiIiDSMRPC9YEREREREREQahSMDiIiIiIiIiDQMgwFEREREREREGobBACIiIiIiIiINw2AAERERERERkYZhMICIiIiIiIhIwzAYQERERERERKRhGAwgIiIiIiIi0jAMBhBRgSUnJ2P69OlwcnKCnp4erKys0KtXL4SHhxd10+gzcOzYMfTr1w+Ojo7Q09ODgYEBKlWqhPHjxyMyMjLHchs2bICbmxuMjIxgbm6O5s2bIzAw8BO2nD5HDx8+hImJCSQSCYKCgnLMx/5D2V25cgW9evVCmTJlIJPJUKJECdStWxfjxo1DXFycSn72HwKAkydP4ttvv0W5cuWgr68PW1tbeHp6Ys+ePTmWYd/RTMnJyfjxxx+hq6sLb2/vPPPv378fTZo0gampKUxMTNCwYUNs37491zJPnz5F//79YWNjAz09PZQvXx6TJ09GfHx8IW0FgwFEVEBJSUlo2rQp5s6dixkzZuD169c4cOAAQkJCUL16dVy7dq2om0hFaOnSpWjatCmCg4OxfPlyREZG4u7du/D29saiRYtQuXJl/PPPPyrlRo8ejV69eqFFixZ4/Pgxbty4geLFi6Np06bw9/cvgi2hz0X//v3VXrxlx/5D2S1btgz16tWDk5MTTp8+jdjYWBw+fBhaWlqYP38+oqKilPKz/xAALFmyBI0aNcLdu3exYcMGREVF4dixYzA1NUW7du0wYsQIlTLsO5opMDAQVapUwbJly5CWlpZnfj8/P7Rq1QoVKlTAP//8g/DwcNSpUwedOnWCr6+v2jJ3795FtWrVcOrUKezatQvR0dHw8/PD8uXL0bBhQ7x586ZwNkYQERXAhAkTBACxZMkSpeX37t0TWlpawsXFRaSnpxdR66iozZ07V+jq6oqHDx+qpE2cOFEAEC1atFBavm/fPgFAdOzYUWl5amqqqFChgtDT0xMREREftd30eVqxYoUwNzcXzs7OAoAIDAxUycP+Q9kFBgYKiUQi5s+fr5J2584d4eTkJB4/fqxYxv5DQmR93yYmJgKAuHDhglJaYmKiKFasmAAgbt++rVjOvqOZ/vjjD2FqaiqWLl0q1q5dKwAILy+vHPOHhIQILS0tUbt2bZGZmamU5uHhIQCIs2fPqpRzc3MTEolEXL9+XWm5v7+/ACAGDx5cKNvDkQFElG+JiYlYunQpdHV14eXlpZTm4OCAZs2a4ebNm9i3b18RtZCKWqlSpdCtWzfY2tqqpLVr1w4AcOTIEWRkZCiWz507FwAwcOBApfw6Ojro27cvkpOT8b///e8jtpo+Rw8fPsT48eOxcOFCWFpa5piP/YeyGzduHExMTDBs2DCVNEdHR9y+fRs2NjaKZew/BADR0dGIjY0FALi4uCil6evrw8HBAQBw9epVxXL2Hc1UtmxZhIaGYsiQIZBIJHnmX7BgATIyMjBgwACV/IMGDQLwti/JnThxAufOnUO9evVQuXJlpbRu3brBxMQEa9aswevXrz9wa/iYABEVwLFjx5CQkICqVavCyMhIJb1hw4YAgN27d3/qptFnomfPnggICFCbZmpqCgCQSCQQQgAAYmNjcfz4cUgkEtSrV0+lDPuU5urfvz8aNGigEnjMjv2HsgsNDcWlS5dQv3596Orq5pmf/YfkSpYsqQhi37x5UyktOTkZ9+/fBwBYWVkBYN/RZG5ubrC2ts53fvl8Ew0aNFBJk/eTffv2IT09XbFc3m/UlZHJZKhduzbS0tKwf//+ArVdHQYDiCjf5PMBlC1bVm16uXLllPIRZXf79m0AWX/8tLW1AQAhISHIzMxEiRIl1AaY5H3q/v37SEhI+HSNpSK1atUqnD9/HitWrMg1H/sPZXf69GkAgJ2dHQ4dOoTGjRvDxMQEBgYGqFatGubPn690ws3+Q9lt2LABlpaW6Nu3L86ePYukpCTcvXsXPXv2xOvXr1G7dm3FxRn7DuXHkydP8OrVKwDqz51tbGygq6uL5ORkpfmUPuX5NoMBRJRvz58/BwAUK1ZMbbq5ublSPqLs1q9fD4lEgqlTpyqW5bdPCSHw4sWLj99IKnKPHj3CuHHjMG/ePLWPm2TH/kPZ3b17FwBw8OBBdO7cGQMGDMCDBw9w79491KtXD+PGjUP79u0Vjymx/1B2jRo1woULF+Ds7Iz69evDwMAAjo6OOHHiBIYMGaKYhBJg36H8kfcTPT096Ovrq6RLJBKYmZkp5c3+709xvs1gABHlW2JiIgDkOPxSJpMp5SOSO3DgAHbv3o0xY8bAw8NDsTy/fSp7XvpvGzBgAGrXro0BAwbkmZf9h7KTz64dHh6OuXPnonv37jA3N4e1tTWWLVuG6tWrY9++fVizZg0A9h9StmfPHlSvXh1hYWE4deoU4uLicOvWLfTq1QspKSlKd/jZdyg/8uongPpz5095vs1gABHlm4GBAQAgNTVVbXpKSopSPiIg626dl5cXvvvuO8yePVspLb99Knte+u9avXo1Tp06hdWrV+crP/sP5aRHjx4qy+TzT/zxxx8A2H/orQcPHqBr165ITU3Fvn370KBBAxgZGcHZ2Rnz5s3D1atXUaVKFTx8+BAA+w7lT179BFB/7vwpz7cZDCCifCtVqhQA5Dh7aXR0NIC3E+wQPXjwAJ6enmjYsCH++OMPxRBLufz2KYlEkuuM8vTle/z4McaOHYtZs2bB3t4+X2XYfyg7+ZDa4sWLqz1Jlj9/K382l/2H5DZv3ozExEQ0adJE5buWSCTo1q0boqKiFO+EZ9+h/JD3k+TkZCQlJamkCyEQExMDQPnc+VOebzMYQET5Vq1aNQBZQzDVCQsLAwBUqVLlUzWJPmP37t1Do0aN8NVXX2Hr1q1qh7tVrlwZUqkUkZGRaidZkvcpBwcHGBoafvQ2U9E5cuQIYmNjMXz4cEgkEqXP8ePHAQCNGzdWLAsKCmL/ISWVKlUCkPtdOACK13ux/5Cc/Lwmp4sr+ezxly5dAsC+Q/ljY2OD4sWLA1B/7vzkyROkpqZCX18fFSpUUCz/lOfbDAYQUb41btwYhoaGuH79uto/fvKZnOXvkyfNFRoaikaNGqF58+ZYt26d0oiAefPm4dGjRwAAExMTuLu7QwiBs2fPqtTDPqU5vL29IYRQ+3F3dwcABAYGKpZ5eHiw/5CSZs2aQSqVIjY2Vu0dtYiICABvgwbsPyQnv2B7+vSp2nT5ch0dHQDsO5R/bdu2BQCcOXNGJU3eT1q2bKl4y1JeZVJTU3Hx4kXo6OigZcuWH9w+BgOIKN8MDAwwbNgwpKSkYN26dUppYWFhOHLkCFxcXNCqVasiaiF9Dq5duwYPDw989913WLlyJaRS5T8148ePV7yzGQAmTJgAAFi5cqVSvrS0NPj7+0NPTw8jRoz4+A2nLxL7D8lZWVmha9euAICAgACV9PXr1wMA+vfvr1jG/kPA24v2wMBAvHz5UilNCIHNmzcDADw9PRXL2XcoP8aOHQstLS2sXr0aQgilNHnfGT9+vNLyRo0awc3NDWfOnMHNmzeV0jZv3ozY2Fj069cvx7cNFIggIiqAxMREUa9ePWFsbCy2b98uEhMTxbVr10S1atWEiYmJuHr1alE3kYrQxYsXhbm5uTAwMBBdunRR+wEgAgMDlcqNHDlSABDTpk0TMTEx4smTJ6JHjx5CIpGINWvWFM3G0GfD3d1dbb+RY/8huZcvXwonJydhbGwsNm/eLBITE8XTp0/F0KFDBQAxYMAAlTLsPySEEGPGjBEARK1atcSZM2dEfHy8uH37tujRo4cAIFxdXUV0dLRSGfYdzebv7y8ACC8vr1zzLVy4UAAQgwcPFi9evBBRUVFi7NixAoDw8fFRW+aff/4RJUqUEM7OziI4OFgkJSWJvXv3CnNzc1GlShURExNTKNvAYAARFVhSUpLw9fUVFSpUEDKZTJQsWVL06NFD3L9/v6ibRkVMfmKU10fdRd369etFnTp1hIGBgTAxMRHNmjUTR48e/fQbQZ+F8PDwHPuPu7u7Sn72H5KLiYkREyZMEOXLlxe6urrC1NRUeHh4iE2bNuVYhv2HhBDizz//FC1atBDFixcXWlpawsjISNSsWVPMnDlTxMfHqy3DvqN5cju/8ff3V1vm77//Fh4eHsLY2FgYGRmJ+vXri23btuW6nsePH4t+/foJKysroaurK8qWLSsmTpwo4uLiCm1bJP9uEBERERERERFpCM4ZQERERERERKRhGAwgIiIiIiIi0jAMBhARERERERFpGAYDiIiIiIiIiDQMgwFEREREREREGobBACIiIiIiIiINw2AAERERERERkYZhMICIiIiIiIhIwzAYQERERERERKRhGAwgIiIiIiIi0jAMBhARERFRnpKTkxEQEPDR6j958iRu3Ljx0eonIiJlDAYQERERAfDw8IBEIinQx8PDo8Dradu2LaysrHDnzp33bmtQUJBSOz7mRToAhIeHo06dOjhy5IjS8mfPnqFTp04wNjaGqakpunbtihcvXqit49GjRzA2NsacOXPUpkskEtSvXx9+fn6F3XwiIlKDwQAiIiKif/n4+EAIofSRe3e5j4/Pe60jLCwMkZGReP369Xu308PD44PaUBCRkZFo2rQpypYtqxJ0+PbbbxESEoJLly4hODgYV65cwXfffae2nu+//x7ly5fHmDFj1KY3bNgQu3btwtixY/G///2vsDeDiIjeoV3UDSAiIiLSJJcuXUJcXBxKlChR1E3Jl+HDh+PJkycIDAyEtvbbU8fAwECcO3cOGzZsgKOjIwBgypQp6N27N44fPw53d3dF3i1btuDAgQM4d+6cUh3vatKkCXr16oXx48fD09MTFStW/HgbRkSk4TgygIiIiAjA+PHj0a5du3znb9euHcaPH1/g9ejp6X0xgYDQ0FBs2bIFXbp0gZ2dnVKa/Pn+qlWrKpbJ/5392f/o6GiMHDkSw4cPR61atfJc5+TJk5GamooZM2YUxiYQEVEOGAwgIiIiAtC6dWvUqFEj3/lr1KiB1q1b49y5c0rP7/v6+iIwMBDu7u4wNjZWLM/rOf/nz59jzpw5aNSoEaysrKCrqws7OzsMGDAAjx8/LtC2CCGwbt061KtXDxYWFjAxMUHVqlUxYsQInD59Ot/1rF+/HgDg6empkpaZmQkg61l/Oak069QyPT1dsWz8+PGQyWT5vrh3cnKCnZ0dtm/fjvj4+Hy3lYiICobBACIiIqIP4ObmBiEE/P39AQCnTp3CzJkzsWTJEjx//lwxeiCv5/x37dqFSZMmoWXLlrh+/Tpev36NdevW4ezZs6hduzaePn2a7zb9+OOP6N+/P/r164f79+/j6dOnmDlzJrZs2YIePXrku56jR48CAFxcXFTSnJycAAA3b95ULAsNDQUAODs7AwCOHz+OtWvXYtmyZTA0NMz3el1dXZGamoqTJ0/muwwRERUMgwFEREREhejChQvYsWMHKleuDENDQ4wYMQJdunTJs1zx4sUxcOBATJ48GSVKlICRkRE8PDzw+++/4/nz55g7d26+27BixQpUqVIF/fv3h5mZGYyMjNCmTZsCDb0XQigu7m1sbFTSPT09UalSJfzyyy+IjIxEVFQUZs2aBVdXVzRr1gwpKSkYOHAgOnfujFatWuV7vdnXlz3QQEREhYvBACIiIqJC1KpVK5iamir+v3Tp0ti8eXOe5Tp27Ijly5erLJc/h1+Qu+QSiQR37tzBpUuXlJb37NkTx48fz1cdMTExSExMBACYmZmppGtpaWHHjh3Q09NDqVKlYGlpCZlMhu3bt0NLSws///wzIiMjsWjRIgDAn3/+ifr168Pa2hr169fHnj17cly3fH0FfTyCiIjyj8EAIiIiokJUpkyZ9yonhMC2bdvQuHFj2NjYQEtLCxKJBFpaWgBQoFcRjhs3DgkJCahTpw6aNm2K3377DY8ePYK+vr7KRIA5SUhIUPxbV1dXbR5nZ2ecO3cOUVFRiIqKwtmzZ+Hk5IQbN25gzpw5mDt3LiwtLXH06FF8++23cHNzw6FDh1C3bl20b98eJ06cUFuvTCZTaQMRERUuBgOIiIiICpGBgcF7lRs6dCg6d+4MPT09HDx4EElJSRBCQAgB4O2EffkxadIkHDx4EC1btsTJkycxdOhQ2NnZoXnz5rh+/Xq+6sj+CsCMjIxc85qZmSlGQ2RmZmLAgAGoX78++vbtCwCYMWMG7OzsMH/+fLi6umL+/PmwtbXFtGnT1NYnX19uryEkIqIPw2AAERERURF7+vQpli9fDqlUij/++AOurq453o3Pr+bNm2Pv3r2IjIzE77//joYNG+Lw4cOoV68ewsLC8iyf/VGH5OTkfK/3t99+w5UrV7By5UrFmwauX7+OGjVqKP5fKpWiRo0aOQYmkpKSAKh/PIGIiAoHgwFERERERSwiIgJCCBQvXhzm5uZKafIL44LYtWuX4rV8pqam6NmzJ06cOIHevXsjMTERu3fvzrMOfX19WFtbAwAiIyPztd7Hjx/jhx9+wJQpU+Do6KhYnpmZqXjcQU5bWzvHEQfy9ZUvXz5f6yUiooJjMICIiIioiMnnGYiMjMSzZ8+U0t7n9XrffPON2okC5ZMR5vdRhjp16gAA7t+/n6/88scRJk6cqLS8UqVKuHHjhtKyGzduqH1lIQDcu3dPaf1ERFT4GAwgIiIiKmKlS5dG165dIYRA586dERISgoSEBBw7dgyDBw9+rzpHjhyJQ4cOITY2FvHx8Thy5AgWLlwIe3t7dO7cOV91tG3bFgAQHBycZ97t27dj7969WLlyJXR0dJTSJk6ciNu3b2PBggV48+YN/Pz8cPv2bUyaNEmlnpSUFFy/fh1ly5ZF5cqV89VOIiIqOImQz0pDRERERAry59vf9e6pU0REBMqWLauSz8vLCwEBAYr/DwoKQuPGjVXyBQYGwsPDA6mpqfDz88O6desQFhYGmUyGOnXqYNKkSWjatKkiv4+PDzw8PHKt6+DBg9i6dSvOnj2LJ0+eIDMzE2XKlEG7du0wZswYlChRIl/7ICkpCWXKlIGNjQ2uXLmS4z558+YNKlasiG+++QZLly5Vm2f16tWYPXs2Hjx4ADs7O0yePFkxwWB2W7ZsQdeuXbFw4UKMGjUqX+0kIqKCYzCAiIiIiHK0fv16eHl5YevWrejUqdNHXVd6ejpcXFygq6uLy5cvq4wwICKiwsP3tRARERFRjnr37o1r165h0KBBKFu2LGrVqvVR1pOWloaePXsiISEBBw4cYCCAiOgjYzCAiIiIiHI1f/58VKhQAUOGDMGFCxc+yjpWr16N+Ph4BAcHo1SpUh9lHURE9BYfEyAiyibt5XNkREcXap1a5ubQKan+xNbe3h4PHjzIsayBgQGcnZ3Ru3dvDBs2TOXVXNk9ffoUS5YswaFDhxAWFoa4uDiYmZnByckJbm5u8PT0RJMmTXi3jUhDpGXEIUMkFm6lmTLo6ZrlmPwhx7SEhAQYGhoq/p/HtIKJiYmBn58f7O3t4e3t/dHWc+TIEXh6eir+39/f/6Our7Dt2rULV69ehbe3N+zt7QtUds6cOfj555+xatUqdO3a9eM08BP7kP1R2CIiIhAQEIBq1aqhQ4cORdoWTcGRAURE2WRERyOi6zeFWqf95j9zDAZEREQAeDtRWXh4uOKPcVxcHIKDgzF8+HCMGjUKp06dwrZt29TWs2rVKowYMQJubm749ddfUbt2bejo6ODRo0fYvXs35s2bh/nz56NYsWKIiooq1O0jos9ThkjEw/jfC7XOMka9ck3/kGNa9kAAj2kFFxMTg2nTpsHd3f2jXpw3a9YMQgj4+vpi2rRpH209H8uuXbuwbt06eHh4FPji98GDB4iPj8ejR48+TuOKwIfsj8IWERGBadOmwcvLi8GAT4SvFiQi+kwZGxujcePG+OOPPwBkvbbr9OnTKvmWLl2KgQMHol27djh69Cg8PT1hZmYGQ0NDODs7Y8KECQgKCoKpqSlev379qTeDiAgAj2n05VuyZAkePXqE8ePHF3VTiAoFgwFERJ+5KlWqwMTEBABw7tw5pbS7d+9i9OjRMDY2xrJlyyCVqj+sV6pUCcOGDfvobSUiyguPafSlkkgkKF26dFE3g6jQMBhARPQFkE/vIpPJlJbPmzcPaWlp6NChA4oVK5ZrHf369cPYsWPztT49PT1IJBJIJBJ4eHggLCwMHTp0UNyda9asGW7cuKG2bFpaGhYuXIjq1avDwMAAJiYmqF+/PtatW6eUz97eXrEO+Uc+xDgiIkIlLbuTJ0+ibdu2sLCwgEwmg4ODA8aMGaMyXLhFixZKdSQmJmLcuHEoU6YMtLW1IZFI4Ovrm+f+OHnyJNq0aQMbGxsYGBjAyckJ3bt3x44dO5Cenq6UVwgBf39/1KtXD8bGxjA0NESNGjXg5+enklcuJCQEXbt2RalSpaCrqwtbW1sMGDAADx8+VMo3ePBgpe25d+8efH19YW9vD5lMhgoVKmDFihV5bg9RUfvUxzQA2L9/P/r06YOKFSvC0NAQxsbGqF+/Pn7/Xf2jFHFxcZg+fTpcXFxgZGSEkiVLokGDBvj5559x7949pbxhYWEYMGAAypUrB319fdjZ2aFly5ZYvnw5otXMQ1PYxzB7e3uULVsWAHD8+HGlMkFBQe+1brmrV6+iVatWMDExgbGxMb766iscPXo0v7s9T+fOnVNqr6+vL44cOYJ69erBwMAAFhYW8PLyQkxMjNryUVFRmDBhAhwdHaGnp4dixYqhefPm2Ldvn1K+gIAASCQSxd+ixo0bK/2dy0v2NmZ/DOND25+bs2fP4ptvvkHJkiUhk8lgb2+PwYMHqzymUK1atRy35d22FWR/eHh4KJbZ29sjPDwc3bt3V7SnUqVKWLRoETIzMxX1+vn5qf27DgDe3t5K9b3bzsaNGwMA1q1bl2MdVMgEEREpJN0OFbeqORXqJ+l2aJ7rBSAAiPDwcJW0q1evCgBCIpGIK1euKKXZ2NgIAGLp0qWFtAfeCg8PFwCEq6uraNy4sTh79qyIi4sTe/fuFcbGxsLGxkYkJiYqlUlJSRFNmzYVUqlUzJ8/X7x580ZERkaKn376SQAQQ4cOVcq/YMECAUAMHjxYbRvat28vRo4cqbRs5cqVQiqViq+//lrcvXtXJCcni/3794tSpUoJBwcH8fz5c5V67OzsBADRrl07sWLFCvHq1SsRGhoqSpYsKXx8fHLdD/v37xdSqVQMHTpUPHr0SCQlJYng4GBRt25dtd9Zr169BAAxYcIEERkZKd68eSMWL14stLS0RNu2bUVmZqZS/r179wo9PT1Rs2ZNcfXqVZGcnCzOnDkjnJycRIkSJURoqGr/cXd3FwBEmzZtxG+//SaioqLEvXv3RKNGjQQA8ddff+W6TaQZktKeizvRcwv1k5Sm+vtS53M8pllaWoqqVauKc+fOiaSkJPHo0SMxdepUAUD89NNPSnkzMzNFgwYNhJWVlTh27JhISEgQkZGRws/PT0ilUuHl5aXI+/TpU2FhYSHc3d3FjRs3RHJysggPDxd9+vQRAIS/v79S3R/rGCY/Zru7u+e4Dwq67rNnzwoDAwNhZ2cnjh8/LpKTk8WtW7eEp6en8PDwULt978vf318AEE2bNhXfffeduHv3rnj9+rWYO3euACC++eYblTIRERHC3t5eWFlZiX379omkpCTx8OFD0b9/fwFATJs2TaWMl5eXACACAwML3MbAwEABQOn7/5D252bNmjVCS0tLtG/fXvFdnTp1Sjg6OgoLCwtx6dIlpfy5ff/ytqn7e5ef/QFAmJubC2dnZ7F9+3aRmJgoHj9+rOjjAwcOVCkj/zul7hgAQNjZ2aksz23/0sfBYAARUTafUzAgNjZWHDt2TLi4uAipVCpmzpypVCYuLk5Rbs+ePYW9KxQnFgDEhQsXlNK+//57tRedU6ZMEQCEt7e3Sn1ff/21ACAOHz6sWBYVFSVkMpkwMTER8fHxSvkfP34stLW1lS6GQ0NDhY6OjjA1NRUxMTFK+Tdt2iQAiB49eqisW34iPXXqVKXlCxcuFNu2bct1P3z77bcCgIiNjVVa/s8//6h8Z6tXrxYAhIeHh0o9gwYNEgDEqlWrFMtevXolTE1NhZaWlrh3755S/rNnzwoAokGDBip1yU+yhgwZorT8/Pnz73XSSf9Nn1swoKiPaU2bNlUJPgghRLt27YSurq54+fKlYtn169cFADFmzBiV/N27d1e6WPnf//4nAIjdu3cr5cvIyBClS5dWulj+mMewvIIBBV13ZmamqFixogAgjh8/rpT/1atXQl9f/6MEAywsLERSUpJSmrOzs5BKpSI6Olpp+VdffSUAiEOHDiktz8zMFG5ubgKAOHr0qFLaxw4GFKT9Obl165bQ0dER9vb2IiUlRSnt5s2bQiKRCAcHB5GWlqZY/rGDAQDEb7/9prQ8IyNDVK1aVe13wGDAl4GPCRARfUbKli2rGBZnYmKCJk2awMXFBSEhIfjhhx+U8r5580bxb319/Y/WJhsbG9SuXVtpWcWKFQEA//zzj2JZRkYGli5dCgAYNGiQSj09e/YEAKxcuVKxrFixYujYsSNiY2OxefNmpfxr1qxB/fr1FesCgGXLliEtLQ1du3aFqampUv6OHTtCJpNh69atSvsmu27duin9/6hRo9CxY0f1G/4v+SMKGzduVFpeoUIF3L17V+n50UWLFgHI//avW7cOb968QbNmzeDg4KCU383NDQ4ODjh9+jRCQ0PVtu2bb5TffKHueyEqSp/TMe3IkSOoVq2ayvKqVasiNTUV58+fVyyT/+6PHDmCly9fKuVfvHgx5syZo5J3+/btSE1NVSyXSqU4d+4cvvvuO8WyojiGve+6z549i1u3bsHa2hqNGjVSym9hYYGWLVvma70F1aJFC+jp6Sktq1SpEjIzM5Uez7hy5QpOnjyJUqVKKb3qEMj6Tnr1ynrzhfy4/Knkt/25kX9XXbp0ga6urkpdtWrVwv3797Fnz55Ca3d+yP+OyUmlUkWfXL169SdtCxUOBgOIiD4j4eHhEEIgNTUVFy5cQIUKFbBt2zYcOnRIJa+ZmZni34mJhfwu8WxsbGxUlhkZGQHIei+43J07dxAdHQ2pVIqqVauqlClTpgwA4MKFC0rLBw4cCED5IjkjIwOrV69WuaiWTzZWvXp1lfq1tbVhbW2NtLQ0XLlyRe22yNtQEMOGDYNMJsP3338PV1dXTJ8+HVevXgUAlC9fHtraWW/pTUhIQEhISI7tk6/76tWrSEtLy3N7spd5d5/JvfvdqPteiIrS53RMe/ToEUaMGIFKlSrByMhIEaT4+eefAUDpzQQuLi5o0aIFrl+/jrJly6Jbt27YsmULYmNjUaxYMZQsWVKRt2vXrihdujTWr18POzs7jBgxAkePHkV6ejpsbGxgbGysyFsUx7D3XfelS5cAAM7Ozmrr+1ivocvv35yzZ88CgFLAODsXFxelfJ9KftufG3mbK1WqpDa9KLatWLFiSn1ZTr7/L1++/MnaQoWHwQAios+Qjo4OateujW3btkEqlWLChAkqdxQMDQ0VJx0f853H6u7Qye+EiX8nAQOgmCQrMzMTBgYGKhMAuru7AwBevHihVFejRo3g7OyMCxcu4Pr16wCyJvpKTExUuqOWfR3vTqQn/4SHh6tdh5yBgUGBt9/DwwNXrlxB//798fjxY/j4+KB69epwdXXFjh07FPmyTw7l7Oys0jY7OzsAWRMsyi865Nsza9YstdsTGBiY6/a8+92o+16IPgdFfUz7559/UKVKFaxatQoTJ05EREQERNbjsvDx8QEApUnQJBIJ9u7di+XLl6NSpUrYvHmzYpLPfv36KU22V7x4cVy7dg0//fQTZDIZFi9ejGbNmqF06dKYOXOm0sShRXEMe991y0cIGBoaqq1P3YVhYSjo3xz5hfa75MvVTeD4MeW3/bn5HLctp+9b3j/eZ4JEKnoMBhARfcaqVq2K3r17Iy0tDTNmzFBJb926NYD83R1ISEjA8+fPkZSUVOjtBABzc3MAWSf9GRkZihPtdz/q1i8fHSCfCX/FihXw9vZWmWlcvo7169fnWL8QAl26dCnUbatYsSJWrVqFyMhI7N+/Hz169MDt27fRsWNHxeMN2e9qPnz4MNf2WVpaKm3P9OnTc80/ceLEQt0eoqJSVMe0+fPnIyYmBoMHD4aXlxeKFy+eZxktLS0MGjQIFy9eREREBObNmwdra2usXbsWHh4eShf5xYoVw/Tp0xEREYHg4GCMHz8eycnJmDJlitIrEIvqGPY+65Yf03K6mx0XF1fobSwI+fbEx8erTZcvl+f7krzPtr371p3sCmOkTU7ft7x/ZP8bmFt7PuZIRio4BgOIiD5zP/74I7S0tLBx40aEhYUppU2cOBG6urr466+/lIa4qtO+fXvY2tri2bNnH6WdTk5OMDc3R1paGh4/fqw2z+XLl1XeKw4AXl5ekMlk2LhxI+7cuYMDBw6ofe7ezc0NABR3sN714sULHDhwoFCHyZ86dUrxWiMdHR20aNECGzZsUDzWIA8GGBoaonLlyrm279atW4q7/UDe2xMbG4sDBw7g1atXhbItRJ+DojimyX9jjo6OKmnqggmvXr3CgQMHFP9vZ2eHsWPHIiQkBE5OTrhx44bi9aq3b99GcHCwIm/NmjUxZ84cXLhwAVpaWkrzoXzMY1huF4Pvs+6aNWsCyNo+dR48eFDgNhYm+faEhoaqveN+8+ZNAEC9evWUlue1nz4H8m2Tb8O71G2bfESCur6T099kIP/74/Xr12qDE/I5beT9Ja/2FEZbqPAwGEBE9JlzcHBAjx49kJ6ejpkzZyqllStXDkuXLkVcXBy+//57pWGu2W3duhVHjx7FoEGDUK5cuY/STi0tLQwdOhRA1uR/74qKikKTJk1w8OBBlTT5RIJv3rzBt99+i0aNGqFChQoq+b7//nvo6Ohg/fr1Snfl5MaNG4dhw4YV6uRjU6ZMUUyMmJ18IrLsw3ZHjhwJQP32p6SkoG3btli/fr1imZeXF0xNTbFz5061Qyznzp2LLl26qIyQIPqSFcUxTf6s/bVr11TSTp48qbLsxo0baNOmDWJjY5WW6+vrKwIK8t/+5s2b8f3336vUUaFCBRgaGiodIz7mMaxYsWIAlIMb8+fPV0wwWNB1u7m5wcXFBU+fPsXx48eV8r5+/Rr79+8vcBsLU40aNdCwYUO8ePFC5e+KEAK///47AGDEiBFKae/up6SkJLi6uuLYsWOfoNX5I/+utmzZgpSUFKW00NBQBAcHo2zZsmjTpo1ieYkSJWBubo67d+8iIyNDsVwIgT///DPHdRVkf7w7kW5mZqYi2NWvXz+lNCcnJwBZQfDssj9el1dbAGDMmDEq3yEVHgYDiIi+AFOmTIGWlhbWr1+vclenf//+WLt2Lfbs2YNmzZrhyJEjiI2NVUxoN27cOPTs2RO9evX66LMq//TTT2jatClmzZqF2bNn48GDB0hMTMTJkyfRvHlzODo6YvTo0WrLyh8VCA0NVTsqAMgarr906VKEh4ejTZs2uHz5MhITExEeHo6RI0dix44dWLlyJaTSwv3ztmTJEqxZswYvX75EcnIyrl27hnHjxkEmkykCAEDWyVDv3r3x+++/Y/z48bh79y6SkpJw+fJltGnTBkIIxWRlQNaM3Bs3bkRKSgo8PT1x+vRpxMfH4+nTp5g5cyZmz56NpUuXfrRnc4mKyqc+pg0fPhx6enpYs2YNVqxYgejoaDx79gyjRo1SO1oJyJrItHPnzrh69SqSkpIQFRWFlStX4sCBA+jQoYPSKIPg4GCMHTsWERERSE1NxcOHDzFixAjExsZiwoQJinwf8xhmbGyMSpUq4datW4iIiMCzZ8+wdu1axVDygq5bIpFg7dq1MDQ0hJeXF06cOIHU1FT8888/6NKlS64TCPbs2RMSiQS7du0q8HYUxPr161GmTBn06dMH+/btQ0pKCh4/foyBAwfi3Llzir9J2cnvuh85cgRJSUnw9/fHnTt3Plqg/H3Iv6tHjx6hU6dOuHfvHlJTU3HmzBl8++23MDMzw9atW6Gjo6MoI3+Dwps3bzBhwgS8evUKjx49wvfff5/rd5Xf/WFpaYmNGzfizz//RHJyMp4+fYoBAwbg2rVrGDBggMobHXr27AmpVIrp06fjxo0biI2NxZYtW3KcHBPImpTXwsICwcHBePnyJe7du4eNGzcqHq2jj6BQX1RIRPSFS7odKm5VcyrUT9Lt0BzXJ3939Lsfde/77dmzp1Kes2fPKqU/ffpUTJo0SVStWlUYGxsLbW1tYW1tLdq3by/+/PPPAu0Hde2Sv0taXXuzv0c4NTVVLFq0SNSsWVMYGBgIExMTUblyZTFjxgzx5s2bXNfr7OwsSpYsKVJTU3PNd/r0adG+fXtRvHhxIZPJhL29vejRo4e4du2aUj75+5Nza29erl+/LiZOnChq1KghSpQoIfT09ES5cuVEr169REhIiNoy69evFw0aNBDGxsbCwMBAODs7iwkTJojnz9W/o/3GjRuie/fuolSpUkJHR0eULl1atG/fXpw4cUIpn4+Pj8q2yN8prW5b1b1TmjRHUtpzcSd6bqF+ktLU92G5z/WYJoQQV69eFW3btlX8zhwcHMSECRPE2LFjldohhBCJiYli/fr1on379sLBwUHo6+sLCwsLUbt2bbF06VKld78/e/ZMzJ8/X7i7u4vSpUsLXV1dYW1tLZo3by7+/vtvtW35WMew4OBgxbHHwsJCdOrUSbx8+fK91p19v7Vq1UoYGRkJAwMDUatWLbFp0yaV41H246GHh4cAIG7cuJHn9xIeHp7jNsrfO5/98+776SMjI8XYsWOFg4OD0NXVFWZmZqJZs2Ziz549ateXmZkpJk6cKEqXLi309PSEs7Oz2LhxY57tVNdGHx+fD25/bk6dOiXatWsnLCwshI6OjrC1tRUDBgwQERERavMnJSWJESNGCEtLSyGTyUTVqlXFtm3bhL+/v1Ib9u/fX6D9IW93ZGSk6Nevn7C0tBS6urrC2dlZ+Pn5iYyMDLXt2b59u6hcubLQ1dUVpUqVEqNHjxaJiYlKbenSpYtSmUOHDonq1asLAwMDYWlpKfr37y8SExPzvc+oYCRCcMphIiK5tJfPkVHIs/NqmZtDp2SpQq2TiCgvaRlxyBCFO1mXlsQAOlocqUI5i4qKgo2NDVxdXZXmUqAvl/yNOPL5c+i/Q7uoG0BE9DnRKVmKF+5E9J+go2UMHfDCnT6djIwMfP/995BKpYq3wxDR54tzBhARERER0Qe7fPkyLl26hIMHD6rMLk9Enx8+JkBERERERIUiMzOz0CdxpaLh4eGh8iYJd3d3BAUFFU2DqNAxGEBERERERESkYRi2IyIiIiIiItIwDAYQERERERERaRgGA4iIiIiIiIg0DIMBRERERERERBqGwQAiIiIiIiIiDcNgABEREREREZGGYTCAiIiIiIiISMMwGEBERERERESkYRgMICIiIiIiItIwDAYQERERERERaRjtom4AEdHnRMS9BpJiC7dSfRNIjIupTbK3t8eDBw9yLGpgYABnZ2f07t0bw4YNg5aWVo55nz59iiVLluDQoUMICwtDXFwczMzM4OTkBDc3N3h6eqJJkybQ0dH54E2iDxcTEwM/Pz/Y29vD29u7qJvzyaWmpqJ+/fqIiYnBuXPnULx48UJfR8OGDXH69GkAgJ2dHSIiIgp9HZ+zZwlRiCrk45mFvgmsDC1yTOcxreh8qmPKkSNH4Onpqfh/f3//Qllf27ZtERwcjKCgIDg5OX1wfZ8DPz8/xMTEwNfXt6ibgqCgIAQFBaFDhw6oVq1aUTeHPhMSIYQo6kYQEX0uxMsIZG74oVDrlPb8BZKS9rnmkUgkAIDw8HDY22fljYuLQ3BwMIYPH46bN2+iY8eO2LZtm9ryq1atwogRI+Dm5oYffvgBtWvXho6ODh49eoTdu3dj3rx5iIyMRLFixRAVFVWYm0fvKSIiAmXLloW7uzuCgoKKujmfXEJCAsqVK4f4+HiEhobCzs7uo61LIpFoZDDgxqtwfL1jTKHWefC7BXAtXjbPfDymfXqf+pji6+uLadOmFVowwMXFBXfu3MHJkydRr169D2/gZ0AeHPscLrcK+/ui/waODCAi+kwZGxujcePG+OOPP1C1alVs374dp0+fRoMGDZTyLV26FMOGDUPnzp2xadMmSKVvnwBzdnaGs7Mz2rRpg/r16+P169efejOI1DI0NMT9+/eRlpYGc3Pzom4OfQI8plFuLl26hLi4OJQoUaKom0KkMThnABHRZ65KlSowMTEBAJw7d04p7e7duxg9ejSMjY2xbNkypZPm7CpVqoRhw4Z99LYSFYSRkREDARqIxzRSR09Pj4EAok+MwQAioi+AfIihTCZTWj5v3jykpaWhQ4cOKFZM/bwEcv369cPYsWPzvc5r166ha9eusLOzg76+PhwcHPDNN99g/fr1SExMVMm/e/duNGnSBGZmZtDX10elSpUwbdo0JCQkqK0/IiIC/fv3h62tLWQyGaysrNCtWzfcvHlTKd+sWbMgkUgUn6CgICxZsgROTk6QyWQoU6YMfv7550IZhtmiRQuldSUkJGDkyJGwtraGTCZD5cqV8eeff+ZYft++fWjevDmKFSsGPT09ODo6YsKECSrDmO3t7VG2bNZQ6+PHj6tsX368fPkSAwcOhKWlJfT09ODk5ISZM2fi8OHDSvX5+vpi165dOa7D19dXKU2dyMhIjB49Gg4ODpDJZChevDjatm2LM2fOKOXbvHmzUl0BAQHYvn076tSpA0NDQ0gkEtjb2yMgICDPbY6Pj4ePjw8qVqwIPT09mJmZoWnTpti7d6/aNqalpeGXX35B+fLlIZPJYGtrizFjxuTY/6hoFcUxbf/+/ejTpw8qVqwIQ0NDGBsbo379+vj999/V5o+Li8P06dPh4uICIyMjlCxZEg0aNMDPP/+Me/fuKeUNCwvDgAEDUK5cOejr68POzg4tW7bE8uXLER0drVL3yZMn0bZtW1hYWEAmk8HBwQFjxoxROVa8e0xKTEzEuHHjUKZMGWhrayt+4wU5puR33XJXr15Fq1atYGJiAmNjY3z11Vc4evRofnd7noKCglSOG3If8/i/YcMGfPXVVzAxMYG+vj4qV66MGTNmKP19i4mJUTme5tTu7PvZ29sbEolEMX+GujqyL/P29kZgYCCaNGkCU1NTGBoaokGDBjhw4IBSmzt06KAoI3/8Rs7e3l6pvnfbOW3aNABAnz59cqyDNA+DAUREn7lr164hLi4OEokEDRs2VEr7+++/AQBubm551lO2bFnMmzcvX+u8fv063NzckJGRgWPHjiEmJgZ79+5FWloavLy8cOHCBaX8P/30E9q3b4/SpUvj5s2biIqKwtSpUzF37lw0btwYSUlJSvmDg4NRo0YNBAYGYtOmTYiLi8OhQ4fw4MED1KlTBydOnFDknTRpEoQQ8PLyAgDMnj0bUVFROHHiBMLCwlCvXj1MnToVixcvzte25ebAgQMQQiieX/f29kaDBg1w69YtBAcHQwiBjh074urVqyplp02bhtatW6Ns2bK4fv26YjKv9evXo1atWkqTqkVERCA8PBwA4O7uDiGE4uPh4ZFnO1+/fo0GDRpg3bp1Svvj1atXGDVqFADAy8sLQgj4+vqiQ4cOSvswO19fX6Vtfld4eDhq1qyJ33//Hf/73/8QGxuL8+fPQyKRoFGjRtiyZYsib9euXSGEgI+PD4Cs4MDWrVuxceNGPHv2DN27d1fs15zaAwDR0dFo0KAB5syZgx9++AGvX7/GzZs34eDggLZt26rtx127dsWPP/6I9u3b49GjR7h16xZcXV3RuXPnPPcnfVpFcUwDsi6Crly5goCAAERFReHWrVvw9PRE7969MXXqVKW8QgjFxfySJUvw8uVLhIaGonPnzvD19cWMGTMUeZ89e4Y6derg7t272LNnD2JiYnD8+HFYWVnh+++/x19//aVU96pVq+Dh4YG0tDScP38esbGxWLp0KTZt2oS6devixYsXirzvHpO6desGR0dHXLlyBSEhIShZsiSA/B9TCrJuIGvkRoMGDRAaGoq9e/fi1atXWLVqFWbPno3jx4/ne9/nxsPDQ+m4kd3HOv737dsXvXr1QpMmTXD//n28evUKkydPxqxZs9CwYUO8efMGAGBmZgYhBPz9/XNst7rjWEBAgNL3lv37kAcDhBAIDAwEAJw9exZjx47F7Nmz8fLlS5w7dw7p6elo1aoVNm7cqKh3165dOQY+IiIiFPWpa6d8//r7+yvaomnzqJAqBgOIiD5TcXFxCAwMRI8ePSCVSjFjxgylGYDj4+Px5MkTAECZMmUKdd3r1q1DcnIypk6dqrgbXLFiRWzZskVl5u4jR45gxowZcHBwgL+/P2xsbGBgYICuXbtiypQpuHjxImbOnKnIn5aWhs6dOyM6Ohpr165Fw4YNoauri8qVK2PLli1ITU1F7969kZ6errZtUqkUPj4+sLS0hI2NDf73v/8BgNLdpMJSp04ddO7cGaampqhcuTKmTp2KzMxMrF+/XinfsWPH4OvriwYNGmDFihUoXbo09PT00KpVKwQEBCAiIgK9e/cutHb98MMPuHfvHn788Ud4e3vD0NAQlpaWWLhwIYyMjAptPQDQu3dvPHr0CPPnz0fr1q0VdxI3b94MU1NTDBw4MMfntv/55x9s2rQJFSpUgImJCcaNG4e2bdvmuc4RI0bg+vXrmDRpEnr16gUDAwPY2NhgxYoVcHZ2xuTJk3H79m1F/s2bN2Pnzp1o3Lgx5s+fj5IlS8LIyAh9+/aFq6troe0L+jBFeUwDAFdXVwQEBKBu3brQ09ND6dKlMW3aNLRr1w6zZ89GZGSkIu+NGzdw+vRpdOvWDY0bN4aBgQGKFy+OkSNHomvXrkr1bt++HVFRURg7dixcXFwgk8lgb2+P1atXo3Tp0kp5b926haFDh8LY+P/t3XlUVVXjN/DvZbrg5SoBFiGKSE7gI5WSlhZoWI6RyxFFySGVtRL9BWGDJlo++fhoLQcepxXoTRNFn0xMwSQE9ZEUQYTAtATDQBGSQCAuw37/8D03Dvcyiory/ax114J99jl7n33O3XD22YMae/fu1fVkGTlyJL744gv8+uuvDfZ2ePbZZzFv3jzY2Nigb9+++OCDD5p8jzc3bSEEZs+ejbKyMmg0GrzyyitQKpXo06cP9uzZgx9//LFJ6bam1qj/w8LCEB4ejmnTpmHFihXo3LkzVCoVpk2bhrVr1yIlJQUBAQH36QwMu3LlCnbt2gV3d3dZLzQzMzP4+/tzbgy6b9gYQETUhjg5Oem673Xs2BHDhw+Hq6sr0tLS8OGH8lUOpDcXAGBhYdGq+ZC6i0dERKCmpkYXrlKpcPnyZQwaNEgXJv0zNnfuXL1lwnx9fQHcfRsliYqKQlZWFnr37g0PDw9Z/K5du8LDwwPXrl3DsWPHDOZt/Pjxst+feuopWFtb4/Lly809zUbVTatv374AoJfW+vXrAfx9vrW9/vrreOqpp5CQkICUlJR7zlNlZaWuW7OPj4/edkN5aKnk5GScOnUKlpaWmD59umxbhw4dMH78eBQXFyMiIsLg/pMnT5bdE88991yjb/Bu3bqFPXv2AADmz58v26ZQKDB9+nRUVVUhLCxMFy79fL/Lg5qvrdRpwN2GS0NLqrm5uUGr1coebqU68Pjx48jPz5fF37hxI9asWaMXd//+/dBqtbpwIyMjJCYmYsKECbqwzZs3o7KyElOnTkWnTp1kx504cSKUSiX27dsnK4va6t7jixcvxsSJExs67RanfebMGWRmZsLe3h6vvPKKLL6NjQ1GjRrVpHRbU2vU/1J9PWPGDL1t06dPh7GxMXbv3o2CgoJ7y2wzPP/883BxcZGF2dvbY9iwYSgpKZH1wCJqTWwMICJqQ7KysiCEgFarxdmzZ9GzZ09ERkYafDC2srLS/WxoDP+9mDt3Ljp16qQbg/3BBx/gf//7H4QQ6N69u+wfdWkCsOeee07vOPb29jAxMUF+fr6uO2JD8YG/3wjWHYog6dKli16YpaXlfRkbXjct6a173bTOnDkDAHr/zAF3HxSkRgQp3r24dOkSysrKYGJigmeeeUZve2uOAZWuVb9+/WBior8AUWPXqiVvd8+dO4fq6mrY2dnBzs6uSWmeP38ewN2Z5uvimNiHq63UaQCQk5ODgIAAuLi4wNLSUtdI8cknnwCA7O2rq6srRo4ciYsXL8LJyQk+Pj7Yu3cviouLYW1treueD9wdouLg4ACNRgNHR0cEBAQgNjYWVVVV6NKlC9RqtS5uQ/WfiYkJ7O3tUVlZWW/D4b30mGhu2g19r4CH89261/q/tLQUFy9eBGC4vlar1ejatSuqq6vrrdfuh/qGaUl/O5KTkx9YXqh9YWMAEVEbZGpqCnd3d0RGRsLIyAjBwcF6E1apVCrdP0Y5OTmtmn6fPn2QlpaGxYsXo7S0FKtXr8aQIUPg7OyMrVu3yuJKk2PVnehKoVDA2NhY191fGosqxa872Zz0kcZm1h27KjH0xrC+ie/uVd20pHTqjtmUzqm+LvpSuKGJxJpLemtnYWFh8LxrP3jcKym/iYmJBq+VNAa1vmvVoUOHFqd548YNg2lK43NrpymViUql0jtea5YHtdzDrtMuX76M/v37Y/v27ViyZAmys7N146al+7h2LyiFQoHDhw9jy5YtcHFxQUREBKZOnQo7OzvMmTNHNtmera0tUlNTsWzZMiiVSmzcuBFeXl5wcHDAqlWrZEOepPt7wYIFBu9vadx/a36nWpp2Q98r4OF8t+61/i8qKtL9/CDq66aqryylsq+db6LWxMYAIqI2zM3NDTNnzkRlZaVswirJmDFjADTtjXNpaSlu3LihN5lffbp27YovvvgCeXl5SEhIwIIFC5CXl4cFCxZg9erVunjS0nAJCQmySZLqfqShBVL82bNnNxh/8+bNTcpnWyCd0507dwxul8JbYxk96e1peXm57OFFUlJSUu++Df3TbOhNrJTf4cOHN3itjh492syzqJ+UZo8ePRpMMzMzU7ePVCaG3g42VB704D2sOm3dunUoKirCggUL4OfnB1tb20b3MTY2xvz583Hu3DlkZ2dj7dq1sLe3R1hYGDw9PWUP+dbW1li5ciWys7ORlJSE9957D3/99ReWLl0qWwJRur81Gk2D9/eUKVMazV9zNTfthr5XwKP53ard+6Q59XVz687mqq8spbKvne+G3I8eNfR4Y2MAEVEb99FHH+nGMF69elW2bcmSJTAzM8O3337b6ARD3t7e6Nq1K/Ly8hpNMzk5GRkZGQDujnt9+eWXsXnzZt1M37XHiEuzfktvlerKzs5GTEyM7sG1sfiVlZWIjo5u9TeD95N0TnWXRQQge3B98cUXdeEt7c3Qu3dvqFQqVFVV6b1ZBSBbtaAu6a1a3X/uKyoqDI6PbexaAXfHVLfmfA3u7u4wNjbG9evX651E8vTp00hNTdX9PmDAAACQNRBIGioPejgeRp0m3cO9evXS22aoMaGgoEC2rJujoyMCAwORlpaG3r17Iz09Henp6QDuDt1JSkrSxR0wYADWrFmDs2fPwtjYuFn15c2bNxEdHd2iYU+N1SnNTVv6XtWerLO2R/G7pVKp8I9//AOA4fq6pKQEOTk5MDY2xgsvvKALr6/uBIDr16/Xm15T6/n6ylL6Oyxdi9r5uV95ofaFjQFERG2cs7OzbtK02rPyA3ffnoaGhqKkpAT+/v4G3xQDwL59+xAbG4v58+ejR48ejaa5YcMGg8s8SZNv1e6qKs26XHtCN4kQAjNnzsTq1athZHT3T864cePg5OSEhIQEgw+zGo0GY8aMQWVlZaP5bCukMpDeuNUWHR2NmzdvYujQobKxutIa6rUfRNatW9foZGCmpqa6lQmkifZq27VrV7379u7dG4D+Q/N///tfg8tVPf/88xg6dCiysrIMLlkVGxuLESNGNOlhrKk6d+4MHx8faLVag+u///zzz/Dw8JAt7zhnzhwAzS8PejgeRp0mjbWv3YgkOXnypF5Yeno6xo4di+LiYlm4hYWFrkFBqgcjIiLg7++vd4yePXtCpVLJ6kt/f3+YmppCo9EYbOwKCgrCO++806IJFBurU5qb9uDBg+Hq6orc3Fy9ZQT/+OOPVu0R9CAtWrQIwN1Vc+ratWsXqqur4ePjI+s9Ul/def369QbnFqh7TRISEtCvXz+9ngDJycl6x87NzcWJEyegVqv1lkjt3bs3CgoKZCtgAMCBAweanBcAePnllw3Ws9R+sDGAiOgRsHTpUhgbG0Oj0ei91Zk7dy7CwsIQFRUFLy8vHD9+HMXFxSgtLUVaWhqCgoLg6+uLGTNm6GZRbooDBw5gzZo1yM3NRUVFBS5fvgx/f38YGRkhKChIF8/LywtLly5FfHw8Zs6ciYyMDJSXlyMzMxPTpk1DRkaGbsUB4O7D7L59+9CxY0eMGjUKMTExKCoqwq1bt7BlyxYsXLgQn376aZP+wW/ML7/8AoVCYXAG8dYklUFiYiLefvttXL9+HRUVFTh69Chmz56Nbt266f3DpVar4eLigszMTGRnZyMvLw9hYWFNGkqwatUq9OrVC5999hl27tyJsrIy5OfnIzAwsMGlBSdOnIgOHTpg/fr1OHPmDO7cuYNjx45h586dePrppw3uo9Fo0K1bN/j4+CAyMhIFBQW4ffs2IiIiMGXKFMybN09vVYh7tWHDBvTv3x+LFi3C9u3bkZubi5KSEhw5cgSjR4/GyJEjMW3aNF38yZMnY9KkSThx4gQCAwNx69YtlJaWYseOHfj+++9bNW/UOh50nbZw4UKYm5vjyy+/xNatW3H79m3k5eVh8eLFuon16qqursbkyZNx4cIFlJeXo7CwENu2bUN0dDTefPNNWS+DpKQkBAYGIjs7G1qtFr/99hsCAgJQXFyM4OBgXby+ffsiNDQUWVlZGDt2LJKTk1FWVoasrCwsWrQIBw4cwLZt23SNp83RWJ3S3LQVCgXCwsKgUqng5+eHhIQEaLVaXL58GVOmTGlwAkFfX18oFAocPHiw2edxv82ZMwczZszA3r17sWzZMty6dQtlZWXYs2cPgoOD0b9/f9nfLODu8BY3NzfExMRg9+7dKCkpQVpaGubMmaPrcWGItC06Ohp37tzB9u3bUVFRoTdHwKBBgzBv3jwkJSVBq9UiPT0d48ePh1arxebNm3UP8hKpQXjRokXIzc3FzZs3ERISoreij6G8xMXFoaysDIcPH8apU6fg5OTU9MKjx48gIiKdmuJCUXMzq3U/xYX1pufo6CgA6H3i4uL04vr6+srinDlzRrY9NzdXvP/++8LNzU2o1WphYmIi7O3thbe3t/jmm2+aVQ6//vqrWLlypRg8eLCws7MTZmZmolu3bmL8+PHi9OnTBvc5fPiwGDFihLCyshLm5ubC2dlZ+Pv7i6tXrxqMn52dLebPny+6desmTE1NhZ2dnRgxYoQ4dOiQLF54eLhe+Tg6OgohhFi+fLneNj8/P92+J06cEADExIkTm3Tefn5+9R7Pw8NDb1t4eLhs/6ioKOHl5SWsrKyEmZmZcHZ2FoGBgeLWrVsG00tKShJDhgwRarVa2NjYiEmTJon8/Pwm5TU/P1/MmzdPPPnkk0KpVAoXFxexadMmERcXp1cOtcXFxYlBgwYJpVIpbGxsxFtvvSUKCgpk9+KgQYNk+9y6dUsEBQWJZ555RpiZmQlbW1sxdOhQodFoRE1NjezYhu7n5cuXy45n6JoCEFlZWbo4d+7cEStXrhSurq7C3NxcPPHEE2LgwIFiw4YNoqKiQu+8tFqt+Oc//ymcnZ1199OsWbPEzZs3ZWlMmDChSeX7OMi9UyDSbl1t1U/unYIG02yrdZoQQly4cEGMGzdO2NnZCVNTU+Hs7CyCg4NFYGCgLB9CCFFWViY0Go3w9vYWzs7OwsLCQtjY2Ah3d3cRGhoquwfz8vLEunXrhIeHh3BwcBBmZmbC3t5evPbaa+K7774zmJfTp08Lb29vYWtrK5RKpejevbuYPn26SE1NlcUzVCfV/a7U1pQ6palp1y630aNHC0tLS9GhQwcxcOBAsWfPHr36Ny0tTbePp6enACDS09MbvS711RtxcXH3VP83pKamRuzcuVO89NJLwtLSUiiVSuHq6ipWrFgh7ty5Y3Cfa9euCW9vb6FWq4VKpRJeXl7iwoULeteovLxct09+fr6YNGmSsLW1FSqVSgwZMkScO3dO79z9/PxESkqKGDVqlOjUqZOwsLAQL730kjhy5IjBvFRXV4sVK1aIrl27ClNTU9GrVy/xn//8R68sN2/eLNvv888/F05OTkKpVAonJyfx+eefN6m86PGlEMJAv0AiIqLHQFBQENatW4eoqCiMHTv2YWfngThx4gSGDRsGPz8/7Nix42Fnh4jamcLCQnTp0gX9+vWTzaVA+lhf08PGYQJERPRYOnfuHDZt2oTJkye3m4YAIqKHqbq6WjecrO4ytETU9rAxgIiIHkurV6/GjBkzDE4SRURErS85ORnnz59HTEyM3gz4RNT2cJgAERE9lmpqalo0CdejzNDSUcuXL0dISMiDzwwRtUvtse5tCdbX1BawMYCIiIiIiIionWGzHREREREREVE7w8YAIiIiIiIionaGjQFERERERERE7QwbA4iIiIiIiIjaGTYGEBEREREREbUzbAwgIiIiIiIiamfYGEBERERERETUzrAxgIiIiIiIiKidYWMAERERERERUTvDxgAiolpEVQmE9mbrfqpK6k2ve/fuUCgU9X5UKhUGDBiA9evXo7q6usG85+bm4sMPP8TAgQNhbW0NU1NTdO7cGUOHDkVQUBBiYmJQWVnZ2kXWpn399ddQq9VYu3btfTn+8ePHZddrx44d9yWd++XgwYMICQlBdnb2w85Km1NUVISQkJD7fk0f9XuIiIgeXSYPOwNERG1KTRmQ91XrHvPpGQDUBjdJD2EKhQIAkJWVhe7duwMASkpKkJSUhIULF2Lx4sU4deoUIiMjDR5n+/btCAgIwODBg/HZZ5/B3d0dpqamyMnJwaFDh7B27VqsW7cO1tbWKCwsbN3za8N+//133LlzB1lZWffl+F5eXhBCICQkBCtWrLgvadxPBw8exM6dO+Hp6am77+iuoqIirFixAh4eHnjrrbfuWzqP+j1ERESPLvYMICJqo9RqNYYNG4avv/4aALB//36cPn1aL15oaCjmzZuHN954A7GxsRgxYgSsrKygUqnQp08fBAcH48SJE+jUqRP++OOPB30aD9V7772HnJwcbNq06WFnhYiIiKhNYWMAEVEb179/f3Ts2BEAkJiYKNt25coV/N///R/UajU2b94MIyPD1bqLiwveeeed+57XtsjBwUHX84KIiIiI7mJjABHRI0AIAQBQKpWy8LVr16KyshJvvvkmrK2tGzzGnDlzEBgY2Kx0IyMj4e7uDgsLC1hbW2Ps2LFISUmBp6enbJwzADz77LO63z09PWXHqR03JCTEYFqHDh3C8OHDYWVlBQsLC7i4uGDFihUoLS2VxevTp4/uWN27d0dhYSHefvttPP300zAyMtKNu649H0Pd/EhOnjyJcePGwcbGBkqlEs7Oznj33XfrHUpx4cIFjB49Gh07doRarcbLL7+M2NjYZpVpQxITE/XK6vjx43jxxRfRoUMH2NjYwM/PD0VFRQb3LywsRHBwMHr16gVzc3NYW1vjtddew5EjR2TxduzYAYVCgZ07dwIAhg0b1mhZ1XX06FHMmjULffv2hUqlglqtxksvvYSvvjI8zKakpAQrV66Eq6srLC0t8eSTT2LIkCH45JNP8Msvv8jiXr16FW+//TZ69OgBCwsLODo6YtSoUdiyZQtu376td+ymXseRI0fKyresrAxBQUHo1q0bTExMdGXevXt3ODk5AQDi4+Nl+5w4caJFaUvu9z1ERETUVGwMICJq41JTU1FSUgKFQoGhQ4fKtn333XcAgMGDBzd6HCcnp2ZNpLdhwwZMnjwZnTp1QmpqKvLy8hAUFARfX19cvXoVwN05DqSGigsXLtQ7Nl8IgfDw8HrTWrZsGby9veHg4ICffvoJhYWF+Pjjj/Hvf/8bw4YNQ3l5uS7upUuXdGlWVFRgwoQJeP3113Hp0iWcPn0a5ubmAO7Ox9DQXAHbt2+Hp6cnKisr8eOPP6K4uBihoaHYs2cPBg0ahJs3b8riJyYmYsiQIcjIyMDhw4dRUFCA7du341//+hfi4+ObUKKNGzx4sKysTp06hS1btuCrr77C77//jg8++AAajQazZ8/W2/fatWsYOHAgdu3ahfXr16OoqAipqalwdHTEmDFjsHLlSl3ct956C0II+Pn5AQDi4uIghIAQQu9htz6zZs1CSkoKduzYgcLCQmRmZmLEiBGYOXMmPv74Y1lcIYTuYX7Tpk3Iz89HRkYGJk+ejJCQEHz66ae6uHl5eXjhhRdw5coVREVFoaioCPHx8Xj66afh7++Pb7/9Vnbs5lzH6OhoCCHg6OgIAPDx8UGvXr2QkpKCtLQ0PPnkkwDk946Hh4eubIQQssaStngPERERNZkgIiKdmooboib73637qbjRaLoABACRlZWlCysuLhY//PCDcHV1FUZGRmLVqlWyfUpKSnT7RUVFtWo5ZGdnCzMzM2FlZSWKiopk244cOWIwv0IIkZWVJQAIDw8PvWOGh4cLAGL58uWy8O+//14AEM7OzqKqqkq27bPPPhMAxEcffaR3PCkPYWFhsvAPP/xQnDhxosH8ZGRkCFNTU9GpUye989uzZ48AIKZPn64Lq6mpEX379hUARHx8vCx+QUGBsLCwEABEeHi4Xj5bQiorGxsbUV5eLtvWp08fYWRkJG7fvi0Lf/nllwUAcezYMVl4TU2NGDx4sAAgYmNjZdv8/PwEABEXF9fsPL766qsiJSVFL/yNN94QZmZmIj8/Xxd28eJFAUC8++67evGnTZsm/Pz8dL9v2LBBABCHDh2SxauurhYODg6yMm7udZQ4OjoKAOLjjz+WhX/xxRciMjJSCNHwvdyStB/0PURERNQY9gwgImpDnJycdN2RO3bsiOHDh8PV1RVpaWn48MMPZXH//PNP3c8WFhatmo/du3dDq9Vi9OjR6NSpk2zbyJEjYWNj02ppbdiwAQAwd+5cGBsby7b5+voCuPsG1hCFQoGpU6fKwlatWgUPD48G09y8eTMqKysxdepUvfObOHEilEol9u3bpyvjM2fOIDMzE/b29njllVdk8W1sbDBq1KhGzrJlRo4cqevpIHFxcUFNTY2sa31KSgpOnjwJOzs7jBgxQhZfoVBgxowZAID169e3Wt6OHz+OZ599Vi/czc0NWq0WP/74oywP0j75+fmy+Bs3bsSaNWv04u7fvx9arVYXbmRkhMTEREyYMEEX1tzrWJePj4/s98WLF2PixIkNnXaL035Y9xAREVF92BhARNSGSN3utVotzp49i549eyIyMhLHjh3Ti2tlZaX7uaysrFXzcf78eQB3x+fXpVAodN2sW4M0KeJzzz2nt83e3h4mJibIz8/XLcNYm62tbYsaQhpK08TEBPb29qisrERKSgqAhssDwH1blq9Lly56YZaWlgAgm0vhzJkzAIC+ffsaPI6rq6ssXmvIyclBQEAAXFxcYGlpqWvE+uSTTwBAtnKFq6srRo4ciYsXL8LJyQk+Pj7Yu3cviouLYW1treueDwBTp06Fg4MDNBoNHB0dERAQgNjYWFRVVaFLly5Qq/9eprO517Gubt26tfj8H5V7iIiIqD5sDCAiaoNMTU3h7u6OyMhIGBkZITg4WG+SNZVKpXtYzMnJadX0pbeZKpXK4PbaD2T3SpoQru7kbgqFAsbGxqiqqgIAvfHXANChQ4d7SnPBggV6aSoUCt14cSnNB1ketRlq6JDenIv/P28C8Pf5SA0FdUnhhibfa4nLly+jf//+2L59O5YsWYLs7GzdmPrly5cDAGpqamR5Pnz4MLZs2QIXFxdERERg6tSpsLOzw5w5c2ST7dna2iI1NRXLli2DUqnExo0b4eXlBQcHB6xatUp3P9Q+n6Zex7paev+0JO2HdQ8RERHVh40BRERtmJubG2bOnInKykrZJGuSMWPGAGjaG9/S0lLcuHFDNhlffaReB3Vn8peUlJQYDG9oCb/6ei888cQTAICEhATZRG11P4MGDWo0300lpanRaBpMc8qUKQBaXh4PinQ+d+7cMbhdCpfi3at169ahqKgICxYsgJ+fH2xtbRvdx9jYGPPnz8e5c+eQnZ2NtWvXwt7eHmFhYfD09JQ95FtbW2PlypXIzs5GUlIS3nvvPfz1119YunSpbInM5l7H1vS43UNERNT+sDGAiKiN++ijj2BsbIzdu3frZvGXLFmyBGZmZvj2229l3bIN8fb2RteuXZGXl9domgMGDAAAZGZmGtz+22+/GQyX3mQbeuC5fv26wX2klRDqm/k/OzsbMTExsjfN96qxNG/evIno6GjdeUjlcenSJYPxr1271mp5awnpfDIyMmQ9BiQ//fQTAODFF1+UhTfUeNMQqdx69eqlt81QY1NBQQGio6N1vzs6OiIwMBBpaWno3bs30tPTkZ6eDuBuGSclJeniDhgwAGvWrMHZs2dhbGyMiIgI3bbmXsfmaKxsHrd7iIiI2h82BhARtXHOzs6YPn06qqqqsGrVKtm2Hj16IDQ0FCUlJfD396/3gXnfvn2IjY3F/Pnz0aNHj0bT9PX1hVKpxNGjR/UmX4uOjkZBQYHB/Tp37ownnngCV65cQXV1tS5cCIFvvvnG4D4BAQEAgLCwML1tQgjMnDkTq1evhpFR6/3J8vf3h6mpKTQajeyNtCQoKAjvvPOOrnFj8ODBcHV1RW5urt4ScH/88QeOHj3aanlrieeffx5Dhw7FzZs3ERMTI9smhMBXX30F4O+yllhbWwP4+wG+vLwc/fr1ww8//NBgetJY+9TUVL1tJ0+e1AtLT0/H2LFjUVxcLAu3sLDQNShIXfYjIiLg7++vd4yePXtCpVLJuvY39zo2R92yAe72iJAmGHzc7iEiImp/2BhARPQIWLp0KYyNjaHRaPTeRM6dOxdhYWGIioqCl5cXjh8/juLiYpSWliItLQ1BQUHw9fXFjBkzmjybfNeuXXVdwSdOnIgrV65Aq9UiPj4eK1euhIuLi8H9pJnr//zzTwQHB6OgoAA5OTnw9/evd4I0Ly8vLF26FPHx8Zg5cyYyMjJQXl6OzMxMTJs2DRkZGboVB1pL3759ERoaiqysLIwdOxbJyckoKytDVlYWFi1ahAMHDmDbtm26BgiFQoGwsDCoVCr4+fkhISEBWq0Wly9fxpQpUxqc/M3X1xcKhQIHDx5s1XOoS6PRoFu3bpg1axaOHDmCiooKXL9+HfPmzUNiYiKWLVuGV199VbaP9Hb7+PHjKC8vR3h4OH7++edGG4wWLlwIc3NzfPnll9i6dStu376NvLw8LF68WDexXl3V1dWYPHkyLly4gPLychQWFmLbtm2Ijo7Gm2++KetlkJSUhMDAQGRnZ0Or1eK3335DQEAAiouLERwcrIvX3OvYHGq1Gi4uLsjMzER2djby8vIQFhamGx7wIO8hIiKi++L+rlxIRPRoqam4IWqy/926n4ob9aYnrXde92No3XdfX19ZnDNnzsi25+bmivfff1+4ubkJtVotTExMhL29vfD29hbffPNNi8pj//79YuDAgUKpVApra2sxYcIEcfXqVeHh4SEAiKysLL19ysvLRUBAgHjqqaeEUqkUbm5uIjIyUoSHh8vyf/ToUdl+hw8fFiNGjBBWVlbC3NxcODs7C39/f3H16lVZPCntup+6DJVt7fXshRDi9OnTwtvbW9ja2gqlUim6d+8upk+fLlJTUw2Wx4ULF8To0aOFpaWl6NChgxg4cKDYs2ePWL58uSydtLQ03T6enp4CgEhPT2+0vKW17et+srKyRFxcnF64o6OjbP9bt26JwMBA4ezsLMzMzISVlZXw8vISUVFRBtOrqakRS5YsEQ4ODsLc3Fz06dNH7N69u9F8SmUxbtw4YWdnJ0xNTYWzs7MIDg4WgYGBetelrKxMaDQa4e3tLZydnYWFhYWwsbER7u7uIjQ0VFRUVOiOm5eXJ9atWyc8PDyEg4ODMDMzE/b29uK1114T3333ncG8NPU6+vn51Vu+hiQlJYkhQ4YItVotbGxsxKRJk0R+fn6L0q5dbs29h4iIiO4HhRAGBhcSEbVToqoEqGndZfpg1AEKk8drpnBPT0/Ex8cjKyuLbzQbUFhYiC5duqBfv36ycfBERERED5vJw84AEVFbcveh/fF6cKeHo7q6Gv7+/jAyMsLWrVsfdnaIiIiIZDhnABER0X2QnJyM8+fPIyYmRjeTPBEREVFbwWECRETUZCEhIVixYoVeOP+UGFZTU9OqqyAQERERtRY2BhARERERERG1M3xdQURERERERNTOsDGAiIiIiIiIqJ1hYwARERERERFRO8PGACIiIiIiIqJ2ho0BRERERERERO0MGwOIiIiIiIiI2hk2BhARERERERG1M2wMICIiIiIiImpn2BhARERERERE1M6wMYCIiIiIiIionWFjABEREREREVE7w8YAIiIiIiIionaGjQFERERERERE7QwbA4iIiIiIiIjaGTYGEBEREREREbUzbAwgIiIiIiIiamfYGEBERERERETUzrAxgIiIiIiIiKidYWMAERERERERUTvDxgAiIiIiIiKidoaNAURERERERETtDBsDiIiIiIiIiNqZ/wdoU4SL+4CByAAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 1012.5x375 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saved: fig_conversion_funnel.png\n"
     ]
    }
   ],
   "source": [
    "plot_pipeline_funnel(discovery_summary_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d9b2e6a2",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-block alert-info\">\n",
    "<b>Because we use heuristics to quickly match the root cause entity in the trajectories for cost purposes in this lab, the numbers above are lower than the overall results. However, the general trends shown are still valid. \n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "40ad0bc0",
   "metadata": {},
   "source": [
    "# Putting it All Together"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f0d8e7c5",
   "metadata": {},
   "source": [
    "## Discussion Questions"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d3a044ca",
   "metadata": {},
   "source": [
    "- What do you think is driving the performance of each model?\n",
    "- Where do you think each model is struggling, and how muchimpact does that have on overall performance?\n",
    "- What improvements do you think would have the mostefficient impact on model performance?"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a96759a6",
   "metadata": {},
   "source": [
    "## Aggregated Results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "id": "6c0952c1",
   "metadata": {},
   "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>F1</th>\n",
       "      <th>R</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>OpenAI-GPT-OSS-120B</td>\n",
       "      <td>0.164021</td>\n",
       "      <td>0.174603</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 model        F1         R\n",
       "0  OpenAI-GPT-OSS-120B  0.164021  0.174603"
      ]
     },
     "execution_count": 87,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "f1_df = perf_df[perf_df['metric'] == 'RC Entity F1'][['model', 'performance']].rename(columns={'performance': 'F1'})\n",
    "r_df = perf_df[perf_df['metric'] == 'RC Entity Rec'][['model', 'performance']].rename(columns={'performance': 'R'})\n",
    "\n",
    "f1_df.merge(r_df, on='model')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "51d25841-ee28-44dc-a534-8003cc6e03d7",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "SRE",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.13.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}