intervening_probe_interact_column.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "b429e6d2-101b-45df-9958-85b10e8aa81b",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "05efae22-ca7c-44df-9bf2-8551c74d9224",
   "metadata": {},
   "outputs": [],
   "source": [
    "# make deterministic\n",
    "from mingpt.utils import set_seed\n",
    "set_seed(44)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "796b304f-0f64-46d4-ae69-b03553c08034",
   "metadata": {},
   "outputs": [],
   "source": [
    "import math\n",
    "import time\n",
    "import numpy as np\n",
    "from copy import deepcopy\n",
    "import pickle\n",
    "import seaborn as sns\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "from torch.utils.data.dataloader import DataLoader\n",
    "from torch.nn import functional as F\n",
    "from torch.utils.data import Subset\n",
    "from tqdm import tqdm\n",
    "from matplotlib import pyplot as plt\n",
    "\n",
    "from data import get_othello, plot_probs, plot_mentals\n",
    "from data.othello import permit, start_hands, OthelloBoardState, permit_reverse\n",
    "from mingpt.dataset import CharDataset\n",
    "from mingpt.model import GPT, GPTConfig, GPTforProbeIA\n",
    "from mingpt.utils import sample, intervene, print_board\n",
    "from mingpt.probe_model import BatteryProbeClassification, BatteryProbeClassificationTwoLayer\n",
    "\n",
    "plt.rc('text', usetex=True)\n",
    "plt.rc('font', **{'size': 14.0})\n",
    "plt.rc('text.latex', preamble=r'\\usepackage{lmodern}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "837026e8-1834-4a9b-ae5b-4a6ff857c9eb",
   "metadata": {},
   "outputs": [],
   "source": [
    "championship = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "be98be3c-f484-4a15-bbe9-a640ecb37bac",
   "metadata": {},
   "outputs": [],
   "source": [
    "mid_dim = 128\n",
    "how_many_history_step_to_use = 99\n",
    "exp = f\"state_tl{mid_dim}\"\n",
    "if championship:\n",
    "    exp += \"_championship\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "71378fe3-d9c2-401a-b1a0-9445f6ffd242",
   "metadata": {},
   "source": [
    "## Load a game from intervention benchmark and select an intervention configuration"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "8de77e19-451a-4f3b-89fd-4efdba144882",
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"intervention_benchmark.pkl\", \"rb\") as input_file:\n",
    "    dataset = pickle.load(input_file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "a2e08851-7508-4a56-993a-0f67be5ecb01",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'intervention_position': 'e6', 'intervention_from': 2.0, 'intervention_to': 0.0}\n"
     ]
    }
   ],
   "source": [
    "case_id = 777\n",
    "wtd = {\n",
    "    \"intervention_position\": permit_reverse(dataset[case_id][\"pos_int\"]), \n",
    "    \"intervention_from\": dataset[case_id][\"ori_color\"], \n",
    "    \"intervention_to\": 2 - dataset[case_id][\"ori_color\"], \n",
    "}\n",
    "completion = dataset[case_id][\"history\"]\n",
    "print(wtd)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "0944ee4b-452a-4d83-a4ff-8bafb2cc0171",
   "metadata": {},
   "outputs": [],
   "source": [
    "wtd_list = [wtd]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4b89df37-9dea-4261-94e4-b602ec8cfffb",
   "metadata": {},
   "source": [
    "## Load nonlinear probes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "f8e85d97-4f7a-4c90-99b2-9b04cae1154d",
   "metadata": {},
   "outputs": [],
   "source": [
    "probes = {}\n",
    "layer_s = 4\n",
    "layer_e = 9\n",
    "for layer in range(layer_s, layer_e):\n",
    "    p = BatteryProbeClassificationTwoLayer(torch.cuda.current_device(), probe_class=3, num_task=64, mid_dim=mid_dim)\n",
    "    load_res = p.load_state_dict(torch.load(f\"./ckpts/battery_othello/{exp}/layer{layer}/checkpoint.ckpt\"))\n",
    "    p.eval()\n",
    "    probes[layer] = p"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "96259d73-16a9-4ec2-94e2-1dbc294648b3",
   "metadata": {},
   "source": [
    "## Load trained models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "a8d0f8a5-b06b-454e-a1b2-a97d719f3d17",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 1/1 [00:00<00:00, 36.48it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset created has 1 sequences, 61 unique words.\n"
     ]
    }
   ],
   "source": [
    "# othello = get_othello(ood_perc=.2, data_root=\"data/othello_pgn\", wthor=False)\n",
    "othello = get_othello(ood_perc=0., data_root=None, wthor=False, ood_num=1)\n",
    "train_dataset = CharDataset(othello)\n",
    "\n",
    "mconf = GPTConfig(61, 59, n_layer=8, n_head=8, n_embd=512)\n",
    "\n",
    "models = {}\n",
    "for layer in range(layer_s, layer_e):\n",
    "    model = GPTforProbeIA(mconf, probe_layer=layer)\n",
    "    # model = GPT(mconf)\n",
    "    load_res = model.load_state_dict(torch.load(\"./ckpts/gpt_synthetic.ckpt\" if not championship else \"./ckpts/gpt_championship.ckpt\"))\n",
    "    if torch.cuda.is_available():\n",
    "        device = torch.cuda.current_device()\n",
    "        model = model.to(device)\n",
    "    _ = model.eval()\n",
    "    models[layer] = model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cfbc45bc-e369-44f4-bc5e-54bfec187825",
   "metadata": {},
   "source": [
    "### Check the partial game progression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "5f90954b-6e49-4b63-9dca-5851c748ba54",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--------------------\n",
      "[]\n",
      "a                \n",
      "b                \n",
      "c                \n",
      "d       O X      \n",
      "e       X O      \n",
      "f                \n",
      "g                \n",
      "h                \n",
      "  1 2 3 4 5 6 7 8\n",
      "--------------------\n",
      "--------------------\n",
      "['e6']\n",
      "a                \n",
      "b                \n",
      "c                \n",
      "d       O X      \n",
      "e       X X X    \n",
      "f                \n",
      "g                \n",
      "h                \n",
      "  1 2 3 4 5 6 7 8\n",
      "--------------------\n",
      "--------------------\n",
      "['e6', 'd6']\n",
      "a                \n",
      "b                \n",
      "c                \n",
      "d       O O O    \n",
      "e       X X X    \n",
      "f                \n",
      "g                \n",
      "h                \n",
      "  1 2 3 4 5 6 7 8\n",
      "--------------------\n",
      "--------------------\n",
      "['e6', 'd6', 'c3']\n",
      "a                \n",
      "b                \n",
      "c     X          \n",
      "d       X O O    \n",
      "e       X X X    \n",
      "f                \n",
      "g                \n",
      "h                \n",
      "  1 2 3 4 5 6 7 8\n",
      "--------------------\n",
      "--------------------\n",
      "['e6', 'd6', 'c3', 'f3']\n",
      "a                \n",
      "b                \n",
      "c     X          \n",
      "d       X O O    \n",
      "e       O X X    \n",
      "f     O          \n",
      "g                \n",
      "h                \n",
      "  1 2 3 4 5 6 7 8\n",
      "--------------------\n",
      "--------------------\n",
      "['e6', 'd6', 'c3', 'f3', 'c4']\n",
      "a                \n",
      "b                \n",
      "c     X X        \n",
      "d       X X O    \n",
      "e       O X X    \n",
      "f     O          \n",
      "g                \n",
      "h                \n",
      "  1 2 3 4 5 6 7 8\n",
      "--------------------\n"
     ]
    }
   ],
   "source": [
    "ab = OthelloBoardState()\n",
    "ab.update(completion, prt=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "8dd21ce5-6c66-419f-a4d0-2a5c12896862",
   "metadata": {},
   "outputs": [],
   "source": [
    "partial_game = torch.tensor([train_dataset.stoi[s] for s in completion], dtype=torch.long).to(device)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "10e84e85-9c75-402c-9026-6dee99e4db46",
   "metadata": {},
   "source": [
    "### Check pre-intervention ground-truth legal next-steps and the predicted ones"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "d7b0c20c-895c-4385-bb69-0ff45e39715a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['b4', 'c6', 'd3', 'e7', 'f4', 'f6']"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pre_intv_valids = [permit_reverse(_) for _ in ab.get_valid_moves()]\n",
    "pre_intv_valids"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "9ad9b4d1-5fd5-48db-add4-fa03305ad661",
   "metadata": {},
   "outputs": [],
   "source": [
    "pre_intv_pred, _ = model(partial_game[None, :])  # [B, T, F=512]\n",
    "# no history activations used here, that's why the prediction map is different to across layers\n",
    "pre_intv_pred = pre_intv_pred[0, -1, 1:]\n",
    "padding = torch.zeros(2).cuda()\n",
    "pre_intv_pred = torch.softmax(pre_intv_pred, dim=0)\n",
    "pre_intv_pred = torch.cat([pre_intv_pred[:27], padding, pre_intv_pred[27:33], padding, pre_intv_pred[33:]], dim=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "f71efd0e-2614-41d5-9ac0-7d9bdf212c36",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAesAAAGVCAYAAADe29kQAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAxOAAAMTgF/d4wjAABA50lEQVR4nO3df3wc9X0n/pdiMMbB1miNQxpUm50NoS4E7JUIcFelV2sXpy3XNPGuBAnXb9LHWesk32t47BErJkf6KH4c6q5hz0lrxyvnEfJoSQBpIUmP3mG0dguEqx1Jw49AFBJ7ZPtEwcHenbXBGGM094e6g0a70uyMZnc/s3o985jHozvz0WdeTAfe+/nMj23SdV0HERERCesD9Q5AREREc2OxJiIiEhyLNRERkeBYrImIiATHYk1ERCQ4FmsiIiLBsVgTEREJ7oJ6ByAiIqq3ZDIJWZaRy+UgyzJCodCsbTVNw8DAADRNw/DwMLq7uxGJRGz1Z2d/AIs1EREtcLFYDNFo1CiY0WgUsixDluWy7fv6+pBIJABMFW6/3w9ZlhEMBivqz+7+AE6DExHRAjcwMGAa2YbDYaTT6bJtVVWFpmnGZ0mS0NXVhb6+vor7s7O/IhZrIiJasBRFgc/nM63z+XzIZrNl2/t8PmMKfLriZ6v+7O6viMWaiIgWrFwuB0mSTOskSSopxtO35fN509+MjIwYU+BW/dndX1HdrllfsPjyeu2aiIgqcP7cq3Xb97snVFf6+Zu/+zFSqZTxOR6PIx6PG59nK5K5XK6i/lVVhaqq2Lp1a0X9Od0fbzAjIqKGNbM4l1NuSnvmVPVsotEo9u3bZxotW/XnZH8s1kREJJ7J92qym+LjU9OVm6oup7e3F4lEwpgCr6Q/p/vjNWsiIhKPPunOYmF6oS3SNM3yuef+/n6Ew2GjnaIoFfXndH8s1kREtKB1dXWZ7sYeGhpCLBYzPquqatqezWbh8/nQ3t4OTdOgqipGRkYq7s9qezlNuq7rzv7x5oc3mBERia2uN5i9NuZKPxf+1pqK2iWTSQSDQSiKgmAwaBrpJpNJDA8PY3BwEKqqIhAIlPx9IpHAli1bKuqvku0zsVgTEVFZ9SzW5/71ZVf6WfyRq13pp944DU5ERCQ43g1ORETimbS+OWwhYbEmIiLxVHAn90LCYk1EROKp0XPWXsFr1kRERILjyJqIiMTDaXATFmsiIhIPbzAz4TQ4ERGR4DiyJiIi4eicBjfxxMh6184E3jj+Ml4Zexad6zvm3d5uf04wc20yO9mPCLmZmefHfDLYbV+rY+2qyUl3lgYh/Mh6184EfD4JKy+7Gp3rO7D3iYdx5VU3YXz8mKP2dvtjZnEzezU3M/P8aLTMVH2uvRv8sccew2c/+9mK21f6bvDz517FipVrUCicAjB14hUKBWy9615H7e325wQz1yazV3MzM88Pr2Su57vB3/nVT13p56KP/Z4r/dTbvKbB9+/fj+7ubixatAjRaNStTIZ1a69BPq8ZJxkAKMqLWD/LNI5Ve7v9MbO4mb2am5l5fjRa5qqZfM+dpUHYLtbPP/88vvSlL2HFihUIh8PI5/PYvXs33nvP/YPil1cjl9NM68bHj0FqXu6ovd3+nGDm2mR2sh8RcjMzz4+5eDFz1eiT7iwNoqJr1keOHEE6nUZ/fz/y+TyCwSC+/vWvo7+/H08++WRVA2rTvhEW+XyS4/Z2+3OCmWuT2cl+RMjNzDw/5uLFzFR9liPrK6+8EoFAAIqi4K//+q+Rz+cxMjKCr33ta7XIV/Yb4Mxvinba2+3PCWauTWYn+xEhNzPz/JiLFzNXBe8GN7Es1idOnMC6desQjUbR3d2N5uZmRztKpVJobW01lsn33rT8m3H1aMk3QL9/VdlvipW0t9ufE8xcm8xO9iNCbmbm+TEXL2auGk6Dm1gW63w+jz179mBkZARXXHEFrr/+etx///04cuSIrR3F43FMTEwYywcWXWL5N889/xJaWiT4/auMdcHgtdi//xlH7e325wQz1yazV3MzM8+PRstMtWH70a1sNov+/n5kMhk0NTUhmUxi06ZNWL7c3g0LlT66VXxG8NbbYli39ho8ufcRfOLGPzSeESyehDOfKZytvdV2NzBzbTJ7NTcz8/zwSua6Prr14l5X+rno2g2u9FN3+jwMDg7q4XBYb2pq0m+++WZbf7vowo9UvKT7/17P5fL6yOgL+s0buk3bktv/Vh8aeqri9pVsd2Nh5tpk9mpuZub54YXM9fT28//oytIoXHkpiqZpGBwcxKZNmyr+m0pH1kREVB/1HFmfff5xV/pZsvYWV/qpN9feYGYXizURkdhYrMUh/LvBiYhoAWqgx67cwGJNRETiaaDHrtwwr3eDExERUfVxZE1EROJpoB/hcAOLNRERiYfT4CacBiciIhIcR9ZERCQe3g1uwmJNRETi4TS4CYs1ERGJhyNrE16zJiIiEhxH1kREJB6OrE1YrImISDi6XtvnrJPJJGRZRi6XgyzLCIVCc7ZXVRXpdBqBQAA9PT229pXNZiHLMmRZBgAoioJgMDjn33AanIiIFrRYLIZgMIhIJIKenh6k02moqjpr+2w2C1VVZ20TjUahKAo0TTMtRYqiIBAIoKmpCeFwGJIkWWZksSYiIvFMTrqzVGBgYMA0kg6Hw0in07O2D4VCCIVC8Pl8Zbdns1m0tbWhpaXFtBSLuyRJ0HUduq5jaGjIGGHPhcWaiIjEo0+6s1hQFKWk6Pp8PmSzWcfRE4mEUYx1Xcfhw4crLsqz4TVrIiIST41uMMvlciXT0JIkmaat7erq6jJ9zmazJde1M5kMgKlr37IsIxKJzNknizUJ5/T3vlDvCLYt+/Pv1zsCEZWRSqWQSqWMz/F4HPF43Pg8W1HO5XKO9zm9+Pf29iKRSJi2t7e3Q5Zlo11bWxtkWZ7zJjMWayIiEo9LbzCbWZzLmVmwNU2b9Xq0HYqiYMWKFSXrZxbl9vZ29PX1YXBwcNa+eM2aiIjEU6MbzIqPa01Xbmrcib6+vrKPgCmKYvocCATmvPscYLEmIqIFrNzUs6Zpls9ZVyKTyZTcVKaqKjo7O03rTp48aXnzGYs1ERGJp0Z3gwNTN4RNv/t7aGgIsVjM+Kyqatm7w3O53KzXvKc/pjWdLMvYunWraV0mkylZNxOvWRMRkXhq+LrRdDqNZDIJYGqKure31zTSzWQyGB4eNkbbiqIgm80aL0cBgEgkYvobTdNmHS1HIhEkk0lIkoTR0VGk02nLN5g16bquz+uf0qELFl9ej92SB/BucCIxnD/3at32/fb//rYr/Vz8h3/hSj/1xpE1ERGJhz/kYcJiTURE4nHp0a1GwWJNRETi4cjahHeDExERCY4jayIiEg+nwU1YrImISDycBjfhNDgREZHgOLImIiLxcBrchMWaiIjEw2lwE09Mg+/amcAbx1/GK2PPonN9x7zb2+3PCWauTWYA2PaPo+jY/hPc8rf/GwfU4xX9zQP/5xXsyL5Ydtups+dwZ+ZfjD7HXsu7GReAN4+1FzM72Y8Iub2Y2XU1+tUtrxB+ZL1rZwI+n4SVl12NzvUd2PvEw7jyqpswPn7MUXu7/TGzuJmBqUJdePscnvnap3FAPY7NP3gGj///n0JryyWztj84/hsU3j6Hjev8JdtPnT2Hz313H77w767CfZGbcOrsOVfzAt481l7M7NXcXsxM1Sf8u8HPn3sVK1auQaFwCsDUiVcoFLD1rnsdtbfbnxPMPL/Mdt4NvnZbBk9/7U+wfMliAFPFeNlFF+KO0LVz/t2dmX9Bq/TBknY7si/i9Dvv4u4/bqs4A2Dv3eAiHetGzuzV3CJlruu7wR/5K1f6ubj7L13pp94cT4OfOnUKp06dcjNLiXVrr0E+rxknGQAoyotYP8s0jlV7u/0xs7iZAWDstTyWLbnQKNQAsOa3WnBw/DeO+3z0uXFEgnP/rux8ePFYezGzV3N7MXPVcBrcxLJY79+/H93d3abf2ty6dSskSUJLSwtuvfXWqoXzy6uRy2mmdePjxyA1L3fU3m5/TjBzbTIDwET+LTRfvNi0rlX6IE6/867D/t7E6bPvovD2OcQefBod23+Cbf846kZUgxePtRczO9mPCLm9mJlqY85i/dxzzyEcDkPXdRw+fBjd3d3Yv38/0uk0RkdH8etf/xqHDh3C/fffX7WAWqF09O7zSY7b2+3PCWauTWYAWHbRhSXrCm87u848kX8LADA0NoH07Z/EQ/+5EwfHf4OMos4r40xePNZezOxkPyLk9mLmquDI2mTOYt3b24uNGzdiYGAAAwMD8Pv96O3txZ49e7Bu3TrIsoz+/n7s3r27agHLfQOc+U3RTnu7/TnBzLXJDKDsKHrmaLvivs5O9fXFf3cVAKC15RJsDMp4dNTdYu3FY+3FzE72I0JuL2auCn3SnaVBzFmsR0ZGcNdddxmfe3p6oCgKNm7caKwLBoM4efKk5Y5SqRRaW1uNZfK9Ny3/Zlw9WvIN0O9fVfabYiXt7fbnBDPXJjMAtLZ8sGQUPaG9VXa0XWl/y5ZcaLqTfD7T6uV48Vh7MbOT/YiQ24uZq4YjaxPLa9Zr1641/m9ZllHu5nGfz2e5o3g8jomJCWP5wKLyj9ZM99zzL6GlRYLfv8pYFwxei/37n3HU3m5/TjBzbTIDUzeTnT77Liby73/xG3stjxv8H5pXf9Mf15rQ3sLl0gfnnbXIi8fai5m9mtuLmak25izWK1asKFnX0tJSsq6pqcm9RDP073kQffd+A8DUnY3RyC3o3/Ogsd3vX2U6Ea3aW21nZu9kBoCNQT927Ps5gKlC/eQvJhBpe/9u7on8m6ZibuULN30M9zw+avzt9579pTEt7hYvHmsvZvZqbi9mrgpdd2dpEHO+FKV4U9l0+Xy+ZJ2quntNb7ovf6XXePuOOn4Mt9622fQwf8+m27Fu7cfxqT+6raL2VtuZ2TuZAeDuP24z3mDWKn0Q2zfeaJrGzoyqGHtdQ/r2TwKYeo760efGjevTjz43jq92ftx4XOuO0LVGf80XL8Y3/7gNN8qXuZrZi8fai5m9mtuLmauigaaw3TDnS1FisVjFo2a7N5lV+lIUWnjsvBRFFHZeikLkFXV9KcoDW1zp5+IvJl3pp97mHFmn0+la5SAiInofR9Ymwr8bnIiIFqAGeuzKDY5fN0pERES1wZE1EREJR59snDu53cBiTURE4uE1axMWayIiEg+vWZvwmjUREZHgOLImIiLx8Jq1CYs1ERGJh9esTTgNTkREJDiOrImISDwcWZuwWBMRkXga6Bez3MBiTURE4qnxyDqZTEKWZeRyOciyjFAoNGd7VVWRTqcRCATQ09Nj2pbNZiHLMmR56tf8FEVBMBic1/5YrImIaEGLxWKIRqNGwYxGo6ZiO1M2mwUwVbADgUDJdkVREA6HAQChUKjkR7Hs7g/gDWZERCSiSd2dpQIDAwOmkW04HJ7zVydDoRBCoRB8Pl/Z7ZIkQdd16LqOoaGhkiJsd38AizUREYlIn3RnsaAoSknR9fl8xujZbU73x2lwIiJasHK5HCRJMq2TJAmaps2r30wmA2BqqlyWZUQikXntj8WahLPsz79f7wgLxukffqneEWxb9rnv1DsC1UKN3mA2W5HM5XKO+2xvb4csy0ZRbmtrgyzLCAaDjvfHaXAiIhKOPjnpypJKpdDa2mosqVSqZF8zC6imabNej65EMBg0jZ7b29vR19c3r/1xZE1ERA0rHo8jHo/Pur34+NR05aaq7Zj5qFYgEMDIyMi89seRNRERiadGd4PPfP4ZmBrpWj33PBtVVdHZ2Wlad/LkSeOOcKf7Y7EmIiLx1OhucADo6uoy3Y09NDSEWCxmfFZVtezd2rlcrmRKW5ZlbN261bQuk8mY1lntrxxOgxMRkXhq+BOZ6XQayWQSwNQUdm9vr+nZ6Ewmg+HhYWP0qygKstksstksVFUFAEQiEeNvIpEIkskkJEnC6Ogo0um0aURttb9ymnS9Pi9gvWDx5fXYLRFNw7vBaS7nz71at32/dc/nXenng9/8gSv91BtH1kREJB7+6pYJizUREYmnhtPgXsAbzIiIiATHkTUREYmnwju5FwoWayIiEg+nwU1YrImISDg6bzAz4TVrIiIiwXFkTURE4uE0uAmLNRERiYfF2oTT4ERERILzRLHetTOBN46/jFfGnkXn+o55t7fbnxPMXJvMTvYjQm4vZt724wPo2PYIbrn/Rzhw6LWK/uaBp1/GjicU07qhl45i7Tf+vmTp2PZINWJ78lh7MbPravhDHl4g/DT4rp0J+HwSVl52NTrXd2DvEw/jyqtuwvj4MUft7fbHzOJm9mpuL2be9uMDKLz9Dp65uxsHDr2GzQ9k8fh//VO0+pbN2v7g4ddQOHMOG6+/0rQtfM1qPP/f/5Np3QNPv4zCmXdcy1vkxWPtxcxVwWlwE+F/yOP8uVexYuUaFAqnAEydeIVCAVvvutdRe7v9OcHMtcns1dwiZa70hzzWfuPv8fR/68byixcDmCrGy5Ysxh2fKv1t3unufOgptLYsm7PdRO70vxX/z1SUxc4PeYh0rL2YuZ4/5PFm/E9c6eeS1D+40k+9WU6Df/e738V3v/vdOdts3rwZ+/fvdy1U0bq11yCf14yTDAAU5UWsn2Uax6q93f6YWdzMXs3txcxjr57EsiWLjUINAGs+4sPBw5VNhVvZ9uMD+OqGuYu+E1481l7MXC36pO7K0igsi3U6nbb8nc2enh709va6FqrIL69GLqeZ1o2PH4PUvNxRe7v9OcHMtcnsZD8i5PZi5on8m2heuti0rtW3DKfPnpt33wcOvYZfvJpD+JrV8+5rJi8eay9mrppJ3Z2lQVhes1YUBe3t7XO2CQaDUBRlzjZOadO+ERb5fJLj9nb7c4KZa5PZyX5EyO3FzMuWLC5ZVzgz/2I99NLRkmvabvLisfZi5qrgG8xMLEfWzc3Nlp0UCgXLdqlUCq2trcYy+d6bFQUs9w1w5jdFO+3t9ucEM9cms5P9iJDbi5nLjaJnjradeHT419jwcfdH1UVePNZezEzVZ1msQ6EQBgcH52wzODhoOfqOx+OYmJgwlg8susQy3Lh6tOQboN+/quw3xUra2+3PCWauTWYn+xEhtxczt7ZcUjKKnsidLjvatqP4+Neay1fMq5/ZePFYezFz1XAa3MSyWH/961/Hpk2b8E//9E9lt+/btw+xWAzJZNL1cM89/xJaWiT4/auMdcHgtdi//xlH7e32x8ziZvZqbi9mXnP5Cpw+ew4TudPGurF/zeGGwG/Nq9+J3Gms+YhvvvFm5cVj7cXMVcNibWJZrIPBIHbv3o3Ozk5s2LAB9913Hx577DHcd9992LBhA26++Wbs3r0ba9eurUrA/j0Pou/ebwCYurMxGrkF/XseNLb7/atMJ6JVe6vtzOydzF7N7cXMG6+/Ejv2Tt2XMvbqSTz586OIfOL9a80TudOmYl6JidybpjvMq8GLx9qLman6KnqDWU9PD3K5HK644grs3r0bkUgEu3fvht/vx6FDh7Bp06aqBfzyV3qRy2l44/jL2L17O269bbPpYf6eTbfjOzsTFbe32s7M3sns1dxezHz3n96I5osvQse2R3DPjw9g+22fNL0QJfOzX2Pbjw8Yn3c8oaBj2yPIvnQM33/mZXRsewSZn/3K1OdE/vSsL1VxixePtRczV4Ou664sjUL4l6IQUfVU+lIUkdh5KQrNTz1finJq082u9LN8z5Ou9FNvwr9ulIiIFqAGut7shoqmwYmIiKh+OLImIiLhNNKrQt3AYk1EROJhsTbhNDgREZHgOLImIiLx8NXgJizWREQkHF6zNuM0OBERkeA4siYiIvFwZG3CYk1EROLhNWsTFmsiIhIOr1mb8Zo1ERGR4DiyJiIi8dR4GjyZTEKWZeRyOciyjFAoNGd7VVWRTqcRCATQ09Nj2qZpGgYGBqBpGoaHh9Hd3Y1IJGJsz2azkGUZsiwDABRFQTAYnHN/LNZERCScWk6Dx2IxRKNRo0BHo1FTMZ0pm80CmCrYgUCgZHtfXx8SiamfMdU0DX6/H7IsGwVZURSEw2EAQCgUQjqdtszIaXAiIlrQBgYGTCPpcDg8ZwENhUIIhULw+Xwl21RVhaZpxmdJktDV1YW+vj7TuuLvbQ8NDc36pWA6FmsiIhLPpEuLBUVRSoquz+czRs92+Xw+Ywp8upmf7eI0eIM7vfev6h3BtmUb/rLeERaMZZ/7Tr0jEJWl1+iadS6XgyRJpnWSJDkurpIkIZ/Pm9aNjIyUXAPPZDIApkbisiybrmmXw2JNRETicalYp1IppFIp43M8Hkc8Hjc+z1aUc7mcK/tXVRWqqmLr1q3Guvb2dsiybHxJaGtrM13TLofFmoiIGtbM4lxOuSnrctejnYhGo9i3b59p9D6zKLe3t6Ovrw+Dg4Oz9sNr1kREJBx90p3FSvFxrenKTY070dvbi0QiUVKcFUUxfQ4EAlBVdc6+WKyJiEg8NbrBrNzUs6Zpls9ZW+nv70c4HDb6KRZoVVXR2dlpanvy5EnLO8JZrImIaEHr6uoy3f09NDSEWCxmfFZVtezd4blcruw172w2C5/Ph/b2dmiaBlVVMTIyAmBqJD/9+jUwdbPZzHUz8Zo1EREJp1Z3gwNAOp1GMpkEMDUC7u3tNY10M5kMhoeHTaPkbDaLbDZrTF9HIhHIsgxVVY0XnkxXfElKsW0ymYQkSRgdHUU6nbZ8g1mTrut1eVv6BYsvr8duFxw+ukVETp0/92rd9v2bzt93pZ8P7XvKlX7qjdPgREREguM0OBERCaeW0+BewGJNRETi0ZvqnUAoLNZERCQcjqzNeM2aiIhIcBxZExGRcPRJToNPx2JNRETC4TS4GafBiYiIBMeRNRERCUfn3eAmnhhZ79qZwBvHX8YrY8+ic33HvNvb7c8JL2be9oO96Ih/G7f8t34cGDtS0d88sPcgdjxW+oagU2fO4s7+nxj9DY2+4nLa93nxWDNzbTI72Y8Iub2Y2W21+tUtrxB+ZL1rZwI+n4SVl12NzvUd2PvEw7jyqpswPn7MUXu7/S2UzNt+sBeFt87imdRf4MDYEWz+1iAe37YJrSulWdsfHDuKwpmz2Ph715Vs/1r/P+DGNatxX8+ncerMWXzu3r/DsqUX4cY1V7iWGfDmsWbm2mT2am4vZq4G3mBmJvy7wc+fexUrVq5BoXAKwNSJVygUsPWuex21t9ufEyJlrvTd4Gs3b8fTqf+C5UuXAJgqxssuXoI7Pjv3+3nv7P8JWi+VTO1OnTmLT8b/Bs/v/pqxLvPMCxg79jru/vwGyyx23g0u0rFmZrEyezW3SJnr+W7w/3t9p3WjCvz28D5X+qm3iqfBjxw5gu7ublx55ZVYtGgRrrzySnR3d+OFF16oWrh1a69BPq8ZJxkAKMqLWD/LNI5Ve7v9LZTMY8eOY9nSi4xCDQBrVn0YB395dF79njpz1vi/J97Q5tVXOV481sxcm8xeze3FzNWi6+4sjaKiYr19+3bIsgxd19HT04OBgQFs2bIFLS0tWLduHe66666qhPPLq5HLaaZ14+PHIDUvd9Tebn9OeDHzxBsamqcVagBovbQZp6cVWzuWL12CjR3X4ls/egqnzpzF2LHjyCqv4Is33+BGXIMXjzUz1yazk/2IkNuLmatFn2xyZWkUltes9+3bh76+PoyOjmLdunWmbZs2bcKWLVvQ3t6O66+/Hp/5zGdcD6hN+0ZY5PNJjtvb7c8JL2ZeNqNYA0DBYbEGgLs/vwF39v8En4z/DQDgobv+bNbr3/PhxWPNzLXJ7GQ/IuT2YmaqPsuRdTKZxODgYEmhLpJlGQMDA9i9e7fr4QCU/QY485uinfZ2+3PCi5nLjaJnjrbtiO0YwNWrP4znd38Nj2/bhHse3FvxHeZ2ePFYM3NtMjvZjwi5vZi5GjiyNrMs1iMjI+jsnPtCfygUwsjIyJxtUqkUWltbjWXyvTctw42rR0u+Afr9q8p+U6ykvd3+nPBi5taVUskoeuJEoexouxIHxo7gF8dexxc33GD0v7HjOnzrR0/PO+t0XjzWzFybzE72I0JuL2auFl6zNrMs1i0tLRV1ZNUuHo9jYmLCWD6w6BLLPp97/iW0tEjw+1cZ64LBa7F//zOO2tvtzwkvZl6z6jKcPvOO6SawsWOv44bfWe3aPqrBi8eamWuT2au5vZiZasOyWDc1VTaNUGk7u/r3PIi+e78BYOrOxmjkFvTvedDY7vevMp2IVu2tti/UzBs7rsWOH0293GTs2HE8OfoKIh3vPz898YZW8R3dN665As1Ll+CBvQcBTN0V/v29B7FxWn9u8eKxZubaZPZqbi9mrgZOg5tZ3mB2+PBhdHd3W3akqqorgWb68ld6jbfvqOPHcOttm00P8/dsuh3r1n4cn/qj2ypqb7V9oWa++/MbjDeYtV4qYfumPzHdEDb1nPRxpO/oAgDseOwpPPrTF3D6zDsAgEd/+gK++pnfNwr8D+/6M3yt/x/wvb0H0bx0Cb6w4QZT8XeLF481M9cms1dzezFzNfB1o2aWL0WJxWIVj5rt3GRW6UtRaH4qfSmKSOy8FIWIqqeeL0U59LvWL1CqxEd/sdeVfurNcmSdTqdrkYOIiIhmIfy7wYmIaOGZ5DS4CYs1EREJh9eszSp+NzgRERHVB0fWREQknEZ67MoNLNZERCScRnr7mBtYrImISDgcWZvxmjUREZHgOLImIiLh8NEtMxZrIiISDh/dMuM0OBERkeA4siYiIuHwbnAzFmsiIhJOra9ZJ5NJyLKMXC4HWZYRCoXmbK+qKtLpNAKBAHp6emz3Z3d/LNZERCScWl6zjsViiEajRsGMRqOQZRmyLJdtn81mAUwV7EAgYLs/u/sDeM2aiIgWuIGBAdPINhwOz/mLk6FQCKFQCD6fz1F/dvcHsFgTEZGAdN2dxYqiKCVF1+fzGaNnu6z6c7o/ToMTEZFwanXNOpfLQZIk0zpJkqBpWlX6c7o/FusGt2zDX9Y7ApGrTn/ntnpHsG3Zlx6qd4QFK5VKIZVKGZ/j8Tji8bjxebYimcvlHO3Pqj+n+2OxJiIi4bh1g9nM4lzOzAKqadqs16MrYdWfk/2xWBMRkXBqNQ1efHxqunJT1W7153R/vMGMiIgWrGAwWLJO0zTL556d9ud0fyzWREQkHN2lpRJdXV2mu7GHhoYQi8WMz6qqlr1bO5fLlb0GbdWf1fZyOA1ORETCqeUbzNLpNJLJJICpR6t6e3tNLyjJZDIYHh42Rr+KoiCbzSKbzUJVVQBAJBIx/saqP6vt5TTpen3ewHrB4svrsVsi8jjeDV4758+9Wrd9P/vhiCv9/PvXM670U2+cBiciIhIcp8GJiEg4k/UOIBgWayIiEo6O2v7qlug4DU5ERCQ4jqyJiEg4k3W59VlcLNZERCScSU6Dm7BYExGRcHjN2ozXrImIiATHkTUREQmHj26ZeWJkvWtnAm8cfxmvjD2LzvUd825vtz8nmLk2mZ3sR4TczFy782Pb3hfQ8a0ncEv/Phw48kZFf/PAwUPY8c+/KLvt1Nl3cedPRow+x17XXEw7xavH2k06mlxZGoXwI+tdOxPw+SSsvOxqdK7vwN4nHsaVV92E8fFjjtrb7Y+Zxc3s1dzMXLvzY9veF1A4+y6e+eqncODIG9g8cACP96xHq/TBWdsfPHoChbffxcbrVpVsP3X2XXzu757GFz7xUdz36XacOvuuq3kB7x5rqi7h3w1+/tyrWLFyDQqFUwCmTrxCoYCtd93rqL3d/pxg5tpk9mpuZp5fZjvvBl+b/J94+i8+heVLLgQwVYyXXXQh7vgPvzvn3935kxG0Ni8tabfjn3+B0++8i7s3XFdxBsDeu8FFOtb1fDf4E5fd6ko/nzr+sCv91JvlNPiRI0cq6uixxx6bb5YS69Zeg3xeM04yAFCUF7F+lmkcq/Z2+2NmcTN7NTcz1+78GHtdw7KLLjQKNQCsuUzCwaMnHPf56AvHELlutRvxyvLqsa6GSZeWRmFZrAOBQMm67u7uknXRaNSdRNP45dXI5TTTuvHxY5Calztqb7c/J5i5Npmd7EeE3Mxcu/NjonAGzRdfaFrXKi3F6XecTV1PaG/h9DvvonD2XcQe+Rd0fOsJbNv7ghtRDV491lR9lsW63Cx5uR/hrtZsujbtG2GRzyc5bm+3PyeYuTaZnexHhNzMXLvzY9lFF5asK7zttFifAQAMvfKvSHffhIf+vw4cPHoCmeePzivjTF491m7jDWZmlsW6qamyf9hK29lV7hvgzG+Kdtrb7c8JZq5NZif7ESE3M9fu/Cg3ip452rbb1xdv+CgAoFX6IDZetxqPvuBusfbqsXbbZJM7S6Oo2aNbqVQKra2txjL53puWfzOuHi35Buj3ryr7TbGS9nb7c4KZa5PZyX5EyM3MtTs/WpuXloyiJ7QzZUfblfa37KILTXeSz2davRyvHutqmESTK0ujqFmxjsfjmJiYMJYPLLrE8m+ee/4ltLRI8Pvff4QiGLwW+/c/46i93f6cYObaZPZqbmau3fmx5sMSTr/zLia0t4x1Y8c13LD60nn1N/1xrQntDC5vXjrvrEVePdZUfRVds96wYYNpyefzJeuqpX/Pg+i79xsApu5sjEZuQf+eB43tfv8q04lo1d5qOzN7J7NXczNz7c6Pjdetwo6nxgBM3R3+5C9fQ2Tt+3dzT2hvmYq5lS98IoB7/u2msgntLXzvwCFjWtwtXj3WbtNdWhqFZbHu6emB3+83LeXWbdq0qSoBv/yVXuRyGt44/jJ2796OW2/bbHqYv2fT7fjOzkTF7a22M7N3Mns1NzPX7vy4e8N1aF5yITq+9QTu2fsitn+6zTSNnXn+KLbtfdH4vOOff4GObz2B7Cuv4fs/O4yObz1huoHsjv/wu0Z/mwcO4JufuhY3XrHS1cxePdZu46NbZsK/FIWIaDo7L0URhZ2Xooikni9FeezDn3Oln8++/kNX+qk34V83SkREC89klZ4w8ioWayIiEk4jXW92A4s1EREJp5GuN7uhZo9uERERkTMcWRMRkXAa6e1jbmCxJiIi4TTS28fcwGlwIiIiwXFkTUREwuHd4GYs1kREJBxeszbjNDgREZHgOLImIiLh8DlrMxZrIiISDq9Zm7FYExGRcGp9zTqZTEKWZeRyOciyjFAo5Gr76bLZLGRZhizLAABFURAMBuf8G16zJiKiBS0WiyEYDCISiaCnpwfpdBqqqjpuH41GoSgKNE0zLUWKoiAQCKCpqQnhcBiSJFlmZLEmIiLh1PL3rAcGBkwj43A4jHQ67bh9NptFW1sbWlpaTEuxoEuSBF3Xoes6hoaGjBH2XFisiYhIOLUq1oqiwOfzmdb5fD5ks1nH7ROJhFGMdV3H4cOHKy7Ks+E1ayIiWrByuVzJNLQkSaZpa7vtu7q6TNuz2Sx6enpM6zKZDABAVVXIsoxIJDJnThZrEs7p/3V3vSPYtuyPttU7giOnd3XXO4Jty770UL0jUA3oNbrBbK6i7LT99GLe29uLRCJhatve3g5Zlo12bW1tkGV5zpvMOA1ORETCcWsaPJVKobW11VhSqVTJvmYWYE3TSqa6nbRXFAUrVqwoWR8MBk0Fvb29HX19fbPuD2CxJiIiAblVrOPxOCYmJowlHo+b9lN8/Gq6clPdTtr39fWVfaRLURTT50AgMOfd5wCLNRERLWDlpp41TZv1uWk77TOZTMlNZaqqorOz07Tu5MmTljefsVgTEZFwdJeWSnR1dZnu5h4aGkIsFjM+q6pq2m7Vvvg3AEpG3LIsY+vWraZ1mUymZN1MvMGMiIiEU8s3mKXTaSSTSQBTU9S9vb2mkW4mk8Hw8LAxerZqD0yNtmcbLUciESSTSUiShNHRUaTTacs3mDXpul6XV7BesPjyeuyWPIB3g9eOJ+8G//Ij9Y6wYJw/92rd9v2tVbe70s9Xjz3oSj/1xpE1EREJh7+6ZcZiTUREwmGxNuMNZkRERILjyJqIiITD37M2Y7EmIiLh1Pr3rEXHYk1ERMLhNWszXrMmIiISHEfWREQkHF6zNmOxJiIi4UyyXJt4Yhp8184E3jj+Ml4Zexad6zvm3d5uf04wc20yA8C2H2bRcecu3PKX38OBXx6t6G8eGBrGjh8/Y1o3pPwKa7/yP0qWjjt3uZ7Zi8d6296fo+Pbe3FL/z/hwJETFf3NAwcPY8dTY2W3nTr7Lu78yajR59jxgptxDV481l7MTNUl/Mh6184EfD4JKy+7Gp3rO7D3iYdx5VU3YXz8mKP2dvtjZnEzA1OFunDmLJ6578s48Muj2Pw3j+Hxv/oiWi+VZm1/8JVjKLx1Fht/7+OmbeHgx/B88GOmdQ8MDaPw1llXM3vxWG/b+3MUzp7DM3+xAQeOnMDmwYN4fNMfoFVaOmv7g0dPoHD2HDZet6pk+6mz7+Jzf/dTfOGGAO77dBtOnX3XtazTefFYezFzNfAGMzPh3w1+/tyrWLFyDQqFUwCmTrxCoYCtd93rqL3d/pxg5vlltvNu8LVf+R94evuXsHzpEgBTxXjZ0otwx5/OPXq487uPo/XS5jnbTZzQ/q34/7llDjvvBhfqWFf4bvC12/8RT/+Xm7F8yYUAporxsiUX4I7fXzPn3935k1G0SktL2u14agynz57H3Rs+Pstfzs7Ou8FFOtZezFzPd4Pfs/rzrvTzzaM/cKWfehN6Gnzd2muQz2vGSQYAivIi1s8yjWPV3m5/zCxuZgAYO3Ycyy6+yCjUALBm1Ydw8JfujBi2PbQPX/3077nSV5EXj/XY8QKWXXSBUagBYM2Hm3HwaGVT4eU8+sIxRNaWjrjd5MVj7cXM1TLp0tIoLIv1ihUrStbt37+/KmFm8surkctppnXj48cgNS931N5uf04wc20yA8DEiQKaP7jEtK710macfvudefd94JdH8YujxxGeMS0+X1481hPaGTQvWWxa19q8FKfPnnfc3+l3zqPw9ruIDRxEx7f3Ytven7sR1cSLx9qLmak2LIt1Pp8vWReNRqsSphxt2jfCIp9Pctzebn9OMHNtMgPAsosvKlnnxjXmIeXXJde03eLFY71sSentLYWz5xz1NaGdAQAMvfIa0l034KE/68DBoyeQecH9a6hePNZezFwNk03uLI3Cslg3NZX+09byMne5b4AzvynaaW+3PyeYuTaZAZQdRc8cbTvx6LM/xwaXR9VFXjzW5UbRM0fbFff1ztTNZF+8IQAAaJWWYuN1q/DoC5XdyW+HF4+1FzNXwyR0V5ZG4eiadbkCbiWVSqG1tdVYJt970/JvxtWjJd8A/f5VZb8pVtLebn9OMHNtMgNTU94zR9ETJwplR9t2FB//WrPqsnn1U44Xj3WrtLRkFD1ROFN2tF1pf8suusB0J3mr5HxafTZePNZezEy1UbMbzOLxOCYmJozlA4susfyb555/CS0tEvz+929ECQavxf79zzhqb7c/J5i5NpmBqWJ6+u13MHFCM9aNHfsNbvid+d24NHGigDW//aF5pivPi8d6zWXNOP3OeWP6GgDGXi/ghtWXzqu/6Y9rTWhncPksj4E55cVj7cXM1aK7tDQKy2Kt6zo2bNhgWvL5fMm6DRs2VCVg/54H0XfvNwBM3dkYjdyC/j0PGtv9/lWmE9GqvdV2ZvZOZgDY+O8/jh0//imAqbvDn1R+hci0a80TJzRTMa/ExIkClrswlT4bLx7rjdeuMl5uMna8gCdf+VdErlttbJ/QzpiKuZUvfELGPXtfNP72ewcP4YufCLiaGfDmsfZi5mrg3eBmlsW6p6cHfr/ftJRb5/f7qxLwy1/pRS6n4Y3jL2P37u249bbNpof5ezbdju/sTFTc3mo7M3snMwDc/bkQmpcuQcedu3DPD7PY/p//2PRClMxPf45tD+0zPu/48TPouHMXss/9Gt8fGkHHnbuQ+emLpj4nThTQuqLZ9axFXjzWd2/4OJqXLEbHt/finr0vYvuftJmmsTMvHMW2J9+/o3vHU2Po+PZeZH/1Or7/MxUd395ruoHsjt9fY/S3eeAgvrnhWtx4hbOR+ly8eKy9mJmqT/iXotDCY+elKKKw81IUkVT6UhSR2HkpCs1PPV+K0nvFba70kzjykCv91JvwrxslIqKFp5GuN7uBxZqIiITTSNeb3SD060aJiIiII2siIhJQI73QxA0s1kREJByWajNOgxMREQmOI2siIhIObzAzY7EmIiLh6JwIN2GxJiIi4XBkbcZr1kRERILjyJqIiITDR7fMWKyJiEg4LNVmnAYnIiISHEfWREQknFpPgyeTSciyjFwuB1mWEQqFHLfPZrOQZRmyLAMAFEVBMBic1/5YrImISDi1vBs8FoshGo0aBTMajZqKrd32iqIgHA4DAEKhENLp9Lz2B3AanIiIFriBgQHTyDYcDpcUWDvtJUmCruvQdR1DQ0MlRdju/gAWayIiEpDu0v+sKIoCn89nWufz+ZDNZl1p79bfcxqchLPsj7bVO8KCsezLj9Q7AlFZtZoGz+VykCTJtE6SJGiaNq/2mUwGAKCqKmRZRiQScbS/IhZrIiISjluvG02lUkilUsbneDyOeDxufJ6rKJdTSfv29nbIsmwU5ba2NsiyjGAwaHt/RSzWRETUsGYW53JmFlBN00qmqu20n3nnd3t7O/r6+jA4OOhofwCvWRMRkYAmXVqsFB+fmq7cVLWd9oqimLYHAgGoqupof0UcWRMRkXAm9do8Zz1zFAxMjXRne+7Zqr2qqujs7EQ+nze2nzx50rgj3O7+ijiyJiKiBa2rq8t0N/bQ0BBisZjxWVVV0/a52suyjK1bt5r6z2QypnVW+yunSddr9PVlhgsWX16P3RIRUYXOn3u1bvu+ffVnXennwaOPVdQumUwiGAwabxubPtJNJpMYHh42rjlbtVdVFZlMBpIkYXR01PQClEr+vhwWayIiKquexfpzqz/jSj8/PPojV/qpN16zJiIi4bj16Faj4DVrIiIiwXFkTUREwqnlD3l4AYs1EREJp9Y/kSk6ToMTEREJjiNrIiISDm8wM2OxJiIi4fCatRmnwYmIiATHkTUREQmnTu/rEpYnRta7dibwxvGX8crYs+hc3zHv9nb7c4KZa5PZyX5EyM3MPD/mk8Fu+1odazdNQndlaRTCj6x37UzA55Ow8rKr0bm+A3ufeBhXXnUTxsePOWpvtz9mFjezV3MzM8+PRstcDbxmbSb8u8HPn3sVK1auQaFwCsDUiVcoFLD1rnsdtbfbnxPMXJvMXs3NzDw/vJK5nu8G/4+rbnGln/957HFX+qk3oafB1629Bvm8ZpxkAKAoL2L9LNM4Vu3t9sfM4mb2am5m5vnRaJmrRXfpf43Csljv2bMH3d3dJet9Ph8WLVpkLPfff7/r4fzyauRymmnd+PgxSM3LHbW3258TzFybzE72I0JuZub5MRcvZq4WXrM2syzW/f39ZYs1ABw6dAiHDh3CI488gocfftj1cACgTftGWOTzSY7b2+3PCWauTWYn+xEhNzPz/JiLFzNT9VkWa1VVZ/1RbL/fD7/fj3A4DFVV5+wnlUqhtbXVWCbfe7OigOW+Ac78pminvd3+nGDm2mR2sh8RcjMzz4+5eDFzNei67srSKCyL9Wz/sKOjoxW1K4rH45iYmDCWDyy6xDLcuHq05Bug37+q7DfFStrb7c8JZq5NZif7ESE3M/P8mIsXM1fLpEtLo7As1qFQCIODgyXr/X6/8X8PDg7OOvqej+eefwktLRL8/lXGumDwWuzf/4yj9nb7Y2ZxM3s1NzPz/Gi0zNXCG8zMLB/dUhQF7e3t2LdvH/7gD/6gZPujjz6Krq4ujI6OYu3atRXvuNJHt4rPCN56Wwzr1l6DJ/c+gk/c+IfGM4LFk3DmM4Wztbfa7gZmrk1mr+ZmZp4fXslcz0e3bv7tT7nSz5P/9wlX+qk3y5F1MBjE7t270dnZiQ0bNuC+++7DY489hvvuuw/XX389urq6MDAwYKtQ2/Hlr/Qil9PwxvGXsXv3dtx622bTSdaz6XZ8Z2ei4vZW25nZO5m9mpuZeX40WuZq4N3gZhW/FEVVVXz961+HoihQVRXBYBDt7e1IJBJobm62veNKR9ZERFQf9RxZd7be7Eo/+yaedKWfeqv4daOyLGNgYKCaWYiIiKgM4d8NTkREC08jTWG7gcWaiIiE00h3cruBxZqIiIQz2UAvNHGD5d3gREREVF8cWRMRkXA4rjZjsSYiIuHwBjMzToMTEREJjiNrIiISDkfWZizWREQknEb6eUs3cBqciIhIcBxZExGRcDgNbsZiTUREwqn1G8ySySRkWUYul4MsywiFQo7ba5qGgYEBaJqG4eFhdHd3IxKJGNuz2SxkWYYsywCmfoo6GAzOuT8WayIiEk4tr1nHYjFEo1Gj4EajUVMxtdu+r68PicTUz5hqmga/3w9Zlo2CrCgKwuEwACAUCiGdTltm5DVrIiJa0AYGBkwj43A4PGcBnau9qqrQNM3YJkkSurq60NfXZ1qn6zp0XcfQ0NCsXwqmY7EmIiLhTEJ3ZbGiKAp8Pp9pnc/nQzabddTe5/MZU+DTzfxsF6fBiYhIOLWaBs/lcpAkybROkqRZi6tVe0mSkM/nTdtHRkZKroFnMhkAUyNxWZZN17TLYbEmIqKGlUqlkEqljM/xeBzxeNz4PFdRLsdue1VVoaoqtm7daqxrb2+HLMtG0W9razNd0y6HxZqIiITj1qNbM4tzOeWmrGdOdTttH41GsW/fPtNofGZRbm9vR19fHwYHB2fdJ69ZExGRcHSX/mel+PjVdOWmup207+3tRSKRKCnOiqKYPgcCAaiqOmdOFmsiIhLOpK67slgpN/Wsadqsz1lX2r6/vx/hcNhYXyzQqqqis7PT1PbkyZOWd4SzWBMR0YLW1dVluvt7aGgIsVjM+Kyqqmm7VftsNgufz4f29nZomgZVVTEyMgJgamQ+/fo1MHWz2cx1MzXpdXpb+gWLL6/HbomIqELnz71at31ffdkNrvTz8vGDFbVLJpMIBoPG28Smj5STySSGh4dN15Rna6+qKgKBQEn/iUQCW7ZsMdpkMhlIkoTR0VHTC1Zmw2JNRERl1bNYr/nQJ1zpZ+w3P3Oln3rjNDgREZHg+OgWEREJp9Y/5CE6FmsiIhJOJXdyLyScBiciIhIcR9ZERCQcToObsVgTEZFwOA1uxmJNRETC4cjajNesiYiIBMeRNRERCUfXJ+sdQSieGFnv2pnAG8dfxitjz6Jzfce829vtzwlmrk1mJ/sRITcz8/yYTwa77Wt1rN00Cd2VpVEIP7LetTMBn0/CysuuRuf6Dux94mFcedVNGB8/5qi93f6YWdzMXs3NzDw/Gi0zVZ/w7wY/f+5VrFi5BoXCKQBTJ16hUMDWu+511N5uf04wc20yezU3M/P88Ermer4bfJXv4670cyz3c1f6qTfLafAjR45UvLht3dprkM9rxkkGAIryItbPMo1j1d5uf8wsbmav5mZmnh+NlrlaOA1uZjkNLssympqa5myj6zqamprw3nvvuRYMAPzyauRymmnd+PgxSM3LHbW3258TzFybzE72I0JuZub5MRcvZq6WOk36CsuyWB8+fNj0Wdd1fPSjHy1ZXy3atG+ERT6f5Li93f6cYObaZHayHxFyMzPPj7l4MTNVn2Wx9vv9JetaWlrKrq+Gct8AZ35TtNPebn9OMHNtMjvZjwi5mZnnx1y8mLka+AYzs5o9upVKpdDa2mosk++9afk34+rRkm+Afv+qst8UK2lvtz8nmLk2mZ3sR4TczMzzYy5ezFwtukv/axQ1K9bxeBwTExPG8oFFl1j+zXPPv4SWFgl+/ypjXTB4Lfbvf8ZRe7v9OcHMtcns1dzMzPOj0TJTbTh6dGvFihU4efLkvHZc6aNbxWcEb70thnVrr8GTex/BJ278Q+MZweJJOPOZwtnaW213AzPXJrNXczMzzw+vZK7no1uXNf+OK/0cL/zSlX7qTrdw8803lyxNTU1l19ux6MKPVLyk+/9ez+Xy+sjoC/rNG7pN25Lb/1YfGnqq4vaVbHdjYebaZPZqbmbm+eGFzPV06fKPubI0CsuR9ebNmysu/Lt37664baUjayIiqo96jqxXNl/lSj9vFF5xpZ96E/4NZkREVB/1LNaXLv+YK/2cOPUrV/qpN+HfDU5ERAsPH90yY7EmIiLh1GnSV1g1e3SLiIiInOHImoiIhNNIP8LhBhZrIiISDqfBzTgNTkREJDiOrImISDi8G9yMxZqIiITTSD/C4QYWayIiEg5H1ma8Zk1ERCQ4jqyJiEg4vBvcjMWaiIiEU+tr1slkErIsI5fLQZZlhEKhebWf7/aZWKyJiGhBi8ViiEajRsGMRqOQZRmyLDtqP9/t5fBXt4iIqKx6/urW4otaXenn3DsTlm1aWlqQz+eNz/39/Th8+DASiYSj9vPdXg5vMCMiIuHouu7KYkVRFPh8PtM6n8+HbDbrqP18t8+GxZqIiBasXC4HSZJM6yRJgqZpjtrPd/ts6nbNulrTK6lUCvF4vCp9V5MXczNzbXgxM+DN3MwsjnddqhGpVAqpVMr4HI/HTcdrrqJcjlX7+W6fTcONrKf/P8VLvJibmWvDi5kBb+Zm5sYTj8cxMTFhLOW+2MwsoJqmlUxV22k/3+3lNFyxJiIiqlTx8anpyk1VV9p+vttnw2JNREQLVjAYLFmnadqszz1btZ/v9tk0XLH26rUbL+Zm5trwYmbAm7mZeWHq6uoy3Y09NDSEWCxmfFZV1bTdqv18t5dTt+esiYiIRJFMJhEMBqEoCoLBoGmkm0wmMTw8jMHBwYrau7F9JhZrIiIiwTXcNDgREVGjaZhinUwm0dvbW+8Ytqiqira2NjQ1NaGtrc3yDTYi0DQN0WgUgUAAgUAAmUym3pEqpmkampqaLK8NiaJ4bkxfvHSOtLS0IBAIQFGUekeaVSaTKTnGTU1NaGlpqXc0SzOPs5f+XST7PF+sY7EYAoEA+vr66h3FtlgshkQiAV3XsXXrVoTDYaH/wwZMFZDu7m4cPnwYg4ODiEajwmcu6u3tLXsnpsiGhoZMr060uq5Vb5qmoa2tDeFwGPl8HqOjo3P+OEG9RSKRktdTJhIJ9PT01DuapWg0iuuvv944zr29vZ74MkfOeL5Yp9NpHD58WPj/iM2kKIrpZ9EikQgikQjS6XSdk80tkUggEokAmHoEQZZlT/wHong3p9fOE6/p6+tDKBQyip0kSZbPj4pEVVWk0+k5f1BBBJqmIZvNYsuWLQCmjnNvb6/pBihqLJ4v1l4VDAZLpu1lWYaqqnVKVJlioS5SVdUTBbC3t9cz099e1t/f7+njXJzt8orpb8I6fPhw/YJQ1bFY19HM6UFVVYWeMpxO0zSEw2GEQiHhp5YVRTGNQrwkkUh45pqkqqrQNA25XA7hcBgtLS2eKtzZbBYjIyMlX0hFJEkSenp60NvbC03ToCgKMpmM5+7bocrV7Yc8yKxYUMbHx+sdxdL0m/m88G2+t7fXU6OlmcbHx5HL5RAIBDA6Oirsl6PirNDg4CCGhoagqirC4TD6+/s9cQ14cHDQEzmL0um0cYMZAOHvD6D54chaAMW7Ovft2+eJ63tbtmyBrusYHR1FOBwW+pp1cbTkpf8IFxWLniRJkGVZ+HsailOyxS9ysiwjFosJnXm6/v5+dHd31ztGxcLhMK6//nrouo7Dhw9j06ZNQv+7SPPDkbUAOjs7kU6nhR0xzab41p3e3l6Mjo7WO05Zg4OD0DTNGH0UC8rAwADGx8eF/nI0c5Tk8/mEvqdBlmXji8X0dVa/0yuCYpHzyr+DxS+hQ0NDAN7/YiTyv4s0PxxZ11k4HEYikfDETVrl/qMrcrEDpqYKdV1HPp9HPp9HT08Penp6kM/nhc8+83iLfk9DMBiEpmmm3KJnLlJV1TOFmhYmFus6CofDiMVinijUwPsvcSk+V62qqufv/hWVoiimY53JZJDNZoW/gWjLli3YtGkTgKnzo6+vT/jMwNS9F1a/JyySUCgEn8+HZDIJYOqLXSKR4L+LjUz3uC1btuiSJOkAdAC6JEl6Op2udyxLo6OjRubpSzAYrHe0OQ0NDemhUEiXJEmXZdkTx3q6np4evaenp94xKjI4OKjLsqxLkqQHg0F9dHS03pEq0tPTY5wfg4OD9Y5TkUgk4pnzoiifz3v630Wyhz/kQUREJDhOgxMREQmOxZqIiEhwLNZERESCY7EmIiISHIs1ERGR4FisiYiIBMdiTUREJDgWayIiIsGxWBMREQmOxZqIiEhw/w/xB98T7yH+GwAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 800x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig=plt.figure(figsize=(10, 6), dpi= 80, facecolor='w', edgecolor='k')\n",
    "vv = 0.2\n",
    "sns.heatmap(pre_intv_pred.detach().cpu().numpy().reshape(8, 8), vmin=0., vmax=vv, \n",
    "            yticklabels=list(\"ABCDEFGH\"), xticklabels=list(range(1,9)), square=True, \n",
    "            annot=True, fmt=\".2f\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c0e559c0-78b1-4114-af0f-521db6b7107d",
   "metadata": {},
   "source": [
    "## Check post-intervention ground-truth next steps\n",
    "0 for white; 1 for blank; 2 for black"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "554c0a8c-8746-4089-a888-bb28fa1aaf5c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--------------------\n",
      "['e6', 'd6', 'c3', 'f3', 'c4']\n",
      "a                \n",
      "b                \n",
      "c     X X        \n",
      "d       X X O    \n",
      "e       O X O    \n",
      "f     O          \n",
      "g                \n",
      "h                \n",
      "  1 2 3 4 5 6 7 8\n",
      "--------------------\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "['b3', 'b4', 'c6', 'd3', 'f4']"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "htd = {\"lr\": 1e-3, \"steps\": 1000, \"reg_strg\": 0.2}\n",
    "for wtd in wtd_list:\n",
    "    move = permit(wtd[\"intervention_position\"])\n",
    "    r, c = move // 8, move % 8\n",
    "    ab.state[r, c] = wtd[\"intervention_to\"] - 1\n",
    "ab.__print__()\n",
    "post_intv_valids = [permit_reverse(_) for _ in ab.get_valid_moves()]\n",
    "post_intv_valids"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f6bd486f-ec23-4700-bf5a-9d85eed8813d",
   "metadata": {},
   "source": [
    "## Intervene and observe how the world representation changes along the way"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "83ef294d-ad0b-4256-bd16-0ef533250e6f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "e6 Sucessfully intervened!\n",
      "e6 Sucessfully intervened!\n",
      "e6 Sucessfully intervened!\n",
      "e6 Sucessfully intervened!\n",
      "e6 Sucessfully intervened!\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x3200 with 10 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axs = plt.subplots(layer_e - layer_s, 2, figsize=(8 * (1), 8 * (layer_e - layer_s)), dpi= 80, facecolor='w', edgecolor='k')\n",
    "# two rows for the intervened layer layer_s, one for the rest\n",
    "if len(axs.shape) == 1:\n",
    "    axs = axs[:, None]\n",
    "\n",
    "p = probes[layer_s]\n",
    "whole_mid_act = models[layer_s].forward_1st_stage(partial_game[None, :])  # [B, T, F=512]\n",
    "\n",
    "# intervene at the earlest interested layer \n",
    "mid_act = whole_mid_act[0, -1]\n",
    "pre_intv_logits = p(mid_act[None, :])[0].squeeze(0)  # [64, 3]\n",
    "plot_mentals(axs[0, 0], pre_intv_logits)\n",
    "axs[0, 0].set_title(f\"Pre-intervention Probe Result \\n at the {layer_s}-th Layer\")\n",
    "labels_pre_intv = pre_intv_logits.detach().argmax(dim=-1)\n",
    "new_mid_act = mid_act.clone()\n",
    "for wtd in wtd_list:\n",
    "    new_mid_act = intervene(p, new_mid_act, labels_pre_intv, wtd, htd, plot=True)\n",
    "    pre_intv_logits = p(new_mid_act[None, :])[0].squeeze(0)  # [64, 3]\n",
    "    labels_pre_intv = pre_intv_logits.detach().argmax(dim=-1)\n",
    "post_intv_logits = p(new_mid_act[None, :])[0].squeeze(0)  # [64, 3]\n",
    "plot_mentals(axs[0, 1], post_intv_logits)\n",
    "axs[0, 1].set_title(f\"Post-intervention Probe Result \\n at the {layer_s}-th Layer\")\n",
    "# swap in \n",
    "whole_mid_act[0, -1] = new_mid_act\n",
    "\n",
    "for i, layer in enumerate(range(layer_s, layer_e - 1)):  # 4, 5, 6, 7, indices of the layers to be passed\n",
    "    p = probes[layer+1]\n",
    "    whole_mid_act = models[layer_s].forward_2nd_stage(whole_mid_act, layer, layer+1)[0]  # [1, T, F=512]\n",
    "    \n",
    "    # intervene the output of the features freshly out\n",
    "    mid_act = whole_mid_act[0, -1]\n",
    "    pre_intv_logits = p(mid_act[None, :])[0].squeeze(0)  # [64, 3]\n",
    "    plot_mentals(axs[i+1, 0], pre_intv_logits)\n",
    "    axs[i+1, 0].set_title(f\"Post-intervention Probe Result \\n at the {layer+1}-th Layer\")\n",
    "    labels_pre_intv = pre_intv_logits.detach().argmax(dim=-1)\n",
    "    new_mid_act = mid_act.clone()\n",
    "    for wtd in wtd_list:\n",
    "        new_mid_act = intervene(p, new_mid_act, labels_pre_intv, wtd, htd, plot=True)\n",
    "        pre_intv_logits = p(new_mid_act[None, :])[0].squeeze(0)  # [64, 3]\n",
    "        labels_pre_intv = pre_intv_logits.detach().argmax(dim=-1)\n",
    "    post_intv_logits = p(new_mid_act[None, :])[0].squeeze(0)  # [64, 3]\n",
    "    plot_mentals(axs[i+1, 1], post_intv_logits)\n",
    "    axs[i+1, 1].set_title(f\"Post-intervention Probe Result \\n at the {layer+1}-th Layer\")\n",
    "    # swap in \n",
    "    whole_mid_act[0, -1] = new_mid_act"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b931979c-6224-42e9-8756-3b04706a0091",
   "metadata": {},
   "source": [
    "## Compare post-intervention prediction heatmap with pre-intervention ones\n",
    "Underscored tiles are the ground-truth"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "3529ff1c-b645-477f-ab71-b42ee4eb5546",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Post-intervention Prediction Heatmap')"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1600x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axs = plt.subplots(1, 2, figsize=(20, 6), dpi= 80, facecolor='w', edgecolor='k')\n",
    "\n",
    "plot_probs(axs[0], pre_intv_pred, [permit(_) for _ in pre_intv_valids])\n",
    "axs[0].set_title(f\"Pre-intervention Prediction Heatmap\")\n",
    "\n",
    "tb_resumed = whole_mid_act\n",
    "post_intv_pred, _ = models[layer_s].predict(tb_resumed)\n",
    "post_intv_pred = post_intv_pred[0, -1, 1:]\n",
    "post_intv_pred = torch.softmax(post_intv_pred, dim=0)\n",
    "post_intv_pred = torch.cat([post_intv_pred[:27], padding, post_intv_pred[27:33], padding, post_intv_pred[33:]], dim=0)\n",
    "\n",
    "plot_probs(axs[1], post_intv_pred, [permit(_) for _ in post_intv_valids])\n",
    "axs[1].set_title(f\"Post-intervention Prediction Heatmap\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "othello",
   "language": "python",
   "name": "othello"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}