src/final/beetl_sleep_mixup_vat_v3.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "f2JZC8x0mZRd",
        "outputId": "0d6aa6f1-3ca9-4f4a-d794-0cb80c1ef1e5"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Fri Oct  1 14:16:14 2021       \n",
            "+-----------------------------------------------------------------------------+\n",
            "| NVIDIA-SMI 470.63.01    Driver Version: 460.32.03    CUDA Version: 11.2     |\n",
            "|-------------------------------+----------------------+----------------------+\n",
            "| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n",
            "| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n",
            "|                               |                      |               MIG M. |\n",
            "|===============================+======================+======================|\n",
            "|   0  Tesla K80           Off  | 00000000:00:04.0 Off |                    0 |\n",
            "| N/A   48C    P8    32W / 149W |      0MiB / 11441MiB |      0%      Default |\n",
            "|                               |                      |                  N/A |\n",
            "+-------------------------------+----------------------+----------------------+\n",
            "                                                                               \n",
            "+-----------------------------------------------------------------------------+\n",
            "| Processes:                                                                  |\n",
            "|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |\n",
            "|        ID   ID                                                   Usage      |\n",
            "|=============================================================================|\n",
            "|  No running processes found                                                 |\n",
            "+-----------------------------------------------------------------------------+\n"
          ]
        }
      ],
      "source": [
        "!nvidia-smi"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "JBvEXtUsx-Iv",
        "outputId": "29a8ed9e-d3af-4dee-e71c-f4e0983a73a1"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Collecting mne\n",
            "  Downloading mne-0.23.4-py3-none-any.whl (6.9 MB)\n",
            "\u001b[K     |████████████████████████████████| 6.9 MB 2.0 MB/s \n",
            "\u001b[?25hRequirement already satisfied: numpy>=1.15.4 in /usr/local/lib/python3.7/dist-packages (from mne) (1.19.5)\n",
            "Requirement already satisfied: scipy>=1.1.0 in /usr/local/lib/python3.7/dist-packages (from mne) (1.4.1)\n",
            "Installing collected packages: mne\n",
            "Successfully installed mne-0.23.4\n",
            "Collecting braindecode\n",
            "  Downloading Braindecode-0.5.1-py3-none-any.whl (97 kB)\n",
            "\u001b[K     |████████████████████████████████| 97 kB 2.2 MB/s \n",
            "\u001b[?25hCollecting skorch\n",
            "  Downloading skorch-0.10.0-py3-none-any.whl (128 kB)\n",
            "\u001b[K     |████████████████████████████████| 128 kB 12.1 MB/s \n",
            "\u001b[?25hRequirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from braindecode) (1.19.5)\n",
            "Requirement already satisfied: matplotlib in /usr/local/lib/python3.7/dist-packages (from braindecode) (3.2.2)\n",
            "Requirement already satisfied: mne in /usr/local/lib/python3.7/dist-packages (from braindecode) (0.23.4)\n",
            "Requirement already satisfied: scipy in /usr/local/lib/python3.7/dist-packages (from braindecode) (1.4.1)\n",
            "Requirement already satisfied: h5py in /usr/local/lib/python3.7/dist-packages (from braindecode) (3.1.0)\n",
            "Requirement already satisfied: pandas in /usr/local/lib/python3.7/dist-packages (from braindecode) (1.1.5)\n",
            "Requirement already satisfied: cached-property in /usr/local/lib/python3.7/dist-packages (from h5py->braindecode) (1.5.2)\n",
            "Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.7/dist-packages (from matplotlib->braindecode) (0.10.0)\n",
            "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib->braindecode) (2.4.7)\n",
            "Requirement already satisfied: kiwisolver>=1.0.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib->braindecode) (1.3.2)\n",
            "Requirement already satisfied: python-dateutil>=2.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib->braindecode) (2.8.2)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.7/dist-packages (from cycler>=0.10->matplotlib->braindecode) (1.15.0)\n",
            "Requirement already satisfied: pytz>=2017.2 in /usr/local/lib/python3.7/dist-packages (from pandas->braindecode) (2018.9)\n",
            "Requirement already satisfied: scikit-learn>=0.19.1 in /usr/local/lib/python3.7/dist-packages (from skorch->braindecode) (0.22.2.post1)\n",
            "Requirement already satisfied: tqdm>=4.14.0 in /usr/local/lib/python3.7/dist-packages (from skorch->braindecode) (4.62.3)\n",
            "Requirement already satisfied: tabulate>=0.7.7 in /usr/local/lib/python3.7/dist-packages (from skorch->braindecode) (0.8.9)\n",
            "Requirement already satisfied: joblib>=0.11 in /usr/local/lib/python3.7/dist-packages (from scikit-learn>=0.19.1->skorch->braindecode) (1.0.1)\n",
            "Installing collected packages: skorch, braindecode\n",
            "Successfully installed braindecode-0.5.1 skorch-0.10.0\n"
          ]
        }
      ],
      "source": [
        "!pip install mne\n",
        "!pip install braindecode"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LHx5IcpFmc4e",
        "outputId": "45be23b0-fe47-420e-f36c-ed75d25d750a"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Drive already mounted at /gdrive; to attempt to forcibly remount, call drive.mount(\"/gdrive\", force_remount=True).\n",
            "/gdrive\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/gdrive')\n",
        "%cd /gdrive"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "yQV_qPGgvys-"
      },
      "outputs": [],
      "source": [
        "DATA_FOLDER = \"/home/data/\"\r\n",
        "OUTPUT_FOLDER = \"/home/output/\"\r\n",
        "\r\n",
        "GDRIVE_DATA_FOLDER = \"/gdrive/My Drive/Projects/BEETL/data/sleep/\"\r\n",
        "GDRIVE_OUTPUT_FOLDER = \"/gdrive/MyDrive/Projects/BEETL/output/\""
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "1I0MRUXImkdI"
      },
      "outputs": [],
      "source": [
        "!mkdir {OUTPUT_FOLDER}\r\n",
        "!mkdir {DATA_FOLDER}\r\n",
        "!unzip -qq {GDRIVE_DATA_FOLDER}SleepSource.zip -d {DATA_FOLDER}\r\n",
        "!unzip -qq {GDRIVE_DATA_FOLDER}LeaderboardSleep.zip -d {DATA_FOLDER}\r\n",
        "!unzip -qq {GDRIVE_DATA_FOLDER}finalSleep.zip -d {DATA_FOLDER}"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "WvaKITJanDYm",
        "outputId": "b36347a6-4f35-4629-b598-6d0f8f3b472d"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "finalSleep  LeaderboardSleep  SleepSource\n",
            "sleep_target  testing\n",
            "sleep_target  testing\n"
          ]
        }
      ],
      "source": [
        "!ls {DATA_FOLDER}\n",
        "!ls {DATA_FOLDER}LeaderboardSleep/\n",
        "!ls {DATA_FOLDER}finalSleep/"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0pKksrpU_BgV"
      },
      "source": [
        "## setup"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "458BBEygxACF"
      },
      "outputs": [],
      "source": [
        "import os\n",
        "import pickle\n",
        "import numpy as np\n",
        "from tqdm import tqdm\n",
        "import matplotlib.pyplot as plt"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "UN8BExC-xtNk"
      },
      "outputs": [],
      "source": [
        "SOURCE_DATA_FOLDER = DATA_FOLDER + \"SleepSource/\"\n",
        "LEADERBOARD_TARGET_DATA_FOLDER = DATA_FOLDER + \"LeaderboardSleep/sleep_target/\"\n",
        "LEADERBOARD_TEST_DATA_FOLDER = DATA_FOLDER + \"LeaderboardSleep/testing/\"\n",
        "FINAL_TARGET_DATA_FOLDER = DATA_FOLDER + \"finalSleep/sleep_target/\"\n",
        "FINAL_TEST_DATA_FOLDER = DATA_FOLDER + \"finalSleep/testing/\""
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iTtgjHnb_C71"
      },
      "source": [
        "## data_utils"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "3VTwOdosGN3B"
      },
      "outputs": [],
      "source": [
        "import mne"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ykSu_8O1wo_x"
      },
      "outputs": [],
      "source": [
        "def load_data(data_folder, load_labels=True):\n",
        "    fn_list = sorted(os.listdir(data_folder))\n",
        "    print(f\"Loading data from folder: {data_folder} ({len(fn_list)} files) - Load labels {load_labels}\")\n",
        "\n",
        "    data_map = {}\n",
        "    subject_list = []\n",
        "    sample_counter = 0\n",
        "\n",
        "    for fn in tqdm(fn_list):\n",
        "        if fn.endswith(\"X.npy\"):\n",
        "            code = fn.split(\"_\")[1][:-4]\n",
        "        elif fn == \"headerInfo.npy\":\n",
        "            meta = np.load(data_folder + fn, allow_pickle=True)\n",
        "            print(meta)\n",
        "            continue\n",
        "        else:\n",
        "            continue\n",
        "\n",
        "        eeg = np.load(data_folder + fn, allow_pickle=True)\n",
        "\n",
        "        if load_labels:\n",
        "            label_fn = fn.replace(\"X\", \"y\")\n",
        "            label = np.load(data_folder + label_fn, allow_pickle=True)\n",
        "        else:\n",
        "            label = None\n",
        "\n",
        "        s_part, r_part = code.split(\"r\")\n",
        "        subject = int(s_part[1:])\n",
        "        repetition = int(r_part[:-1])\n",
        "        \n",
        "        subject_list.append(subject)\n",
        "\n",
        "        if subject not in data_map.keys():\n",
        "            data_map[subject] = {}\n",
        "\n",
        "        data_map[subject][repetition] = {\"eeg\": eeg, \"label\": label}\n",
        "        sample_counter += len(eeg)\n",
        "\n",
        "    subject_list = np.unique(subject_list)\n",
        "    print(f\"Loaded total {sample_counter} samples for subjects: {subject_list}\")\n",
        "    return data_map, subject_list"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EkhPIbeMmDsy"
      },
      "source": [
        "### preprocessing"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "1v55ovoxRJm_"
      },
      "outputs": [],
      "source": [
        "def prepare_window_data(data, subject_list=None):\n",
        "    window_data = []\n",
        "    window_labels = []\n",
        "\n",
        "    if subject_list is None:\n",
        "        subject_list = data.keys()\n",
        "\n",
        "    for s in tqdm(subject_list):\n",
        "        for r in data[s].keys():\n",
        "            eeg = data[s][r][\"eeg\"]\n",
        "            label = data[s][r][\"label\"]\n",
        "\n",
        "            window_data.extend(eeg)\n",
        "            \n",
        "            if label is not None:\n",
        "                window_labels.extend(label)\n",
        "\n",
        "    return window_data, window_labels"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "v3_UvY6eKqUZ"
      },
      "outputs": [],
      "source": [
        "import mne"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ktgIfT6E8J8l"
      },
      "outputs": [],
      "source": [
        "def print_stats(desc, data):\n",
        "    print(f\"{desc} mean: {np.mean(data)}, std: {np.std(data)}, min: {np.min(data)}, max: {np.max(data)}\")\n",
        "\n",
        "def normalize(data, mean_value, std_value, desc=\"\"):\n",
        "    data = np.array(data)\n",
        "    data = (data - mean_value) / std_value\n",
        "    print_stats(desc, data)\n",
        "    return list(data)\n",
        "\n",
        "def filter_freq(data, f_min, f_max, FS):\n",
        "    return mne.filter.filter_data(np.array(data, dtype=np.float64), FS, f_min, f_max, method=\"iir\", verbose=False)\n",
        "\n",
        "def downsample(data, FS, FS_new):\n",
        "    return mne.filter.resample(data, down=FS/FS_new)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zr8X6O4Cm7Xh"
      },
      "source": [
        "## runner_utils"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "fPjpxd2vmyvU"
      },
      "outputs": [],
      "source": [
        "import os\n",
        "import torch\n",
        "import random\n",
        "import numpy as np\n",
        "\n",
        "def seed_everything(seed):\n",
        "    os.environ[\"PYTHONHASHSEED\"] = str(seed)\n",
        "    random.seed(seed)\n",
        "    np.random.seed(seed)\n",
        "    torch.manual_seed(seed)\n",
        "    torch.cuda.manual_seed(seed)\n",
        "    torch.backends.cudnn.deterministic = True"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "z8hVD0tG_z8l"
      },
      "source": [
        "## dataset"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "1aPOZBfP8O1M"
      },
      "outputs": [],
      "source": [
        "def add_noise(data):\n",
        "    if np.random.rand() > 0.5:\n",
        "        data += np.random.normal(0, 0.01, (data.shape))\n",
        "    return data\n",
        "\n",
        "def handle_mixup(idx, data, label, mixup_data, supervised_mixup_data, mixup_idx, mixup_rate):\n",
        "    mixup_data = mixup_data\n",
        "    # use supervised mixup data if available\n",
        "    if supervised_mixup_data is not None:\n",
        "        mixup_data = supervised_mixup_data[label]\n",
        "    # if mixup condition is set and id is not mixable\n",
        "    if mixup_idx is not None and not mixup_idx[idx]:\n",
        "        mixup_data = None\n",
        "    # do mixup if available\n",
        "    if mixup_data is not None:\n",
        "        data = mixup(data, mixup_data, mixup_rate)\n",
        "    \n",
        "    return data\n",
        "\n",
        "def mixup(data, mixup_data, mixup_rate):\n",
        "    mixup_rate = np.random.rand() * mixup_rate\n",
        "    idx = np.random.randint(0, len(mixup_data))\n",
        "    return (1-mixup_rate)*data + mixup_rate*mixup_data[idx]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "7dEidv1V_wDZ"
      },
      "outputs": [],
      "source": [
        "class SleepDataset(torch.utils.data.Dataset):\n",
        "    def __init__(self, data, labels, transform=None, mixup_data=None, supervised_mixup_data=None, mixup_idx=None, mixup_rate=None):\n",
        "        self.data = data\n",
        "        self.labels = labels\n",
        "        self.transform = transform\n",
        "        \n",
        "        self.supervised_mixup_data = supervised_mixup_data\n",
        "        self.mixup_data = mixup_data\n",
        "        self.mixup_idx = mixup_idx\n",
        "        self.mixup_rate = mixup_rate\n",
        "\n",
        "    def __len__(self):\n",
        "        return len(self.data)\n",
        "\n",
        "    def __getitem__(self, idx):\n",
        "        data = self.data[idx]\n",
        "\n",
        "        if self.labels is not None:\n",
        "            label = self.labels[idx].astype(np.int64)\n",
        "        else:\n",
        "            label = -1\n",
        "        \n",
        "        ### DATA AUGMENTATION ###\n",
        "        data = handle_mixup(idx, data, label, self.mixup_data, self.supervised_mixup_data, self.mixup_idx, self.mixup_rate)\n",
        "\n",
        "        if self.transform:\n",
        "            data = self.transform(data)\n",
        "        ### END DATA AUGMENTATION ###\n",
        "\n",
        "        data = data.astype(np.float32)\n",
        "\n",
        "        return {\"eeg\": data, \"label\": label}"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HINToD0wVG4Q"
      },
      "source": [
        "## runner"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "duip5LdjmWHV"
      },
      "source": [
        "### Focal loss"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "7g5R7c1rmYOn"
      },
      "outputs": [],
      "source": [
        "import torch.nn.functional as F\n",
        "\n",
        "class FocalLoss(torch.nn.Module):\n",
        "    \"\"\" Focal loss with optional label smoothing.\n",
        "\n",
        "    Calculates mean focal loss over given data with optional label smoothing.\n",
        "\n",
        "    Attributes\n",
        "    ----------\n",
        "        alpha: int = 1\n",
        "        gamma: int = 2\n",
        "        logits: bool = True\n",
        "        reduce: bool = True\n",
        "        ls: float = None\n",
        "            label smoothing ratio, if None label smoothing will not be applied\n",
        "        classes: int = None\n",
        "            number of classes required for label smoothing\n",
        "    \"\"\"\n",
        "\n",
        "    def __init__(\n",
        "        self,\n",
        "        alpha: int = 1,\n",
        "        gamma: int = 2,\n",
        "        logits: bool = True,\n",
        "        reduce: bool = True,\n",
        "        ls: float = 0.05,\n",
        "        classes: int = 6,\n",
        "    ):\n",
        "        super(FocalLoss, self).__init__()\n",
        "        self.alpha = alpha\n",
        "        self.gamma = gamma\n",
        "        self.logits = logits\n",
        "        self.reduce = reduce\n",
        "        self.ls = ls\n",
        "        self.classes = classes\n",
        "\n",
        "    def forward(self, inputs, targets):\n",
        "        targets = F.one_hot(targets, num_classes=6)\n",
        "\n",
        "        if self.ls is not None:\n",
        "            targets = (1 - self.ls) * targets + self.ls / self.classes\n",
        "\n",
        "        if self.logits:\n",
        "            BCE_loss = F.binary_cross_entropy_with_logits(inputs, targets, reduction=\"none\")\n",
        "        else:\n",
        "            BCE_loss = F.binary_cross_entropy(inputs, targets, reduction=\"none\")\n",
        "\n",
        "        pt = torch.exp(-BCE_loss)\n",
        "        F_loss = self.alpha * (1 - pt) ** self.gamma * BCE_loss\n",
        "\n",
        "        if self.reduce:\n",
        "            return torch.mean(F_loss)\n",
        "        else:\n",
        "            return F_loss"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "43a4q46BYy8d"
      },
      "source": [
        "### VAT"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "VU_cmbORYz65"
      },
      "outputs": [],
      "source": [
        "# implementation from: https://github.com/lyakaap/VAT-pytorch\r\n",
        "\r\n",
        "import contextlib\r\n",
        "import torch\r\n",
        "import torch.nn as nn\r\n",
        "import torch.nn.functional as F\r\n",
        "\r\n",
        "\r\n",
        "@contextlib.contextmanager\r\n",
        "def _disable_tracking_bn_stats(model):\r\n",
        "\r\n",
        "    def switch_attr(m):\r\n",
        "        if hasattr(m, 'track_running_stats'):\r\n",
        "            m.track_running_stats ^= True\r\n",
        "            \r\n",
        "    model.apply(switch_attr)\r\n",
        "    yield\r\n",
        "    model.apply(switch_attr)\r\n",
        "\r\n",
        "\r\n",
        "def _l2_normalize(d):\r\n",
        "    d_reshaped = d.view(d.shape[0], -1, *(1 for _ in range(d.dim() - 2)))\r\n",
        "    d /= torch.norm(d_reshaped, dim=1, keepdim=True) + 1e-8\r\n",
        "    return d\r\n",
        "\r\n",
        "\r\n",
        "class VATLoss(nn.Module):\r\n",
        "\r\n",
        "    def __init__(self, xi=10.0, eps=1.0, ip=1):\r\n",
        "        \"\"\"VAT loss\r\n",
        "        :param xi: hyperparameter of VAT (default: 10.0)\r\n",
        "        :param eps: hyperparameter of VAT (default: 1.0)\r\n",
        "        :param ip: iteration times of computing adv noise (default: 1)\r\n",
        "        \"\"\"\r\n",
        "        super(VATLoss, self).__init__()\r\n",
        "        self.xi = xi\r\n",
        "        self.eps = eps\r\n",
        "        self.ip = ip\r\n",
        "\r\n",
        "    def forward(self, model, x):\r\n",
        "        with torch.no_grad():\r\n",
        "            pred = F.softmax(model(x), dim=1)\r\n",
        "\r\n",
        "        # prepare random unit tensor\r\n",
        "        d = torch.rand(x.shape).sub(0.5).to(x.device)\r\n",
        "        d = _l2_normalize(d)\r\n",
        "\r\n",
        "        with _disable_tracking_bn_stats(model):\r\n",
        "            # calc adversarial direction\r\n",
        "            for _ in range(self.ip):\r\n",
        "                d.requires_grad_()\r\n",
        "                pred_hat = model(x + self.xi * d)\r\n",
        "                logp_hat = F.log_softmax(pred_hat, dim=1)\r\n",
        "                adv_distance = F.kl_div(logp_hat, pred, reduction='batchmean')\r\n",
        "                adv_distance.backward()\r\n",
        "                d = _l2_normalize(d.grad)\r\n",
        "                model.zero_grad()\r\n",
        "    \r\n",
        "            # calc LDS\r\n",
        "            r_adv = d * self.eps\r\n",
        "            pred_hat = model(x + r_adv)\r\n",
        "            logp_hat = F.log_softmax(pred_hat, dim=1)\r\n",
        "            lds = F.kl_div(logp_hat, pred, reduction='batchmean')\r\n",
        "\r\n",
        "        return lds"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4iFeq1WGmNWO"
      },
      "source": [
        "### train, validate, predict"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "o3_fBKRWVJbq"
      },
      "outputs": [],
      "source": [
        "import torch\n",
        "import numpy as np\n",
        "\n",
        "def train_epoch(args, model, loader, criterion, optimizer, scheduler, epoch):\n",
        "    losses = []\n",
        "    targets_all = []\n",
        "    outputs_all = []\n",
        "\n",
        "    vat_loss = VATLoss(xi=10.0, eps=1.0, ip=1)\n",
        "\n",
        "    model.train()\n",
        "    t = tqdm(loader)\n",
        "\n",
        "    for i, sample in enumerate(t):\n",
        "        optimizer.zero_grad()\n",
        "\n",
        "        eeg = sample[\"eeg\"].to(args.device)\n",
        "        target = sample[\"label\"].to(args.device)\n",
        "\n",
        "        lds = vat_loss(model, eeg)\n",
        "        output = model(eeg)\n",
        "        loss = criterion(output, target) + args.alpha * lds\n",
        "        loss.backward()\n",
        "        optimizer.step()\n",
        "\n",
        "        if scheduler is not None:\n",
        "            scheduler.step()\n",
        "\n",
        "        losses.append(loss.item())\n",
        "        \n",
        "        target = target.cpu().numpy()\n",
        "        output = output.detach().cpu().numpy()\n",
        "\n",
        "        targets_all.extend(target)\n",
        "        outputs_all.extend(output)\n",
        "\n",
        "        output_loss = np.mean(losses)\n",
        "        output_score = np.mean(targets_all == np.argmax(outputs_all, axis=1))\n",
        "\n",
        "        t.set_description(\n",
        "            f\"Epoch {epoch}/{args.epochs} - Train loss: {output_loss:0.4f}, score: {output_score:0.4f}\"\n",
        "        )\n",
        "\n",
        "    return targets_all, outputs_all, output_score, output_loss\n",
        "\n",
        "\n",
        "def validate(args, model, loader, criterion, desc=\"Valid\"):\n",
        "    losses = []\n",
        "    targets_all = []\n",
        "    outputs_all = []\n",
        "\n",
        "    t = tqdm(loader)\n",
        "    model.eval()\n",
        "    with torch.no_grad():\n",
        "        for i, sample in enumerate(t):\n",
        "            eeg = sample[\"eeg\"].to(args.device)\n",
        "            target = sample[\"label\"].to(args.device)\n",
        "\n",
        "            output = model(eeg)\n",
        "            loss = criterion(output, target)\n",
        "\n",
        "            losses.append(loss.item())\n",
        "            targets_all.extend(target.cpu().numpy())\n",
        "            outputs_all.extend(output.detach().cpu().numpy())\n",
        "\n",
        "            output_loss = np.mean(losses)\n",
        "            output_score = np.mean(targets_all == np.argmax(outputs_all, axis=1))\n",
        "\n",
        "            t.set_description(\n",
        "                f\"\\t  - {desc} loss: {output_loss:0.4f}, score: {output_score:0.4f}\"\n",
        "            )\n",
        "\n",
        "    return targets_all, outputs_all, output_score, output_loss\n",
        "\n",
        "\n",
        "def predict(args, model, loader):\n",
        "    outputs_all = []\n",
        "\n",
        "    t = tqdm(loader)\n",
        "    model.eval()\n",
        "    with torch.no_grad():\n",
        "        for i, sample in enumerate(t):\n",
        "            eeg = sample[\"eeg\"].to(args.device)\n",
        "\n",
        "            output = model(eeg)\n",
        "            outputs_all.extend(output.detach().cpu().numpy())\n",
        "    return outputs_all"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gMPUxGe0X4L_"
      },
      "source": [
        "## model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "SjtnkFJrhTp8"
      },
      "outputs": [],
      "source": [
        "import torch\n",
        "import torch.nn as nn"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "__H9q1GtX5Fe"
      },
      "outputs": [],
      "source": [
        "def get_linear(in_features, out_features):\n",
        "    return nn.utils.weight_norm(nn.Linear(in_features, out_features))\n",
        "\n",
        "class BaseEEGClassifier(nn.Module):\n",
        "    def __init__(self):\n",
        "        super(BaseEEGClassifier, self).__init__()\n",
        "\n",
        "        N_channels = [64, 128, 256]\n",
        "        \n",
        "        self.c1 = nn.Sequential(nn.Conv1d(2, N_channels[0], 401, 50, 200),\n",
        "                                nn.BatchNorm1d(N_channels[0]),\n",
        "                                nn.SiLU(),\n",
        "                                nn.Dropout(0.2),\n",
        "                                nn.MaxPool1d(3,2,1),\n",
        "                                nn.Conv1d(N_channels[0], N_channels[1], 7, 2, 3),\n",
        "                                nn.BatchNorm1d(N_channels[1]),\n",
        "                                nn.SiLU(),\n",
        "                                nn.Dropout(0.2),\n",
        "                                nn.Conv1d(N_channels[1], N_channels[2], 5, 1, 2),\n",
        "                                nn.BatchNorm1d(N_channels[2]),\n",
        "                                nn.AdaptiveMaxPool1d(1),\n",
        "                                )\n",
        "        \n",
        "        self.c2 = nn.Sequential(nn.Conv1d(2, N_channels[0], 51, 5, 25),\n",
        "                                nn.BatchNorm1d(N_channels[0]),\n",
        "                                nn.SiLU(),\n",
        "                                nn.Dropout(0.2),\n",
        "                                nn.MaxPool1d(9,3,4),\n",
        "                                nn.Conv1d(N_channels[0], N_channels[1], 7, 2, 3),\n",
        "                                nn.BatchNorm1d(N_channels[1]),\n",
        "                                nn.SiLU(),\n",
        "                                nn.Conv1d(N_channels[1], N_channels[2], 5, 1, 2),\n",
        "                                nn.BatchNorm1d(N_channels[2]),\n",
        "                                nn.AdaptiveMaxPool1d(1),\n",
        "                                )\n",
        "        \n",
        "        self.linear = get_linear(N_channels[2]*2, 1024)\n",
        "        self.act = nn.SiLU()\n",
        "        self.dropout = nn.Dropout(0.2)\n",
        "        \n",
        "    def forward(self, x):\n",
        "        x1 = self.c1(x)\n",
        "        x2 = self.c2(x)\n",
        "\n",
        "        x = torch.cat([x1, x2], dim=1)\n",
        "        x = x.flatten(1,-1)\n",
        "        x = self.act(x)\n",
        "        x = self.dropout(x)\n",
        "        x = self.linear(x)\n",
        "        x = self.act(x)\n",
        "        return x\n",
        "\n",
        "class EEGClassifier(nn.Module):\n",
        "    def __init__(self):\n",
        "        super(EEGClassifier, self).__init__()\n",
        "\n",
        "        self.base = BaseEEGClassifier()\n",
        "        self.fc = get_linear(1024, 6)\n",
        "\n",
        "    def features(self, x):\n",
        "        x = self.base(x)\n",
        "        return x\n",
        "\n",
        "    def forward(self, x):\n",
        "        x = self.features(x)\n",
        "        x = self.fc(x)\n",
        "        return x"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tgpJaPvEXQWE"
      },
      "source": [
        "## main"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "zTMVQpxbYi7o"
      },
      "outputs": [],
      "source": [
        "import seaborn as sns\n",
        "from torch.utils.data import DataLoader, WeightedRandomSampler\n",
        "from sklearn.metrics import confusion_matrix"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6vqs7U7vqiyd"
      },
      "source": [
        "### args"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "wI4A8-pCYsUW"
      },
      "outputs": [],
      "source": [
        "import json\n",
        "\n",
        "class base_args(object):\n",
        "    def __init__(self, model_name, seed=42):\n",
        "        self.model_name = model_name\n",
        "        self.seed = seed\n",
        "\n",
        "        self.lr = 1e-3\n",
        "        self.epochs = 5\n",
        "        self.batch_size = 120\n",
        "        self.num_workers = 2\n",
        "        self.device = ('cuda' if torch.cuda.is_available() else 'cpu')\n",
        "        self.alpha = 0.01\n",
        "        self.phase = \"base\"\n",
        "\n",
        "    def toJSON(self):\n",
        "        return json.dumps(self, default=lambda o: o.__dict__, sort_keys=True, indent=4)\n",
        "\n",
        "    def get_model_name(self):\n",
        "        return f\"{self.model_name}_{self.phase}\""
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "u8MwnK-aAM5r"
      },
      "source": [
        "### experiment helper functions"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "nOcEgm77xoR1"
      },
      "outputs": [],
      "source": [
        "def train_model(args, model, \n",
        "          train_data, train_labels, \n",
        "          valid_data, valid_labels, \n",
        "          train_weights=None, sample_rate=None, \n",
        "          use_scheduler=True,\n",
        "          history = {\"Train\": {\"Score\": [], \"Loss\": []}, \n",
        "                     \"Valid\": {\"Score\": [], \"Loss\": []}},\n",
        "          mixup_data=None, supervised_mixup_data=None, mixup_idx=None, mixup_rate=None):\n",
        "    \n",
        "    train_dataset = SleepDataset(train_data, train_labels, transform=add_noise, \n",
        "                                 mixup_data=mixup_data, supervised_mixup_data=supervised_mixup_data, \n",
        "                                 mixup_idx=mixup_idx, mixup_rate=mixup_rate)\n",
        "    \n",
        "    if train_weights is not None:\n",
        "        train_sampler = WeightedRandomSampler(weights = train_weights, num_samples=int(len(train_labels)*sample_rate))\n",
        "        train_loader = DataLoader(train_dataset, num_workers=args.num_workers, batch_size=args.batch_size, sampler=train_sampler, drop_last=True)\n",
        "    else:\n",
        "        train_loader = DataLoader(train_dataset, num_workers=args.num_workers, batch_size=args.batch_size, shuffle=True, drop_last=True)\n",
        "\n",
        "    model = model.to(args.device)\n",
        "    criterion = FocalLoss()\n",
        "    optimizer = torch.optim.AdamW(model.parameters(), lr=args.lr, weight_decay=5e-4)\n",
        "\n",
        "    if use_scheduler:\n",
        "        scheduler = torch.optim.lr_scheduler.OneCycleLR(\n",
        "                        optimizer,\n",
        "                        max_lr=args.lr,\n",
        "                        epochs=args.epochs,\n",
        "                        steps_per_epoch=len(train_loader),\n",
        "                        div_factor=10, \n",
        "                        final_div_factor=10,\n",
        "                        pct_start=0.1,\n",
        "                        anneal_strategy=\"cos\",\n",
        "                    )\n",
        "    else:\n",
        "        scheduler = None\n",
        "    \n",
        "    for epoch in range(1, args.epochs+1):\n",
        "        _, _, train_score, train_loss = train_epoch(args, model, train_loader, criterion, optimizer, scheduler, epoch)\n",
        "        _, _, valid_score, valid_loss = validate_model(args, model, valid_data, valid_labels, show_plot=False)\n",
        "\n",
        "        history[\"Train\"][\"Loss\"].append(train_loss)\n",
        "        history[\"Train\"][\"Score\"].append(train_score)\n",
        "        history[\"Valid\"][\"Loss\"].append(valid_loss)\n",
        "        history[\"Valid\"][\"Score\"].append(valid_score)\n",
        "\n",
        "    torch.save(model.state_dict(), f\"{OUTPUT_FOLDER}best_model_{args.get_model_name()}.pt\")\n",
        "\n",
        "\n",
        "def validate_model(args, model, valid_data, valid_labels, desc=\"Target\", show_plot=True):\n",
        "    criterion = FocalLoss()\n",
        "\n",
        "    valid_dataset = SleepDataset(valid_data, valid_labels)\n",
        "    valid_loader = DataLoader(valid_dataset, num_workers=args.num_workers, batch_size=args.batch_size, shuffle=False)\n",
        "    targets_all, outputs_all, output_score, output_loss = validate(args, model, valid_loader, criterion, desc)\n",
        "    \n",
        "    if show_plot:\n",
        "        cf_mat = confusion_matrix(targets_all, np.argmax(outputs_all, axis=1), normalize=\"true\")\n",
        "        plt.figure()\n",
        "        sns.heatmap(cf_mat, annot=True)\n",
        "        plt.show()\n",
        "\n",
        "    return targets_all, outputs_all, output_score, output_loss\n",
        "\n",
        "\n",
        "def get_prediction(args, model, data):\n",
        "    model.to(args.device)\n",
        "    model.load_state_dict(torch.load(f\"{OUTPUT_FOLDER}best_model_{args.get_model_name()}.pt\"))\n",
        "    \n",
        "    dataset = SleepDataset(data, None)\n",
        "    loader = DataLoader(dataset, num_workers=args.num_workers, batch_size=args.batch_size, shuffle=False)\n",
        "\n",
        "    output = predict(args, model, loader)\n",
        "    return output\n",
        "\n",
        "\n",
        "def ensemble(args_array, dataset_name=\"test\", phase=None):\n",
        "    ensemble_output = None\n",
        "\n",
        "    for args in args_array:\n",
        "        if phase is None:\n",
        "            fn = f\"{OUTPUT_FOLDER}{args.get_model_name()}-{dataset_name}_output.npy\"\n",
        "        else:\n",
        "            fn = f\"{OUTPUT_FOLDER}{args.model_name}_{phase}-{dataset_name}_output.npy\"\n",
        "            \n",
        "        output = np.load(fn)\n",
        "        if ensemble_output is None:\n",
        "            ensemble_output = output\n",
        "        else:\n",
        "            ensemble_output += output\n",
        "    \n",
        "    return ensemble_output\n",
        "    \n",
        "\n",
        "def predict_and_save(args, model, target_data, test_data):\n",
        "    target_output = get_prediction(args, model, target_data)\n",
        "    np.save(f\"{OUTPUT_FOLDER}{args.get_model_name()}-target_output.npy\", target_output)\n",
        "    \n",
        "    test_output = get_prediction(args, model, test_data)\n",
        "    np.save(f\"{OUTPUT_FOLDER}{args.get_model_name()}-test_output.npy\", test_output)\n",
        "\n",
        "\n",
        "def plot_history(history):\n",
        "    fig, axes = plt.subplots(2,1, figsize=(22,6))\n",
        "    axes[0].plot(history[\"Train\"][\"Score\"], label=\"Train score\")\n",
        "    axes[0].plot(history[\"Valid\"][\"Score\"], label=\"Valid score\")\n",
        "    axes[0].legend()\n",
        "    axes[1].plot(history[\"Train\"][\"Loss\"], label=\"Train loss\")\n",
        "    axes[1].plot(history[\"Valid\"][\"Loss\"], label=\"Valid loss\")\n",
        "    axes[1].legend()\n",
        "    fig.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bzo9LCObFmVa"
      },
      "source": [
        "### competition data loaders"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "XqHbajaLFr78"
      },
      "outputs": [],
      "source": [
        "def get_phase_1_data():\n",
        "    source_data, source_subjects = load_data(SOURCE_DATA_FOLDER)\n",
        "    source_data, source_labels = prepare_window_data(source_data, source_subjects)\n",
        "    # calculate stats of source data and normalize it\n",
        "    source_data = np.array(source_data)\n",
        "    source_mean = np.mean(source_data)\n",
        "    source_std = np.std(source_data)\n",
        "    source_data = (source_data - source_mean) / source_std\n",
        "    print(f\"Source mean: {source_mean}, std: {source_std}, min: {np.min(source_data)}, max: {np.max(source_data)}\")\n",
        "    source_data  = list(source_data)\n",
        "    # load and normalize target data\n",
        "    lb_target_data, lb_target_subjects = load_data(LEADERBOARD_TARGET_DATA_FOLDER)\n",
        "    lb_target_data, lb_target_labels = prepare_window_data(lb_target_data, lb_target_subjects)\n",
        "    lb_target_data = normalize(lb_target_data, source_mean, source_std, \"Leadeboard target\")\n",
        "    # load and normalize test data\n",
        "    lb_test_data, lb_test_subjects = load_data(LEADERBOARD_TEST_DATA_FOLDER, load_labels=False)\n",
        "    lb_test_data, lb_test_labels = prepare_window_data(lb_test_data, lb_test_subjects)\n",
        "    lb_test_data = normalize(lb_test_data, source_mean, source_std, \"Leadeboard test\")\n",
        "\n",
        "    return source_data, source_labels, lb_target_data, lb_target_labels, lb_test_data, lb_test_data\n",
        "\n",
        "\n",
        "def get_phase_2_data():\n",
        "    source_data, source_subjects = load_data(SOURCE_DATA_FOLDER)\n",
        "    source_data, source_labels = prepare_window_data(source_data, source_subjects)\n",
        "    # add target data from phase 1 to source data\n",
        "    lb_target_data, lb_target_subjects = load_data(LEADERBOARD_TARGET_DATA_FOLDER)\n",
        "    lb_target_data, lb_target_labels = prepare_window_data(lb_target_data, lb_target_subjects)\n",
        "    # source_data += lb_target_data\n",
        "    # source_labels += lb_target_labels\n",
        "    # calculate stats and normalize\n",
        "    source_data = np.array(source_data)\n",
        "    source_mean = np.mean(source_data)\n",
        "    source_std = np.std(source_data)\n",
        "    source_data = (source_data - source_mean) / source_std\n",
        "    print(f\"Source mean: {source_mean}, std: {source_std}, min: {np.min(source_data)}, max: {np.max(source_data)}\")\n",
        "    source_data  = list(source_data)\n",
        "    # load test data from phase 1 and normalize it\n",
        "    lb_test_data, lb_test_subjects = load_data(LEADERBOARD_TEST_DATA_FOLDER, load_labels=False)\n",
        "    lb_test_data, lb_test_labels = prepare_window_data(lb_test_data, lb_test_subjects)\n",
        "    lb_test_data = normalize(lb_test_data, source_mean, source_std, \"Leadeboard test\")\n",
        "    # load and normalize target data\n",
        "    fn_target_data, fn_target_subjects = load_data(FINAL_TARGET_DATA_FOLDER)\n",
        "    fn_target_data, fn_target_labels = prepare_window_data(fn_target_data, fn_target_subjects)\n",
        "    fn_target_data += lb_target_data\n",
        "    fn_target_labels += lb_target_labels\n",
        "    fn_target_data = normalize(fn_target_data, source_mean, source_std, \"Target\")\n",
        "    # load and normalize test data\n",
        "    fn_test_data, fn_test_subjects = load_data(FINAL_TEST_DATA_FOLDER, load_labels=False)\n",
        "    fn_test_data, fn_test_labels = prepare_window_data(fn_test_data, fn_test_subjects)\n",
        "    fn_test_data = normalize(fn_test_data, source_mean, source_std, \"Final test\")\n",
        "\n",
        "    return source_data, source_labels, fn_target_data, fn_target_labels, fn_test_data, fn_test_data"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JMXg2fEUFsk3"
      },
      "source": [
        "### training pipeline function"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "s1TkNDtYSVI6"
      },
      "outputs": [],
      "source": [
        "def supervised_run(args, model, train_data, train_labels, target_data, target_labels, test_data, mixup_data=None, supervised_mixup_data=None):\n",
        "    print(args.toJSON())\n",
        "\n",
        "    history = {\"Train\": {\"Score\": [], \"Loss\": []}, \n",
        "               \"Valid\": {\"Score\": [], \"Loss\": []}}\n",
        "\n",
        "    ############################\n",
        "    ### USE ONLY SOURCE DATA ###\n",
        "    ############################\n",
        "    args.phase = \"base\"\n",
        "\n",
        "    train_model(args, model, \n",
        "                train_data, train_labels, \n",
        "                target_data, target_labels, \n",
        "                mixup_data=mixup_data, mixup_rate=0.,\n",
        "                history=history)\n",
        "    # validate\n",
        "    model.load_state_dict(torch.load(f\"{OUTPUT_FOLDER}best_model_{args.get_model_name()}.pt\"))\n",
        "    print(\"\\n###### PHASE 1 FINISHED ##########\")\n",
        "    validate_model(args, model, target_data, target_labels, \"Target\")\n",
        "    # make predictions and save results\n",
        "    predict_and_save(args, model, target_data, test_data)\n",
        "\n",
        "\n",
        "    ##############################\n",
        "    ### SOURCE DATA WITH SUPERVISED MIXUP ###\n",
        "    ##############################\n",
        "    args.lr = 1e-4\n",
        "    args.epochs = 10\n",
        "    args.phase = \"mixup\"\n",
        "\n",
        "    train_model(args, model, \n",
        "                train_data, train_labels, \n",
        "                target_data, target_labels, \n",
        "                supervised_mixup_data=supervised_mixup_data, mixup_rate=0.6,\n",
        "                history=history)\n",
        "    # validate\n",
        "    model.load_state_dict(torch.load(f\"{OUTPUT_FOLDER}best_model_{args.get_model_name()}.pt\"))\n",
        "    print(\"\\n###### PHASE 2 FINISHED ##########\")\n",
        "    validate_model(args, model, target_data, target_labels, \"Target\")\n",
        "    # make predictions and save results\n",
        "    predict_and_save(args, model, target_data, test_data)\n",
        "\n",
        "    ###############################\n",
        "    ### FINETUNE ON TARGET DATA WITH MIXED UP SOURCE ###\n",
        "    ###############################\n",
        "    args.lr = 1e-3\n",
        "    args.epochs = 20\n",
        "    args.phase = \"mixup_finetuned\"\n",
        "    # prepare extended train data with sampler settings\n",
        "    extended_train_data = train_data + target_data\n",
        "    extended_train_labels = train_labels + target_labels\n",
        "    train_weights = [0.25] * len(train_labels) + [0.75] * len(target_labels)\n",
        "    mixup_idx = [True] * len(train_labels) + [False] * len(target_labels)\n",
        "    train_sample_rate = 0.5\n",
        "\n",
        "    train_model(args, model, \n",
        "                extended_train_data, extended_train_labels,\n",
        "                target_data, target_labels,\n",
        "                train_weights = train_weights, sample_rate=train_sample_rate,\n",
        "                mixup_data=mixup_data, mixup_idx=mixup_idx, mixup_rate=0.4,\n",
        "                use_scheduler=True, history=history)\n",
        "    \n",
        "    # validate\n",
        "    model.load_state_dict(torch.load(f\"{OUTPUT_FOLDER}best_model_{args.get_model_name()}.pt\"))\n",
        "    print(\"\\n###### PHASE 3 FINISHED ##########\")\n",
        "    validate_model(args, model, target_data, target_labels, \"Target\")\n",
        "    # make predictions and save results\n",
        "    predict_and_save(args, model, target_data, test_data)\n",
        "    # plot training history\n",
        "    plot_history(history)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FEjmRhhGF4yL"
      },
      "source": [
        "### training"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KydYNDzJJF6r",
        "outputId": "8cd2f1ab-53d0-40ed-d683-a54f707cc03d"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Loading data from folder: /home/data/SleepSource/ (157 files) - Load labels True\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "\r  0%|          | 0/157 [00:00<?, ?it/s]"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "<Info | 8 non-empty values\n",
            " bads: []\n",
            " ch_names: Fpz-Cz, Pz-Oz\n",
            " chs: 2 EEG\n",
            " custom_ref_applied: False\n",
            " highpass: 0.5 Hz\n",
            " lowpass: 100.0 Hz\n",
            " meas_date: 1991-09-26 15:00:00 UTC\n",
            " nchan: 2\n",
            " projs: []\n",
            " sfreq: 100.0 Hz\n",
            " subject_info: 2 items (dict)\n",
            ">\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 157/157 [01:52<00:00,  1.39it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Loaded total 90545 samples for subjects: [ 0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23\n",
            " 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38]\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 39/39 [00:00<00:00, 1583.43it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Loading data from folder: /home/data/LeaderboardSleep/sleep_target/ (25 files) - Load labels True\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "\r  0%|          | 0/25 [00:00<?, ?it/s]"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "<Info | 8 non-empty values\n",
            " bads: []\n",
            " ch_names: Fpz-Cz, Pz-Oz\n",
            " chs: 2 EEG\n",
            " custom_ref_applied: False\n",
            " highpass: 0.5 Hz\n",
            " lowpass: 100.0 Hz\n",
            " meas_date: 1990-03-13 15:09:00 UTC\n",
            " nchan: 2\n",
            " projs: []\n",
            " sfreq: 100.0 Hz\n",
            " subject_info: 2 items (dict)\n",
            ">\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 25/25 [00:19<00:00,  1.29it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Loaded total 15442 samples for subjects: [0 1 2 3 4 5]\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 6/6 [00:00<00:00, 1502.44it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Source mean: -2.175762164676705e-07, std: 1.6836217270029333e-05, min: -14.7172265485441, max: 12.426637935405019\n",
            "Loading data from folder: /home/data/LeaderboardSleep/testing/ (24 files) - Load labels False\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 24/24 [00:32<00:00,  1.35s/it]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Loaded total 25748 samples for subjects: [ 6  7  8  9 10 11 12 13 14 15 16 17]\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 12/12 [00:00<00:00, 2770.50it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Leadeboard test mean: 0.014119353426048422, std: 0.8645048345254686, min: -11.985021370728651, max: 11.832680288053835\n",
            "Loading data from folder: /home/data/finalSleep/sleep_target/ (20 files) - Load labels True\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 20/20 [00:20<00:00,  1.03s/it]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Loaded total 16568 samples for subjects: [0 1 2 3 4]\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 5/5 [00:00<00:00, 997.46it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Target mean: 0.031174427319106832, std: 0.9668677818534862, min: -11.866229841258413, max: 12.010867582259191\n",
            "Loading data from folder: /home/data/finalSleep/testing/ (18 files) - Load labels False\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 18/18 [00:32<00:00,  1.79s/it]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Loaded total 25756 samples for subjects: [ 5  6  7  8  9 10 11 12 13]\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 9/9 [00:00<00:00, 2231.28it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Final test mean: 0.04296289931340894, std: 0.8598280113936251, min: -12.282000194404242, max: 11.713888758583598\n"
          ]
        }
      ],
      "source": [
        "source_data, source_labels, target_data, target_labels, test_data, mixup_data = get_phase_2_data()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "MFMIOOcXi8s_",
        "outputId": "9e377104-6232-46f7-81d5-ac067ede64e2"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "0 (13819, 2, 3000)\n",
            "1 (3311, 2, 3000)\n",
            "2 (9739, 2, 3000)\n",
            "3 (1393, 2, 3000)\n",
            "4 (559, 2, 3000)\n",
            "5 (3189, 2, 3000)\n"
          ]
        }
      ],
      "source": [
        "tmp = np.array(target_data)\n",
        "supervised_mixup_data = {}\n",
        "for c in np.unique(target_labels):\n",
        "    supervised_mixup_data[c] = tmp[target_labels == c]\n",
        "    print(c, np.shape(supervised_mixup_data[c]))\n",
        "\n",
        "del tmp"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "oY5ZcUnXF6W_",
        "outputId": "43c6399f-f198-4520-a92c-2eb57bd6d531"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "{\n",
            "    \"alpha\": 0.01,\n",
            "    \"batch_size\": 120,\n",
            "    \"device\": \"cuda\",\n",
            "    \"epochs\": 5,\n",
            "    \"lr\": 0.001,\n",
            "    \"model_name\": \"eeg-classifier_seed-42\",\n",
            "    \"num_workers\": 2,\n",
            "    \"phase\": \"base\",\n",
            "    \"seed\": 42\n",
            "}\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "\r  0%|          | 0/754 [00:00<?, ?it/s]/usr/local/lib/python3.7/dist-packages/torch/nn/functional.py:652: UserWarning: Named tensors and all their associated APIs are an experimental feature and subject to change. Please do not use them for anything important until they are released as stable. (Triggered internally at  /pytorch/c10/core/TensorImpl.h:1156.)\n",
            "  return torch.max_pool1d(input, kernel_size, stride, padding, dilation, ceil_mode)\n",
            "Epoch 1/5 - Train loss: 0.0521, score: 0.7450: 100%|██████████| 754/754 [02:37<00:00,  4.78it/s]\n",
            "\t  - Target loss: 0.0413, score: 0.7987: 100%|██████████| 267/267 [00:09<00:00, 29.02it/s]\n",
            "Epoch 2/5 - Train loss: 0.0401, score: 0.8061: 100%|██████████| 754/754 [02:39<00:00,  4.74it/s]\n",
            "\t  - Target loss: 0.0416, score: 0.7923: 100%|██████████| 267/267 [00:09<00:00, 29.09it/s]\n",
            "Epoch 3/5 - Train loss: 0.0379, score: 0.8176: 100%|██████████| 754/754 [02:39<00:00,  4.71it/s]\n",
            "\t  - Target loss: 0.0422, score: 0.7889: 100%|██████████| 267/267 [00:09<00:00, 27.94it/s]\n",
            "Epoch 4/5 - Train loss: 0.0359, score: 0.8286: 100%|██████████| 754/754 [02:40<00:00,  4.70it/s]\n",
            "\t  - Target loss: 0.0407, score: 0.7973: 100%|██████████| 267/267 [00:09<00:00, 28.68it/s]\n",
            "Epoch 5/5 - Train loss: 0.0343, score: 0.8376: 100%|██████████| 754/754 [02:39<00:00,  4.72it/s]\n",
            "\t  - Target loss: 0.0408, score: 0.7994: 100%|██████████| 267/267 [00:09<00:00, 28.91it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\n",
            "###### PHASE 1 FINISHED ##########\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "\t  - Target loss: 0.0408, score: 0.7994: 100%|██████████| 267/267 [00:09<00:00, 27.17it/s]\n"
          ]
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 267/267 [00:05<00:00, 48.72it/s]\n",
            "100%|██████████| 215/215 [00:04<00:00, 47.35it/s]\n",
            "Epoch 1/10 - Train loss: 0.0322, score: 0.8506: 100%|██████████| 754/754 [02:38<00:00,  4.74it/s]\n",
            "\t  - Target loss: 0.0384, score: 0.8102: 100%|██████████| 267/267 [00:08<00:00, 29.87it/s]\n",
            "Epoch 2/10 - Train loss: 0.0317, score: 0.8536: 100%|██████████| 754/754 [02:38<00:00,  4.74it/s]\n",
            "\t  - Target loss: 0.0387, score: 0.8093: 100%|██████████| 267/267 [00:09<00:00, 29.37it/s]\n",
            "Epoch 3/10 - Train loss: 0.0314, score: 0.8568: 100%|██████████| 754/754 [02:39<00:00,  4.72it/s]\n",
            "\t  - Target loss: 0.0395, score: 0.8103: 100%|██████████| 267/267 [00:09<00:00, 28.91it/s]\n",
            "Epoch 4/10 - Train loss: 0.0310, score: 0.8561: 100%|██████████| 754/754 [02:40<00:00,  4.70it/s]\n",
            "\t  - Target loss: 0.0392, score: 0.8087: 100%|██████████| 267/267 [00:09<00:00, 27.41it/s]\n",
            "Epoch 5/10 - Train loss: 0.0305, score: 0.8608: 100%|██████████| 754/754 [02:40<00:00,  4.71it/s]\n",
            "\t  - Target loss: 0.0374, score: 0.8134: 100%|██████████| 267/267 [00:09<00:00, 27.59it/s]\n",
            "Epoch 6/10 - Train loss: 0.0301, score: 0.8617: 100%|██████████| 754/754 [02:40<00:00,  4.70it/s]\n",
            "\t  - Target loss: 0.0376, score: 0.8129: 100%|██████████| 267/267 [00:09<00:00, 28.08it/s]\n",
            "Epoch 7/10 - Train loss: 0.0299, score: 0.8619: 100%|██████████| 754/754 [02:41<00:00,  4.66it/s]\n",
            "\t  - Target loss: 0.0377, score: 0.8155: 100%|██████████| 267/267 [00:09<00:00, 27.39it/s]\n",
            "Epoch 8/10 - Train loss: 0.0298, score: 0.8634: 100%|██████████| 754/754 [02:41<00:00,  4.67it/s]\n",
            "\t  - Target loss: 0.0372, score: 0.8161: 100%|██████████| 267/267 [00:09<00:00, 27.59it/s]\n",
            "Epoch 9/10 - Train loss: 0.0296, score: 0.8644: 100%|██████████| 754/754 [02:42<00:00,  4.65it/s]\n",
            "\t  - Target loss: 0.0390, score: 0.8060: 100%|██████████| 267/267 [00:10<00:00, 26.63it/s]\n",
            "Epoch 10/10 - Train loss: 0.0298, score: 0.8637: 100%|██████████| 754/754 [02:43<00:00,  4.61it/s]\n",
            "\t  - Target loss: 0.0370, score: 0.8170: 100%|██████████| 267/267 [00:10<00:00, 25.92it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\n",
            "###### PHASE 2 FINISHED ##########\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "\t  - Target loss: 0.0370, score: 0.8170: 100%|██████████| 267/267 [00:10<00:00, 24.37it/s]\n"
          ]
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 267/267 [00:05<00:00, 47.40it/s]\n",
            "100%|██████████| 215/215 [00:04<00:00, 46.95it/s]\n",
            "Epoch 1/20 - Train loss: 0.0398, score: 0.8065: 100%|██████████| 510/510 [01:46<00:00,  4.81it/s]\n",
            "\t  - Target loss: 0.0351, score: 0.8268: 100%|██████████| 267/267 [00:09<00:00, 27.35it/s]\n",
            "Epoch 2/20 - Train loss: 0.0403, score: 0.8038: 100%|██████████| 510/510 [01:45<00:00,  4.86it/s]\n",
            "\t  - Target loss: 0.0344, score: 0.8341: 100%|██████████| 267/267 [00:09<00:00, 26.79it/s]\n",
            "Epoch 3/20 - Train loss: 0.0397, score: 0.8042: 100%|██████████| 510/510 [01:45<00:00,  4.85it/s]\n",
            "\t  - Target loss: 0.0337, score: 0.8347: 100%|██████████| 267/267 [00:09<00:00, 27.58it/s]\n",
            "Epoch 4/20 - Train loss: 0.0388, score: 0.8110: 100%|██████████| 510/510 [01:45<00:00,  4.82it/s]\n",
            "\t  - Target loss: 0.0317, score: 0.8457: 100%|██████████| 267/267 [00:09<00:00, 27.55it/s]\n",
            "Epoch 5/20 - Train loss: 0.0384, score: 0.8133: 100%|██████████| 510/510 [01:45<00:00,  4.84it/s]\n",
            "\t  - Target loss: 0.0322, score: 0.8460: 100%|██████████| 267/267 [00:09<00:00, 27.50it/s]\n",
            "Epoch 6/20 - Train loss: 0.0376, score: 0.8170: 100%|██████████| 510/510 [01:45<00:00,  4.83it/s]\n",
            "\t  - Target loss: 0.0315, score: 0.8502: 100%|██████████| 267/267 [00:09<00:00, 27.37it/s]\n",
            "Epoch 7/20 - Train loss: 0.0372, score: 0.8206: 100%|██████████| 510/510 [01:45<00:00,  4.84it/s]\n",
            "\t  - Target loss: 0.0311, score: 0.8471: 100%|██████████| 267/267 [00:09<00:00, 28.18it/s]\n",
            "Epoch 8/20 - Train loss: 0.0365, score: 0.8249: 100%|██████████| 510/510 [01:45<00:00,  4.84it/s]\n",
            "\t  - Target loss: 0.0300, score: 0.8540: 100%|██████████| 267/267 [00:10<00:00, 25.58it/s]\n",
            "Epoch 9/20 - Train loss: 0.0362, score: 0.8267: 100%|██████████| 510/510 [01:45<00:00,  4.85it/s]\n",
            "\t  - Target loss: 0.0297, score: 0.8570: 100%|██████████| 267/267 [00:09<00:00, 27.59it/s]\n",
            "Epoch 10/20 - Train loss: 0.0353, score: 0.8304: 100%|██████████| 510/510 [01:45<00:00,  4.83it/s]\n",
            "\t  - Target loss: 0.0288, score: 0.8683: 100%|██████████| 267/267 [00:09<00:00, 26.93it/s]\n",
            "Epoch 11/20 - Train loss: 0.0351, score: 0.8313: 100%|██████████| 510/510 [01:45<00:00,  4.82it/s]\n",
            "\t  - Target loss: 0.0280, score: 0.8672: 100%|██████████| 267/267 [00:09<00:00, 26.71it/s]\n",
            "Epoch 12/20 - Train loss: 0.0342, score: 0.8366: 100%|██████████| 510/510 [01:45<00:00,  4.85it/s]\n",
            "\t  - Target loss: 0.0263, score: 0.8781: 100%|██████████| 267/267 [00:09<00:00, 27.93it/s]\n",
            "Epoch 13/20 - Train loss: 0.0338, score: 0.8391: 100%|██████████| 510/510 [01:44<00:00,  4.86it/s]\n",
            "\t  - Target loss: 0.0259, score: 0.8794: 100%|██████████| 267/267 [00:09<00:00, 27.33it/s]\n",
            "Epoch 14/20 - Train loss: 0.0329, score: 0.8431: 100%|██████████| 510/510 [01:45<00:00,  4.84it/s]\n",
            "\t  - Target loss: 0.0258, score: 0.8783: 100%|██████████| 267/267 [00:09<00:00, 26.91it/s]\n",
            "Epoch 15/20 - Train loss: 0.0327, score: 0.8441: 100%|██████████| 510/510 [01:45<00:00,  4.85it/s]\n",
            "\t  - Target loss: 0.0254, score: 0.8809: 100%|██████████| 267/267 [00:09<00:00, 26.93it/s]\n",
            "Epoch 16/20 - Train loss: 0.0325, score: 0.8481: 100%|██████████| 510/510 [01:44<00:00,  4.88it/s]\n",
            "\t  - Target loss: 0.0242, score: 0.8889: 100%|██████████| 267/267 [00:09<00:00, 27.89it/s]\n",
            "Epoch 17/20 - Train loss: 0.0321, score: 0.8496: 100%|██████████| 510/510 [01:44<00:00,  4.86it/s]\n",
            "\t  - Target loss: 0.0247, score: 0.8846: 100%|██████████| 267/267 [00:09<00:00, 27.22it/s]\n",
            "Epoch 18/20 - Train loss: 0.0315, score: 0.8519: 100%|██████████| 510/510 [01:45<00:00,  4.84it/s]\n",
            "\t  - Target loss: 0.0243, score: 0.8868: 100%|██████████| 267/267 [00:09<00:00, 26.99it/s]\n",
            "Epoch 19/20 - Train loss: 0.0315, score: 0.8529: 100%|██████████| 510/510 [01:45<00:00,  4.84it/s]\n",
            "\t  - Target loss: 0.0240, score: 0.8880: 100%|██████████| 267/267 [00:09<00:00, 27.56it/s]\n",
            "Epoch 20/20 - Train loss: 0.0315, score: 0.8528: 100%|██████████| 510/510 [01:44<00:00,  4.87it/s]\n",
            "\t  - Target loss: 0.0234, score: 0.8928: 100%|██████████| 267/267 [00:09<00:00, 28.20it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\n",
            "###### PHASE 3 FINISHED ##########\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "\t  - Target loss: 0.0234, score: 0.8928: 100%|██████████| 267/267 [00:09<00:00, 27.39it/s]\n"
          ]
        },
        {
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAV0AAAD4CAYAAABPLjVeAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdd1gU19fA8e/dBdQkYsFCFfvP3jWaaGwRG1iiYmzRaGKKLdZobIlRY4oxGtM0UdFoFLsUg71giWBBEZSooLKAKAIaS4Bl3j8gK0tddNldee8nzzxhZs7MnBnGw907szNCURQkSZIk01CZOwFJkqT/T2TRlSRJMiFZdCVJkkxIFl1JkiQTkkVXkiTJhKyKegOpd649V7dH2Lp0NHcKhZamTTN3CoX2XJ0UksmkpWjEs66jMDXHukL1Z95eYcmWriRJkgkVeUtXkiTJpNK15s4gX7LoSpJUvFh4d5ssupIkFSuKkm7uFPIli64kScVLuiy6kiRJpiNbupIkSSYkL6RJkiSZkGzpSpIkmY4i716QJEkyIXkhTZIkyYRk94IkSZIJyQtpkiRJJmThLV2LfOBN4Mlg3N98h+6eI/l1nXeO+TFxtxg1fjp93/qAEWOnERd/G4BTp0PoN3yMbmjWsRf7jxw3Sc5durQnJOQAoaGHmTLlgxzzbWxsWLduOaGhhzlyZAdVqjgD0KlTW44d8yUoKIBjx3xp3/4Vo+bl5taB0NAjhIcFMnXqmFzzWr/+J8LDAjkW6IOrq7Nu3rRpYwkPCyQ09AhdurTXW06lUhF0KoAd271yrHPJt/NIvBtRqDy7unXgYugRLoUFMi2PPDes/4lLYYEcz5bnx9PGcikskIuhR3DLkmde61zr9T0XQ49w7ux+Vq5YjJVVRtujbNkybNn8K2dO7+XEMV/q1/9fofbhaRS03+ba7tMc75UrFhMTHcK5s/v11rVh/U8EB+0hOGgPVyJOEhy0p+h2DDK+BmzoYAYWV3S1Wi3zF//AT4s/Z9f6X/Dfd4irkdf1Yr5Z/iu9unVm+9qf+ODtwXz38xoAWjVvzFavH9jq9QOrvl9EyRIleKVVsyLPWaVS8d13n9O793CaNn2dAQN6UadOLb2YESMGkpiYTIMG7fn++99YsGA6AAkJifTvP5KWLbvy7ruTWLVqiVHzWrZ0AR4eQ2nUuCNvDuxD3br6eY18exBJicnUrdeWpctWsnDhTADq1q3FQM/eNG7SCXf3IXy/bCEq1ZPTZfy4dwi/9HeObTZv1ohy5co+VZ7uHkNp2LgjA/PIMzExmTr12vLdspV8kSVPT8/eNGrSiZ5Z8sxvnX/8sZ36DV6jSdPOlCpVklEjBwMw4+NxhIRcpFnzLowYOYEli+cVaj8Ky5D9Ntd2C3u8Adau9aan+5Ac2xs85ANatHSjRUs3tm/3Z8cO/6LdwfR0wwczsLiieyE8girOjrg4OWBtbU33zu05cPSkXszVyBu0at4EgFbNGnPw6Ikc69lz8CjtWregVMmSRZ5zy5ZNuHo1iqiom6SmprJ5sw/u7l30Ytzdu7B+/VYAtm3zp0OHVwEICblIbGw8AGFhEZQsWRIbGxuj5NWqZVOuXo0iMvIGqampbPLeiYdHV70YDw831q3bDMDWrX506tg2c3pXNnnvJCUlhaiom1y9GkWrlk0BcHJyoHv3zqxa9YfeulQqFYsWzWb6jPnPlKe39056ZcuzVx559vLoincueea3zt1/HtCtNyjoHM7ODgDUrVubgwePAXD58lVcXZ2pVKlCofbF2Pttru0W9ngDHA38i7uJSfluu39/DzZu2lkEe/WEomgNHsyhwKIrhKgjhPhYCLEsc/hYCFG3qBKKv30H+0oVdeOVK1Ug/naCXsz/alVn3+GMfxz7Dh/nwcNHJCXf04vZve8I3bt0KKo09Tg62hMdHasb12hicXKyzyUmBshozd+7dx87u3J6MX379uDcuVBSUlKMk5fTk23q8nK0zxFzM0teycn3sLMrh5NjzmUdM/dp8eLPmDFjPunZWgpjPnwbX989xMXFFzrPm1m2Fa2JxdHAPB0dc1nWyd6gdVpZWTFkSD8CAg4CcP5CGH379ACgZYsmuLo64+zkUKh9KQxDcjTXdgt7vA3Rru3L3Iq/zZUrkUbYi3wo6YYPZpBv0RVCfAxsBARwKnMQwB9CiOn5LDdaCBEshAj+de0feYU9tSlj3iH47AX6jxhD8LkLVK5op/fR9/adu/x9LZJXX25u9G0Xlbp1azF//nTGjp1h7lTy1aPH69yOv8OZsxf0pjs4VKZfP3eW/7DKTJkV3vLvF3L06F8EHjsFwJdfLadMWVuCg/YwZsxIzp4LRWvh93w+TwYO7MOmIm7lAhbfvVDQ3QujgPqKoqRmnSiE+Ba4CCzKbSFFUVYAK6Dwr+upVLGC7sIYwK34O1SqaJctxo6lX8wG4OHDR+w7FIht6Zd08/88cITOr72CtZVpbs6IiYnTfUSFjI/fGk1cLjGOaDRxqNVqbG1Lk5CQmBlvz6ZNK3jnnUlERt4wXl6ajG3q5RUTlyPGxdkRjSYWtVpNmTK2JCQkoonJuWyMJg53jy64u7vRrVsnSpYsga1tabzWLGPjpp3UqFGVS+EZn0BeeKEU4WGB1K3X1qA8XbJsy9nJgRgD84yJyWXZzGOf3zpnz5pIxYp2fPDhO7pp9+//wzvvTtKNX4k4ybVr+tcTjMmQ/TbXdp/meOdHrVbTt093WrXubrwdyctzfvdCOuCYy3SHzHlG16BObW5ExxAdE0dqaiq79x+mY9vWejGJScm6j7Yr122ib083vfm79x6ix+sdiiK9XAUHh1CzZjVcXV2wtrZmwAAP/Pz26sX4+e1jyJB+ALzxRg8OH864q6JMGVu2bVvN7NlfcuJEsFHzCgo+R82a1ahaNSOvgZ698fXVv3Ls67uHYcMGANCvX08OHjqmmz7Qszc2NjZUrepCzZrVOBV0llmzFlGtegtq1W7NkKEfcvDgMYaPGM/u3ftxqdKUWrVbU6t2ax4+fGRQwc0tT0/P3vhky9Mnjzx9fPfgmUue+a1z5NuDcOvSgSFDx6AoT9oEZcrYYm1tDcCokYM5GvgX9+//U9jDbjBD9ttc2y3s8S7I653bcfnyFTSa2AJjn5k21fDBDApqCn4E7BdC/A3czJxWBagJjC2ShKzUfDLxA96bNAutVktfdzdqVndl+cq11K9Tm47tWhN09jzf/bwGIQTNGzdg1uQPdctrYm8RF3+HFk0bFkV6udJqtUycOAcfn7Wo1Wq8vLwJD/+b2bMncebMefz89rFmzSZWrVpCaOhhEhOTGDYs4/C9//5watSoyowZ45kxYzwAHh7DuJ2tH/tp85rw0Sz8/DagVqlY47WJsLAI5s6dwunTIfj67mXV6o2sWbOM8LBAEhOTGDI041iGhUWweYsP50MOkqbVMn7CzBx9uMbyX57+2fL8dO4UgrPk6bVmGZcy8xycJc8tW3y4kEueua0T4McfFnH9ejSBR3cBsGOHP/MXfEfdOrVYteo7FEUhLOwy746eUiT7W9B+F7WiOt6/r/uB9q+1oUKF8kRdC+azed+wes1GADw9exf5BTQdC+8SEln/0ucaIIQKaAU4ZU7SAEGKgZf+5NuAi558G7BUXBjjbcCPT/xh8OlVss0gk78NuMBOTyXj3RcnC4qTJEmyCBbe0pVfA5YkqXiRRVeSJMl0FDNdIDOULLqSJBUvFn7LmCy6kiQVL7J7QZIkyYRkS1eSJMmEZEtXkiTJhGRLV5IkyYTSLPvLQrLoSpJUvMiWriRJkgnJPl1JkiQTki1dSZIkE/r/3tJ9tdHbRb0Jozpq19TcKRTaPvVLBQdZmNmxB82dQqE8j09Fe9Gm6N8PaJGM2NIVQnQDlgJq4FdFURZlm18F8ALKZsZMVxQl3zdvypauJEnFi5HuXhBCqIEfgC5ANBAkhNilKEpYlrBZgLeiKD8JIeoB/kDV/NZrcW8DliRJeiaKYviQv1bAFUVRrimKkkLG+yJ7Z98aYJv5cxkghgLIlq4kScVLIfp0hRCjgdFZJq3IfMcjZLy44WaWedHAy9lW8SmwRwgxDngReL2gbcqiK0lS8VKIopv1JbpPaRCwRlGUxUKINsA6IUSDzJc/5EoWXUmSihfjXUjTAC5Zxp0zp2U1CugGoCjKCSFESaACEJ/XSmWfriRJxYtWa/iQvyCglhCimhDCBngT2JUt5gbQGUAIURcoCdzOb6WypStJUvFipPt0FUVJE0KMBQLIuB1slaIoF4UQ84BgRVF2AZOBlUKIiWRcVBuhFPC2X1l0JUkqXoz45YjMe279s02bk+XnMODVwqxTFl1JkooX+TVgSZIk01HSLfv7g7LoSpJUvPx/f/aCJEmSSRV8V4JZyaIrSVLxIlu6kiRJJmThRdcivxzRukMrNh9dx9Zj63lr7OAc85u+3Ii1ASs5fmM/nXq2zzH/xZdewCd4M1MWTDBFugDYdmhKg8M/0CDwJ+zHvJFnXNkebWgRvYMXGtUAQFhbUXXxOOrtW0q9PUso3aaBSfKt1r4R7x74mvcOL6b1Bx455jcZ0omRAV/wtv8ChmyZjV0tRwAcGlfnbf8FvO2/gJG7F1C7awuj5+bm1oHQ0COEhwUydeqYHPNtbGxYv/4nwsMCORbog6urs27etGljCQ8LJDT0CF266J8bKpWKoFMB7NjupZvWsWNbTv31J8FBezh0cDs1alQ1+v7kp6tbBy6GHuFSWCDTctlXU+j8+msEn9nL2ZADTJz0Xo75NjY2rPZaxtmQA+w/uJUqVZwAaNa8EUeP+3D0uA+BJ3xx93DTLVOmTGnW/r6coDN7OHU6gJatTPjIVOM98KZIWFzRValUTFv4EROGTGNgh+F07d2ZarVc9WLiNPHM++gL9mzfn+s63ps2inN/nTdFuhlUKqrMf4+IYfO42HEc5Xu3o2Qt55xhL5ak8kh3/jlzWTetwuAuAIS9PoGIQZ/iPPttEKJI0xUqgdvnw/Ee/hUrX59GvV6tdUX1P2E7T7Cq6wxW95jJXz/70XnWUABuX45mjcdsVveYyabhX9N14dsItfFOI5VKxbKlC/DwGEqjxh15c2Af6tatpRcz8u1BJCUmU7deW5YuW8nChTMBqFu3FgM9e9O4SSfc3Yfw/bKFqFRPchs/7h3CL/2tt67ly7/greFjadHSjY0bd/DJDNP9of5vX909htKwcUcG5rKvpshh8bef0v+NkbRq0ZV+Azz4X52aejFvDR9AUlIyTRt34scfVvPZ5x8DEB4WQYd2fWj3igf9+rzNd8vmo1arAVj01Rz27T1Cy2ZuvNranYjLV0y3U+nphg9mYHFFt37TukRHaYi5EUtaahp7dh7gta5t9WJio+O4En6N9FwOWp2GtSlfsRwnDweZKmVebFKLf6NiSblxCyU1jbs7Aynrlv1hROA0dQhxP25D+TdVN61ULRfuHb8AQFpCMtp7D3ixcc0cyxqTQ5MaJEbdIvnmbdJTtYT5nKRWl+Z6MSn/PNL9bP1CCZTMx3inPU5B0WYcd6sS1kZ/unerlk25ejWKyMgbpKamssl7Jx4eXfViPDzcWLduMwBbt/rRqWPbzOld2eS9k5SUFKKibnL1ahStWma0sJycHOjevTOrVv2hty5FUbAtXRoA2zKliYm9Zdwdykf2ffX23kmvbPta1Jq3aMy1a9eJirpJamoq27b40rOn/oOyevR8nQ3rtwGwY/tu2ndoA8CjR4/RZl60KlmyBP99EcvW9iVefbUla728AUhNTSU5+b6pdgnSFcMHM7C4olvRvgK3Yp48KyI+9jYVHSoYtKwQgglzP2TZvJ+KKr1c2TiUJyX2jm48JS4BG4fyejEvNKiOjWMFkg+c1pv+MDyKsl1aglqFjUslXmhYA2tHw/b3aZW2L8f92Lu68fuxdyltXy5HXLO3Xue9I4vpOONN9s1dq5vu0KQGo/YuYlTAFwTMXK0rwsbg6GRPdPSTR5JqNLE4OdrniLmZGaPVaklOvoedXTmcHHMu6+iUsezixZ8xY8b8HH+o33tvCrt2rSPyWjBDhvTjq6+WG21fCpJ1PwCiNbE4ZtvXIs/BsTKa6FjduEYTh4NjZb0YB0d7XYxWq+Ve8n3K22WcL81bNOZk0G6O/+XPxAmz0Wq1uLq6cOfOXX78+SuOHtvF98sX8sILpUy3U8Z79kKReOqiK4TI8z08QojRQohgIURw/MPYvMKMrv+IPhw/8Bfxsfk+b8L0hMBl7khuzludY9adjftIjU2gnv9iXD4dxYPTl8CIRexZnFm7j19em8yhRRt5ZVwf3fTYc1f5rct0vHrNofWHHqhLWJsxy4L16PE6t+PvcObshRzzJkx4l169hlGtegu8vDbxzddzzZDh8+t0cAitW3anY/u+TJr8PiVK2GBlZUXjJvX57df1tHu1Fw8ePmLi5PdNlpOSnm7wYA7PcvfCZ0DOKoL+MypbObYvVBv+dtwdKjtW0o1XcqjI7SytyPw0bF6fJi83ot/w3rzwYimsrK159OARPyx8lsdlFiwl9i42WVrjNvZ2pGRpSapfKkXJ/1Xhf5vnA2BdsSw1V83kysgFPDx/lZufrdLF1tmxiMfXsj89zrjuxyVSOktLvLRDee7HJeYZH7brJG7zc/6NTbgSQ+rDx1Ss7UzchUij5BajicPZ+Un/spOTA5qYuBwxLs6OaDSxqNVqypSxJSEhEU1MzmVjNHG4e3TB3d2Nbt06UbJkCWxtS+O1ZhmTp3xKo4b1OBV0FoDNm3fh67veKPthiP/24z/OTg7EZNvXIs8h5hZOzg66cScne2Jj9LtYYmPicHLOyE2tVmNbpjR3E/TPl4jLV3nw4CH16v0PjSYWjSaO08EhAOzcsZuJk0xXdM3VbWCofFu6QojzeQwXgMr5Lfu0ws5dwqWaM44u9lhZW+HWuxNH9xwzaNk5Y+fTq6UnfV5+k6XzfsJ/S0CRF1yAByF/U7KaAzYulRDWVpTv3Zakvad087X3HxLS6C0utBnNhTajeXA2QldwVSVtUJUqAYBtu8YoaVoe/x1dpPnGhlyjfDV7yrhURGWtpp5Ha67sPaMXU67qk19vzU5NSIzKKAZlXCrqLpzZOtlRvoYjydHG+2QRFHyOmjWrUbWqC9bW1gz07I2v7x69GF/fPQwbNgCAfv16cvDQMd30gZ69sbGxoWpVF2rWrMapoLPMmrWIatVbUKt2a4YM/ZCDB48xfMR4EhOTKVPGllq1qgPweufXuJTtQltRyr6vnp698cm2r0XtzOnz1KhRFVdXZ6ytrXmjvzv+/voXqP399zN4SMYdOX36dufI4RMAuLo66y6cubg4Uqt2da7fiCY+/g4aTSw1a1UDoH2HV7h8yYQX0pR0wwczKKilWxnoCmRvBgngeFEkpNVq+Xrmdyzb8A0qtQqfjf5ci4hi9NSRhIdc4uie49RtXIevfvsc27KladflFUZPeZs3O44oinQMTDqdG7NXUnv9XFCpSdi0j8cRN3GcMogHIVdI3pv3RT2rCmWpvX4uSno6qXF3iZzwXZGnq2jT2TPHi4FrpyHUKs57H+bO3xraTepH7PlIruw7Q/Phbri2rU96qpbH9x7gN+kXAJxb1Kb1hx6kp2pRFIU9s9bwKPEfo+Wm1WqZ8NEs/Pw2oFapWOO1ibCwCObOncLp0yH4+u5l1eqNrFmzjPCwQBITkxgy9EMAwsIi2LzFh/MhB0nTahk/YWauF1uzbuv9D6bivWkF6ekKiYlJvDt6stH2pSD/7at/tn01Ja1Wy5TJn7FtxxrUahW/r9vCpfC/+WTWR5w9c4Hd/vtZ5+XNil8XczbkAImJSYwckXGHR+s2LZg4+T1SU9NQ0tOZPHGurgU8bfJn/PrbEqxtrImKvMmYD6aZbqcsvKUr8nv0oxDiN2C1oiiBuczboChKzptosyls94K5/ajKeUHJ0slXsBe95+okzvQ8voI9+Z+rz3y/5IM5bxr863px3saivT8zF/m2dBVFGZXPvAILriRJksnJRztKkiSZkIV3L8iiK0lSsWKuW8EMJYuuJEnFi2zpSpIkmZAsupIkSSYkH2IuSZJkOvIdaZIkSaYki64kSZIJybsXJEmSTEi2dCVJkkxIFl1JkiTTMeZD9YtCkRfds3dM+Eg3I2hvZdkP5M5NYtgWc6dQaJ/WOWLuFAolLd2yb0PKzaPUf82dgnnIlq4kSZLpyFvGJEmSTEkWXUmSJBOy7C5dWXQlSSpelDTLrrqy6EqSVLxYds2VRVeSpOJFXkiTJEkyJdnSlSRJMh1Lb+mqzJ2AJEmSUaUXYiiAEKKbEOKyEOKKEGJ6HjGeQogwIcRFIcSGgtYpW7qSJBUrSppx1iOEUAM/AF2AaCBICLFLUZSwLDG1gBnAq4qiJAohKhW0XtnSlSSpWFHSDR8K0Aq4oijKNUVRUoCNQO9sMe8CPyiKkgigKEp8QSuVRVeSpOKlEN0LQojRQojgLMPoLGtyAm5mGY/OnJZVbaC2EOKYEOKkEKJbQenJ7gVJkooVA1qwT2IVZQWw4hk2ZwXUAjoAzsARIURDRVGS8lpAtnQlSSpWjNi9oAFcsow7Z07LKhrYpShKqqIokUAEGUU4T2Ytum5uHQgNPUJ4WCBTp47JMd/Gxob1638iPCyQY4E+uLo66+ZNmzaW8LBAQkOP0KVLe73lVCoVQacC2LHdSzftt1+XEHH5BMFBewgO2kPjxvWfOf8uXdpz9tx+zl84xOTJH+Sav9fa5Zy/cIhDh3dQpUpG/p06tSXwmA+nTv1J4DEf2rdvo1tm958bOXtuPydO+nPipD8VK9o9c565CQwKwWPUFHqMmMSvm3blmB9z6zbvfLyQN96fzttT5xN3O0Fv/j8PHtJ5yFgWLF9TJPn9p0uX9pw/f5CLF48wZcqHOebb2Niwbt0PXLx4hCNHdurOkfLlyxIQsJE7d8JZsmSe3jL9+3sQFBTAmTP7mD9/xjPn2NWtAxdDj3ApLJBpeZzHG9b/xKWwQI5nO48/njaWS2GBXAw9gluW8zivdX74wQguhQWSlqLBzq7cU+fs5taB0AuHCQsLZOqUPP7t/f4jYWGBBB7N9m9v6hjCwgIJvXBY79/e2LGjOHtmH+fO7mfcuFG66f3e6Mm5s/t5/OgGzZo1euqcDaVohcFDAYKAWkKIakIIG+BNIPs/lh1ktHIRQlQgo7vhWn4rNVvRValULFu6AA+PoTRq3JE3B/ahbl39PxAj3x5EUmIydeu1ZemylSxcOBOAunVrMdCzN42bdMLdfQjfL1uISvVkV8aPe4fwS3/n2Ob0GfNp0dKNFi3dCAm5+Mz5f7tkHn37jKB5sy4MGNCLOnVq6sUMH+FJUlIyjRp2YPn3v/H5/Iw7ThISEunffxStWnVj9LuT+fW3Jfr7PfIj2rTuQZvWPbidrdgZg1abzoIf1vDj/GnsXPkVuw+e4Or1aL2Yb1ZuwOP1tmz7eRHvD+nL0tWb9OYvX7uF5g3qGD23rFQqFUuXzqd37+E0adIZT89e1Kmjf46MGDGQpKRk6td/je+//1VXRB8//pfPPlvM9OkL9OLLly/LF198Qvfug2jW7HXs7SvSseOrz5TjsqULcPcYSsPGHRmYx3mcmJhMnXpt+W7ZSr7Ich57evamUZNO9MxyHue3zuMnguja/U2iom7myKUwOS9dOh+PXsNo3LgjAwf2pm624/r222+SmJRMvXptWbZsJQsXfJKRc52MnJs06YS7x1CWLVuASqWifr3/MWrkIF551Z3mLdzo0eN1atSoCsDFsMt4DnyXo0f/euqcC8NYLV1FUdKAsUAAEA54K4pyUQgxTwjRKzMsAEgQQoQBB4GpiqLk+4/WbEW3VcumXL0aRWTkDVJTU9nkvRMPj656MR4ebqxbtxmArVv96NSxbeb0rmzy3klKSgpRUTe5ejWKVi2bAuDk5ED37p1ZteqPIs2/RYsmXLt6naiom6SmprJliw/u7m56Me493Vj/+1YAtm/3p0OHVwAICblIXGzGRc6wsAhKliyJjY1Nkeab1YXLV6niWBkXh0pYW1vRvUNrDp44rRdz7bqGlzM/DbRqXE9v/sW/I0lITOaV5g2LNM+WLZvonSObN/vg4aF/jD083Pj994yHuG/b5q8roA8fPuL48SD+/fexXny1alW4ciWKO3fuAnDgQCB9+nR/6hyzn8fe3jvple087pXHedzLoyveuZzH+a3z3LmLXM/2B7Kwsh9Xb++duR5XXc7b/Oio+7fnliPnli2bUKdOTU6dOsejR4/RarUcPXJSd1wvXbpCRES+jT+jUtKFwUOB61IUf0VRaiuKUkNRlAWZ0+YoirIr82dFUZRJiqLUUxSloaIoGwtaZ4FFVwhRRwjRWQjxUrbpBV6ly4+jkz3R0TG6cY0mFidH+xwxNzNjtFotycn3sLMrh5NjzmUdnTKWXbz4M2bMmE96Lm8EnTfvY86c3ss3X3/6zEXO0bEy0Rr9HBwcK+cZo9VquXfvfo6PhH36dCfkXCgpKSm6ab/8/DUnTvrz8fRxz5RjXuIT7mKfpduicoXy3LqTqBdTu3oV9h0LAmD/sWAePHxM0r37pKen882K9Ux+d3CR5JaVY26/5xzH+ElMXsc4q6tXr1OrVnVcXZ1Rq9V4eLjh7Oz49DlmOUcBojWxOBp4Hjs65rKsk71B63wWTo4ORN+M1Y1rNHE4Ojlki7EnOjr2Sc73MnN2ctBNB9BEx+Hk6MDFsMu0bduK8uXLUqpUSbp16/RMx/VZGLFPt0jkW3SFEOOBncA4IFQIkfUetYX5LKe7DSM9/YFxMjVAjx6vczv+DmfOXsgxb+asL2jQ4DVat+lJ+fJlmTo1Z/+gqdWtW4vP509n3LhPdNNGjpxAq1bd6PL6AF59pSWDB79hltymjB5C8IVwBnz4CcEXwqlUoRwqlYqNPvto17KxXtF+niQlJTN+/EzWrfuB/fu3cP16NFrt8/cqHktz6dIVvv7mR/z9NuDr8zsh5y+a7bgqijB4MIeCbhl7F2iuKMo/QoiqwBYhRFVFUZYCeWac9TYMaxunXL8IHaOJ0/tL6OTkgGca06oAACAASURBVCYmLkeMi7MjGk0sarWaMmVsSUhIRBOTc9kYTRzuHl1wd3ejW7dOlCxZAlvb0nitWcbwEeOJi8v4OJ+SksIar01Mmvh+Abuev5iYWzg76ecQG3Mr15gYTRxqtRpb29IkJGS0KB2d7Plj4y+8+84kIiNv6Jb5bx3//PMAb+9dNG/RmA0btj1TrtlVsiuvd2Hs1p27VK5QLltMOb6bMxGAh48eszfwFLYvvUhI+N+cCb3MJt99PHz0mNS0NF4oVZKJo940ao4AMbn9nnMc44wYTS7HOC/+/vvw998HwKhRg9E+w4sM/ztH/+Ps5ECMgedxTEwuy2oyli1onc9CExOLs8uTlq2Tkz0xmthsMXE4Ozs8ydk2M2dNLM7OWZZ1tkcTk7HsmjUbWbMm49P15/M+JjrbOk3FXC1YQxXUvaBSFOUfAEVRosi4StddCPEt+RRdQwQFn6NmzWpUreqCtbU1Az174+u7Ry/G13cPw4YNAKBfv54cPHRMN32gZ29sbGyoWtWFmjWrcSroLLNmLaJa9RbUqt2aIUM/5ODBYwwfMR4Ae/sn387r3asbF8MuPUv6nD4dQo2aVXF1dcba2pr+/T3w89urF+Pnv5chQ/sB0LdvDw4fPg5AmTK2bNu6mjlzvuTkySd9pWq1WvfR2MrKim7dOxEWFvFMeeamwf+qc10TR3RcPKmpaew+dJIOrZvrxSQm39d10fy6cRd93ToA8OX0Mez9fRkBa5cy+d3BeHRuVyQFFyA4OETvHBkwwANfX/1j7Ou7l6FD+wPwxhs9OHToeIHr/e+OkLJlyzB69DBWr376/v/s57GnZ298sp3HPnmcxz6+e/DM5Tw2ZJ3PIvtx9fTsnetx1eX8Rk8O6f7t7c2Rc1DQOeDJcXVxcaRPn+5s3LjDaDkXRrpWGDyYQ0Et3VtCiCaKopwDyGzxugOrgGe6iqLVapnw0Sz8/DagVqlY47WJsLAI5s6dwunTIfj67mXV6o2sWbOM8LBAEhOTGDI0o0sgLCyCzVt8OB9ykDStlvETZubah5vVWq/lVKxYHoTgfMhFPhyT67MrCpX/5Elz2LlrLWq1mrVrvQkP/5tZsydy5swF/P324bXGm19/+5bzFw6RmJjE8Lcy+mjfe/8tqtdwZcaMCcyYMQGAXh7DePDgITt3rcXaygqVWs2hg8dYXQQXBK3Uaj4ZM4L3P/kSbXo6fd3aU7OqM8u9tlC/djU6tmlO0Pkwlq7ahBCC5g3rMHPMCKPnURCtVstHH83Gx2cdarUaL69NhIdHMGfOJE6fvoCf317WrNnEqlXfcfHiEe7eTeKtt8bqlr98+RilS5fGxsYaD4+uuLsP5dKlv1m8+FMaNqwHwMKF33HlSuQz5Tjho1n4ZzuPP507heAs57HXmmVcyjyPB2c5j7ds8eFCLudxbusEGDtmJFMmf4i9fUXOnt7H7j8P8N77U5/quPr5rkelVuG1ZhNh4RHMnTOF02cycl69eiNrVi8lLCyQxLtJDB2WmXN4Rs4hIQfQpmmZMGGWLudNG1dgZ1eO1NQ0xk+YSXLyPSCjkbNkyedUrFienTu8CDl/EXf3oU99zAtiyAUycxKKkvdj0IQQzkCaoig5PtsIIV5VFOVYQRvIq3vBUtnIV7CbhG0d8/RVP63n8RXsKmHZxSc3Kf9GP3PSUU26GFxzqp7ba/KDlG9LV1GUPO9NMaTgSpIkmVo+7UiLIJ+9IElSsWLp3Quy6EqSVKyY61YwQ8miK0lSsaI1010JhpJFV5KkYkW2dCVJkkxI9ulKkiSZkLx7QZIkyYRkS1eSJMmEtOmW/UIcWXQlSSpWZPeCJEmSCaXLuxckSZJMR94yJkmSZEL/77sXLHz/c/g3LdXcKRSaS+Oie0xeUUk+t87cKRRK+WbDzZ1CoT2P57IxyO4FSZIkE5J3L0iSJJmQpX+6lkVXkqRiRXYvSJIkmZC8e0GSJMmELPxlwLLoSpJUvCjP9qLyIieLriRJxUqa7F6QJEkyHdnSlSRJMiHZpytJkmRCsqUrSZJkQrKlK0mSZEJa2dKVJEkyHQt/Ww+W/WSIPHR168DF0CNcCgtk2tQx5k7HIJaQc8fObTkWvJuTZwMYN/HdHPNtbKxZsfpbTp4NYPf+TbhUcQLApYoTUXHn2H90O/uPbuerJZ/mWHbtHz9y+MSuIs0/8EwoHh/Mpud7M/lty+4c82PiE3hn9rf0G/8ZI2d+Q9ydRAAuXbvJ0GmL6Dt2Lv3Gf8afR4OKNM8uXdpz9tx+zl84xOTJH+SYb2Njg9fa5Zy/cIhDh3dQpYozAJ06tSXwmA+nTv1J4DEf2rdvk2NZ780rCQoKKNL882MJ53FB0hEGD+bw3LV0VSoVy5YuoFuPQURHx3LyhD8+vnsID//b3KnlyRJyVqlULFo8B88+I4nR3CLg4GYC/A8QcfmqLmbwW/1JSrpH66Zd6dOvB7M/m8zotycBcD3yBp3b9c113T08uvDgwcMizV+rTWfhLxtY8dlEKtuVY9CUhXRo1ZgaVRx1MYtXb8ajY2t6d3qFv85fYtm6bSycOIqSJWxY8NHbuDpWJj4hiTcnz+eVpvWxfekFo+epUqn4dsk8PNyHotHEcfToLvz89nLp0hVdzPARniQlJdOoYQf69/fg8/nTGf7WWBISEunffxRxsfHUq1ebnbvWUqtma91yvXp35cE/RXuc82MJ57EhLP2BN89dS7dVy6ZcvRpFZOQNUlNT8fbeSS+PruZOK1+WkHOz5o2IvHaD61HRpKamsmObP916dtaL6dajM94bdgDgsyOAtrm0tLJ74cUXeH/MCJZ8/VOR5P2f0L8jqWJfCWf7ilhbW9GtXUsOngrRi7l2M5aXG9YBoFXD/3Hwr4z5VZ0q4+pYGYBKdmUpX8aWxHv3iyTPFi2acO3qdaKibpKamsqWLT64u7vpxbj3dGP971sB2L7dnw4dXgEgJOQicbHxAISFRVCyZElsbGwAePHFFxg37h2+/PL7IsnbEJZwHhsivRBDQYQQ3YQQl4UQV4QQ0/OJ6yeEUIQQLQpaZ4FFVwjRSgjRMvPnekKISUKIHgbkWyQcney5GR2jG4/WxOLoaG+udAxiCTnbO1YmRhOrG4/RxGHvUFkvxsGhEprMGK1Wy/179ylfviwAVVyd2Xd0G9v91vFym+a6ZabPHM9Py1fz6NHjIs3/VkISlSuU141XtitLfEKiXkztai7sO3kWgP0nz/Lg0WOS7v2jF3MhIpLUtDRc7CsWSZ6OjpWJ1jz5XWs0sTg4Vs4zRqvVcu/efezsyunF9OnTnZBzoaSkpAAwZ85kli37lYcPi/Y458cSzmNDpAth8JAfIYQa+AHoDtQDBgkh6uUSVxqYAPxlSH75Fl0hxFxgGfCTEOILYDnwIjBdCDEzn+VGCyGChRDB6ekPDMlDsmC34uJpVr8Tr7d7g7kzF/HTr9/wUukXqd+wDlWrVWG37z5zpwjA5BH9OR0agedHnxMcGkElu7KoVE9O8dt3k/hkySrmjR+hN93S1K1bi8/nT2fcuE8AaNSoHtWqV8Fnl/n6cp8n2kIMBWgFXFEU5ZqiKCnARqB3LnGfA18CBv1FLKhPtz/QBCgBxAHOiqLcE0J8Q0ZVX5DbQoqirABWAFjZOBm1iyVGE4eL85N+PGcnB2Ji4oy5CaOzhJzjYm7h6OSgG3d0sicu9pZeTGxsPE5ODsTG3EKtVlPatjR37yYBkJKS8f/z5y4SFXmTGjWr0aRZAxo3bUDQ+f1YWampULE823zX8ob7W0bPv7JdWW7duasbv5WQRKVsrcNKdmVZMiPjwtXDR4/Zd+KMrt/2n4ePGPP594wb2ofG/6tu9Pz+ExNzC2enJ7/r/45nbjExmjjUajW2tqVJyGy1OzrZ88fGX3j3nUlERt4AoNXLzWjWrBFh4YFYWampWNGO3X9upHu3N4tsP3JjCeexIQpz94IQYjQwOsukFZn1C8AJuJllXjTwcrblmwEuiqL4CSGmGrLNgv7cpymKolUU5SFwVVGUewCKojzCTPcgBwWfo2bNalSt6oK1tTWenr3x8d1jjlQMZgk5nz1zgeo1XKni6oS1tTV93uhBgP8BvZgA/wN4Du4DgEefrgQeOQmAnV05XcvQtaoz1Wu4cj3qJl6/baRxnddo2agzvboN4dqVqCIpuAD1a1Xlemw80bfukJqaxp9Hg+jQqrFeTOK9+6SnZ5yWv27ZTd/OrwKQmprGR1/8hEfHNri92jzHuo3p9OkQatSsiqurM9bW1vTv74Gf3169GD//vQwZ2g+Avn17cPjwcQDKlLFl29bVzJnzJSdPntbF/7ryd2rWeJl6ddvyeucBXPk70uQFFyzjPDZEYe5eUBRlhaIoLbIMKwreQgYhhAr4FphcmPwKaummCCFeyCy6urNVCFEGMxVdrVbLhI9m4e+3AbVKxRqvTYSFRZgjFYNZQs5arZYZUz5n47bfUKtV/PH7Vi5fusK0T8YRcjaUgN0H2bBuC8tXfMXJswEkJSbz3siMOxdav9qSaZ+MIy01jXQlnWkTPyUpMdmk+Vup1XwyehAffPod2vR0+nR+lZpVHPlh/U7q1XSl48tNCLoQwbJ12xECmtWrzcz3BwEQcCyYMxcjSL7/D7sOZBS4z8e/TZ3qLkbPU6vVMnnSHHbuWotarWbtWm/Cw/9m1uyJnDlzAX+/fXit8ebX377l/IVDJCYmMfytcQC89/5bVK/hyowZE5gxYwIAvTyGcft2gtHzfBqWcB4bwogfrTVA1pPEOXPaf0oDDYBDIqN/2B7YJYTopShKcF4rFUo+7ysWQpRQFOXfXKZXABwURblQUNbG7l6QcrIrVdrcKRTajb9+MXcKhSLfBmwaaSmaZ755dq3TUINrzlua3/PcnhDCCogAOpNRbIOAwYqiXMwj/hAwJb+CCwW0dHMruJnT7wB38ltWkiTJHIz1EVxRlDQhxFggAFADqxRFuSiEmAcEK4ryVN8Geu6+HCFJkpQfrRG/aKYoij/gn23anDxiOxiyTll0JUkqVuRTxiRJkkxIFl1JkiQTsvBXpMmiK0lS8SJbupIkSSZkwNd7zUoWXUmSihVLf4i5LLqSJBUrsntBkiTJhGTRlSRJMiFLf+6ALLqSJBUrsk9XkiTJhOTdC1KRS3hUNO/7KkqdOn9m7hQK5c6OaeZOodBKu+f6joFiL93COxhk0ZUkqViRF9IkSZJMyLLbubLoSpJUzMiWriRJkgmlCctu68qiK0lSsWLZJVcWXUmSihnZvSBJkmRC8pYxSZIkE7LskiuLriRJxYzsXpAkSTIhrYW3dWXRlSSpWJEtXUmSJBNSZEtXkiTJdCy9pasydwJPo6tbBy6GHuFSWCDTpo4xdzoGed5ytsR8X+7Qkj+OeLEpcB1DxwzKMb/xy41Y9ecvHL6+lw49X9Obd+TGXtbsWcGaPSv4cvV8U6XMsbAoen/uhcdna1i1JyjH/Ni793hn2VYGfrmBAV/8ztGLkQBoEu7x8qTleC5aj+ei9czfuN9kOefHEs+L7NJRDB7M4blr6apUKpYtXUC3HoOIjo7l5Al/fHz3EB7+t7lTy9PzlrMl5qtSqZi8YAIfDZpKfOxtfvX/icA9x4n6+7ou5pbmFgsmfsmg9z1zLP/v4xRGuI02Zcpo09P5YvMhfh7Tl8plX2LI1xtp37A6NRzsdDErA4Jwa1oLz3aNuBqbwNifd7L7s2oAOFcoi/f0ISbNOT+WeF7kxrI7F57Dlm6rlk25ejWKyMgbpKam4u29k14eXc2dVr6et5wtMd+6TesQHaUh5kYsaalp7N95gHZdX9GLiYu+xdXwayjplvEBM/T6LVwqlMG5QhmsrdR0bV6bQxeu6cUIAQ8epwDwz+MUKpZ5yRypGsQSz4vcpKEYPJhDoYuuEGJtUSRiKEcne25Gx+jGozWxODramzGjgj1vOVtivhXtKxAfE68bj4+9Q0X7igYvb1PCht/8f2KFz3LadX21KFLMIT7pH+zLldaNVy77EvFJ/+jFvN+9NX5Bl3Cb/Rtjf9rJ9P7tdfM0CckM/HIDo5Zu4cwVjUlyzo8lnhe5UQrxnznk270ghNiVfRLQUQhRFkBRlF55LDcaGA0g1GVQqV40QqqS9PT6vTyIO3F3cKziwDLvxVy7FInmekzBCxaxP09fptfL9XirczNCImOZtW4PW2YMpaLtC/w5byRlXyxF2I1bTFzpy9ZPhvJSqRLmTtniWcbnnLwV1NJ1Bu4B3wKLM4f7WX7OlaIoKxRFaaEoSgtjF9wYTRwuzo5PEnRyICYmzqjbMLbnLWdLzPd23B0qOVbSjVdyqMDtuNsGL38n7g4AMTdiOXviHLUa1DR6jtlVKvsScYlPXqV0K+kfKpXV7z7YfuIibs1qAdC4mgP/pqaR9OARNtZWlH2xFAD1qlTGuUIZrt9OKvKc82OJ50VuLL2lW1DRbQGcBmYCyYqiHAIeKYpyWFGUw0WdXG6Cgs9Rs2Y1qlZ1wdraGk/P3vj47jFHKgZ73nK2xHwvnbuEczUnHFzssbK2onPvTgTuOWHQsqXLvIS1jTUAZcrZ0rBlA6Iirhew1LOrX6UyN24nobmTTGqaloDTEbRvWF0vxqFcaf66fBOAa3F3SUnVUu6lUty9/xBtZt909J1kbtxOwtmuTJHnnB9LPC9yk16IwRzy7V5QFCUdWCKE2Jz5/1sFLVPUtFotEz6ahb/fBtQqFWu8NhEWFmHOlAr0vOVsiflqteksmfU93274ErVKje+m3URGRPHOlBFcCokgcO9x6jT+H1/8No/SZV7i1S5teGfyCIZ2GolrLVemLZpIuqKgEoLfl/+hd9dDUbFSq5g+oAMf/LiDdEWhd+t61HSw40e/E9SrUpkODaszqW875v2xn/UHz4KAz4Z2QQjBmasafvQ7iZVahUoIZg3sRJkXSxZ5zvmxxPMiN1rFsu9fEEohEhRC9AReVRTlE0OXsbJxsuwjIJnFyxX/Z+4UCmXf6v7mTqHQnse3AaelaMSzrmOwa1+Da86G69ufeXuFVahWq6IofoBfEeUiSZL0zOTXgCVJkkzoeb97QZIk6blizK8BCyG6CSEuCyGuCCGm5zJ/khAiTAhxXgixXwjhWtA6ZdGVJKlYMdYtY0IINfAD0B2oBwwSQtTLFnYWaKEoSiNgC/BVQfnJoitJUrGiVRSDhwK0Aq4oinJNUZQUYCPQO2uAoigHFUV5mDl6kozvNuRLFl1JkoqVwnQvCCFGCyGCswxZn4rkBNzMMh6dOS0vo4DdBeUnL6RJklSsFOZCmqIoK4AVz7pNIcRQMr5M1r6gWFl0JUkqVox4y5gGcMky7pw5TY8Q4nUyvrXbXlGUfwtaqSy6kiQVK0Z8OHkQUEsIUY2MYvsmMDhrgBCiKfAL0E1RlPicq8hJFl1JkoqVwnzLtoD1pAkhxgIBgBpYpSjKRSHEPCBYUZRdwNfAS8BmIQTAjbyevvgfWXQlSSpWjPkKdkVR/AH/bNPmZPn59cKuUxZdSZKKFXO9+8xQsuhKklSsGKt7oagUedG1Vj9fdd3WppS5Uyi0iiXLmjuFQvvr9mVzp1Aoz+MTu+7//p65UzAL2dKVJEkyIfmUMUmSJBOy9IeYy6IrSVKxIrsXJEmSTEgWXUmSJBP6f3/3giRJkinJlq4kSZIJybsXJEmSTEirWPZb0mTRlSSpWJF9upIkSSYk+3QlSZJMSPbpSpIkmVC67F6QJEkyHdnSlSRJMiFLv3vBIl/B3qVLe0JCDhAaepgpUz7IMd/GxoZ165YTGnqYI0d2UKVKxqvmO3Vqy7FjvgQFBXDsmC/t279SpHl27NyWY8G7OXk2gHET380lT2tWrP6Wk2cD2L1/Ey5VMt7e7FLFiai4c+w/up39R7fz1ZJPdctYW1vzzdJ5HD/9J4FB/vTs5VYkub/asTU+xzbhf3Izo8YNyzG/eesmeO/14pwmkC7uHfXm/fzHEo5H7OWH378pkty6unXgYugRLoUFMm3qmBzzbWxs2LD+Jy6FBXI80AdXV2fdvI+njeVSWCAXQ4/g1uXJi1lXrlhMTHQI587uz7G+MR++TeiFw4ScO8CiL2YWyT7lpaB9NbVjf8fQ+7tdeCzZyaojF3PMj016wDur9jHwB38GLPfjaETGexqTHv7LO6v20ebzTXzhG2TqtPWkK4rBgzlYXEtXpVLx3Xef07PnEDSaOAIDd+Hru49Ll/7WxYwYMZDExGQaNGjPgAEeLFgwnWHDxpKQkEj//iOJjY2nXr3a+Piso0aNl4ssz0WL5+DZZyQxmlsEHNxMgP8BIi5f1cUMfqs/SUn3aN20K3369WD2Z5MZ/fYkAK5H3qBzu7451vvRlPe5czuBV5p3QwhBuXJliiT3WYum8K7neOJi4tkUsJqDAUe5FhGli4nV3GLWhM8Z8cHgHMuv/nE9JUuVxPOtPkWS27KlC+jWYxDR0bGcPOGPj+8ewsOf/P5Hvj2IxMRk6tRri6dnL75YOJPBQz6gbt1aeHr2plGTTjg6ViZg90bq1m9Heno6a9d68+OPq1m9eqne9jq0f4VeHl1p1rwLKSkpVKxoZ/R9epZ9NSVtejpf+ATx84hOVLZ9gSE//0n7Os7UqPTkHFx5OBS3BlXwbFWbq/HJjF13kN2TnShhpWZM50ZcuZXElfhks+T/H0vvXrC4lm7Llk24ejWKqKibpKamsnmzD+7uXfRi3N27sH79VgC2bfOnQ4dXAQgJuUhsbMYLOcPCIihZsiQ2NjZFkmez5o2IvHaD61HRpKamsmObP916dtaL6dajM94bdgDgsyOAtu3bFLjeQUPfYNm3K4CM+w3v3k0yeu4Nm9XjRmQ00ddjSEtNY/eOvXTq9ppeTMzNWCLCrpCenvME/utoMA//eWj0vABatWzK1atRREbeIDU1FW/vnfTy6KoX08vDjXXrNgOwdasfnTq2zZzeFW/vnaSkpBAVdZOrV6No1bIpAEcD/+JuYs5j+d57b/HV1z+QkpICwO3bCUWyX7kxZF9NKTQ6ARe70jiXL421lZquDV05FH5TL0YADx6nAvDP4xQqls546H8pGyuaulbCxkpt6rRzsPSWbqGKrhCirRBikhCiaD7zAo6O9kRHx+rGNZpYnJzsc4mJAUCr1XLv3n3s7MrpxfTt24Nz50J1/5iMzd6xMjGaJ3nGaOKwd6isF+PgUAlNZoxWq+X+vfuUL5/xlocqrs7sO7qN7X7reLlNcwBsy5QG4OOZE9h7ZCsrvb4rkpZXJfuKxMU8eVv0rZh4KtlXNPp2noajkz03M3+3ANGaWBwd7fOM0Wq1JCffw86uHI6OuSyb7dzJrlat6rRt24rjgT4c2LeFFs0bG3Fv8mfIvppS/L1H2Jd5QTdeucwLxN9/pBfzfqdG+IVE4vb1NsauO8T0ni1MnWaBlEL8Zw75Fl0hxKksP78LLAdKA3OFENPzWW60ECJYCBGclvaP0ZI1VN26tZg/fzpjx84w+bYNcSsunmb1O/F6uzeYO3MRP/36DS+VfhErtRonZweCTp2ly2v9CD51jrnzp5k73WLNykpNuXJleaWtBx9Pn88fG342d0oW7c/zUfRqVoM9U99g+bAOzNp6PNdPQ+akVbQGD+ZQUEvXOsvPo4EuiqJ8BrgBQ/JaSFGUFYqitFAUpYWV1UuFSigmJg5nZwfduJOTAxpNXC4xjgCo1WpsbUuTkJCYGW/Ppk0reOedSURG3ijUtgsjLuYWjk5P8nR0sicu9pZeTGxsPE6ZMWq1mtK2pbl7N4mUlFQSMz/qnj93kajIm9SoWY27d5N4+OAhfrv2AOCz408aNq5n9Nzj425j71hJN17ZsRLxcbeNvp2nEaOJwyXzdwvg7ORATExcnjFqtZoyZWxJSEgkJiaXZbOdO9lpomPZsWM3AEHB50hPT6dChfLG2p18GbKvplTJthRxyU+6jW4lP6RSaf13Bm4/fRW3BlUAaFylIv+mpZP08F+T5lkQRVEMHsyhoKKrEkKUE0LYAUJRlNsAiqI8ANKKIqHg4BBq1qyGq6sL1tbWDBjggZ/fXr0YP799DBnSD4A33ujB4cPHAShTxpZt21Yze/aXnDgRXBTp6Zw9c4HqNVyp4uqEtbU1fd7oQYD/Ab2YAP8DeA7OuNjk0acrgUdOAmBnVw6VKuPQu1Z1pnoNV65HZfSd7fnzIK+2awVAu/Zt9C7MGUvo2XCqVHfBqYoDVtZWdO/ThYMBR42+nacRFHyOmjWrUbVqxu/f07M3Pr579GJ8fPcwbNgAAPr168nBQ8d00z09e2NjY0PVqi7UrFmNU0Fn893ezl0BdOiQcZdLrVrVsbGx4c6du0WwZzkZsq+mVN/JjhsJ99Ek/kNqmpaAC9dpX8dZL8ah7Av8dTXjD8O1+GRS0rSUe7GEOdLNUzqKwYM5iPyqvRAiCkgno/9cAV5VFCVWCPESEKgoSpOCNlCqlGuh96xr1458/fUc1Go1Xl7efPXVcmbPnsSZM+fx89tHiRIlWLVqCY0b1ycxMYlhw8YSFXWTjz8ex9SpH3LlSqRuXR4ewwp1caQwbwPu3OU1Pl/0CWq1ij9+38p33/zCtE/GEXI2lIDdBylRwoblK76iYaO6JCUm897ISVyPiqZnLzemfTKOtNQ00pV0vl64nD1/HgTA2cWR5b98mdl6u8uEDz9Bk6WPOzdP8zbgdp3b8PHnE1GrVWz/w5cV361hzLR3uRhyiUMBR2nQpC7frf4S27KlSXmcwp34BPq0z7iTwWvnz1Sr6coLL5YiKfEecyYu4Pihvwq1/UuJN/Oc171bJxYv/gy1SsUar018sWgZn86dQvDpEHx991KiRAm81iyjSebvf/DQD3WfamZMH8+I4QNJ02qZPHkufwZkHNff1/1A+9faUKFCeW7dusNn875h9ZqNWFtb8+vKxTRuXJ+Uee1KYwAABOxJREFUlFQ+/vhzXRE3hdz21Vie5m3ARyM0fO1/mvR0hd7NavBuhwb8uD+Eeo52dKjrzNX4ZObtPMmjf9NACD7q2pRXamZ8muu+eAcP/k0lVZtO6ZLW/DS8s96dD4Yo5TlHFDrpbJzK1Te45mgSLz7z9gor36Kb50JCvABUVhQlsqDYpym65iRfwW4a+RVdyTiex1ewG6PoOpStZ3DNiU0KM3nRfar7dBVFeQgUWHAlSZJMzdLv07W4L0dIkiQ9C0v/GrAsupIkFSvyIeaSJEkmJB/tKEmSZEKypStJkmRC8nU9kiRJJiRbupIkSSYk716QJEkyIXkhTZIkyYQsvXvB4h5iLkmS9CyM+TxdIUQ3IcRlIcSV3B5nK4QoIYTYlDn/LyFE1YLWKYuuJEnFirEe7SiEUAM/AN2BesAgIUT2Z62OAhIVRakJLAG+LCg/WXQlSSpWjPi6nlbAFUVRrimKkgJsBHpni+kNeGX+vAXoLITI9yE6Rd6n++jR9SJ7io8QYrSiKCuKav3G9rzlC89fzs9bviBzNra0FI3BNUcIMZqMFzT8Z0WW/XICsj4OLxrI/qZbXYyiKGlCiGTADriT1zaf95bu6IJDLMrzli88fzk/b/mCzNlssr7l5v/au38XuaowjOPfhySirL8akYUNaBFsLFSCTcRCUCSGYGERQbGwVIhYCJb6B4h/gEnhDxQhCosBNZAFCWh+bLJRNytikcIgbCGiWyqPxT2CyubemYTccw88H7jsnWGKh2F45+57znunHDf8i6T1ohsRcaNcAXb/6/FSeW7b10jaCdwB9P5qQopuRMT2zgJ7JN0r6SbgELD8v9csAy+U82eAkx5YoWt9n+4ke0o9WssL7WVuLS8k8ySVHu3LwBfADuCo7XVJbwLnbC8DR4D3JP0E/EpXmHtd08/1RETEtUl7ISJiRCm6EREjarLoDo3mTY2ko5I2JX1fO8ssJO2WtCLpkqR1SYdrZxoi6WZJZyRdLJnfqJ1pFpJ2SLog6bPaWWYh6bKk7yStSTpXO0+LmuvpltG8H4HH6TYrnwWetX2parAekh4FtoB3bd9fO88QSYvAou3zkm4DVoGnJ/4eC1iwvSVpF3AKOGz7m8rRekl6FdgL3G77QO08QyRdBvbavurm/+jX4pXuLKN5k2L7K7qVzSbY/sX2+XL+B7BBN3kzWe5slYe7yjHpKwpJS8BTwDu1s8R4Wiy6243mTbogtKzcNelB4HTdJMPKv+prwCZwwvbUM78NvAZM+67b/2XgS0mrZYQ25tRi0Y2RSLoVOAa8Yvv32nmG2P7L9gN0k0MPS5psK0fSAWDT9mrtLHN6xPZDdHfeeqm0zmIOLRbdWUbz4jqVvugx4APbn9TOMw/bvwErwJO1s/TYBxwsPdKPgMckvV830jDbV8rfTeBTunZfzKHFojvLaF5ch7IodQTYsP1W7TyzkHSXpDvL+S10C60/1E11dbZft71k+x66z/BJ289VjtVL0kJZWEXSAvAE0MSOnClpruja/hP4ZzRvA/jY9nrdVP0kfQh8Ddwn6WdJL9bONGAf8Dzd1ddaOfbXDjVgEViR9C3dF/MJ201sw2rI3cApSReBM8Bx259XztSc5raMRUS0rLkr3YiIlqXoRkSMKEU3ImJEKboRESNK0Y2IGFGKbkTEiFJ0IyJG9DdU75ekGpgHCgAAAABJRU5ErkJggg==",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 267/267 [00:05<00:00, 49.23it/s]\n",
            "100%|██████████| 215/215 [00:04<00:00, 48.38it/s]\n"
          ]
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1584x432 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "args_array = []\n",
        "args = base_args(\"eeg-classifier_seed-42\")\n",
        "seed_everything(args.seed)\n",
        "model = EEGClassifier()\n",
        "supervised_run(args, model, source_data, source_labels, target_data, target_labels, test_data, mixup_data=mixup_data, supervised_mixup_data=supervised_mixup_data)\n",
        "args_array.append(args)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "hakoBOz7Poz5",
        "outputId": "67df2086-1722-4022-c571-edc93de6377e"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "{\n",
            "    \"alpha\": 0.01,\n",
            "    \"batch_size\": 120,\n",
            "    \"device\": \"cuda\",\n",
            "    \"epochs\": 5,\n",
            "    \"lr\": 0.001,\n",
            "    \"model_name\": \"eeg-classifier_seed-2021\",\n",
            "    \"num_workers\": 2,\n",
            "    \"phase\": \"base\",\n",
            "    \"seed\": 2021\n",
            "}\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "Epoch 1/5 - Train loss: 0.0520, score: 0.7444: 100%|██████████| 754/754 [02:40<00:00,  4.70it/s]\n",
            "\t  - Target loss: 0.0421, score: 0.7869: 100%|██████████| 267/267 [00:09<00:00, 29.62it/s]\n",
            "Epoch 2/5 - Train loss: 0.0402, score: 0.8066: 100%|██████████| 754/754 [02:40<00:00,  4.70it/s]\n",
            "\t  - Target loss: 0.0419, score: 0.7846: 100%|██████████| 267/267 [00:08<00:00, 29.73it/s]\n",
            "Epoch 3/5 - Train loss: 0.0379, score: 0.8167: 100%|██████████| 754/754 [02:40<00:00,  4.70it/s]\n",
            "\t  - Target loss: 0.0406, score: 0.7972: 100%|██████████| 267/267 [00:08<00:00, 30.07it/s]\n",
            "Epoch 4/5 - Train loss: 0.0361, score: 0.8262: 100%|██████████| 754/754 [02:41<00:00,  4.67it/s]\n",
            "\t  - Target loss: 0.0409, score: 0.7935: 100%|██████████| 267/267 [00:08<00:00, 29.71it/s]\n",
            "Epoch 5/5 - Train loss: 0.0345, score: 0.8359: 100%|██████████| 754/754 [02:41<00:00,  4.67it/s]\n",
            "\t  - Target loss: 0.0413, score: 0.7929: 100%|██████████| 267/267 [00:09<00:00, 28.77it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\n",
            "###### PHASE 1 FINISHED ##########\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "\t  - Target loss: 0.0413, score: 0.7929: 100%|██████████| 267/267 [00:09<00:00, 27.01it/s]\n"
          ]
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 267/267 [00:05<00:00, 48.54it/s]\n",
            "100%|██████████| 215/215 [00:04<00:00, 48.07it/s]\n",
            "Epoch 1/10 - Train loss: 0.0325, score: 0.8477: 100%|██████████| 754/754 [02:40<00:00,  4.69it/s]\n",
            "\t  - Target loss: 0.0393, score: 0.8051: 100%|██████████| 267/267 [00:09<00:00, 28.38it/s]\n",
            "Epoch 2/10 - Train loss: 0.0319, score: 0.8506: 100%|██████████| 754/754 [02:43<00:00,  4.62it/s]\n",
            "\t  - Target loss: 0.0391, score: 0.8078: 100%|██████████| 267/267 [00:09<00:00, 27.75it/s]\n",
            "Epoch 3/10 - Train loss: 0.0316, score: 0.8533: 100%|██████████| 754/754 [02:43<00:00,  4.62it/s]\n",
            "\t  - Target loss: 0.0387, score: 0.8069: 100%|██████████| 267/267 [00:09<00:00, 27.87it/s]\n",
            "Epoch 4/10 - Train loss: 0.0311, score: 0.8564: 100%|██████████| 754/754 [02:42<00:00,  4.63it/s]\n",
            "\t  - Target loss: 0.0386, score: 0.8046: 100%|██████████| 267/267 [00:09<00:00, 28.08it/s]\n",
            "Epoch 5/10 - Train loss: 0.0308, score: 0.8569: 100%|██████████| 754/754 [02:41<00:00,  4.66it/s]\n",
            "\t  - Target loss: 0.0376, score: 0.8105: 100%|██████████| 267/267 [00:09<00:00, 28.17it/s]\n",
            "Epoch 6/10 - Train loss: 0.0307, score: 0.8594: 100%|██████████| 754/754 [02:42<00:00,  4.63it/s]\n",
            "\t  - Target loss: 0.0375, score: 0.8114: 100%|██████████| 267/267 [00:09<00:00, 27.93it/s]\n",
            "Epoch 7/10 - Train loss: 0.0302, score: 0.8610: 100%|██████████| 754/754 [02:42<00:00,  4.63it/s]\n",
            "\t  - Target loss: 0.0369, score: 0.8166: 100%|██████████| 267/267 [00:09<00:00, 28.98it/s]\n",
            "Epoch 8/10 - Train loss: 0.0301, score: 0.8611: 100%|██████████| 754/754 [02:43<00:00,  4.62it/s]\n",
            "\t  - Target loss: 0.0369, score: 0.8153: 100%|██████████| 267/267 [00:09<00:00, 27.07it/s]\n",
            "Epoch 9/10 - Train loss: 0.0300, score: 0.8609: 100%|██████████| 754/754 [02:41<00:00,  4.66it/s]\n",
            "\t  - Target loss: 0.0380, score: 0.8101: 100%|██████████| 267/267 [00:09<00:00, 28.54it/s]\n",
            "Epoch 10/10 - Train loss: 0.0300, score: 0.8606: 100%|██████████| 754/754 [02:41<00:00,  4.66it/s]\n",
            "\t  - Target loss: 0.0370, score: 0.8154: 100%|██████████| 267/267 [00:09<00:00, 28.25it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\n",
            "###### PHASE 2 FINISHED ##########\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "\t  - Target loss: 0.0370, score: 0.8154: 100%|██████████| 267/267 [00:10<00:00, 26.67it/s]\n"
          ]
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 267/267 [00:05<00:00, 48.36it/s]\n",
            "100%|██████████| 215/215 [00:04<00:00, 48.23it/s]\n",
            "Epoch 1/20 - Train loss: 0.0399, score: 0.8069: 100%|██████████| 510/510 [01:46<00:00,  4.79it/s]\n",
            "\t  - Target loss: 0.0387, score: 0.8068: 100%|██████████| 267/267 [00:09<00:00, 28.03it/s]\n",
            "Epoch 2/20 - Train loss: 0.0403, score: 0.8027: 100%|██████████| 510/510 [01:46<00:00,  4.77it/s]\n",
            "\t  - Target loss: 0.0349, score: 0.8380: 100%|██████████| 267/267 [00:09<00:00, 27.45it/s]\n",
            "Epoch 3/20 - Train loss: 0.0392, score: 0.8103: 100%|██████████| 510/510 [01:46<00:00,  4.78it/s]\n",
            "\t  - Target loss: 0.0339, score: 0.8290: 100%|██████████| 267/267 [00:09<00:00, 28.12it/s]\n",
            "Epoch 4/20 - Train loss: 0.0385, score: 0.8145: 100%|██████████| 510/510 [01:46<00:00,  4.77it/s]\n",
            "\t  - Target loss: 0.0326, score: 0.8404: 100%|██████████| 267/267 [00:09<00:00, 28.35it/s]\n",
            "Epoch 5/20 - Train loss: 0.0383, score: 0.8154: 100%|██████████| 510/510 [01:45<00:00,  4.84it/s]\n",
            "\t  - Target loss: 0.0340, score: 0.8295: 100%|██████████| 267/267 [00:09<00:00, 28.28it/s]\n",
            "Epoch 6/20 - Train loss: 0.0378, score: 0.8156: 100%|██████████| 510/510 [01:45<00:00,  4.84it/s]\n",
            "\t  - Target loss: 0.0361, score: 0.8271: 100%|██████████| 267/267 [00:09<00:00, 28.63it/s]\n",
            "Epoch 7/20 - Train loss: 0.0375, score: 0.8183: 100%|██████████| 510/510 [01:45<00:00,  4.82it/s]\n",
            "\t  - Target loss: 0.0312, score: 0.8537: 100%|██████████| 267/267 [00:09<00:00, 28.31it/s]\n",
            "Epoch 8/20 - Train loss: 0.0363, score: 0.8245: 100%|██████████| 510/510 [01:45<00:00,  4.83it/s]\n",
            "\t  - Target loss: 0.0305, score: 0.8454: 100%|██████████| 267/267 [00:09<00:00, 28.29it/s]\n",
            "Epoch 9/20 - Train loss: 0.0364, score: 0.8232: 100%|██████████| 510/510 [01:45<00:00,  4.83it/s]\n",
            "\t  - Target loss: 0.0289, score: 0.8654: 100%|██████████| 267/267 [00:09<00:00, 28.35it/s]\n",
            "Epoch 10/20 - Train loss: 0.0357, score: 0.8290: 100%|██████████| 510/510 [01:45<00:00,  4.84it/s]\n",
            "\t  - Target loss: 0.0289, score: 0.8633: 100%|██████████| 267/267 [00:09<00:00, 28.00it/s]\n",
            "Epoch 11/20 - Train loss: 0.0350, score: 0.8324: 100%|██████████| 510/510 [01:45<00:00,  4.85it/s]\n",
            "\t  - Target loss: 0.0292, score: 0.8550: 100%|██████████| 267/267 [00:09<00:00, 28.27it/s]\n",
            "Epoch 12/20 - Train loss: 0.0343, score: 0.8358: 100%|██████████| 510/510 [01:45<00:00,  4.83it/s]\n",
            "\t  - Target loss: 0.0261, score: 0.8798: 100%|██████████| 267/267 [00:09<00:00, 27.52it/s]\n",
            "Epoch 13/20 - Train loss: 0.0340, score: 0.8380: 100%|██████████| 510/510 [01:45<00:00,  4.81it/s]\n",
            "\t  - Target loss: 0.0260, score: 0.8803: 100%|██████████| 267/267 [00:09<00:00, 27.67it/s]\n",
            "Epoch 14/20 - Train loss: 0.0332, score: 0.8416: 100%|██████████| 510/510 [01:45<00:00,  4.82it/s]\n",
            "\t  - Target loss: 0.0258, score: 0.8786: 100%|██████████| 267/267 [00:09<00:00, 27.79it/s]\n",
            "Epoch 15/20 - Train loss: 0.0328, score: 0.8446: 100%|██████████| 510/510 [01:45<00:00,  4.82it/s]\n",
            "\t  - Target loss: 0.0250, score: 0.8835: 100%|██████████| 267/267 [00:09<00:00, 27.50it/s]\n",
            "Epoch 16/20 - Train loss: 0.0320, score: 0.8499: 100%|██████████| 510/510 [01:46<00:00,  4.80it/s]\n",
            "\t  - Target loss: 0.0241, score: 0.8860: 100%|██████████| 267/267 [00:09<00:00, 27.07it/s]\n",
            "Epoch 17/20 - Train loss: 0.0318, score: 0.8502: 100%|██████████| 510/510 [01:46<00:00,  4.79it/s]\n",
            "\t  - Target loss: 0.0238, score: 0.8895: 100%|██████████| 267/267 [00:09<00:00, 26.71it/s]\n",
            "Epoch 18/20 - Train loss: 0.0319, score: 0.8519: 100%|██████████| 510/510 [01:46<00:00,  4.80it/s]\n",
            "\t  - Target loss: 0.0234, score: 0.8933: 100%|██████████| 267/267 [00:09<00:00, 26.83it/s]\n",
            "Epoch 19/20 - Train loss: 0.0315, score: 0.8528: 100%|██████████| 510/510 [01:46<00:00,  4.78it/s]\n",
            "\t  - Target loss: 0.0234, score: 0.8951: 100%|██████████| 267/267 [00:09<00:00, 26.98it/s]\n",
            "Epoch 20/20 - Train loss: 0.0311, score: 0.8542: 100%|██████████| 510/510 [01:46<00:00,  4.78it/s]\n",
            "\t  - Target loss: 0.0232, score: 0.8949: 100%|██████████| 267/267 [00:10<00:00, 26.15it/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\n",
            "###### PHASE 3 FINISHED ##########\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "\t  - Target loss: 0.0232, score: 0.8949: 100%|██████████| 267/267 [00:10<00:00, 26.05it/s]\n"
          ]
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 267/267 [00:05<00:00, 48.54it/s]\n",
            "100%|██████████| 215/215 [00:04<00:00, 47.79it/s]\n"
          ]
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1584x432 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "args = base_args(\"eeg-classifier_seed-2021\", seed=2021)\n",
        "seed_everything(args.seed)\n",
        "model = EEGClassifier()\n",
        "supervised_run(args, model, source_data, source_labels, target_data, target_labels, test_data, mixup_data=mixup_data, supervised_mixup_data=supervised_mixup_data)\n",
        "args_array.append(args)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ib0CDC4DOQsf"
      },
      "source": [
        "### ensemble predictions "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "i53cKXnWN41z",
        "outputId": "e0e39339-3f3d-4ee0-9d92-f2db691dd261"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Base model ensemble target accuracy:  0.7979693845673227\n",
            "Mixup model ensemble target accuracy:  0.8184317400812247\n"
          ]
        }
      ],
      "source": [
        "target_output = ensemble(args_array, \"target\", \"base\")\n",
        "target_pred = np.argmax(target_output, axis=1)\n",
        "print(\"Base model ensemble target accuracy: \", np.mean(target_pred == target_labels))\n",
        "\n",
        "target_output = ensemble(args_array, \"target\", \"mixup\")\n",
        "target_pred = np.argmax(target_output, axis=1)\n",
        "print(\"Mixup model ensemble target accuracy: \", np.mean(target_pred == target_labels))\n",
        "\n",
        "test_output = ensemble(args_array, \"test\")\n",
        "test_pred = np.argmax(test_output, axis=1)\n",
        "np.savetxt(f\"{OUTPUT_FOLDER}answer.txt\", test_pred, delimiter=',',fmt=\"%d\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 338
        },
        "id": "nmizJ9KIrneW",
        "outputId": "5cbfc5e7-c9b4-4dd4-b3d2-d3e2f404cd37"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "(array([10844.,     0.,  2230.,     0.,  9380.,     0.,  1132.,     0.,\n",
              "          270.,  1900.]),\n",
              " array([0. , 0.5, 1. , 1.5, 2. , 2.5, 3. , 3.5, 4. , 4.5, 5. ]),\n",
              " <a list of 10 Patch objects>)"
            ]
          },
          "execution_count": 32,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "plt.hist(test_pred)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xKZF3_UWpEJu"
      },
      "source": [
        "### save results"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "db-5S8ndISNS"
      },
      "outputs": [],
      "source": [
        "import datetime\n",
        "from google.colab import files"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xPePtnX4NGvy",
        "outputId": "0dae9216-60b8-461c-c164-ba21973f61ae"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "2021-10-01-17-09-sleep-mixup_vat_v3\n"
          ]
        }
      ],
      "source": [
        "archive_folder = f\"{datetime.datetime.now():%Y-%m-%d-%H-%M}-sleep-mixup_vat_v3\"\n",
        "print(archive_folder)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 16
        },
        "id": "g4OMglgVgMXz",
        "outputId": "6a1e72f1-601a-4e1a-9b0e-70f93b8a4d12"
      },
      "outputs": [
        {
          "data": {
            "application/javascript": "\n    async function download(id, filename, size) {\n      if (!google.colab.kernel.accessAllowed) {\n        return;\n      }\n      const div = document.createElement('div');\n      const label = document.createElement('label');\n      label.textContent = `Downloading \"${filename}\": `;\n      div.appendChild(label);\n      const progress = document.createElement('progress');\n      progress.max = size;\n      div.appendChild(progress);\n      document.body.appendChild(div);\n\n      const buffers = [];\n      let downloaded = 0;\n\n      const channel = await google.colab.kernel.comms.open(id);\n      // Send a message to notify the kernel that we're ready.\n      channel.send({})\n\n      for await (const message of channel.messages) {\n        // Send a message to notify the kernel that we're ready.\n        channel.send({})\n        if (message.buffers) {\n          for (const buffer of message.buffers) {\n            buffers.push(buffer);\n            downloaded += buffer.byteLength;\n            progress.value = downloaded;\n          }\n        }\n      }\n      const blob = new Blob(buffers, {type: 'application/binary'});\n      const a = document.createElement('a');\n      a.href = window.URL.createObjectURL(blob);\n      a.download = filename;\n      div.appendChild(a);\n      a.click();\n      div.remove();\n    }\n  ",
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "application/javascript": "download(\"download_8d9f70b0-4b3f-41c4-8d08-38b4c47d49a9\", \"answer.txt\", 51512)",
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "files.download(f\"{OUTPUT_FOLDER}answer.txt\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "VQz_uEeBgs9_"
      },
      "outputs": [],
      "source": [
        "%notebook -e {OUTPUT_FOLDER}notebook_run.ipynb"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "FqNPjHHKHZa7"
      },
      "outputs": [],
      "source": [
        "!cp -R {OUTPUT_FOLDER} {GDRIVE_OUTPUT_FOLDER}{archive_folder}"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "collapsed_sections": [],
      "name": "beetl-sleep-mixup_vat_v3.ipynb",
      "provenance": [],
      "toc_visible": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}