hw2_q3.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "execution": {
          "iopub.execute_input": "2023-12-15T17:36:01.123055Z",
          "iopub.status.busy": "2023-12-15T17:36:01.122784Z",
          "iopub.status.idle": "2023-12-15T17:36:31.454465Z",
          "shell.execute_reply": "2023-12-15T17:36:31.453410Z",
          "shell.execute_reply.started": "2023-12-15T17:36:01.123030Z"
        },
        "id": "y8_zCYtOhLvO",
        "outputId": "8ff7c8a9-44f8-4739-f984-7ff9d4423ca5",
        "trusted": true
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: ml_collections in /usr/local/lib/python3.10/dist-packages (0.1.1)\n",
            "Requirement already satisfied: textdistance[extras] in /usr/local/lib/python3.10/dist-packages (4.6.1)\n",
            "Requirement already satisfied: absl-py in /usr/local/lib/python3.10/dist-packages (from ml_collections) (1.4.0)\n",
            "Requirement already satisfied: PyYAML in /usr/local/lib/python3.10/dist-packages (from ml_collections) (6.0.1)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.10/dist-packages (from ml_collections) (1.16.0)\n",
            "Requirement already satisfied: contextlib2 in /usr/local/lib/python3.10/dist-packages (from ml_collections) (21.6.0)\n",
            "Requirement already satisfied: jellyfish in /usr/local/lib/python3.10/dist-packages (from textdistance[extras]) (1.0.3)\n",
            "Requirement already satisfied: numpy in /usr/local/lib/python3.10/dist-packages (from textdistance[extras]) (1.23.5)\n",
            "Requirement already satisfied: Levenshtein in /usr/local/lib/python3.10/dist-packages (from textdistance[extras]) (0.23.0)\n",
            "Requirement already satisfied: pyxDamerauLevenshtein in /usr/local/lib/python3.10/dist-packages (from textdistance[extras]) (1.7.1)\n",
            "Requirement already satisfied: rapidfuzz>=2.6.0 in /usr/local/lib/python3.10/dist-packages (from textdistance[extras]) (3.6.1)\n"
          ]
        }
      ],
      "source": [
        "!pip install ml_collections \"textdistance[extras]\""
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2023-12-15T17:36:31.456632Z",
          "iopub.status.busy": "2023-12-15T17:36:31.456320Z",
          "iopub.status.idle": "2023-12-15T17:36:31.477898Z",
          "shell.execute_reply": "2023-12-15T17:36:31.476971Z",
          "shell.execute_reply.started": "2023-12-15T17:36:31.456604Z"
        },
        "id": "vSDn-u-xYZma",
        "trusted": true
      },
      "outputs": [],
      "source": [
        "# hw2/utils.py\n",
        "\n",
        "import enum\n",
        "import pickle\n",
        "\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "\n",
        "\n",
        "def save_obj(obj, path):\n",
        "    with open(path, \"wb\") as f:\n",
        "        pickle.dump(obj, f)\n",
        "\n",
        "\n",
        "def save_txt(str, path):\n",
        "    with open(path, \"w\", encoding=\"utf-8\") as f:\n",
        "        f.write(str)\n",
        "\n",
        "\n",
        "def load_metrics(file_path):\n",
        "    with open(file_path, \"rb\") as f:\n",
        "        metrics = pickle.load(f)\n",
        "\n",
        "    return metrics\n",
        "\n",
        "\n",
        "def plot_metrics(metrics, plot_path, phase=\"train\"):\n",
        "    metric_keys = list(metrics[0][phase].keys())\n",
        "    len_x = len(metrics)\n",
        "    x_label = \"epoch\".title()\n",
        "    x = np.arange(len_x)\n",
        "\n",
        "    for k in metric_keys:\n",
        "        if k == \"n\":\n",
        "            continue\n",
        "\n",
        "        metric_path = plot_path / f\"{k}__{phase}.pdf\"\n",
        "        y = np.array([metrics[i][phase][k] for i in range(len_x)])\n",
        "\n",
        "        if k != \"loss\":\n",
        "            batch = np.array([metrics[i][phase][\"n\"] for i in range(len_x)])\n",
        "            y /= batch\n",
        "\n",
        "        y = y.mean(axis=-1)\n",
        "        _create_plot([y], [x], x_label, k.title(), metric_path, show=True)\n",
        "\n",
        "\n",
        "def cmp_phase_metrics(metrics, key, phases, plot_path, labels):\n",
        "    len_x = len(metrics)\n",
        "    x_label = \"epoch\".title()\n",
        "    x = np.arange(len_x)\n",
        "    y = []\n",
        "\n",
        "    for phase in phases:\n",
        "        y_phase = np.array([metrics[i][phase][key] for i in range(len_x)])\n",
        "\n",
        "        if key != \"loss\":\n",
        "            batch = np.array([metrics[i][phase][\"n\"] for i in range(len_x)])\n",
        "            y_phase /= batch\n",
        "\n",
        "        y_phase = y_phase.mean(axis=-1)\n",
        "        y.append(y_phase)\n",
        "\n",
        "    x = [x] * len(y)\n",
        "    fig_path = plot_path / f\"{key}__{'_'.join(phases)}.pdf\"\n",
        "    labels = [labels[p] for p in phases]\n",
        "    _create_plot(y, x, x_label, key.title(), fig_path, line_labels=labels, show=True)\n",
        "\n",
        "\n",
        "def _create_plot(ys, xs, x_label, y_label, save_path, line_labels=None, show=False):\n",
        "    fig, ax = plt.subplots()\n",
        "    for y, x in zip(ys, xs):\n",
        "        ax.plot(x, y)\n",
        "\n",
        "    ax.set_xlabel(x_label)\n",
        "    ax.set_ylabel(y_label)\n",
        "    ax.grid()\n",
        "    fig.savefig(save_path, format=\"pdf\", bbox_inches=\"tight\")\n",
        "\n",
        "    if line_labels is not None:\n",
        "        ax.legend(line_labels)\n",
        "\n",
        "    if show:\n",
        "        plt.show(fig)\n",
        "\n",
        "    plt.close(fig)\n",
        "\n",
        "\n",
        "class Dataset(str, enum.Enum):\n",
        "    LJSPEECH_STFT = \"ljspeech_stft\"\n",
        "    LJSPEECH_MEL = \"ljspeech_mel\"\n",
        "\n",
        "\n",
        "class DataSplit(str, enum.Enum):\n",
        "    TRAIN = \"train\"\n",
        "    VALIDATION = \"validation\"\n",
        "    TEST = \"test\"\n",
        "\n",
        "\n",
        "class AudioTransformType(str, enum.Enum):\n",
        "    STFT = \"stft\"\n",
        "    LOG_MEL_STFT = \"log_mel_stft\"\n",
        "\n",
        "\n",
        "class EncoderType(str, enum.Enum):\n",
        "    RNN = \"rnn\"\n",
        "    TRANSFORMER = \"transformer\"\n",
        "\n",
        "\n",
        "class DecoderType(str, enum.Enum):\n",
        "    RNN = \"rnn\"\n",
        "    TRANSFORMER = \"transformer\"\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2023-12-15T17:36:31.479932Z",
          "iopub.status.busy": "2023-12-15T17:36:31.479673Z",
          "iopub.status.idle": "2023-12-15T17:36:31.563241Z",
          "shell.execute_reply": "2023-12-15T17:36:31.562577Z",
          "shell.execute_reply.started": "2023-12-15T17:36:31.479907Z"
        },
        "id": "1OA5lnc_OWlc",
        "trusted": true
      },
      "outputs": [],
      "source": [
        "# hw2/config.py\n",
        "\n",
        "from ml_collections import config_dict\n",
        "\n",
        "\n",
        "def get_config():\n",
        "    config = config_dict.ConfigDict()\n",
        "    config.seed = 42\n",
        "    config.n_fft = 400\n",
        "    config.n_vocab = 35\n",
        "    config.text_len = 99\n",
        "    config.audio_freq = 80  # number of mel filters\n",
        "    config.audio_time = 2754  # 10 seconds of sound\n",
        "    config.epochs = 15\n",
        "    config.lr = 1e-3\n",
        "    config.unk_token_idx = 3\n",
        "    config.storage_folder = \"./storage\"\n",
        "    config.speech_to_text = dict(\n",
        "        max_text_len=config.get_ref(\"text_len\"),\n",
        "        encoder_kwargs=dict(\n",
        "            n_layers=2,\n",
        "            freq_dim=config.get_ref(\"audio_freq\"),\n",
        "            time_dim=config.get_ref(\"audio_time\") // 2,\n",
        "            hidden_dim=200,\n",
        "            n_heads=2,\n",
        "            feed_fwd_dim=400,\n",
        "        ),\n",
        "        decoder_type=DecoderType.TRANSFORMER,\n",
        "        decoder_kwargs=dict(\n",
        "            freq_dim=config.get_ref(\"n_vocab\"),\n",
        "            time_dim=config.get_ref(\"text_len\"),\n",
        "            hidden_dim=200,\n",
        "            n_heads=2,\n",
        "            feed_fwd_dim=400,\n",
        "        ),\n",
        "    )\n",
        "\n",
        "    config.data = dict(\n",
        "        data_folder=\"./data/\",\n",
        "        data_split=dict(train=(0, 9825), validation=(9825, 10480), test=(10480, 13100)),\n",
        "        dataloader=dict(\n",
        "            train=dict(batch_size=64, shuffle=True, num_workers=2, prefetch_factor=2),\n",
        "            validation=dict(\n",
        "                batch_size=16, shuffle=False, num_workers=2, prefetch_factor=2\n",
        "            ),\n",
        "            test=dict(\n",
        "                batch_size=16, shuffle=False, num_workers=2, prefetch_factor=2\n",
        "            ),\n",
        "        ),\n",
        "        transforms=dict(\n",
        "            audio=dict(\n",
        "                spectrogram=dict(\n",
        "                    n_fft=config.get_ref(\"n_fft\"),\n",
        "                    hop_length=160,\n",
        "                    power=2,\n",
        "                    center=True,\n",
        "                    normalized=False,\n",
        "                    onesided=None,\n",
        "                ),\n",
        "                mel_scale=dict(\n",
        "                    n_mels=config.get_ref(\"audio_freq\"),\n",
        "                    n_stft=config.get_ref(\"n_fft\") // 2 + 1,\n",
        "                    norm=\"slaney\",\n",
        "                    mel_scale=\"slaney\",\n",
        "                ),\n",
        "                pad=dict(\n",
        "                    l=0,\n",
        "                    t=0,\n",
        "                    r=config.get_ref(\"audio_time\"),\n",
        "                    b=0,\n",
        "                ),\n",
        "                freq_masking=dict(freq_mask_param=27),\n",
        "                time_masking=dict(time_mask_param=80),\n",
        "            ),\n",
        "            text=dict(\n",
        "                token_vectorizer=dict(\n",
        "                    max_len=config.get_ref(\"text_len\"),\n",
        "                    start=\"<\",\n",
        "                    stop=\">\",\n",
        "                    empty=\"@\",\n",
        "                    unk=\"#\",\n",
        "                    n_vocab=config.get_ref(\"n_vocab\"),\n",
        "                )\n",
        "            ),\n",
        "        ),\n",
        "    )\n",
        "\n",
        "    return config.lock()\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2023-12-15T17:36:31.565324Z",
          "iopub.status.busy": "2023-12-15T17:36:31.565059Z",
          "iopub.status.idle": "2023-12-15T17:36:33.511987Z",
          "shell.execute_reply": "2023-12-15T17:36:33.511025Z",
          "shell.execute_reply.started": "2023-12-15T17:36:31.565300Z"
        },
        "id": "pqcDO618mh4I",
        "trusted": true
      },
      "outputs": [],
      "source": [
        "# hw2/data.py\n",
        "\n",
        "import functools\n",
        "import os\n",
        "from typing import Callable, Dict, List, Tuple\n",
        "\n",
        "from ml_collections import config_dict\n",
        "import torch\n",
        "import torchaudio\n",
        "import torchvision\n",
        "\n",
        "\n",
        "class TokenVectorizer(torch.nn.Module):\n",
        "    def __init__(\n",
        "        self,\n",
        "        n_vocab: int,\n",
        "        max_len: int = 50,\n",
        "        start: str = \"<\",\n",
        "        stop: str = \">\",\n",
        "        empty: str = \"@\",\n",
        "        unk: str = \"#\",\n",
        "    ) -> None:\n",
        "        super().__init__()\n",
        "        self._max_len = max_len\n",
        "        self._start = start\n",
        "        self._stop = stop\n",
        "        self._empty = empty\n",
        "        self._unk = unk\n",
        "        self._vocab = [\" \", \",\", \".\", \"?\", \"-\"]\n",
        "        self._vocab += [chr(i + 96) for i in range(1, 27)]\n",
        "\n",
        "        assert self._start not in self._vocab\n",
        "        assert self._stop not in self._vocab\n",
        "        assert self._empty not in self._vocab\n",
        "        assert self._unk not in self._vocab\n",
        "        self._vocab = [self._start, self._stop, self._empty, self._unk] + self._vocab\n",
        "        assert len(self._vocab) == n_vocab\n",
        "        self._unk_idx = 3\n",
        "        self._mapper = {c: i for i, c in enumerate(self._vocab)}\n",
        "        self._inv_mapper = {v: k for k, v in self._mapper.items()}\n",
        "\n",
        "    def forward(self, transcript: str) -> torch.Tensor:\n",
        "        transcript = transcript.lower()[: self._max_len - 2]\n",
        "        transcript = self._start + transcript + self._stop\n",
        "        pad = [self._mapper[self._empty]] * (self._max_len - len(transcript))\n",
        "        return torch.tensor(\n",
        "            [self._mapper.get(c, self._unk_idx) for c in transcript] + pad\n",
        "        )\n",
        "\n",
        "    @property\n",
        "    def invert_mapper(self) -> Dict[int, str]:\n",
        "        return self._inv_mapper\n",
        "\n",
        "    @property\n",
        "    def mapper(self) -> Dict[str, int]:\n",
        "        return self._mapper\n",
        "\n",
        "    @property\n",
        "    def vocab(self) -> List[str]:\n",
        "        return self._vocab\n",
        "\n",
        "\n",
        "class SpeechRecognitionDataset(torch.utils.data.Dataset):\n",
        "    def __init__(\n",
        "        self,\n",
        "        ds: torch.utils.data.Dataset,\n",
        "        get_el: Callable,\n",
        "        audio_transforms: torch.nn.Module,\n",
        "        text_transforms: torch.nn.Module,\n",
        "    ):\n",
        "        self._orig_ds = ds\n",
        "        self._get_el = get_el\n",
        "        self._audio_tf = audio_transforms\n",
        "        self._text_tf = text_transforms\n",
        "\n",
        "    def __getitem__(self, idx):\n",
        "        audio, text = self._get_el(self._orig_ds, idx)\n",
        "        audio = self._audio_tf(audio)\n",
        "        text = self._text_tf(text)\n",
        "        return audio, text\n",
        "\n",
        "    def __len__(self):\n",
        "        return len(self._orig_ds)\n",
        "\n",
        "\n",
        "def _get_ljspeech_el(\n",
        "    ds: torch.utils.data.Dataset, idx: int\n",
        ") -> Tuple[torch.Tensor, torch.Tensor]:\n",
        "    el = ds[idx]\n",
        "    return el[0], el[-1]\n",
        "\n",
        "\n",
        "def _log_mel_stft_transforms(config: config_dict.ConfigDict) -> torch.nn.Module:\n",
        "    def log_mel(x):\n",
        "        log_x = torch.clamp(x, min=1e-10).log10()\n",
        "        log_x = torch.maximum(log_x, log_x.max() - 8.0)\n",
        "        log_x = (log_x + 4.0) / 4.0\n",
        "        return log_x\n",
        "\n",
        "    return torchvision.transforms.Compose(\n",
        "        [\n",
        "            functools.partial(torch.squeeze, dim=0),\n",
        "            torchaudio.transforms.Spectrogram(**config.spectrogram),\n",
        "            torchaudio.transforms.MelScale(**config.mel_scale),\n",
        "            log_mel,\n",
        "            torchvision.transforms.Pad(\n",
        "                (config.pad.l, config.pad.t, config.pad.r, config.pad.b)\n",
        "            ),\n",
        "            lambda x: x[..., :, : config.pad.r],\n",
        "        ]\n",
        "    )\n",
        "\n",
        "\n",
        "def build_dl(\n",
        "    split: DataSplit, config: config_dict.ConfigDict\n",
        ") -> torch.utils.data.DataLoader:\n",
        "    os.makedirs(config.data_folder, exist_ok=True)\n",
        "    orig_ds = torchaudio.datasets.LJSPEECH(config.data_folder, download=True)\n",
        "    get_el = _get_ljspeech_el\n",
        "    audio_tf = _log_mel_stft_transforms(config.transforms.audio)\n",
        "    text_tf = TokenVectorizer(**config.transforms.text.token_vectorizer)\n",
        "\n",
        "    ds = torch.utils.data.Subset(orig_ds, range(*config.data_split.get(split)))\n",
        "    ds = SpeechRecognitionDataset(ds, get_el, audio_tf, text_tf)\n",
        "    dl = torch.utils.data.DataLoader(ds, **config.dataloader.get(split))\n",
        "    return dl\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2023-12-15T17:36:33.513731Z",
          "iopub.status.busy": "2023-12-15T17:36:33.513328Z",
          "iopub.status.idle": "2023-12-15T17:36:33.566090Z",
          "shell.execute_reply": "2023-12-15T17:36:33.565173Z",
          "shell.execute_reply.started": "2023-12-15T17:36:33.513705Z"
        },
        "id": "2VJPt6aLpoBA",
        "trusted": true
      },
      "outputs": [],
      "source": [
        "# hw2/model.py\n",
        "\n",
        "from typing import Any, Optional, Tuple\n",
        "\n",
        "from ml_collections import config_dict\n",
        "import numpy as np\n",
        "import torch\n",
        "\n",
        "\n",
        "class SpeechToText(torch.nn.Module):\n",
        "    def __init__(\n",
        "        self,\n",
        "        max_text_len: int,\n",
        "        encoder_kwargs: config_dict.ConfigDict,\n",
        "        decoder_type: DecoderType,\n",
        "        decoder_kwargs: config_dict.ConfigDict,\n",
        "    ) -> None:\n",
        "        \"\"\"\n",
        "        text_freq_dim is the vocabulary size.\n",
        "        \"\"\"\n",
        "        super().__init__()\n",
        "        self._max_text_len = max_text_len\n",
        "        self._enc = AudioEncoderTransformer(**encoder_kwargs)\n",
        "        self._dec, self._single_input = self._decoder_factory(\n",
        "            decoder_type, decoder_kwargs\n",
        "        )\n",
        "\n",
        "    def forward(\n",
        "        self, audio_feat: torch.Tensor, text_feat: torch.Tensor\n",
        "    ) -> torch.Tensor:\n",
        "        return self._dec(text_feat, self._enc(audio_feat))\n",
        "\n",
        "    @torch.no_grad()\n",
        "    def generate(self, audio_feat: torch.Tensor) -> torch.Tensor:\n",
        "        enc_feat = self._enc(audio_feat)\n",
        "        text = last_token = torch.zeros(\n",
        "            (enc_feat.shape[0], 1), dtype=torch.long, device=enc_feat.device\n",
        "        )  # vocab index of start token `<` = 0\n",
        "        start_token = True\n",
        "\n",
        "        for _ in range(self._max_text_len - 1):\n",
        "            dec_input = last_token if self._single_input else text\n",
        "            logits = self._dec.generate(dec_input, enc_feat, start_token)\n",
        "            tokens = torch.argmax(logits, dim=-2)\n",
        "            last_token = tokens[:, -1:]\n",
        "            text = torch.concat([text, last_token], axis=-1)\n",
        "            start_token = False\n",
        "\n",
        "        return text\n",
        "\n",
        "    def _decoder_factory(\n",
        "        self, decoder_type: DecoderType, kwargs: config_dict.ConfigDict\n",
        "    ) -> torch.nn.Module:\n",
        "        if decoder_type == DecoderType.RNN:\n",
        "            return TextDecoderRecurrent(**kwargs), True\n",
        "        elif decoder_type == DecoderType.TRANSFORMER:\n",
        "            return TextDecoderTransformer(**kwargs), False\n",
        "        else:\n",
        "            raise ValueError(f\"Decoder type `{decoder_type}` is not valid.\")\n",
        "\n",
        "\n",
        "class AudioEncoderTransformer(torch.nn.Module):\n",
        "    def __init__(\n",
        "        self,\n",
        "        n_layers: int = 2,\n",
        "        freq_dim: int = 129,\n",
        "        time_dim: int = 2754,\n",
        "        hidden_dim: int = 64,\n",
        "        n_heads: int = 2,\n",
        "        feed_fwd_dim: int = 128,\n",
        "        **kwargs,\n",
        "    ) -> None:\n",
        "        super().__init__()\n",
        "        del kwargs\n",
        "        self._embed = SpeechEmbedding(freq_dim, hidden_dim)  # audio embed\n",
        "        self.register_buffer(\"_pos_embed\", self._pos_encoding(time_dim, hidden_dim))\n",
        "        self._att_blocks = torch.nn.ModuleList(\n",
        "            [\n",
        "                ResidualAttentionBlock(hidden_dim, n_heads, feed_fwd_dim)\n",
        "                for _ in range(n_layers)\n",
        "            ]\n",
        "        )\n",
        "        self._out_norm = torch.nn.LayerNorm(hidden_dim)\n",
        "\n",
        "    def forward(self, x: torch.Tensor) -> torch.Tensor:\n",
        "        out = x  # (B, F, T)\n",
        "        out = self._embed(out)\n",
        "        out = out + self._pos_embed\n",
        "\n",
        "        for block in self._att_blocks:\n",
        "            out = block(out)\n",
        "\n",
        "        out = self._out_norm(out)\n",
        "        return out  # (B, T, E)\n",
        "\n",
        "    def _pos_encoding(\n",
        "        self, length: int, channels: int, max_timescale: int = 3000\n",
        "    ) -> torch.Tensor:\n",
        "        \"\"\"Returns sinusoids for positional embedding\"\"\"\n",
        "        assert channels % 2 == 0\n",
        "        log_timescale_increment = np.log(max_timescale) / (channels // 2 - 1)\n",
        "        inv_timescales = torch.exp(\n",
        "            -log_timescale_increment * torch.arange(channels // 2)\n",
        "        )\n",
        "        scaled_time = (\n",
        "            torch.arange(length)[:, np.newaxis] * inv_timescales[np.newaxis, :]\n",
        "        )\n",
        "        return torch.cat([torch.sin(scaled_time), torch.cos(scaled_time)], dim=1)\n",
        "\n",
        "\n",
        "class TextDecoderTransformer(torch.nn.Module):\n",
        "    def __init__(\n",
        "        self,\n",
        "        freq_dim: int = 35,\n",
        "        time_dim: int = 50,\n",
        "        hidden_dim: int = 64,\n",
        "        n_heads: int = 2,\n",
        "        feed_fwd_dim: int = 128,\n",
        "        **kwargs,\n",
        "    ) -> None:\n",
        "        super().__init__()\n",
        "        del kwargs\n",
        "        self._embed = torch.nn.Embedding(freq_dim, hidden_dim)\n",
        "        self._pos_embed = torch.nn.Parameter(torch.rand(time_dim - 1, hidden_dim))\n",
        "        self._att = ResidualAttentionLayer(hidden_dim, n_heads, feed_fwd_dim)\n",
        "        self._cross_att = ResidualAttentionBlock(hidden_dim, n_heads, feed_fwd_dim)\n",
        "        self._out_norm = torch.nn.LayerNorm(hidden_dim)\n",
        "        self._classifier = torch.nn.Linear(hidden_dim, freq_dim)\n",
        "        mask = torch.empty(time_dim, time_dim).fill_(-np.inf).triu_(1)\n",
        "        self.register_buffer(\"_mask\", mask, persistent=False)\n",
        "\n",
        "    def forward(self, x: torch.Tensor, latent: torch.Tensor) -> torch.Tensor:\n",
        "        # Embedding layer to convert the input text tokens into dense vectors\n",
        "        embedded = self._embed(x)  \n",
        "\n",
        "        # Add a position embeddings matrix to the token embeddings\n",
        "        embedded = embedded + self._pos_embed[:x.size(1), :]  \n",
        "\n",
        "        # The token embeddings pass through a residual attention layer as input for the queries,keys, and values\n",
        "        attention_output = self._att(embedded, mask=self._mask)  \n",
        "\n",
        "        # A residual cross-attention block will merge and attend the previous attention’s output with the encoder’s output\n",
        "        cross_attention_output = self._cross_att(attention_output, latent)  \n",
        "\n",
        "        # Normalize the residual block output using a normalization layer\n",
        "        normalized_output = self._out_norm(cross_attention_output)  \n",
        "\n",
        "        # The output of the normalization layer passes through a single linear layer (classifier)\n",
        "        output = self._classifier(normalized_output)  \n",
        "\n",
        "        return output.transpose(1, 2)\n",
        "\n",
        "    def generate(\n",
        "        self, x: torch.Tensor, latent: torch.Tensor, start_token: bool = False\n",
        "    ) -> torch.Tensor:\n",
        "        return self.forward(x, latent)\n",
        "\n",
        "\n",
        "class TextDecoderRecurrent(torch.nn.Module):\n",
        "    def __init__(\n",
        "        self,\n",
        "        n_layers: int = 2,\n",
        "        freq_dim: int = 35,\n",
        "        hidden_dim: int = 64,\n",
        "        n_heads: int = 2,\n",
        "        feed_fwd_dim: int = 128,\n",
        "        **kwargs,\n",
        "    ) -> None:\n",
        "        super().__init__()\n",
        "        del kwargs\n",
        "        self._embed = torch.nn.Embedding(freq_dim, hidden_dim)\n",
        "        self._rnn_norm = torch.nn.LayerNorm(hidden_dim)\n",
        "        self._rnn = torch.nn.LSTM(\n",
        "            hidden_dim, hidden_dim, num_layers=n_layers, batch_first=True\n",
        "        )\n",
        "        self._cross_att = ResidualAttentionBlock(hidden_dim, n_heads, feed_fwd_dim)\n",
        "        self._out_norm = torch.nn.LayerNorm(hidden_dim)\n",
        "        self._classifier = torch.nn.Linear(hidden_dim, freq_dim)\n",
        "        self._state = None\n",
        "\n",
        "    def forward(\n",
        "        self, x: torch.Tensor, latent: torch.Tensor, state: Any = None\n",
        "    ) -> torch.Tensor:\n",
        "        return self._forward(x, latent, state)[0]\n",
        "\n",
        "    def generate(\n",
        "        self, x: torch.Tensor, latent: torch.Tensor, start_token: bool = False\n",
        "    ) -> torch.Tensor:\n",
        "        if start_token:\n",
        "            self._state = None\n",
        "\n",
        "        out, self._state = self._forward(x, latent, self._state)\n",
        "        return out\n",
        "\n",
        "    def _forward(\n",
        "        self, x: torch.Tensor, latent: torch.Tensor, state: Any\n",
        "    ) -> Tuple[torch.Tensor, Any]:\n",
        "        \n",
        "        # Embedding layer to convert the input text tokens into dense vectors, also called token embeddings\n",
        "        embedded = self._embed(x)\n",
        "\n",
        "        # The token embeddings pass through a normalization layer followed by LSTM\n",
        "        rnn_input = self._rnn_norm(embedded)\n",
        "        rnn_output, new_state = self._rnn(rnn_input, state)\n",
        "\n",
        "        # A residual cross-attention block will merge and attend the LSTM’s output with the encoder’s output\n",
        "        cross_attention_output = self._cross_att(rnn_output, latent)\n",
        "\n",
        "        # Normalize the residual block output using a normalization layer\n",
        "        normalized_output = self._out_norm(cross_attention_output)\n",
        "\n",
        "        # The output of the normalization layer passes through a single linear layer (classifier)\n",
        "        output = self._classifier(normalized_output)\n",
        "        \n",
        "        return output.transpose(1, 2), new_state\n",
        "\n",
        "\n",
        "class ResidualAttentionBlock(torch.nn.Module):\n",
        "    def __init__(\n",
        "        self,\n",
        "        hidden_dim: int = 64,\n",
        "        n_heads: int = 2,\n",
        "        feed_fwd_dim: int = 128\n",
        "    ) -> None:\n",
        "        super().__init__()\n",
        "        self._att_norm = torch.nn.LayerNorm(hidden_dim)\n",
        "        self._att = MultiHeadAttention(hidden_dim, n_heads)\n",
        "        self._mlp_norm = torch.nn.LayerNorm(hidden_dim)\n",
        "        self._mlp = torch.nn.Sequential(\n",
        "            torch.nn.Linear(hidden_dim, feed_fwd_dim),\n",
        "            torch.nn.GELU(),\n",
        "            torch.nn.Linear(feed_fwd_dim, hidden_dim),\n",
        "        )\n",
        "\n",
        "    def forward(\n",
        "        self, x: torch.Tensor, cross_x: Optional[torch.Tensor] = None\n",
        "    ) -> torch.Tensor:\n",
        "        out = x\n",
        "        out = out + self._att(self._att_norm(out), cross_x)\n",
        "        out = out + self._mlp(self._mlp_norm(out))\n",
        "        return out\n",
        "\n",
        "\n",
        "class ResidualAttentionLayer(torch.nn.Module):\n",
        "    def __init__(\n",
        "        self, hidden_dim: int = 64, n_heads: int = 2, feed_fwd_dim: int = 128\n",
        "    ) -> None:\n",
        "        super().__init__()\n",
        "        self._att_norm = torch.nn.LayerNorm(hidden_dim)\n",
        "        self._att = MultiHeadAttention(hidden_dim, n_heads)\n",
        "\n",
        "    def forward(\n",
        "        self, x: torch.Tensor, mask: Optional[torch.Tensor] = None,\n",
        "    ) -> torch.Tensor:\n",
        "        out = x\n",
        "        out = out + self._att(self._att_norm(out), mask=mask)\n",
        "        return out\n",
        "\n",
        "\n",
        "class MultiHeadAttention(torch.nn.Module):\n",
        "    \"\"\"Simple multi head attn implementation, torch's version has unnecessary overhead.\"\"\"\n",
        "\n",
        "    def __init__(self, hidden_dim: int, n_heads: int):\n",
        "        super().__init__()\n",
        "        self._n_heads = n_heads\n",
        "        self._query = torch.nn.Linear(hidden_dim, hidden_dim)\n",
        "        self._key = torch.nn.Linear(hidden_dim, hidden_dim, bias=False)\n",
        "        self._value = torch.nn.Linear(hidden_dim, hidden_dim)\n",
        "        self._out = torch.nn.Linear(hidden_dim, hidden_dim)\n",
        "\n",
        "    def forward(\n",
        "        self,\n",
        "        x: torch.Tensor,\n",
        "        xa: Optional[torch.Tensor] = None,\n",
        "        mask: Optional[torch.Tensor] = None,\n",
        "    ):\n",
        "        q = self._query(x)\n",
        "        k = self._key(x if xa is None else xa)\n",
        "        v = self._value(x if xa is None else xa)\n",
        "        wv = self._qkv_attention(q, k, v, mask)\n",
        "        return self._out(wv)\n",
        "\n",
        "    def _qkv_attention(\n",
        "        self,\n",
        "        q: torch.Tensor,\n",
        "        k: torch.Tensor,\n",
        "        v: torch.Tensor,\n",
        "        mask: Optional[torch.Tensor] = None,\n",
        "    ):\n",
        "        _, n_seq, hidden_dim = q.shape\n",
        "        scale = (hidden_dim // self._n_heads) ** -0.25\n",
        "        q = q.view(*q.shape[:2], self._n_heads, -1).permute(0, 2, 1, 3) * scale\n",
        "        k = k.view(*k.shape[:2], self._n_heads, -1).permute(0, 2, 3, 1) * scale\n",
        "        v = v.view(*v.shape[:2], self._n_heads, -1).permute(0, 2, 1, 3)\n",
        "\n",
        "        qk = q @ k\n",
        "\n",
        "        if mask is not None:\n",
        "            qk = qk + mask[:n_seq, :n_seq]\n",
        "\n",
        "        qk = qk.float()\n",
        "        w = torch.nn.functional.softmax(qk, dim=-1).to(q.dtype)\n",
        "        return (w @ v).permute(0, 2, 1, 3).flatten(start_dim=2)\n",
        "\n",
        "\n",
        "class SpeechEmbedding(torch.nn.Module):\n",
        "    def __init__(\n",
        "        self, in_dim: int = 129, hidden_dim: int = 64, kernel: int = 3\n",
        "    ) -> None:\n",
        "        super().__init__()\n",
        "        self._embed = torch.nn.Sequential(\n",
        "            torch.nn.Conv1d(in_dim, hidden_dim, kernel_size=kernel, padding=1),\n",
        "            torch.nn.GELU(),\n",
        "            torch.nn.Conv1d(\n",
        "                hidden_dim, hidden_dim, kernel_size=kernel, stride=2, padding=1\n",
        "            ),\n",
        "            torch.nn.GELU(),\n",
        "        )\n",
        "\n",
        "    def forward(self, x: torch.Tensor) -> torch.Tensor:\n",
        "        \"\"\"\n",
        "        x : torch.Tensor, shape = (bs, freq, time), freq can be n_fft (when converting\n",
        "            audio into spectrogram) or n_mels (when converting audio into mel\n",
        "            spectrogram).\n",
        "        \"\"\"\n",
        "        out = x\n",
        "        out = self._embed(out)\n",
        "        out = out.permute(0, 2, 1)\n",
        "        return out\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2023-12-15T17:36:33.568087Z",
          "iopub.status.busy": "2023-12-15T17:36:33.567715Z",
          "iopub.status.idle": "2023-12-15T17:36:33.614167Z",
          "shell.execute_reply": "2023-12-15T17:36:33.613361Z",
          "shell.execute_reply.started": "2023-12-15T17:36:33.568055Z"
        },
        "id": "VL6vg3hwpqls",
        "trusted": true
      },
      "outputs": [],
      "source": [
        "# hw2/trainer.py\n",
        "\n",
        "import os\n",
        "import pathlib as pl\n",
        "from typing import Callable, Dict, List, Tuple, Union\n",
        "\n",
        "from ml_collections import config_dict\n",
        "import numpy as np\n",
        "import textdistance as td\n",
        "import torch\n",
        "\n",
        "\n",
        "class Trainer:\n",
        "    def __init__(self, config: config_dict.ConfigDict, device: str = \"cpu\"):\n",
        "        self._device = torch.device(device)\n",
        "        self._config = config\n",
        "        self._save_step = np.linspace(0, config.epochs, num=10, dtype=np.int32)\n",
        "        self._model = SpeechToText(**config.speech_to_text)\n",
        "        self._model.to(self._device)\n",
        "        self._opt = torch.optim.AdamW(self._model.parameters(), lr=config.lr)\n",
        "        warm_up_sch = torch.optim.lr_scheduler.LinearLR(\n",
        "            self._opt, 1e-2, 1, int(0.15 * config.epochs)\n",
        "        )\n",
        "        decay_sch = torch.optim.lr_scheduler.LinearLR(\n",
        "            self._opt, 1, 1e-2, int(0.85 * config.epochs)\n",
        "        )\n",
        "        self._opt_sch = torch.optim.lr_scheduler.SequentialLR(\n",
        "            self._opt, [warm_up_sch, decay_sch], [int(0.15 * config.epochs)]\n",
        "        )\n",
        "        self._run_loss = 0.0\n",
        "        self._log_train_step_int = 30\n",
        "        self._prev_log_step = 0\n",
        "\n",
        "        self._text_vec = TokenVectorizer(\n",
        "            **config.data.transforms.text.token_vectorizer\n",
        "        )\n",
        "        self._stop_token_idx = self._text_vec.mapper[self._text_vec._stop]\n",
        "        self._empty_token = self._text_vec._empty\n",
        "        self._str_algos = {\n",
        "            \"jaccard_similarity\": td.jaccard.normalized_similarity,\n",
        "            \"cosine_similarity\": td.cosine.normalized_similarity,\n",
        "            \"damerau-levenshtein_similarity\": td.damerau_levenshtein.normalized_similarity,\n",
        "        }\n",
        "        self._metrics = []\n",
        "        self._storage_path = pl.Path(self._config.storage_folder)\n",
        "        self._storage_path /= str(self._config.seed)\n",
        "        os.makedirs(self._storage_path, exist_ok=True)\n",
        "\n",
        "    def train(self) -> None:\n",
        "        train_dl = build_dl(DataSplit.TRAIN, self._config.data)\n",
        "        val_dl = build_dl(DataSplit.VALIDATION, self._config.data)\n",
        "\n",
        "        for epoch in range(self._config.epochs):\n",
        "            metrics = {}\n",
        "            self._prev_log_step = 0\n",
        "            print(f\"[{epoch + 1}]\\ntrain:\")\n",
        "\n",
        "            for step, data in enumerate(train_dl):\n",
        "                audio, text = data[0].to(self._device), data[1].to(self._device)\n",
        "                train_metrics = self._train_step(audio, text)\n",
        "                metrics = self._upt_metrics(metrics, train_metrics, \"train\")\n",
        "                self._log_train(epoch, step, metrics[\"train\"])\n",
        "\n",
        "            print(\"\\nvalidation:\")\n",
        "\n",
        "            for step, data in enumerate(val_dl):\n",
        "                audio, text = data[0].to(self._device), data[1].to(self._device)\n",
        "                val_metrics, log = self._eval_step(audio, text, step + 1 == len(val_dl))\n",
        "                metrics = self._upt_metrics(metrics, val_metrics, \"val\")\n",
        "\n",
        "            self._log_val(step, metrics[\"val\"], log)\n",
        "            self._opt_sch.step()\n",
        "            self._metrics.append(metrics)\n",
        "\n",
        "            if epoch in self._save_step:\n",
        "                model_file = self._storage_path\n",
        "                model_file /= f\"audio_to_text__{epoch}.pt\"\n",
        "                torch.save(self._model.state_dict(), (model_file))\n",
        "\n",
        "        metrics_file = self._storage_path\n",
        "        metrics_file /= f\"train_metrics__{self._config.epochs}.pkl\"\n",
        "        save_obj(self._metrics, metrics_file)\n",
        "\n",
        "    def test(self) -> None:\n",
        "        test_dl = build_dl(DataSplit.TEST, self._config.data)\n",
        "        metrics = {}\n",
        "        log = []\n",
        "\n",
        "        for data in test_dl:\n",
        "            audio, text = data[0].to(self._device), data[1].to(self._device)\n",
        "            batch_metrics, batch_log = self._eval_step(audio, text, True)\n",
        "            metrics = self._upt_metrics(metrics, batch_metrics, \"test\")\n",
        "            log += batch_log\n",
        "\n",
        "        metrics_file = self._storage_path\n",
        "        metrics_file /= f\"test_metrics__{self._config.epochs}.pkl\"\n",
        "        save_obj(metrics, metrics_file)\n",
        "\n",
        "        log_file = self._storage_path\n",
        "        log_file /= f\"test_log__{self._config.epochs}.txt\"\n",
        "        save_txt(\"\".join(log), log_file)\n",
        "\n",
        "    def _train_step(self, audio: torch.Tensor, text: torch.Tensor) -> float:\n",
        "        self._opt.zero_grad()\n",
        "        loss = self._fwd_step(audio, text)\n",
        "        loss.backward()\n",
        "        self._opt.step()\n",
        "        return {\"loss\": loss.item()}\n",
        "\n",
        "    @torch.no_grad()\n",
        "    def _eval_step(\n",
        "        self, audio: torch.Tensor, text: torch.Tensor, log_text: bool = False\n",
        "    ) -> Tuple[Dict[str, float], str]:\n",
        "        loss = self._fwd_step(audio, text).item()\n",
        "        gen_text = self._model.generate(audio)\n",
        "        text, gen_text = text.cpu().numpy(), gen_text.cpu().numpy()\n",
        "        metrics = self._str_sim_algos(self._str_algos, text, gen_text)\n",
        "        metrics[\"loss\"] = loss\n",
        "        metrics[\"n\"] = text.shape[0]\n",
        "\n",
        "        if not log_text:\n",
        "            return metrics, \"\"\n",
        "\n",
        "        return metrics, self._cmp_target_w_gen(text, gen_text)\n",
        "\n",
        "    def _fwd_step(self, audio: torch.Tensor, text: torch.Tensor) -> torch.Tensor:\n",
        "        target = text[:, 1:]\n",
        "        logits = self._model(audio, text[:, :-1])\n",
        "        loss = torch.nn.functional.cross_entropy(logits, target, label_smoothing=0.1)\n",
        "        return loss\n",
        "\n",
        "    def _create_dataset(self, split: DataSplit):\n",
        "        return\n",
        "\n",
        "    def _log_train(self, epoch: int, step: int, metrics: Dict[str, float]) -> None:\n",
        "        if step % self._log_train_step_int != self._log_train_step_int - 1:\n",
        "            return\n",
        "\n",
        "        print(f\"[{epoch + 1}, {step + 1:4d}]\", end=\" \")\n",
        "        n = step - self._prev_log_step + 1\n",
        "\n",
        "        for m, v in metrics.items():\n",
        "            print(f\"{m}: {sum(v[self._prev_log_step:]) / n:.3f}\")\n",
        "            self._prev_log_step = step + 1\n",
        "\n",
        "    def _log_val(\n",
        "        self, step: int, metrics: Dict[str, Union[float, str]], log: str\n",
        "    ) -> None:\n",
        "        n = sum(metrics[\"n\"])\n",
        "\n",
        "        for k, v in metrics.items():\n",
        "            if k == \"n\":\n",
        "                continue\n",
        "\n",
        "            print(k)\n",
        "            v = sum(v)\n",
        "            v = v / n if k != \"loss\" else v / len(metrics[\"n\"])  # loss is already avg.\n",
        "            print(f\"{v:.3f}\", end=\"\\n\\n\")\n",
        "\n",
        "        print(\"compare-text-gen:\")\n",
        "        print(log, end=\"\\n\\n\")\n",
        "\n",
        "    def _upt_metrics(\n",
        "        self, base: Dict[str, float], upt: Dict[str, float], phase: str\n",
        "    ) -> Dict[str, float]:\n",
        "        base_phase = base.get(phase, {})\n",
        "\n",
        "        for k in upt:\n",
        "            if k not in base_phase:\n",
        "                base_phase[k] = []\n",
        "\n",
        "            base_phase[k].append(upt[k])\n",
        "\n",
        "        base[phase] = base_phase\n",
        "        return base\n",
        "\n",
        "    def _str_sim_algos(\n",
        "        self, str_algos: Dict[str, Callable], target: np.ndarray, gen: np.ndarray\n",
        "    ) -> Dict[str, float]:\n",
        "        metrics = {k: 0 for k in str_algos}\n",
        "\n",
        "        for t, g in zip(target, gen):\n",
        "            t, g = self._preprocess_text(t, g)\n",
        "\n",
        "            for k, algo_fn in str_algos.items():\n",
        "                metrics[k] += algo_fn(t, g)\n",
        "\n",
        "        return metrics\n",
        "\n",
        "    def _cmp_target_w_gen(self, target: np.ndarray, gen: np.ndarray) -> str:\n",
        "        cmp = []\n",
        "\n",
        "        for t, g in zip(target, gen):\n",
        "            t = \"\".join([self._text_vec.invert_mapper[c] for c in t])\n",
        "            t = t.replace(\"@\", \"\")\n",
        "            last_idx = np.where(g == 1)[0]\n",
        "            last_idx = last_idx[0] + 1 if last_idx.size > 0 else g.shape[0]\n",
        "            g = g[:last_idx]\n",
        "            g = \"\".join([self._text_vec.invert_mapper[c] for c in g])\n",
        "            cmp += [\"target: \", t, \"\\n\", \"pred:   \", g, \"\\n\\n\"]\n",
        "\n",
        "        return \"\".join(cmp)\n",
        "\n",
        "    def _preprocess_text(\n",
        "        self, target: np.ndarray, gen: np.ndarray\n",
        "    ) -> Tuple[List[str], List[str]]:\n",
        "        last_idx = np.where((target == self._stop_token_idx))[0][0]\n",
        "        target = target[: last_idx + 1]\n",
        "        last_idx = np.where(gen == 1)[0]\n",
        "        last_idx = last_idx[0] if last_idx.size > 0 else len(gen) - 1\n",
        "        gen[last_idx] = 1\n",
        "        gen = gen[: last_idx + 1]\n",
        "        return target.tolist(), gen.tolist()\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2023-12-15T17:36:33.615843Z",
          "iopub.status.busy": "2023-12-15T17:36:33.615467Z",
          "iopub.status.idle": "2023-12-15T17:36:33.624513Z",
          "shell.execute_reply": "2023-12-15T17:36:33.623592Z",
          "shell.execute_reply.started": "2023-12-15T17:36:33.615809Z"
        },
        "id": "XRA52YInum_K",
        "trusted": true
      },
      "outputs": [],
      "source": [
        "# main.py\n",
        "\n",
        "import random\n",
        "\n",
        "\n",
        "def plotting(config):\n",
        "    epochs, seed = config.epochs, config.seed\n",
        "    metrics_path = pl.Path(config.storage_folder) / str(seed)\n",
        "    plot_path = metrics_path / \"plots\"\n",
        "    os.makedirs(plot_path, exist_ok=True)\n",
        "\n",
        "    labels = {\"train\": \"train\", \"val\": \"validation\"}\n",
        "    train_metrics_path = metrics_path / f\"train_metrics__{epochs}.pkl\"\n",
        "    train_metrics = load_metrics(train_metrics_path)\n",
        "    plot_metrics(train_metrics, plot_path, \"train\")\n",
        "    plot_metrics(train_metrics, plot_path, \"val\")\n",
        "    cmp_phase_metrics(train_metrics, \"loss\", [\"train\", \"val\"], plot_path, labels)\n",
        "\n",
        "    test_metrics_path = metrics_path / f\"test_metrics__{epochs}.pkl\"\n",
        "    test_metrics = load_metrics(test_metrics_path)[\"test\"]\n",
        "    mean_test_metrics = {}\n",
        "\n",
        "    for k, v in test_metrics.items():\n",
        "        if k == \"n\":\n",
        "            continue\n",
        "\n",
        "        if k == \"loss\":\n",
        "            mean_test_metrics[k] = sum(v) / len(test_metrics[\"n\"])\n",
        "        else:\n",
        "            mean_test_metrics[k] = sum(v) / sum(test_metrics[\"n\"])\n",
        "\n",
        "    print(\"test:\")\n",
        "    print(mean_test_metrics)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jiS3_B5jum_L"
      },
      "source": [
        "## (a)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "execution": {
          "iopub.execute_input": "2023-12-15T17:36:44.651096Z",
          "iopub.status.busy": "2023-12-15T17:36:44.650733Z",
          "iopub.status.idle": "2023-12-15T17:59:45.904767Z",
          "shell.execute_reply": "2023-12-15T17:59:45.903736Z",
          "shell.execute_reply.started": "2023-12-15T17:36:44.651067Z"
        },
        "id": "QbCwFdthum_M",
        "outputId": "121f6958-7d16-4094-9fe1-eecc9e6e22a5",
        "trusted": true
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Using cuda\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "100%|██████████| 2.56G/2.56G [00:27<00:00, 101MB/s]\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "[1]\n",
            "train:\n",
            "[1,   30] loss: 3.303\n",
            "[1,   60] loss: 2.987\n",
            "[1,   90] loss: 2.823\n",
            "[1,  120] loss: 2.725\n",
            "[1,  150] loss: 2.653\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.234\n",
            "\n",
            "cosine_similarity\n",
            "0.379\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.257\n",
            "\n",
            "loss\n",
            "2.589\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <e the the the the the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <e the the the the the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <e the the the the the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <e the the the the the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <e the the the the the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <e the the the the the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <e the the the the the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <e the the the the the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <e the the the the the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <e the the the the the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <e the the the the the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <e the the the the the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <e the the the the the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <e the the the the the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <e the the the the the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "\n",
            "\n",
            "[2]\n",
            "train:\n",
            "[2,   30] loss: 2.339\n",
            "[2,   60] loss: 2.181\n",
            "[2,   90] loss: 2.085\n",
            "[2,  120] loss: 2.037\n",
            "[2,  150] loss: 1.991\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.435\n",
            "\n",
            "cosine_similarity\n",
            "0.608\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.261\n",
            "\n",
            "loss\n",
            "1.966\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <the prossion of the prossion of the prossion of the prossion of the prossion of the prossion of t>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <the prossion of the prossion of the prossion of the prossion of the prossion of the prossion of t>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <the prosided the prosided the prosided the prosided the prosided the prosided the prosided the pr>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <the prosided the prosided the prosided the prosided the prosided the prosided the prosided the pr>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <the prosided the prosided the prosided the prosided the prosided the prosided the prosided the pr>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <the prossion the prossion the prossion the prossion the prossion the prossion the prossion the pr>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <the prosided the prosided the prosided the prosided the prosided the prosided the prosided the pr>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <the prosided the prosided the prosided the prosided the prosided the prosided the prosided the pr>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <the prossion the prisons of the prisons of the prisons of the prisons of the prisons of the priso>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <the prossion of the prossion of the prossion of the prossion of the prossion of the prossion of t>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <the prison the prison the prison the prison the prison the prison the prison the prison the priso>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <the prossion the prossion the prossion the prossion the prossion the prossion the prossion the pr>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <the prosided the prosided the prosided the prosided the prosided the prosided the prosided the pr>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <the prossion the prossion the prossion the prossion the prossion the prossion the prossion the pr>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <the prossided the prossided the prossided the prossided the prossided the prossided the prossided>\n",
            "\n",
            "\n",
            "\n",
            "[3]\n",
            "train:\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/torch/optim/lr_scheduler.py:149: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.\n",
            "  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "[3,   30] loss: 1.956\n",
            "[3,   60] loss: 1.912\n",
            "[3,   90] loss: 1.881\n",
            "[3,  120] loss: 1.845\n",
            "[3,  150] loss: 1.818\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.408\n",
            "\n",
            "cosine_similarity\n",
            "0.577\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.266\n",
            "\n",
            "loss\n",
            "1.777\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he could the second the could the second the could the second the could the second the could the >\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <and the could the could the could the could the could the could the could the could the could the>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <on the committed to the committed to the committed to the committed to the committed to the commi>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <when the could to the could to the could to the could to the could to the could to the could to t>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and the continue of the constinue of the constinue of the constinue of the constinue of the const>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <in the could the could the could the could the could the could the could the could the could the >\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <she was a contried to the contried to the contried to the contried to the contried to the contrie>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he was a contried to the contried to the contried to the contried to the contried to the contried>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <when second the second the second the second the second the second the second the second the seco>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <the could to the could to the could to the could to the could to the could to the could to the co>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he could the continual and the continual and the continual and the continual and the continual an>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he second the second the second the second the second the second the second the second the second>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <and the contried to the contried to the contried to the contried to the contried to the contried >\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the second the second the second the second the second the second the second the second the se>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <when the constinue of the constinue of the constinue of the constinue of the constinue of the con>\n",
            "\n",
            "\n",
            "\n",
            "[4]\n",
            "train:\n",
            "[4,   30] loss: 1.782\n",
            "[4,   60] loss: 1.777\n",
            "[4,   90] loss: 1.760\n",
            "[4,  120] loss: 1.737\n",
            "[4,  150] loss: 1.730\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.497\n",
            "\n",
            "cosine_similarity\n",
            "0.663\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.269\n",
            "\n",
            "loss\n",
            "1.699\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he was a constitution of the consideral and the consideral and the consideral and the consideral >\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <and the prison of the prison of the prison of the prison of the prison of the prison of the priso>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <who was a law and a law and a law and a law and a law and a law and a law and a law and a law and>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <who was a consideral and the prisoners of the prisoners of the prisoners of the prisoners of the >\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and the consideral and a constitution of the prisoners of the prisoners of the prisoners of the p>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his were the prisoners of the prisoners of the prisoners of the prisoners of the prisoners of the>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <some the constitution of the prisoners of the prisoners of the prisoners of the prisoners of the >\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he was a prisoners of the prisoners of the prisoners of the prisoners of the prisoners of the pri>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he was a states of the prisoners of the prisoners of the prisoners of the prisoners of the >\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there were and the prisoners of the prisoners of the prisoners of the prisoners of the prisoners >\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he had been and the continual and the continual and the continual and the continual and the conti>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he was a stated the consideral and the consideral and the consideral and the consideral and the c>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <a conten and the prisoners of the prisoners of the prisoners of the prisoners of the prisoners of>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the consideration of the prisoners of the prisoners of the prisoners of the prisoners of the p>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which he was a constitution of the prisoners of the prisoners of the prisoners of the prisoners o>\n",
            "\n",
            "\n",
            "\n",
            "[5]\n",
            "train:\n",
            "[5,   30] loss: 1.705\n",
            "[5,   60] loss: 1.691\n",
            "[5,   90] loss: 1.686\n",
            "[5,  120] loss: 1.677\n",
            "[5,  150] loss: 1.670\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.462\n",
            "\n",
            "cosine_similarity\n",
            "0.631\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.281\n",
            "\n",
            "loss\n",
            "1.648\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he had been the common of the common of the common of the common of the common of the common of t>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the president of the president of the president of the president of the president of the presi>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <more of the law of the long of the law of the long of the law of the long of the law of the long >\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <who was a long the common of the president of the president of the president of the president of >\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and a president of the president of the president of the president of the president of the presid>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his was a complete of the president of the president of the president of the president of the pre>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <side the common of the president of the president of the president of the president of the presid>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had been and the common of the president of the president of the president of the president of>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he had been second and the commons of the street of the street of the street of the street >\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there was a common of the president of the president of the president of the president of the pre>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he could had been the character of the character of the character of the character of the charact>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he said the shoots of the shoots of the shoots of the shoots of the shoots of the shoots of the s>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <on the common of the president of the president of the president of the president of the presiden>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the most of the president of the president of the president of the president of the president >\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which he had been second the common of the common of the common of the common of the common of th>\n",
            "\n",
            "\n",
            "\n",
            "[6]\n",
            "train:\n",
            "[6,   30] loss: 1.646\n",
            "[6,   60] loss: 1.638\n",
            "[6,   90] loss: 1.640\n",
            "[6,  120] loss: 1.620\n",
            "[6,  150] loss: 1.616\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.479\n",
            "\n",
            "cosine_similarity\n",
            "0.649\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.293\n",
            "\n",
            "loss\n",
            "1.599\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he chained the state of the state of the state of the state of the state of the state of the stat>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the same of the prison of the prison of the prison of the prison of the prison of the prison o>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <more not all the all one of the police of the police of the police of the police of the police of>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <while had been in the prisoners were in the prisoners were in the prisoners were in the prisoners>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and the prisoners of the second and a second and a second and a second and a second and a second >\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his was a printers of the prisoners of the prisoners of the prisoners of the prisoners of the pri>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <seven the contain the prisoners were a printed that the prisoners were a printed that the prisone>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had a rear and a rear of the prisoners of the prisoners of the prisoners of the prisoners of t>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he said the same of the same of the same of the same of the same of the same of the same of>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there was southead the prison of the prison of the prison of the prison of the prison of the pris>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he defense the change of the shots of the change of the shots of the change of the shots of the c>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he sassembled the shots of the shoots of the shoots of the shoots of the shoots of the shoots of >\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <on the police of the prison of the prison of the prison of the prison of the prison of the prison>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the morning of the prisoners of the prisoners of the prisoners of the prisoners of the prisone>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which he southeaded the prisoners of the prisoners of the prisoners of the prisoners of the priso>\n",
            "\n",
            "\n",
            "\n",
            "[7]\n",
            "train:\n",
            "[7,   30] loss: 1.587\n",
            "[7,   60] loss: 1.581\n",
            "[7,   90] loss: 1.571\n",
            "[7,  120] loss: 1.574\n",
            "[7,  150] loss: 1.548\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.542\n",
            "\n",
            "cosine_similarity\n",
            "0.703\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.310\n",
            "\n",
            "loss\n",
            "1.557\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he he was to the rest were from the front of the first of the record of the record of the record >\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the same of the first was recorded the rest recorded the resolvent was a newgate the ward and >\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <more not leave the last the consideral new were not were not were were not were were the warder w>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <while was relieved with his were more the most of the morning of the morning of the morning of th>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and her her had seen so really seen a second and a second and a second and a second and a second >\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his was was a point was the point was a recorded with a record was a recorded with a record was a>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <so dead did a rear was recorded the condemned by the condition of the condition of the condition >\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had recomment of the rest of the rest were not be made the read with the read with the read wi>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he streets the president in a scring the same of the same of the most of the most of the mo>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there were some of the complete of the remain the consideral consideral was not the consideral wa>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he defended to the distinction of newgate not wall was he was the walls were the walls were the w>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he assassinated the president was not shot was the rest was resent was resent was resent was rese>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald was a recorded with rear was not was not the ward was not the way was a man was a man was >\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the morning in his constitution of constant of the consideration of the consideration of the c>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which was a second the same the same the survice was a recorded and was a recorded and was a reco>\n",
            "\n",
            "\n",
            "\n",
            "[8]\n",
            "train:\n",
            "[8,   30] loss: 1.519\n",
            "[8,   60] loss: 1.494\n",
            "[8,   90] loss: 1.493\n",
            "[8,  120] loss: 1.487\n",
            "[8,  150] loss: 1.457\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.604\n",
            "\n",
            "cosine_similarity\n",
            "0.753\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.349\n",
            "\n",
            "loss\n",
            "1.461\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he cases to his were from from form of the first saw to the contracted to the car to the car to t>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the same of the first for his last of a considerable and called in the came of the work of the>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <moramage oswald the called the house had been in the borough of one of the moment of more more mo>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <while hospital comparison to the most of morning many of many of his printed he said the man he s>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and a particular surplain place of classificate of a man of the man of the man of the man of the >\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his was to contain that he came had part of his saw of march of his came of his came of his came >\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <so dead dead was arrested by the part of the part of building of printer printed.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had recorded his took to be discovered to be arraigned to be carried to be came to be came to >\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he case to strange in a surganism and the convict of the last of more for more for more for>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there was all of the palmpt of the latter of the contain who could carried the possibilities of t>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he differed to the desiate in newgate, showed at the chaplain, and the character of the forgery o>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he sassinated the prisoner of the shot he should be shot to be sentenced to reside to rester to r>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald harrary arrival recovery of the commission of his offense of the fine of the fine of the f>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the morning he had had strange constantly convicted of the consideration of the morning man of>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which was sentenced to state of the state is of a record, and could could for the forger, and for>\n",
            "\n",
            "\n",
            "\n",
            "[9]\n",
            "train:\n",
            "[9,   30] loss: 1.417\n",
            "[9,   60] loss: 1.409\n",
            "[9,   90] loss: 1.391\n",
            "[9,  120] loss: 1.383\n",
            "[9,  150] loss: 1.379\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.630\n",
            "\n",
            "cosine_similarity\n",
            "0.772\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.380\n",
            "\n",
            "loss\n",
            "1.385\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he case to his were from from the friends to be first for the time to the return to the reserve t>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the same him to have for forged from his law of the house of the house of the work in the work>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <more not oswald flooked the hough he with he had been in a book accomple, and moment of home hous>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <while also more people to most to most in november has provided him he same seventeen sixty-three>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and a partial of surplay as which local public and man of the man of the same and sentenced and t>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his was a time his that he companied him him with him with his ready readers.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <socan dead as a great were pounded the pounds of but in a public provident provided.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had a restic to the continue of body of body and her had been he had been for higher.>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he has present in string and so down and the body of down by the same for more for more for>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there were some clothely that the hand, and the hands with the power betters had been for the pos>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he differed to the doom, should not be a house, should not one of the world should be traced the >\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he assassinated the prison the provision should be surving to reside, as the rise to rester as th>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald point her the strong regard of not be one of his offense of his offender.>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the mory he had had not string conviction of the convicts of the rifle of march master sentenc>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which was succeeded to state, as he had a river for great of a great of the forger, but four so f>\n",
            "\n",
            "\n",
            "\n",
            "[10]\n",
            "train:\n",
            "[10,   30] loss: 1.325\n",
            "[10,   60] loss: 1.323\n",
            "[10,   90] loss: 1.315\n",
            "[10,  120] loss: 1.321\n",
            "[10,  150] loss: 1.297\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.661\n",
            "\n",
            "cosine_similarity\n",
            "0.794\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.416\n",
            "\n",
            "loss\n",
            "1.320\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he accessed to his were from the freen of the state and fall to hard to the terres of the are of >\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the same of him it for for a respondent and that he can that he called the ward water.>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <mory nine oswald flooked with he wated he had been a work of one of a moment of work and would an>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <wall oswald he came to most to the news in newgate in his promise promise and so made the south o>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and a paralia, sure as a place of clear contain, on the man of an attend at the assassinate at th>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his was a find his that he compared him a confession mr. reachester.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <six as a great grand were appeared the comparation by no considerations of president.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had a recognized to a good in his inches were at a half day be captain before.>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he case the ascracted and insue and the convicted and down at long for a long as near of ne>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there were some called at the about that he left that he was had not the paractice of carried for>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he differed to the soon yeay again, should already all one all trade, to tried the good been supp>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he assassinated the president#s should have she as a sir resident, were resisted at a rifle and a>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald had been a starter in his own about in washing was about to back to feet.>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the mory and he asked and conveyed conveyed as to the crime of the crime of mr. marked by the >\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which was susuated to scrief as a great it i could a great and could no all the found as >\n",
            "\n",
            "\n",
            "\n",
            "[11]\n",
            "train:\n",
            "[11,   30] loss: 1.258\n",
            "[11,   60] loss: 1.247\n",
            "[11,   90] loss: 1.254\n",
            "[11,  120] loss: 1.247\n",
            "[11,  150] loss: 1.244\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.670\n",
            "\n",
            "cosine_similarity\n",
            "0.801\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.439\n",
            "\n",
            "loss\n",
            "1.296\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he chains to his way from from the other of the salt, and to her to her to the return to the area>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the same time through for still for her resent and that he cased in a wood along the woode.>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <more not also fall the heath he with he had not be a work of our common of a moorning and provide>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <while oswald here commosted to those to him immediately carled him he had a some he southeast and>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and a paralia, sir as least a public could came of a man of amendment and the satisfacture of he >\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his was effect his decoved that he comparily confirmed mr. reachestries of his drivers.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <socan deant of great grand were pare the comparily company in a changes of praction.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had a great hostoned to a good disasing by the borough have have been capital.>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he access the strange in socient and so violent and down at a long blood blood blood in a s>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there were called along the babylon, that lettenting at the pant which had been paralias he was a>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he diffed to the soon to the aimous in him, shown all one of all own to a trial with a good by th>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he assassinated the prison the president chaptain to sir resided, written as it as he as him at h>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald good part staried ready of his objecting his own his about twenty-five teasure.>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the morining he had added as going conviction of the conviction of the commission as a marked >\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which was so seemed to scries had as a great i court and could and could hospital as a so as the >\n",
            "\n",
            "\n",
            "\n",
            "[12]\n",
            "train:\n",
            "[12,   30] loss: 1.206\n",
            "[12,   60] loss: 1.198\n",
            "[12,   90] loss: 1.188\n",
            "[12,  120] loss: 1.192\n",
            "[12,  150] loss: 1.186\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.686\n",
            "\n",
            "cosine_similarity\n",
            "0.812\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.465\n",
            "\n",
            "loss\n",
            "1.233\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he accessive this were frenchering of the stoom into self-side and to the territy of the artistin>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the same time the press were for a receive and executed that in oswald will would have from th>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <more not all the flooked the other half being were being would any one of a momore wood accept an>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <while oswald impertant to hospitable immediate as a marsous and he special as a some man to man t>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and a paralia, sir as lee as a public crime, on them of a man of a federal at one head were seen >\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his which has defined that he companied him self which was a reading as regards.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <socian dealing as a great were part were possibly in a crimes of a craction.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had a great hostonest or to his invictive by a half were account day.>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he accept in strange instructing and hanges of a vice more access of long in the assassinat>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there were calculated at the other hand, at the lettenting and a passerant had debtors of a calcr>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he diffed to the soon to the adewity you and should already or the ward to go a good by the occup>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he assassintated the president#s should she officered to sir resident as it as the ordinal of a f>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald have pared the starter went of him in which one in his about eight twenty-fifty.>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the moring he had addendence to conviction as to the criction of the master nists, actual cris>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which was so seemed to scrief of a great i could a great and could as a so own as the fire for th>\n",
            "\n",
            "\n",
            "\n",
            "[13]\n",
            "train:\n",
            "[13,   30] loss: 1.150\n",
            "[13,   60] loss: 1.150\n",
            "[13,   90] loss: 1.145\n",
            "[13,  120] loss: 1.143\n",
            "[13,  150] loss: 1.148\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.699\n",
            "\n",
            "cosine_similarity\n",
            "0.821\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.490\n",
            "\n",
            "loss\n",
            "1.204\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he access to his way from the french of a stood into self-hored to the souther to the you>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the same time the press to love for a runner, and a seventh one of a wallant.>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <more none oswald faughtlees he went he not be an inner occupate, on the momore process, and accor>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <while oswald appeared to most to hospitan impened as a handless present on the some man to man to>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and a parliame six was a public could crime of griffith, and the man of eighteen thirty-eight >\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his wife a tast of the head companies of for a point mr. records of his dread.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <socan deant of grenacraw were pare the company company, by no examinent.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had acraigned hose of to order is in body are a have been to be admitted.>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he accept in strange insure and so viewed a view and down blow for the forgery shall nation>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there were called a county of the legislative and with an appearant had been part for for a coupl>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he diffed did to the so the so of any young and one of all one to a good with the chimney, and we>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he assassing the president was should officer officer better resident arrised, and the frauds, en>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald had pared it starged we have been out in his own his about feet to east the feet.>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the morings, he added in a conviction of a conviction of the president more common sixty-three>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which one of suscipate described a who appointed and of our and could board, but as the alread fo>\n",
            "\n",
            "\n",
            "\n",
            "[14]\n",
            "train:\n",
            "[14,   30] loss: 1.124\n",
            "[14,   60] loss: 1.114\n",
            "[14,   90] loss: 1.108\n",
            "[14,  120] loss: 1.110\n",
            "[14,  150] loss: 1.111\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.707\n",
            "\n",
            "cosine_similarity\n",
            "0.827\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.506\n",
            "\n",
            "loss\n",
            "1.182\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he accessive the stafford from the unirons to himself, and to her to her terreted the southeasmin>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the same time the press loft of love progre, and a seventh can whole one the wallow with the o>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <mory not all the flooked the other being whether beyond were in in or move one on the movements, >\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <while oswald appeared to most to those in newgate as a marse parts approximately some times in th>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and a paralia, sir as leas of public could crime, on the man when said, after he to >\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his wife a tastified that he competer himself with a great receivers of his dread.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <such again deaso grand her were pare the company by no examiness of pression.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had a great hostone of to ordinary of body are a have have been contined.>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he accept in strange in so ensoe as a violent accepts, long for the forgeries national same>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there were called a wall the other that them had lettenting at the bank who had been part for a c>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he diffed did to the so the so of the enemy, showed on walked to a good with the chemical be wind>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he assassinated the president chaptain should sheriffs hit be sixteen arristics, end quote, excap>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald garraged starge reading of common in his own his bate feet to easily.>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the morings, he added and in conviction accommodation as the creating more common sixteen proc>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which one associated to scries accorded in a great wall, and gost the wall so as the aleran>\n",
            "\n",
            "\n",
            "\n",
            "[15]\n",
            "train:\n",
            "[15,   30] loss: 1.098\n",
            "[15,   60] loss: 1.093\n",
            "[15,   90] loss: 1.092\n",
            "[15,  120] loss: 1.091\n",
            "[15,  150] loss: 1.092\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.705\n",
            "\n",
            "cosine_similarity\n",
            "0.825\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.513\n",
            "\n",
            "loss\n",
            "1.174\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he accessive his wall from the friends to himself, and to have tell her to the southeasmove in yo>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the same time the press lofter of a brink, and a seventh in one of the wallow with a wall.>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <more notice was all the other the numpound he were been in in or one of a momore wood process, en>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <while oswald appeared to most to those in newgate as a handlesperate he sound some times in the m>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and a paralia, sir as leas of public crime of crime, and the sand that he to heavy years when he >\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his wife a tastified that he competered self with a great receivers of his dread.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <such again deaso grand her were pare the company in a new press of pression.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had a great hostoned to ordinary his inviroyed, where appeared him.>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he accept in strange in essagilant and have down at a long blow for the forgeries national >\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there were called a wall the other had that the lettenting at the bank who had been part for othe>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he diffed to the soon the so vany you, should and one walked to a good with the children by the >\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he assassinated the president#s shot of a shot officer to sir resided, and the state of hich in a>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald had particulary reading about in eighteen his own his bate feet to the feet.>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the morings, he added and in violent conviction as to the president more common sixteen prison>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which one associated to scries accorded in a governor of which book as a so own as the arrench th>\n",
            "\n",
            "\n",
            "\n"
          ]
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "test:\n",
            "{'jaccard_similarity': 0.714888785108462, 'cosine_similarity': 0.8323868545583469, 'damerau-levenshtein_similarity': 0.5086983165244969, 'loss': 1.1828280331158056}\n"
          ]
        }
      ],
      "source": [
        "if __name__ == \"__main__\":\n",
        "    # not using absl (no need for command line), get config manually\n",
        "    config = get_config()\n",
        "    config.unlock()  # update hyperparams before `config.lock()`\n",
        "    config.storage_folder = \"./storage/attention-rnn\"\n",
        "    config.speech_to_text.decoder_type = DecoderType.RNN\n",
        "    config.speech_to_text.decoder_kwargs = dict(\n",
        "        n_layers=1,\n",
        "        freq_dim=config.get_ref(\"n_vocab\"),\n",
        "        time_dim=config.get_ref(\"text_len\"),\n",
        "        hidden_dim=200,\n",
        "        n_heads=2,\n",
        "        feed_fwd_dim=400\n",
        "    )\n",
        "    config.lock()\n",
        "\n",
        "    # Manual seed\n",
        "    torch.manual_seed(config.seed)\n",
        "    np.random.seed(config.seed)\n",
        "    random.seed(config.seed)\n",
        "\n",
        "    # Device to use\n",
        "    DEVICE = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n",
        "    print(f\"Using {DEVICE}\")\n",
        "\n",
        "    # Training & testing\n",
        "    trainer = Trainer(config, DEVICE)\n",
        "    trainer.train()\n",
        "    trainer.test()\n",
        "\n",
        "    # Plotting - check './storage/attention-rnn/{seed}/plots\n",
        "    plotting(config)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yOSwaho2um_M"
      },
      "source": [
        "## (b)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "execution": {
          "iopub.execute_input": "2023-12-15T18:01:22.385134Z",
          "iopub.status.busy": "2023-12-15T18:01:22.384568Z",
          "iopub.status.idle": "2023-12-15T18:16:50.079656Z",
          "shell.execute_reply": "2023-12-15T18:16:50.078607Z",
          "shell.execute_reply.started": "2023-12-15T18:01:22.385099Z"
        },
        "id": "nwz_4GeR1GLN",
        "outputId": "60f461ef-7f4c-4758-9d1c-76efe2ce1afe",
        "trusted": true
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Using cuda\n",
            "[1]\n",
            "train:\n",
            "[1,   30] loss: 3.629\n",
            "[1,   60] loss: 3.439\n",
            "[1,   90] loss: 3.277\n",
            "[1,  120] loss: 3.147\n",
            "[1,  150] loss: 3.019\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.132\n",
            "\n",
            "cosine_similarity\n",
            "0.235\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.174\n",
            "\n",
            "loss\n",
            "2.936\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <e o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <e o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <e o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <e o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <e o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <e o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <e o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <e o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <e o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <e o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <e o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <e o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <e o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <e o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <e o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o>\n",
            "\n",
            "\n",
            "\n",
            "[2]\n",
            "train:\n",
            "[2,   30] loss: 2.593\n",
            "[2,   60] loss: 2.413\n",
            "[2,   90] loss: 2.379\n",
            "[2,  120] loss: 2.360\n",
            "[2,  150] loss: 2.350\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.234\n",
            "\n",
            "cosine_similarity\n",
            "0.379\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.261\n",
            "\n",
            "loss\n",
            "2.330\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <the the the the t the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <the the the the the the the the the the the the the the the the the the the the the the the the t>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <the the the the the the the the the the the the the the the the the the the the the the the the t>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <the the the the the the the the the the the the the the the the the the the the the the the the t>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <the the the the t the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <the the the the the the the the the the the the the the the the the the the the the the the the t>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <the the the the the the the the the the the the the the the the the the the the the the the the t>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <the the the the the the the the the the the the the the the the the the the the the the the the t>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <the the the the t the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <the the the the the the the the the the the the the the the the the the the the the the the the t>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <the the the the t the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <the the the the t the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <the the the the the the the the the the the the the the the the the the the the the the the the t>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <the the the the t the the the the the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <the the the the the the the the the the the the the the the the the the the the the the the the t>\n",
            "\n",
            "\n",
            "\n",
            "[3]\n",
            "train:\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/torch/optim/lr_scheduler.py:149: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.\n",
            "  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "[3,   30] loss: 2.334\n",
            "[3,   60] loss: 2.325\n",
            "[3,   90] loss: 2.329\n",
            "[3,  120] loss: 2.316\n",
            "[3,  150] loss: 2.297\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.234\n",
            "\n",
            "cosine_similarity\n",
            "0.379\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.262\n",
            "\n",
            "loss\n",
            "2.297\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <the the the the the the the the t the the t the the the the the t the the the t the the the the t>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <the the the the the the the the t the the t the the the the the t the t the the the the the the t>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <the the the the the the the the t the the t the the the the the t the the the t the the the the t>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <the the the the the the the the t the the t the the the the the t the the the t the the the the t>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <the the the the the the the the t the the t the the the the the t the t the the the the the the t>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <the the the the the the the the t the the t the t the the the the t t the the t the the the the t>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <the the the the the the the the t the the t the the the the the t the t the the the the the the t>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <the the the the the the the the t the the t the the the the the t the t the the the the the the t>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <the the the the the the the the t the the t the the the the the t the t the the the the the the t>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <the the the the the the the the t the the t the the the the the t the t the the the the the the t>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <the the the the the the the the t the the t the the the the the t the t the the the the the the t>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <the the the the the the the the t the the t the the the the the t the t the the the the the the t>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <the the the the the the the the t the the t the the the the the t the t the the the the the the t>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <the the the the the the the the t the the t the the the the the t the t the the the the the the t>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <the the the the the the the the t the the t the the the the the t the the the t the the the the t>\n",
            "\n",
            "\n",
            "\n",
            "[4]\n",
            "train:\n",
            "[4,   30] loss: 2.314\n",
            "[4,   60] loss: 2.263\n",
            "[4,   90] loss: 2.269\n",
            "[4,  120] loss: 2.236\n",
            "[4,  150] loss: 2.203\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.394\n",
            "\n",
            "cosine_similarity\n",
            "0.562\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.288\n",
            "\n",
            "loss\n",
            "2.159\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <it he the the the the was the was the the the the the are the ad the ar the are the ade the adere>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <and secente the the the the was was ound the the as the the the the the the the the wade the the >\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <whe was os of the the the was alle and and and ard the was alll ond ond and and of and and therer>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <whos was of was of the was the the the as ad and as and and a a as a a and as as a and a a as a a>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and the the was the se seat seand of the ad ad the the ad and ad athe and and the and athe and ad>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <is and the the the the the the the the the the the the the the the the the t the the the the the >\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <she cone the the was ard ward and the the the the the the the the the the the the the the the the>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <in the the was the was was ound the the the the are the the the the the the the the the the the t>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <whis the seres the se the ad ad ad ad as and and a the and as and and and aste the and and as all>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <the rous wall the the the the was the the ad the ad the the the the the the the the wase the the >\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <it he the the the sead the the the the the the the the the and wad the the the the the t the t th>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <is and the as and the the the the sead sthe se the the the ase the ase ad the the ad the the the >\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <as allle the the the cone the ar and the ad ad the the the the the the the the the the the t t th>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the the was the the sead seand the the se the ad the the ad the the the the the the the t t th>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <the the secon se the se the the ad the was ad the as ard as ad was as and as as and and and and a>\n",
            "\n",
            "\n",
            "\n",
            "[5]\n",
            "train:\n",
            "[5,   30] loss: 2.141\n",
            "[5,   60] loss: 2.103\n",
            "[5,   90] loss: 2.057\n",
            "[5,  120] loss: 1.987\n",
            "[5,  150] loss: 1.951\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.557\n",
            "\n",
            "cosine_similarity\n",
            "0.714\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.348\n",
            "\n",
            "loss\n",
            "1.932\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he che histion was frecour fre fremint of the sement of arte art al the to the was in the warre t>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the sent of the foustions for the low as and a the and the palow the the polowe the f fore.>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <who walls was of the fite a the neat mat and ward ay ward a the wo the wall the the wart art s re>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <who als was wall the the posime tos of was and the he as and the wart he seand the the o the o th>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and the are of seres was was was low as a low the as and the and the and the at the and o the the>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <is was to the cout the the parte the ard so was was the ware his trecers.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <suchad the presided ware ward are the part of the and the was prion the pres precontion tion.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he and recate stion to the for the was the ware the ard the paction the the bee the beeay.>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <whe hich st cris to the in sure in in the so the pole as and a the was the to the the se seand th>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <the was wall cout and the the the the pand the the part of the cour the of the court the corouse >\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he sectitent to the so the hing hing hing hat hat hand wal wal the warte the the was he hiche hic>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <hese as as neced the pris of the she she start the the che was res resided are and a thecon ant a>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald the are are stare sting wat he and was of whis and the was in the the fich.>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the ware st and anestre to such to the pround the pris as of the art the art the cor to the th>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which his sof the the st for the sessed wart a a prof a the was and the pore the se posest o o th>\n",
            "\n",
            "\n",
            "\n",
            "[6]\n",
            "train:\n",
            "[6,   30] loss: 1.890\n",
            "[6,   60] loss: 1.847\n",
            "[6,   90] loss: 1.813\n",
            "[6,  120] loss: 1.747\n",
            "[6,  150] loss: 1.699\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.638\n",
            "\n",
            "cosine_similarity\n",
            "0.778\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.399\n",
            "\n",
            "loss\n",
            "1.694\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he comstes was way fre frefring onerst to sele for the sall and the compred the was own mone he w>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the sand to the propestes lof the repror he seventy not ith not pallowe the ward.>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <moreing oswal the fout the ging win of he pan whout, and wo the wo wound the wown a prre appppent>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <walls was wall and pered the was the was of new in a mand had have he crespred and the t then man>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and a sere severy severy cole proped of the cript, one and sand and at a sand and a a free w he p>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his watte that the pated the penter companers of pring were ight spripters fred.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <safaing the des a were were were pare pand companions whing a prechans of pration.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had reat his to the tor his no ber his by the dexpent,>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which spresst the strange in so and the some ban a a dembed a sat lowe forning thes nechis n.>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there so comoted the ban the land the of the and the with and parind the pare por por the pop pon>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he defated to the dis of the he many who ad and whow apper, the were two ave the comme ated hain >\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he as as necated the be prison, she his a seret the dith the secor the secon, and a the the the p>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald he pare stared reand rean whing bout he was and he his a be his a peabe he fieen.>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in whore he strom hand strom pand the pass to the prict of the pricte mor the sope butr the the p>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which he so secated the sprives, whire a pon con, and who prope, by propes was cone prowas t t th>\n",
            "\n",
            "\n",
            "\n",
            "[7]\n",
            "train:\n",
            "[7,   30] loss: 1.627\n",
            "[7,   60] loss: 1.571\n",
            "[7,   90] loss: 1.563\n",
            "[7,  120] loss: 1.507\n",
            "[7,  150] loss: 1.480\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.671\n",
            "\n",
            "cosine_similarity\n",
            "0.802\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.461\n",
            "\n",
            "loss\n",
            "1.490\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he exew to is was was on other fring of therst do the sald arre teretarr to the some in he we had>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the saing to hem ight refestence lof prorthe and seved it not kin out call wo ther.>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <mreanion oswald faut the becon not he not be had new ark, who the more of the more hapses, end wa>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <whose wall neper to most the the was on in he mase nemas and harles present he come the on tan th>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and harried senery suplated of latted on new the man nesed the and the ased a wenty wenenty werse>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his wiftested of a the pent the pare hand sor he bread reaters of his treestred.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <suphand deas of grand here were parre but by companity, brachange of a prechan.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had a great of lell the to reand with a read were haver having beeenty.>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he a spress drased in enessil thems of and a backes, lon backs, lon for the for the for fic>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there resmon called impartime, the led hand him to and the a prear the das larre or of ard or bor>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he defat dight the so the and man maning on on on on wall on on of the were to end in the din din>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he esess any the bress when the shout of as accertated res dearres to rest, end the collle nt oll>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald parred in starding read reage in of manining he was a bot he was a deeen.>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the moreans, had and stron the prom pand becharaction the prect of marcess exesed dofrenteren.>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which on so so secrive thas crives wo the ir a revold, and were the and prooked reant ral ferent >\n",
            "\n",
            "\n",
            "\n",
            "[8]\n",
            "train:\n",
            "[8,   30] loss: 1.417\n",
            "[8,   60] loss: 1.386\n",
            "[8,   90] loss: 1.370\n",
            "[8,  120] loss: 1.355\n",
            "[8,  150] loss: 1.349\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.708\n",
            "\n",
            "cosine_similarity\n",
            "0.827\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.513\n",
            "\n",
            "loss\n",
            "1.373\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he exewas to i f prof prom other a of a the salf for terer to reter terent of the so ving win by >\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at this same time proffestes loof pror rear, be and not knot gall would though alow ther.>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <moreinal the fal fout the wen the not be have be a be almom wo the mome would and he watho ed dig>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <while oswald opeard to the most of the was who neme in harmlos pres prosen he some tin ting the d>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and at here lates we as of proxe of a proxid, un the mand whis ad the at a resed wenty wen beers >\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his whiftested of ound, that he compare his for he breaters of his trifes trit.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <suchand deas of gren or were paratly compan ot been no a no a precen.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had a great ostill his in by where his nement wet a reat hat the depen.>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he express drat ins ragain son thems a by ackes, lowed acess, loon be for the for fifficie.>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there was so calotty of them them the led him to and with a parrint desleme or oposidens.>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he diffected to the so vie in ow at a shout at it who a cribt to ewge of a book, and in the an th>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he esessen hat the press not of as irticared to a bet the resis is is intriel neal no punill nal >\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald of pared the starty reeging of commminas ome of in he was a bout feen.>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the moreans, he at istron the chan as to the crect nesiburect is is as exess douctrin.>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which when sose such i scrives who he reve of a could, and who elso espope feret retifieret rent >\n",
            "\n",
            "\n",
            "\n",
            "[9]\n",
            "train:\n",
            "[9,   30] loss: 1.300\n",
            "[9,   60] loss: 1.285\n",
            "[9,   90] loss: 1.272\n",
            "[9,  120] loss: 1.260\n",
            "[9,  150] loss: 1.248\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.724\n",
            "\n",
            "cosine_similarity\n",
            "0.839\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.549\n",
            "\n",
            "loss\n",
            "1.293\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he couse to sits way forring others to bersed to bee sout arturer to terion the sove hout houthe >\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at this same time profestis lotheses lothese be had not geould gat it of gall on though ther.>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <moreinal fall falt falt the wen not he have be aning were could on one of a move would d acked of>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <whilas wald appeared tome to the was ho neme hame mes ade conve or losperson he some timentand ma>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and a parion, sily as of pelit cold a croped, un the mand that the a fires he wenty-ears he wears>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his whiftestipon that that that pered himselforied reaters of his of his trighter.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <such and ges of grender were were pared by companioutions of a precions of an.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had a great of a still hardison by with aver a haver it hat to bady.>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he a spresed in and sunt hems ontimes by acts, long acts, long be for the assonatition.>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there was calotiaby them that led hat legent him to and with nebarint desly bor of a conside cons>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he dit the to the so vade he hame maninal on of a chan of a good good good culpe and he the the d>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he esessing hat the president, shot of as a shit bet the bet the resest and he could nothe lener >\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald oper it a reater it reade reading a mes ome of a could fateen.>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the morians, had in sto strong convich i the crectiver agrectnes of marces docture.>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which on a socccrived thas crives whoted ireve a gooked, but also is was pothereatial fe and of a>\n",
            "\n",
            "\n",
            "\n",
            "[10]\n",
            "train:\n",
            "[10,   30] loss: 1.209\n",
            "[10,   60] loss: 1.201\n",
            "[10,   90] loss: 1.207\n",
            "[10,  120] loss: 1.202\n",
            "[10,  150] loss: 1.203\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.722\n",
            "\n",
            "cosine_similarity\n",
            "0.837\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.553\n",
            "\n",
            "loss\n",
            "1.272\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he che is way for firming others to you hand salf, and the ard terered tentan whout has and whout>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at this same time prefestastas loof rer rear, and seved it a not gat in out houghtr.>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <morina oswald falow the would be had new are, quote, only on the mome womoved wo, end aled.>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <while as wald a peard to thost of those new himas a mespresen, he some times and madessand msent >\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and a parionly serias leas of priculate grophict, a man whis said that the ares he wenty when bri>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his whiftest of of i that that he compare his for it reaters of histright.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <such and esub grand ure were were pany newgate, and no and a no a presshon.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had a great ostill hate to orgess and by reat hat hat the depie.>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he express dragaing and enso thems on by acts, long be for the is as nation.>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there was some calod ibabout hat led hat led han the nemparint dessly bor for oposible conss onss>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he dit at did to the so vabionial on ow all on ow all on of a good good good and the ginetined th>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he asassinated the presside who as shot a cirtic a bet the resist and he it hillong one onerulean>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald a pared istarty reading a bout conious ome was a bout fate feten.>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the morians, evidence of and ching as to the crect the cressisass a marks is doctrin.>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which one a so so succribites care revot ireval, and where a also is was pooked all feant a an th>\n",
            "\n",
            "\n",
            "\n",
            "[11]\n",
            "train:\n",
            "[11,   30] loss: 1.164\n",
            "[11,   60] loss: 1.154\n",
            "[11,   90] loss: 1.151\n",
            "[11,  120] loss: 1.149\n",
            "[11,  150] loss: 1.145\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.741\n",
            "\n",
            "cosine_similarity\n",
            "0.850\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.594\n",
            "\n",
            "loss\n",
            "1.219\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he coused his way for for forring others to hind south and ther turnt the sove yout he ing win wi>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the same time he profestess lof per rear, and subane subben could i allow though thour.>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <morina oswald fall faut the would not be had be any were who mome whom whom wome who would ped wa>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <while as wald appeared to the was ho new himas and hemas and harmless procentimes ontimen to time>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and a parionly, serious leet gropalit cold a grophict, on whis ad that a fires he would bris he w>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his wifed hastipod that he compared himse for his for it readers of histrig.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <such as a wrand grandure were were parred by new in notions of a precion.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had great hostill hate torgess in by rent, whet abrit happod bady.>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he express dragaing and so thems ontimes by acts, on bof for the for the for the forine.>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there were som callety obtty out hat leght imto age age age parint desseriben gor for pok onsose >\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he difected to the so vie in you in ow one of a general on of a gook accupe of a cupe an then the>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he esis nate the preside the preserved to of his artic a bet resist and hight a gillanother lelli>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald apparagerstarter it reeding a bout connious of ane he was a bout fifteen.>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the morins, he at in sto strong conviction of a prectnes of marknes of marks is is dout trirea>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which one sose excribed to scribed are revoth a greval, and who as was was was was pooke feer and>\n",
            "\n",
            "\n",
            "\n",
            "[12]\n",
            "train:\n",
            "[12,   30] loss: 1.121\n",
            "[12,   60] loss: 1.114\n",
            "[12,   90] loss: 1.110\n",
            "[12,  120] loss: 1.112\n",
            "[12,  150] loss: 1.117\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.749\n",
            "\n",
            "cosine_similarity\n",
            "0.855\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.607\n",
            "\n",
            "loss\n",
            "1.195\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he che huse to six wiffering others to hirst wo hintial he sald he ard to the sove younght he and>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the same time he prefestices love per and subance that he gail one wiber.>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <morina oswald falow the would not he have be aning who mome whom whom wome whom whom whoked dil.>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <while all nepured to most of those who new him hamas a me are lisprocen, he some times ontimes id>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and a parily, sinry as leas of prike a concrifict a man whis said that a fiture years hears hears>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his wifed hastified that he compared himse for his for his of his tright.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <such and es of grandure were were pared leapened by notions of a precion.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had great ostill hilty torgess in by whet whet aver it happied.>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he express dracy instragain enso time a by oung acts, long before the for the assassination>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there was some caloty obty that led hat led have age parint is suremarint dessly bouse oposs cons>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he diffected to the so vie in you it shot eight chanow all on of exew but and the and the gigh de>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he esis neated the preside the presserved to officirtic a resisted arrest intried a pitalinglon o>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald aparaterater it reade reading a by the was a moan he was advot feen.>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the morings, he evidence to strong conviction of the crecknes of marces dopturn.>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which one so socceit discrive ithes caut ireval, and quote, but also eas was was was fee retical >\n",
            "\n",
            "\n",
            "\n",
            "[13]\n",
            "train:\n",
            "[13,   30] loss: 1.084\n",
            "[13,   60] loss: 1.085\n",
            "[13,   90] loss: 1.091\n",
            "[13,  120] loss: 1.089\n",
            "[13,  150] loss: 1.078\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.752\n",
            "\n",
            "cosine_similarity\n",
            "0.857\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.619\n",
            "\n",
            "loss\n",
            "1.177\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he che huse ware for firme other a the reself and to reture turntary turnt the sove youthe e ing >\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the same time prefestices love per and sever, and sever, a gaill wibed nout hout houger.>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <morina oswald fall falt the went not be had be any who mome whom whom whom whom whom whoked thole>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <while all nepured to most of those who ne him as a mes a me armlose presen, he some timentimes an>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and a parelay serias leas as a me as a man who man whis said that that a fiture years he he wew b>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his wiftestived of age that he compared himself with breaters of histrit.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <such and es of grandure were were pared by neight by notions of a pression.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had a great of stilliaty torges in by whet aver it happoded,>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he express dragaing and so thems ontimes by acts, ong boabses assassasnation.>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there were som calloty obtty out at led him to and with and a parrint desslem gor for pa poso con>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he diffected to the so vie in you ade you all ong on on ow all of a cupe and the and the on ding >\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he esiss nated the president shot a shot officertic a bet resisted arrest and hill not oner lelli>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald aparaterater it reading about comminase ome was a bout fifteen,>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the morians, eat add and sto stron gin as to the prect the cress as of marcess doctriron.>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which on a so seet discribed has crive ireval ireval, and quote, but ose was pooke feretical f an>\n",
            "\n",
            "\n",
            "\n",
            "[14]\n",
            "train:\n",
            "[14,   30] loss: 1.066\n",
            "[14,   60] loss: 1.061\n",
            "[14,   90] loss: 1.065\n",
            "[14,  120] loss: 1.071\n",
            "[14,  150] loss: 1.055\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.758\n",
            "\n",
            "cosine_similarity\n",
            "0.861\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.626\n",
            "\n",
            "loss\n",
            "1.169\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he coused his wayfth prof proffer a thers to been tiald her to returnt to the sove younght he out>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the same time he prefestices lof per and seve that he gead a not gail wow though thour.>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <morina oswald fall faut the would not be had be any were, quote, on the move wome per haps,>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <while oswald appeared to most of those who nemas a mes a me armlose preson he sometimes in as mad>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and apparily, siry as leas as a supperifict unct on whis said that that at a fitures he would bri>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his wiftestified that the compared himselforet with breaters of histright read.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <such agas a grandure were were were paratly apened, by notions of appression.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had a great ostillity torgisme in by rened, whet aver it happod bady.>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he express dragain ensid he mr. in a vimes by acts, ong bof for the assassination.>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there was some calloty obty out at led hat legime to and a parint disly gar for of ause conse con>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he diffected to the so vie in you at shot eigch at juno of a cupe of a cupe and the beend the dee>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he assinated the preside the presertich shot officertic a resisted arrest and hight hickell onoth>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald apparaterater it revery in about connious of and he was adough fifteen.>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the morings, eat add astrong conviction as to the crecknes of marcts as dopturen.>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which on a so soccusied to scrive ireval ireval, and quote, but also as where reatical f feerent >\n",
            "\n",
            "\n",
            "\n",
            "[15]\n",
            "train:\n",
            "[15,   30] loss: 1.051\n",
            "[15,   60] loss: 1.049\n",
            "[15,   90] loss: 1.058\n",
            "[15,  120] loss: 1.042\n",
            "[15,  150] loss: 1.048\n",
            "\n",
            "validation:\n",
            "jaccard_similarity\n",
            "0.757\n",
            "\n",
            "cosine_similarity\n",
            "0.860\n",
            "\n",
            "damerau-levenshtein_similarity\n",
            "0.625\n",
            "\n",
            "loss\n",
            "1.165\n",
            "\n",
            "compare-text-gen:\n",
            "target: <he accused his wife of preferring others to himself and told her to return to the soviet union wi>\n",
            "pred:   <he che huse to i f prof profer and thers to hinsout f. and to returnt terir to the so hewgathon w>\n",
            "\n",
            "target: <at the same time he professed his love for her and said that he could not get along without her.>\n",
            "pred:   <at the same time the prestices love per and subaness the gat of gat a gail would withouger.>\n",
            "\n",
            "target: <marina oswald thought that he would not be happy anywhere, quote, only on the moon, perhaps, end >\n",
            "pred:   <morina oswald fall faut the would not be had be any were, quote, on the move wome per haps,>\n",
            "\n",
            "target: <while oswald appeared to most of those who knew him as a meek and harmless person, he sometimes i>\n",
            "pred:   <while oswald a peard to most of those who neme ham as a mes a meloss person, he some times ade ma>\n",
            "\n",
            "target: <and, apparently seriously, as a political prophet -- a man who said that after twenty years he wo>\n",
            "pred:   <and apparily, siry as leas as a supperifict unct on the said that that a fiture years he he wod b>\n",
            "\n",
            "target: <his wife testified that he compared himself with great readers of history.>\n",
            "pred:   <his wiftestived of id that he compared himself with breaters of histright read.>\n",
            "\n",
            "target: <such ideas of grandeur were apparently accompanied by notions of oppression.>\n",
            "pred:   <such agas a grandure were were were pared by new notions of a precion.>\n",
            "\n",
            "target: <he had a great hostility toward his environment, whatever it happened to be,>\n",
            "pred:   <he had great ostill hill his in by whet wethere it happoded to bady.>\n",
            "\n",
            "target: <which he expressed in striking and sometimes violent acts long before the assassination.>\n",
            "pred:   <which he express dragain ensid he mr. in a vime a long of for the assassassenation.>\n",
            "\n",
            "target: <there was some quality about him that led him to act with an apparent disregard for possible cons>\n",
            "pred:   <there was some calloty obty out hat led him took and with and a parint dislim gor for pa poso con>\n",
            "\n",
            "target: <he defected to the soviet union, shot at general walker, tried to go to cuba and even contemplate>\n",
            "pred:   <he diffected to the so vie in you ade you all on on ow on of a cupe of a cupe and the ging then t>\n",
            "\n",
            "target: <he assassinated the president, shot officer tippit, resisted arrest and tried to kill another pol>\n",
            "pred:   <he assinated the preside the preserved shot officertic a bet resist and hight a neter to pelonert>\n",
            "\n",
            "target: <oswald apparently started reading about communism when he was about fifteen.>\n",
            "pred:   <oswald apparaterater exerty reading a nevoctiminas ome was adobout fifteen.>\n",
            "\n",
            "target: <in the marines, he evidenced a strong conviction as to the correctness of marxist doctrine,>\n",
            "pred:   <in the morings, have add insto strong ching as to the crectnas of marcys dopturn.>\n",
            "\n",
            "target: <which one associate described as, quote, irrevocable, end quote, but also as, quote, theoretical,>\n",
            "pred:   <which on a so soccusied to scrive ireval ireval, and quote, but also as was was was photherer ret>\n",
            "\n",
            "\n",
            "\n"
          ]
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjcAAAGwCAYAAABVdURTAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAABT+klEQVR4nO3dd3hUZd7G8e+kk5AQICQhEHqvASIIKDaK4OpiRUHIYtnXggJBBSwUG6Ir4lpAVERcURBRcVWKoSu9g/TekhAC6WUyc94/gGg2lGQyyUlm7s915Qpz2tzPcRx+POc557EYhmEgIiIi4iI8zA4gIiIi4kwqbkRERMSlqLgRERERl6LiRkRERFyKihsRERFxKSpuRERExKWouBERERGX4mV2gLJmt9s5efIkgYGBWCwWs+OIiIhIERiGQVpaGhEREXh4XLlvxu2Km5MnTxIZGWl2DBEREXHAsWPHqF279hW3cbviJjAwEDh/coKCgpx6bKvVyqJFi+jZsyfe3t5OPXZF4O7tB50Dd28/6Byo/e7dfii9c5CamkpkZGT+3+NX4nbFzcVLUUFBQaVS3Pj7+xMUFOSWH2p3bz/oHLh7+0HnQO137/ZD6Z+Dogwp0YBiERERcSkqbkRERMSlqLgRERERl6LiRkRERFyKihsRERFxKSpuRERExKWouBERERGXouJGREREXIqKGxEREXEpKm5ERETEpai4EREREZei4kZERERcittNnCkiIuKqDMPgdHoOhgE+nh54e3mc/+1pKdKEk65CxY2IiEgFk5Nn48iZTA4kpnPgdDoHTmec/52YTkau7ZL7XCxy/ix4PPD1Ov/b28uSv8znL+t9vP66zHKJZeeP6ePleeG3Bx4Y7DproU8Zn5O/UnEjIiJSTp3NyL1QvFwoYC4UM0eTM7Ebl97HwwIWiwXb/2yQa7OTawMuU/w4U4CXByNK/V0uT8WNiIiIiWx2gxNns9h/Oo0DiRkFipnkjNzL7hfo60XD0Mo0rFGZhqEB53/XqEydav74eHlgsxtYbXZybXaseRd/G+TabOTmFVyX89dtLmxXYNmF3389xvnfF5bnXdjPZifHaiMzJbkMz2BhKm5ERETKQEZOHoeS/rx8dPFS0sGkDHLz7Jfdr1ZwpQtFzJ8FTMPQAGpU9r3iOBpPDwueHp74eXuWRnMuy2q18vPPP5fpe/4vFTciIiJOlJILqw+e4UhydoGxMCdTsi+7j6+XB/VDAv7siblQyDSoEYC/j/6qLi6dMRERkRJIy7ay+sAZVu5LYvneRI4me8HGjZfcNqSyDw1q/KWACa1MoxqViQiuhKeH+9zNVNpU3IiIiBSDzW6w40QKK/edZsXeJDYdPUveXwbvWjCoWz2ARqGBBcbCNKwRQLC/j4nJ3YeKGxERkauIT8lmxb7TrNh7mt/2J3E201pgfb3q/nRrUoMuDaqSsm8Dd91+Hd7e3ialFRU3IiIi/yPbamPtoWRW7D3Nyn2n2ZuQXmB9oK8XXRpV5/rGNejWuAZ1qvsDFwbTHjQjsfyVihsREXF7hmGwJyGNlXuTWLHvNGsPJRe4g8ligba1g+nWOIRuTWrQNjIYb0/NYFReqbgRERG3dCY9h1X7k1ixN4mV+06TmJZTYH3NKn50a1yDbk1q0LVRdY2XqUBU3IiIiFvIzbOz6ejZC5eakthxMgXjLw/x9fP24NoG1S8UNCE0rFHZreZjciUqbkRExCUZhsHhM5n542ZWHzhTaN6l5jWD8i81dahbtcwfeCelQ8WNiIi4jPScPFbtS8q/s+n42awC66sH+HD9hWLmusYhhAb6mZRUSpOKGxERqdDOpOfw664EFu5MYNW+JHJtfw4E9va0EF23Gt2a1OD6xiG0qBmEhx6W5/JU3IiISIVz8lwWC3fGs3BnPOsOJReYIbt+SAA3NDk/bqZT/eoE+OqvOnej/+IiIlIhHDidzoId8SzaGc/W4ykF1rWqFUSvFuHc2iqcRqEaCOzuykVx88EHH/DWW28RHx9P27Ztee+99+jYseNltz937hwvvPAC8+bNIzk5mbp16zJ58mT69OlThqlFRKQ0GYbBzpOpLNgRz4Kd8exP/PNBehYLXFO3Gj1bhtGrZTiR1fxNTCrljenFzezZs4mNjWXq1Kl06tSJyZMn06tXL/bs2UNoaGih7XNzc+nRowehoaHMnTuXWrVqceTIEYKDg8s+vIiIOJXNbrDhcDILdyawcGc8J879OSDY29NCl4Yh3NoqnO7Nw6gR6GtiUinPTC9uJk2axKOPPsrgwYMBmDp1Kj/99BPTp09n1KhRhbafPn06ycnJ/P777/nzdtSrV++yx8/JySEn588HM6WmpgLnH5FttVovt5tDLh7P2cetKNy9/aBz4O7tB50DR9qfk2dnzcEzLN6VyOJdiSRn/LlvJW8PujUOoWeLMG5qGkKg35/zNZXHc+zu//2h9M5BcY5nMYy/PsKobOXm5uLv78/cuXPp27dv/vKYmBjOnTvHDz/8UGifPn36UK1aNfz9/fnhhx+oUaMG/fv3Z+TIkXh6Fn4+wbhx4xg/fnyh5bNmzcLfX92YIiJmyLHBrnMWtp6x8Mc5C9m2P8fI+HsatKxm0LaaQdMqBj569IwAmZmZ9O/fn5SUFIKCgq64rak9N0lJSdhsNsLCwgosDwsLY/fu3Zfc5+DBgyxZsoQBAwbw888/s3//fp544gmsVitjx44ttP3o0aOJjY3Nf52amkpkZCQ9e/a86skpLqvVyuLFi+nRo4dbzgbr7u0HnQN3bz/oHFyp/ecyrSzZk8iiPxJZtf8MOX+Zuyk00JfuzWvQs0UYHetVrbDzNrn7f38ovXNw8cpLUZh+Waq47HY7oaGhTJs2DU9PTzp06MCJEyd46623Llnc+Pr64utb+Lqst7d3qX3wSvPYFYG7tx90Dty9/aBzcLH98SnZLPrj/C3baw4mY/vLPdt1q/tza8twerYMp11ksEs9f8bd//uD889BcY5lanETEhKCp6cnCQkJBZYnJCQQHh5+yX1q1qyJt7d3gUtQzZs3Jz4+ntzcXHx8NLGZiIiZTmfBtJWH+HX3aTYfPVdgXbPwQG5tFU6vluE0Cw/ULdtSKkwtbnx8fOjQoQNxcXH5Y27sdjtxcXEMGTLkkvt07dqVWbNmYbfb8fA43225d+9eatasqcJGRMRE5zJzeeI/G/n9oBewL395+zrB+QVN3eoB5gUUt2H6ZanY2FhiYmKIjo6mY8eOTJ48mYyMjPy7pwYNGkStWrWYMGECAI8//jjvv/8+Q4cO5amnnmLfvn28/vrrPP3002Y2Q0TErSWmZjPw03XsSUjDw2JwbYPq9G4dQc8WYYQFaf4mKVumFzf9+vXj9OnTjBkzhvj4eKKioliwYEH+IOOjR4/m99AAREZGsnDhQoYPH06bNm2oVasWQ4cOZeTIkWY1QUTErR1LzuTBT9dy5EwmoYG+DG6QwSP3RLv9mBMxj+nFDcCQIUMuexlq2bJlhZZ17tyZNWvWlHIqERG5mn0JaTz46VoSUnOIrFaJz2I6sHPNMrNjiZurmPfaiYiI6bYdP8d9H60mITWHxqGV+eb/ulBX0yBIOVAuem5ERKRiWXPwDI98voH0nDza1K7C54M7UjXAx62fzCvlh4obEREpliW7E3j8P5vIybNzbYNqfDwousC0CCJmU3EjIiJF9sOWE4yYs5U8u0H35qG83789ft6aH0HKFxU3IiJSJF+uPcKL3+/AMODvURH86962FXaaBHFtKm5EROSqpiw7wMQF5+f8e/DaOrx8RyuXmi5BXIuKGxERuSzDMHhz4R6mLDsAwBM3NuTZXk01bYKUaypuRETkkux2g5d+2MGXa48CMKp3Mx67oaHJqUSuTsWNiIgUYrXZGTFnK/O3nsRigdf6tqZ/pzpmxxIpEhU3IiJSQLbVxhNfbmLJ7kS8PCy80y+K29tGmB1LpMhU3IiISL60bCuPfL6BtYeS8fXyYOqDHbipWajZsUSKRcWNiIgAkJyRS8z0dWw/kUJlXy8+jYmmU4PqZscSKTYVNyIiQnxKNg9+upb9ielUC/Dh88EdaV27itmxRByi4kZExM0dTsrgwU/XcvxsFuFBfvznkU40Cq1sdiwRh6m4ERFxY7vjUxn46TpOp+VQr7o/XzzciUjN7C0VnIobERE3tenoWQZ/tp6ULCvNwgOZ+XBHQgP9zI4lUmIqbkRE3NBv+5N4dOYGMnNttK8TzGf/6EgVf83sLa5BxY2IiJtZuDOep2ZtJtdm57pGIXw0sAMBvvrrQFyHPs0iIm7k243Hee7bbdjsBre2DOfdB6Lw9fI0O5aIU6m4ERFxEzN+O8S4H/8A4J4OtXnjrtZ4eXqYnErE+VTciIi4OMMweG/JfiYt3gvA4K71eOm2Fnh4aGZvcU0qbkREXJhhGLz20y4+WXUIgGHdGzP0lsZYLCpsxHWpuBERcVE2u8HoeduYs+E4AGP+1oKHrqtvciqR0qfiRkTEBeXk2Rg+ews/b4/HwwIT727DvdGRZscSKRMqbkREXExmbh7/98VGVu5LwsfTg38/EMWtrWqaHUukzKi4ERFxISlZVh6asZ6NR85SyduTaYM6cH3jGmbHEilTKm5ERCqYzNw8ElJziE/JJiH1/E98ajaJqTlsO3GOY8lZBPl58dnga+hQt5rZcUXKnIobEZFyIs9mJyk9N79YyS9cUnIKFDFp2XlXPE5IZR9mPtSJFhFBZZRcpHxRcSMiUsoMwyA1O+9CofJn0ZKQmlOgiDmdloPdKNoxK3l7El7Fj7AgX8KD/Aj7y0+XhtWpGuBTuo0SKcdU3IiIOEm21cbcDcdZctiDxXO2cfpCL0xCag5ZVluRjuHpYaFGZV/CqvgRHuRboGgJD/IjvIovoUF+BPp66Vk1Ipeh4kZExAlOnMvi/77YwI4TqYAHnIovtE2VSt6EB/kReqG3JbyKH6EXipaLPTDVK/viqScHi5SIihsRkRJadyiZJ77cSFJ6LlX9vWlbJYdr2zYjomoAYYG+54uYQD8q+WiCSpGy4HBxExMTw8MPP0y3bt2cmUdEpEL5z5ojjJu/kzy7QYuaQXzYvy1bf19Kn6718Pb2NjueiFtyeDrYlJQUunfvTuPGjXn99dc5ceKEM3OJiJRruXl2Rs/bzovf7yDPbvC3NjX59vEu1AquZHY0EbfncHHz/fffc+LECR5//HFmz55NvXr16N27N3PnzsVqtTozo4hIuZKYlk3/j9fw1bqjWCww8tZmvPdAO112EiknHC5uAGrUqEFsbCxbt25l7dq1NGrUiIEDBxIREcHw4cPZt2+fs3KKiJQL246f4473fmPDkbME+nkxPeYaHr+xoe5cEilHSlTcXHTq1CkWL17M4sWL8fT0pE+fPmzfvp0WLVrwzjvvOOMtRERMN2/Tce6Zupr41Gwa1gjghye7clOzULNjicj/cHhAsdVqZf78+Xz22WcsWrSINm3aMGzYMPr3709Q0PmnYn733Xc89NBDDB8+3GmBRUTKWp7NzoRfdvPpqkMAdG8eyjv9ogj004BhkfLI4eKmZs2a2O12HnjgAdatW0dUVFShbW666SaCg4NLEE9ExFxnM3J56qvNrNqfBMBTNzdiePcmeOhZNCLllsPFzTvvvMO9996Ln5/fZbcJDg7m0KFDjr6FiIipdsen8ujMDRxLzsLfx5O3721L79Y1zY4lIlfh8JibpUuXXvKuqIyMDB566KEShRIRMdsv209x14e/cyw5i8hqlZj3RBcVNiIVhMPFzeeff05WVlah5VlZWcycObNEoUREzGK3G0xatIfHv9xEZq6Nro2qM//J62gWrhm2RSqKYl+WSk1NxTAMDMMgLS2twGUpm83Gzz//TGio7h4QkYonLdvK8Nlb+HVXIgAPX1ef0b2b4eXplBtLRaSMFLu4CQ4OxmKxYLFYaNKkSaH1FouF8ePHOyWciEhZOXg6nX9+sZH9ien4eHkw4c7W3N2httmxRMQBxS5uli5dimEY3HzzzXz77bdUq1Ytf52Pjw9169YlIiLCqSFFRErT0j2JPP3VZtKy8wgP8uOjgR1oGxlsdiwRcVCxi5sbbrgBgEOHDlGnTh09lVNEKizDMJi6/CBvLtyNYUCHulWZ8mB7QgMvfxeoiJR/xSputm3bRqtWrfDw8CAlJYXt27dfdts2bdqUOJyISGnJyrXx3Lfb+HHrSQAe6BjJuDta4uul+aFEKrpiFTdRUVHEx8cTGhpKVFQUFosFwzAKbWexWLDZbE4LKSLiTMfPZvLPmRv541QqXh4Wxt7Rkgc7qSdaxFUUq7g5dOgQNWrUyP+ziEhFs/rAGZ6ctYnkjFyqB/jw4YD2dGpQ3exYIuJExSpu6tatC5yfV2r8+PG89NJL1K9fv1SCiYg4k2EYzFx9hJf/+wc2u0GrWkF8NDCaWsGVzI4mIk7m0MMbvL29+fbbb52dRUSkVOTk2Rj57TbGzt+JzW5wR9sIvvm/LipsRFyUw0+m6tu3L99//70To4iIOF9iajb3T1vDnA3H8bDA832a8e79UVTy0cBhEVfl8MSZjRs35uWXX+a3336jQ4cOBAQEFFj/9NNPlziciEhJbD56lsf+s5GE1ByC/Lx4r397bmhSw+xYIlLKHC5uPv30U4KDg9m4cSMbN24ssM5isai4ERFTfbPhGC98t4Ncm53GoZX5eFA09UICrr6jiFR4Dhc3ultKRMojq83Oaz/tYsbvhwHo0SKMd/pFUdnX4a87Ealg9H+7iLiMsxm5PPHlJlYfPAPA0FsaM/SWxnh46Pk1Iu6kRMXN8ePHmT9/PkePHiU3N7fAukmTJpUomIhIcWRbbTz0+Xo2Hz2Hv48nk+6L4tZW4WbHEhETOFzcxMXFcccdd9CgQQN2795Nq1atOHz4MIZh0L59e2dmFBG5IsMweH7edjYfPUeVSt7M/r9raRYeZHYsETGJw7eCjx49mmeeeYbt27fj5+fHt99+y7Fjx7jhhhu49957nZlRROSKPlpxkHmbT+DpYeHDAe1V2Ii4OYeLm127djFo0CAAvLy8yMrKonLlyrz88stMnDjRaQFFRK7k1z8SmLhgNwDjbm9B10YhJicSEbM5XNwEBATkj7OpWbMmBw4cyF+XlJRU8mQiIlexJz6NoV9vxjDgwWvrMLBzPbMjiUg54PCYm2uvvZZVq1bRvHlz+vTpw4gRI9i+fTvz5s3j2muvdWZGEZFCzqTn8PDn68nItdG5QXXG3t7S7EgiUk44XNxMmjSJ9PR0AMaPH096ejqzZ8+mcePGulNKREpVbp6dx7/cxPGzWdSt7s+HA9rj7elwR7SIuBiHvw0aNGhAmzZtgPOXqKZOncq2bdv49ttv82cPL6oPPviAevXq4efnR6dOnVi3bt1lt50xYwYWi6XAj5+fn6PNEJEKxjAMxvywg3WHkgn09eLTmGiqBviYHUtEyhHT/6kze/ZsYmNjGTt2LJs2baJt27b06tWLxMTEy+4TFBTEqVOn8n+OHDlSholFxEwzfj/M1+uP4WGBf/dvR6PQQLMjiUg5U6zLUlWrVsViKdqTPpOTk4u03aRJk3j00UcZPHgwAFOnTuWnn35i+vTpjBo16pL7WCwWwsP1cC4Rd7Ni72le+e8fADzfpzk3NQ01OZGIlEfFKm4mT57s1DfPzc1l48aNjB49On+Zh4cH3bt3Z/Xq1ZfdLz09nbp162K322nfvj2vv/46LVteejBhTk4OOTk5+a9TU1MBsFqtWK1WJ7WE/GP+9be7cff2g85Babb/4OkMnpy1CbsBd7ePYFCn2uXyPOszoPb/9bc7Kq1zUJzjWQzDMJz67sVw8uRJatWqxe+//07nzp3zlz/33HMsX76ctWvXFtpn9erV7Nu3jzZt2pCSksK//vUvVqxYwc6dO6ldu3ah7ceNG8f48eMLLZ81axb+/v7ObZCIlIrMPJi03ZPT2RbqBxoMaWHDy/SL6iJSljIzM+nfvz8pKSkEBV35QZ3F6rlJTU3NP+DFHpDLudobO6pz584FCqEuXbrQvHlzPvroI1555ZVC248ePZrY2Nj816mpqURGRtKzZ0+nZ7RarSxevJgePXrg7e3t1GNXBO7eftA5KI3259nsPPzFJk5nJxNRxY+vHutE9cq+Tjl2adBnQO135/ZD6Z2Dq9Udf1XsMTenTp0iNDSU4ODgS46/MQwDi8WCzWa76vFCQkLw9PQkISGhwPKEhIQij6nx9vamXbt27N+//5LrfX198fUt/EXo7e1dah+80jx2ReDu7QedA2e2/7VfdvL7gWT8fTz5JOYawqtWdspxS5s+A2q/O7cfnH8OinOsYhU3S5YsoVq1agAsXbq0eKkuwcfHhw4dOhAXF0ffvn0BsNvtxMXFMWTIkCIdw2azsX37dvr06VPiPCJSvny59ggzfj8MwDv9omgRoTmjROTqilXc3HDDDZf8c0nExsYSExNDdHQ0HTt2ZPLkyWRkZOTfPTVo0CBq1arFhAkTAHj55Ze59tpradSoEefOneOtt97iyJEjPPLII07JIyLlw+oDZxj7w04Anu3VlF4tdYekiBSNw08oBsjOzmbbtm0kJiZit9sLrLvjjjuKdIx+/fpx+vRpxowZQ3x8PFFRUSxYsICwsDAAjh49iofHnyMHz549y6OPPkp8fDxVq1alQ4cO/P7777Ro0aIkTRGRcuTomUwe/3IjeXaDO9pG8MSNDc2OJCIViMPFzYIFCxg0aNAlJ8ks6pibi4YMGXLZy1DLli0r8Pqdd97hnXfeKVZWEak40rKtPPz5es5lWmlbuwpv3tOmyM/XEhGBEjyh+KmnnuLee+/l1KlT2O32Aj/FKWxERC6y2Q2Gfr2FfYnphAX5Mm1QNH7enmbHEpEKxuHiJiEhgdjY2PzLRyIiJfXmgt0s2Z2Ir5cHHw+KJixI88aJSPE5XNzcc889hS4ZiYg4au7G43y04iAA/7q3LW1qB5sbSEQqLIfH3Lz//vvce++9rFy5ktatWxe6//zpp58ucTgRcQ8bjyTz/LztADx9cyNubxthciIRqcgcLm6++uorFi1ahJ+fH8uWLSsw4M9isai4EZEiOXEui//7YiO5Nju3tgxnWPcmZkcSkQrO4eLmhRdeYPz48YwaNarArdoiIkWVkZPHI59vICk9l+Y1g5jUry0eHrozSkRKxuGqJDc3l379+qmwERGH2O0GI+ZsZdepVEIq+/BJTDT+PiV69JaICFCC4iYmJobZs2c7M4uIuJHJv+5lwc54fDw9+GhgB2oFVzI7koi4CIf/mWSz2XjzzTdZuHAhbdq0KTSgeNKkSSUOJyKu6cetJ/n3kvOT3b5+V2s61K1mciIRcSUOFzfbt2+nXbt2AOzYsaPAOj1NVEQuZ9vxczzzzVYA/q9bA+7pUNvkRCLiahwubpwxK7iIuJeE1GwenbmBnDw7NzcL5blbm5kdSURckEYDi0iZyLba+OfMDSSk5tA4tDLv3h+Fp+6MEpFSUKyem7vuuosZM2YQFBTEXXfddcVt582bV6JgIuI6DMPgubnb2Ho8har+3nwacw2Bft5X31FExAHFKm6qVKmSP56mSpUqpRJIRFzPh8sOMH/rSbw8LHw4oAN1qvubHUlEXFixipvPPvvskn8WEbmcBTvieWvhHgBe/nsrOjesbnIiEXF1Tnti1vLly8nIyKBz585UrVrVWYcVkQrsj5OpxM7ZAsA/utSjf6c65gYSEbdQ7OJm4sSJpKen88orrwDnr6X37t2bRYsWARAaGkpcXBwtW7Z0blIRqVCS0nN4dOYGMnNtXNcohBdva252JBFxE8W+W2r27Nm0atUq//XcuXNZsWIFK1euJCkpiejoaMaPH+/UkCJSseTk2Xnsi42cOJdF/ZAAPujfHi9P3ZwpImWj2N82hw4dok2bNvmvf/75Z+655x66du1KtWrVePHFF1m9erVTQ4pIxWEYMPbHP9hw5CyBfl58EhNNFX/dGSUiZafYxU1eXh6+vr75r1evXk2XLl3yX0dERJCUlOScdCJS4Sw7ZeHbTSfxsMAH/dvTsEZlsyOJiJspdnHTsGFDVqxYAcDRo0fZu3cv3bp1y19//PhxqlfX3RAi7mj53tP8cOT818pLf2tBtyY1TE4kIu6o2AOKn3zySYYMGcLKlStZs2YNnTt3pkWLFvnrlyxZkj/nlIi4j/iUbJ6ZuwMDC/2ia/GPLvXMjiQibqrYxc2jjz6Kp6cnP/74I926dWPs2LEF1p88eZKHHnrIaQFFpPyz2w2enbuVc1lWIgMMxtzWXBPoiohpHHrOzUMPPXTZAubDDz8s8PqNN97gscceIzg42JG3EpEKYMbvh1m5Lwk/bw8GNs7Fx0t3RomIeUr9G+j1118nOTm5tN9GREyyJz6NNxbsBmDUrU0Jq2RyIBFxe6Ve3BiGUdpvISImycmzMWz2FnLz7NzUtAb9r6ltdiQRkdIvbkTEdU1atJddp1KpFuDDxHvaaJyNiJQLKm5ExCG/H0hi2sqDALxxV2tCA/1MTiQicp6KGxEptpQsK8/M2YphwAMdI+nZMtzsSCIi+VTciEixjflhBydTsqlX3Z8Xb2tx9R1ERMpQqRc3119/PZUq6fYJEVfxw5YT/LDlJJ4eFib1iyLA16EnSoiIlJoSfSvZ7Xb2799PYmIidru9wLqLUzL8/PPPJXkLESlHTpzL4sXvdwDw1M2NaF+nqsmJREQKc7i4WbNmDf379+fIkSOFbve2WCzYbLYShxOR8sNuNxgxZwtp2XlERQYz5KZGZkcSEbkkh4ubxx57jOjoaH766Sdq1qypW0BFXNwnqw6y5mAy/j6eTO4XhZenhuyJSPnkcHGzb98+5s6dS6NG+tebiKv742Qqby3cA8CYv7WgXkiAyYlERC7P4X96derUif379zszi4iUQ9lWG8Nmb8ZqM+jRIox+10SaHUlE5Ioc7rl56qmnGDFiBPHx8bRu3Rpvb+8C69u0aVPicCJivokLdrM3IZ2Qyr68cVdrXYIWkXLP4eLm7rvvBigwO7jFYsEwDA0oFnERK/ed5rPfDgPw1j1tqF7Z19xAIiJF4HBxc+jQIWfmEJFy5mxGLs98sxWAgdfW5aZmoSYnEhEpGoeLm7p16zozh4iUI4Zh8Px320lIzaFBjQCe79Pc7EgiIkVWrOJm/vz59O7dG29vb+bPn3/Fbe+4444SBRMR83y76QS/7IjHy8PCu/3aUcnH0+xIIiJFVqzipm/fvsTHxxMaGkrfvn0vu53G3IhUXMeSMxk3fycAw3s0oXXtKiYnEhEpnmIVN3+dYuF/p1sQkYrPZjcYPnsL6Tl5RNetymM3NDQ7kohIsTnlEaPZ2dnOOIyImGzq8gNsOHKWyr5evNMvCk8P3fYtIhWPw8WNzWbjlVdeoVatWlSuXJmDBw8C8NJLL/Hpp586LaCIlI1tx8/xzuK9AIy/oyWR1fxNTiQi4hiHi5vXXnuNGTNm8Oabb+Lj45O/vFWrVnzyySdOCSciZSMr18aw2VvIsxv0aR3OXe1rmR1JRMRhDhc3M2fOZNq0aQwYMABPzz/vpGjbti27d+92SjgRKRuv/7yLg6czCAvy5bW+egqxiFRsDhc3J06cuOSkmXa7HavVWqJQIlJ2lu5O5Is1RwD4171tqRrgc5U9RETKN4eLmxYtWrBy5cpCy+fOnUu7du1KFEpEysaZ9ByenbsNgIe61uf6xjVMTiQiUnIOP6F4zJgxxMTEcOLECex2O/PmzWPPnj3MnDmT//73v87MKCKlwDAMRs3bTlJ6Dk3CKvPcrU3NjiQi4hQO99z8/e9/58cff+TXX38lICCAMWPGsGvXLn788Ud69OjhzIwiUgpmrz/G4j8S8PH0YHK/dvh56ynEIuIaHO65Abj++utZvHixs7KISBk5lJTB+B//AOCZXk1oERFkciIREedxuOemQYMGnDlzptDyc+fO0aBBgxKFEpHSY7XZGTZ7C1lWG50bVOeR6/T/q4i4FoeLm8OHD19y/qicnBxOnDhRolAiUnreX7KfrcfOEejnxdv3tcVDTyEWERdT7MtSf50NfOHChVSp8uekejabjbi4OOrVq+eUcCLiXJuOnuX9pfsBeLVvKyKCK5mcSETE+Ypd3FycDdxisRATE1Ngnbe3N/Xq1ePtt992SjgRcZ6MnDyGz96CzW7w96gI/h6lpxCLiGsqdnFzcTbw+vXrs379ekJCQpweSkSc75X//sGRM5lEVPHj5b+3MjuOiEipcfhuqUOHDjkzh4iUooU74/l6/TEsFnj7viiqVPI2O5KISKkp0a3gcXFxxMXFkZiYmN+jc9H06dNLFExEnCMxLZvR87YD8M/rG9C5YXWTE4mIlC6Hi5vx48fz8ssvEx0dTc2aNTXRnkg5ZBgGz83dRnJGLs1rBhHbs4nZkURESp3Dxc3UqVOZMWMGAwcOdGYeEXGi/6w5wrI9p/Hx8uDd+6Pw9dJTiEXE9Tn8nJvc3Fy6dOnizCwi4kT7E9N59addAIy6tRlNwgJNTiQiUjYcLm4eeeQRZs2a5cwsIuIkuXl2hs3eTE6enesbh/CPLvXMjiQiUmaKdVkqNjY2/892u51p06bx66+/0qZNG7y9C959MWnSpCIf94MPPuCtt94iPj6etm3b8t5779GxY8er7vf111/zwAMP8Pe//53vv/++yO8n4urejdvLjhOpBPt786979RRiEXEvxSpuNm/eXOB1VFQUADt27CiwvDiDi2fPnk1sbCxTp06lU6dOTJ48mV69erFnzx5CQ0Mvu9/hw4d55plnuP7664veABE3sO5QMh8uOwDA63e2JizIz+REIiJlq1jFzdKlS50eYNKkSTz66KMMHjwYOD9Q+aeffmL69OmMGjXqkvvYbDYGDBjA+PHjWblyJefOnXN6LpGKKC3byvDZWzAMuLt9bfq0rml2JBGRMlei59z8VWpqKkuWLKFZs2Y0a9asSPvk5uayceNGRo8enb/Mw8OD7t27s3r16svu9/LLLxMaGsrDDz/MypUrr/geOTk55OTkFMgJYLVasVqtRcpZVBeP5+zjVhTu3n4w9xwYhsGY73dw4lwWtYP9eKF3kzLPoc+AzoHa797th9I7B8U5nsPFzX333Ue3bt0YMmQIWVlZREdHc/jwYQzD4Ouvv+buu+++6jGSkpKw2WyEhYUVWB4WFsbu3bsvuc+qVav49NNP2bJlS5FyTpgwgfHjxxdavmjRIvz9/Yt0jOJavHhxqRy3onD39oM552DZKQvfHfbEgsFdtdJZuWRRmWe4SJ8BnQO1373bD84/B5mZmUXe1uHiZsWKFbzwwgsAfPfddxiGwblz5/j888959dVXi1TcFFdaWhoDBw7k448/LvKcVqNHjy4wEDo1NZXIyEh69uxJUFCQU/NZrVYWL15Mjx49Cg2wdgfu3n4w7xws3XOaH9acHxP3bK8mPHpd/TJ777/SZ0DnQO137/ZD6Z2Di1deisLh4iYlJYVq1aoBsGDBAu6++278/f257bbbePbZZ4t0jJCQEDw9PUlISCiwPCEhgfDw8ELbHzhwgMOHD3P77bfnL7s47YOXlxd79uyhYcOGBfbx9fXF19e30LG8vb1L7YNXmseuCNy9/VC252DXqVSGz9mG3YB+0ZE8fmNj058Yrs+AzoHa797tB+efg+Icy+Hn3ERGRrJ69WoyMjJYsGABPXv2BODs2bP4+RXt7gwfHx86dOhAXFxc/jK73U5cXBydO3cutH2zZs3Yvn07W7Zsyf+54447uOmmm9iyZQuRkZGONkekQkpMy+bhGevJyLXRuUF1XunbyvTCRkTEbA733AwbNowBAwZQuXJl6taty4033gicv1zVunXrIh8nNjaWmJgYoqOj6dixI5MnTyYjIyP/7qlBgwZRq1YtJkyYgJ+fH61atSqwf3BwMECh5SKuLttq49GZGzmZkk2DkACmPtgBHy+H/70iIuIyHC5unnjiCTp27MixY8fo0aMHHh7nv1QbNGjAq6++WuTj9OvXj9OnTzNmzBji4+OJiopiwYIF+YOMjx49mn9sETnPbjcYMWcrW4+dI9jfm0//cQ1V/N27C1xE5KIS3QoeHR1NdHR0gWW33XZbsY8zZMgQhgwZcsl1y5Ytu+K+M2bMKPb7iVR07/y6l5+2n8Lb08LUBztQPyTA7EgiIuWGw8WNzWZjxowZxMXFkZiYmD+w96IlS5aUOJyIFDZv03HeW7IfgNfubM21DaqbnEhEpHxxuLgZOnQoM2bM4LbbbqNVKw1iFCkL6w8nM+rb7QA8dkND7ovWIHoRkf/lcHHz9ddfM2fOHPr06ePMPCJyGUfOZPDPmRvItdm5tWU4z/VqanYkEZFyyeGRuj4+PjRq1MiZWUTkMlKyrDw0Yz1nM620rlWFd/pFaaZvEZHLcLi4GTFiBO+++y6GYTgzj4j8D6vNzpNfbuLA6QzCg/z4JCaaSj6eZscSESm3HL4stWrVKpYuXcovv/xCy5YtCz05cN68eSUOJ+LuDMNg7PydrNqfRCVvTz6JiSYsqGgPyRQRcVcOFzfBwcHceeedzswiIv9j+m+HmbX2KBYL/PuBdrSqVcXsSCIi5Z7Dxc1nn33mzBwi8j/idiXw6k9/APB87+b0aBFmciIRkYqhRI/+zcvL49dff+Wjjz4iLS0NgJMnT5Kenu6UcCLu6o+TqTz11WYMA+6/JpJHrjdnlm8RkYrI4Z6bI0eOcOutt3L06FFycnLo0aMHgYGBTJw4kZycHKZOnerMnCJuIzE1m0c+X09mro0uDTUZpohIcTncczN06FCio6M5e/YslSpVyl9+5513FpjlW0SKLivXxqMzN5yfDLNGAFMGdMDbU3OriYgUh8M9NytXruT333/Hx8enwPJ69epx4sSJEgcTcTd2u8GIb7aw9XgKwf7eTI/RZJgiIo5w+J+Edrsdm81WaPnx48cJDAwsUSgRd/T24j38vD0eb08LHz3YgXqaDFNExCEOFzc9e/Zk8uTJ+a8tFgvp6emMHTtWUzKIFNO3G4/zwdIDAEy4qw2dNBmmiIjDHL4s9fbbb9OrVy9atGhBdnY2/fv3Z9++fYSEhPDVV185M6OIS1t3KJlR87YB8MSNDbmnQ22TE4mIVGwOFze1a9dm69atfP3112zbto309HQefvhhBgwYUGCAsYhc3uGkDP7viw1YbQa9W4XzTE9NhikiUlIOFzfZ2dn4+fnx4IMPOjOPiNtIybTy0OfnJ8NsU7sKk+7TZJgiIs7g8Jib0NBQYmJiWLx4MXa73ZmZRFye1WbniVkbOXg6g5pV/PhkkCbDFBFxFoeLm88//5zMzEz+/ve/U6tWLYYNG8aGDRucmU3EJRmGwZgfdvLb/jP4+5yfDDNUk2GKiDiNw8XNnXfeyTfffENCQgKvv/46f/zxB9deey1NmjTh5ZdfdmZGEZfy6apDfLXuwmSY97ejZYQmwxQRcaYSP/o0MDCQwYMHs2jRIrZt20ZAQADjx493RjYRl/PrHwm89vMuAF7o05zumgxTRMTpSlzcZGdnM2fOHPr27Uv79u1JTk7m2WefdUY2EZey82QKT399fjLMBzrW4eHrNBmmiEhpcPhuqYULFzJr1iy+//57vLy8uOeee1i0aBHdunVzZj4Rl3B+MswNZOba6NqoOi//vaUmwxQRKSUOFzd33nknf/vb35g5cyZ9+vTB21tz4IhcSlaujUdmbuDUhckwP+yvyTBFREqTw8VNQkKC5pASuQq73SB2zha2HU+hqr83n/1Dk2GKiJQ2h//5GBgYyIEDB3jxxRd54IEHSExMBOCXX35h586dTgsoUpH9a9EeftlxYTLMgdHUra7JMEVESpvDxc3y5ctp3bo1a9euZd68eaSnpwOwdetWxo4d67SAIhXVNxuO8eGy85NhvnFXGzrWr2ZyIhER9+BwcTNq1CheffVVFi9ejI+PT/7ym2++mTVr1jglnEhFte5wMs9/tx2AJ29qyN2aDFNEpMw4XNxs376dO++8s9Dy0NBQkpKSShRKpCI7nQVPztqK1WbQp3U4I3poMkwRkbLkcHETHBzMqVOnCi3fvHkztWrVKlEokYoqJcvKtN2enMuy0rZ2Fd6+V5NhioiUNYeLm/vvv5+RI0cSHx+PxWLBbrfz22+/8cwzzzBo0CBnZhSpEKw2O099vZXEbAs1q/jxsSbDFBExhcPFzeuvv06zZs2IjIwkPT2dFi1a0K1bN7p06cILL7zgzIwiFcK7v+5j9cFkfD0MPhrQTpNhioiYxOHn3Pj4+PDxxx8zZswYtm/fTnp6Ou3ataNx48bOzCdSIaw5eIYPlu0H4P6GdprX1DOgRETM4nBxc1FkZCSRkZH5r7dt20Z0dDS5ubklPbRIhXAuM5fhs7dgGHB3+wja+x41O5KIiFtz+jPgDcPAZrM5+7Ai5ZJhGDz/3XZOpWRTPySAl/o0MzuSiIjb0wQ3IiUwZ8Mxft4ej5eHhXfvjyLAt8SdoSIiUkIqbkQcdOB0OuPm/wHAM72a0qZ2sLmBREQEcGDMTWpq6hXXp6WlORxGpKLIybPx9FebybLa6NKwOv+8voHZkURE5IJiFzfBwcFYLJd/KJlhGFdcL+IK3l60l50nUwn292bSfXpQn4hIeVLs4mbp0qWlkUOkwli57zTTVhwEYOLdbQivoufZiIiUJ8Uubm644YbSyCFSIZxJz2HEnK0ADOhUh14tw01OJCIi/8spA4pvu+22S84zJeJKDMNg5LfbSEzLoVFoZV68rYXZkURE5BKcUtysWLGCrKwsZxxKpNz6z5oj/LorER9PD/59fzvNGyUiUk7pVnCRItgTn8arP+0CYFTvZrSICDI5kYiIXI5Tipu6devi7e3tjEOJlDvZ1vO3fefk2bmxaQ0Gd61ndiQREbkCpzxOdceOHc44jEi59MYvu9mTkEZIZR/euqetHnUgIlLOOVzcrFix4orru3Xr5uihRcqNJbsTmPH7YQDeuqctNQJ9zQ0kIiJX5XBxc+ONNxZa9td/0WryTKnoEtOyefabbQAM7lqPm5qFmpxIRESKwuExN2fPni3wk5iYyIIFC7jmmmtYtGiRMzOKlDm73WDEnK2cycilWXggI2/VbN8iIhWFwz03VapUKbSsR48e+Pj4EBsby8aNG0sUTMRM0387xMp9Sfh6efDeA+3w89Zt3yIiFYXTbwUPCwtjz549zj6sSJnZcSKFiQt2A/DS31rQOCzQ5EQiIlIcDvfcbNu2rcBrwzA4deoUb7zxBlFRUSXNJWKKzNw8hn69GavNoEeLMAZ0qmN2JBERKSaHi5uoqCgsFguGYRRYfu211zJ9+vQSBxMxwyv/3cWB0xmEBvoy8e42uu1bRKQCcri4OXToUIHXHh4e1KhRAz8/zZAsFdOCHfF8te4oFgu80y+KagE+ZkcSEREHOFzc1K1b15k5REx1KiWLUfPOX2r9Z7cGdG0UYnIiERFxVImeUJyRkcHy5cs5evQoubm5BdY9/fTTJQomUlZsdoPY2Vs5l2mlda0qjOjR1OxIIiJSAg4XN5s3b6ZPnz5kZmaSkZFBtWrVSEpKwt/fn9DQUBU3UmF8tOIAqw+ewd/Hk3fvj8LHS/PJiohUZA5/iw8fPpzbb7+ds2fPUqlSJdasWcORI0fo0KED//rXv5yZUaTUbDl2jkmL9gIw7o6WNKhR2eREIiJSUg4XN1u2bGHEiBF4eHjg6elJTk4OkZGRvPnmmzz//PPOzChSKtJzzt/2nWc3uK11Te7tUNvsSCIi4gQOFzfe3t54eJzfPTQ0lKNHjwLnn1x87Ngx56QTKUVjf9jJkTOZRFTx4/U7W+u2bxERF+HwmJt27dqxfv16GjduzA033MCYMWNISkriiy++oFWrVs7MKOJ087ee5NtNx/GwwOT721HF39vsSCIi4iQO99y8/vrr1KxZE4DXXnuNqlWr8vjjj3P69GmmTZvmtIAiznYsOZMXvtsOwJCbGtGxfjWTE4mIiDM51HNjGAahoaH5PTShoaEsWLDAqcFESkOezc7w2VtIy86jfZ1gnr6lsdmRRETEyRzquTEMg0aNGjltbM0HH3xAvXr18PPzo1OnTqxbt+6y286bN4/o6GiCg4MJCAggKiqKL774wik5xPW9v3Q/G46cpbKvF+/e3w4vT932LSLiahz6Zvfw8KBx48acOXOmxAFmz55NbGwsY8eOZdOmTbRt25ZevXqRmJh4ye2rVavGCy+8wOrVq9m2bRuDBw9m8ODBLFy4sMRZxLVtOJzMv+P2AfDana2IrOZvciIRESkNDv+z9Y033uDZZ59lx44dJQowadIkHn30UQYPHkyLFi2YOnUq/v7+l51888Ybb+TOO++kefPmNGzYkKFDh9KmTRtWrVpVohzi2lKyrAz9egt2A+5sV4u/R9UyO5KIiJQSh++WGjRoEJmZmbRt2xYfHx8qVapUYH1ycvJVj5Gbm8vGjRsZPXp0/jIPDw+6d+/O6tWrr7q/YRgsWbKEPXv2MHHixEtuk5OTQ05OTv7r1NRUAKxWK1ar9arvURwXj+fs41YU5bX9hmHw/LztnDiXRe2qlXipT9NSy1hez0FZcff2g86B2u/e7YfSOwfFOZ7FMAzDkTf5/PPPr7g+Jibmqsc4efIktWrV4vfff6dz5875y5977jmWL1/O2rVrL7lfSkoKtWrVIicnB09PTz788EMeeuihS247btw4xo8fX2j5rFmz8PfXZQl3sO60hS/3e+KBwdBWNuoFmp1IRESKKzMzk/79+5OSkkJQUNAVt3W456YoxUtpCQwMZMuWLaSnpxMXF0dsbCwNGjTgxhtvLLTt6NGjiY2NzX+dmppKZGQkPXv2vOrJKS6r1crixYvp0aMH3t7u99yU8tj+I2cyef7D1YCNobc05okbG5Tq+5XHc1CW3L39oHOg9rt3+6H0zsHFKy9FUaJZwQ8cOMBnn33GgQMHePfddwkNDeWXX36hTp06tGzZ8qr7h4SE4OnpSUJCQoHlCQkJhIeHX3Y/Dw8PGjVqBEBUVBS7du1iwoQJlyxufH198fX1LbTc29u71D54pXnsiqC8tN9qszNi7nYycm10rF+NIbc0wdOjbJ5CXF7OgVncvf2gc6D2u3f7wfnnoDjHcnhA8fLly2ndujVr165l3rx5pKenA7B161bGjh1bpGP4+PjQoUMH4uLi8pfZ7Xbi4uIKXKa6GrvdXmBcjQjAO4v3svV4CkF+XkzuF1VmhY2IiJjL4eJm1KhRvPrqqyxevBgfH5/85TfffDNr1qwp8nFiY2P5+OOP+fzzz9m1axePP/44GRkZDB48GDg/cPmvA44nTJjA4sWLOXjwILt27eLtt9/miy++4MEHH3S0KeKCfj+QxJTlBwB44+42RARXusoeIiLiKhy+LLV9+3ZmzZpVaHloaChJSUlFPk6/fv04ffo0Y8aMIT4+nqioKBYsWEBYWBgAR48ezZ+gEyAjI4MnnniC48ePU6lSJZo1a8Z//vMf+vXr52hTxMWczcgldvZWDAP6RUfSp3VNsyOJiEgZcri4CQ4O5tSpU9SvX7/A8s2bN1OrVvGeITJkyBCGDBlyyXXLli0r8PrVV1/l1VdfLdbxxX0YhsGoeduIT82mQUgAY25vYXYkEREpYw5flrr//vsZOXIk8fHxWCwW7HY7v/32G8888wyDBg1yZkaRIvt6/TEW7kzA29PCu/e3I8C3RGPmRUSkAirRrODNmjUjMjKS9PR0WrRoQbdu3ejSpQsvvviiMzOKFMmKvacZ/+NOAJ7t1ZTWtauYnEhERMzg8D9rfXx8+Pjjj3nppZfYsWMH6enptGvXjsaNNcuylK2jZzJ55ac/WPzH+UcKXNcohEeuK93n2YiISPlV4j77OnXqUKdOHWdkESmWjJw8Pli6n09WHiLXZsfTw8LAa+syomcTPHTbt4iI23K4uDEMg7lz57J06VISExOx2+0F1s+bN6/E4UQuxW43+H7LCd74ZTeJaeefb3RdoxDG3N6CJmGaW0FExN05XNwMGzaMjz76iJtuuomwsDAsFv1LWUrf1mPnGPfjTjYfPQdAnWr+vHhbc3q00GdQRETOc7i4+eKLL5g3bx59+vRxZh6RS0pMy+bNBXuYu/E4AP4+njx5UyMevq4+ft6eJqcTEZHyxOHipkqVKjRooEGbUrpy8mzM+O0w7y3ZT3pOHgB3tavFyN7NCAvyMzmdiIiURw4XN+PGjWP8+PFMnz6dSpX0aHtxLsMwWLI7kVf++weHz2QC0LZ2Fcbe0ZL2daqanE5ERMozh4ub++67j6+++orQ0FDq1atXaLbOTZs2lTicuKf9iem88t8/WL73NAAhlX0ZeWtT7m5fW3dBiYjIVTlc3MTExLBx40YefPBBDSgWp0jJsvLvuH18/vth8uwG3p4WHrquPkNuakSgX9GnuhcREffmcHHz008/sXDhQq677jpn5hE3ZLMbzNlwjH8t3MOZjFwAbmkWyot/a0H9kACT04mISEXjcHETGRlJUFCQM7OIG1p/OJlx83ey82QqAA1rBPDS31pwY9NQk5OJiEhF5XBx8/bbb/Pcc88xdepU6tWr58RI4g5Onstiwi+7+XHrSQAC/bwY1r0JgzrXxdvT4SnPREREHC9uHnzwQTIzM2nYsCH+/v6FBhQnJyeXOJy4nmyrjWkrDjJl2QGyrDYsFrj/mkhG9GxKSGVfs+OJiIgLcLi4mTx5shNjiKszDINfdsTz2k+7OHEuC4Br6lVl7O0taVVLs3eLiIjzlOhuKZGi2HUqlfE/7mTNwfO9eTWr+DG6T3Nub1NTd9mJiIjTlXhWcIDs7Gxyc3MLLNNgYzmbkcvbi/cwa+1R7Ab4ennwfzc05LEbGuDv45SPnoiISCEO/w2TkZHByJEjmTNnDmfOnCm03mazlSiYVFx5NjtfrjvEO7/uIyXLCsBtrWsyqnczIqv5m5xORERcncPFzXPPPcfSpUuZMmUKAwcO5IMPPuDEiRN89NFHvPHGG87MKBWEYRjsSbHw/oer2ZeYAUCz8EDG3t6Szg2rm5xORETchcPFzY8//sjMmTO58cYbGTx4MNdffz2NGjWibt26fPnllwwYMMCZOaUEbHaDbKuNbKuNrIu/c+1k59nIyv1z2fnlNrKs9gLbZ+XayM6zn//912NYbWTn73/+eIbhCWRQ1d+bET2bcv81kXjp1m4RESlDDhc3ycnJ+bOCBwUF5d/6fd111/H44487J50UyycrDzJ34/G/FCB2sqw2cvPsZZbBw2IwoFNdRvRsSrC/T5m9r4iIyEUOFzcNGjTg0KFD1KlTh2bNmjFnzhw6duzIjz/+SHBwsBMjSlEcTspgwi+7sdmNK27n6+VBJR9P/Lw8z//29qSSt8eF35745a/zoNKFZb4Xfp/f/vzy/O0vLK/k7Ykndn5bFkff25oVeu6RiIhIWXG4uBk8eDBbt27lhhtuYNSoUdx+++28//77WK1WJk2a5MyMUgTvL92PzW7QtVF1hndvgt//FB5+3h74eXmW6qzaVqsVH89SO7yIiEiROFzcDB8+PP/P3bt3Z/fu3WzcuJFGjRrRpk0bp4STojlyJoPvNp8A4NlezYiKDDY3kIiIiIkcKm7sdjszZsxg3rx5HD58GIvFQv369bnnnnto3bq1szPKVby/5HyvzY1Na6iwERERt1fs21gMw+COO+7gkUce4cSJE7Ru3ZqWLVty5MgR/vGPf3DnnXeWRk65jCNnMph3oddm6C2NTU4jIiJivmL33MyYMYMVK1YQFxfHTTfdVGDdkiVL6Nu3LzNnzmTQoEFOCymX98GFsTY3NKlBuzpVzY4jIiJiumL33Hz11Vc8//zzhQobgJtvvplRo0bx5ZdfOiWcXNnRM5l8u+lCr0139dqIiIiAA8XNtm3buPXWWy+7vnfv3mzdurVEoaRoLvbadGtSg/bqtREREQEcKG6Sk5MJCwu77PqwsDDOnj1bolBydceSM/l203FAY21ERET+qtjFjc1mw8vr8kN1PD09ycvLK1EouboPlu4nz25wfeMQOtRVr42IiMhFxR5QbBgG//jHP/D19b3k+pycnBKHkis7lpzJ3I3ne22GaayNiIhIAcUubmJiYq66je6UKl0fLvtrr001s+OIiIiUK8Uubj777LPSyCFFdCw5k282aKyNiIjI5RR7zI2Y68NlB8izG1zXKIToeuq1ERER+V8qbiqQ42cz+WbDMUDPtREREbkcFTcVyMVem66NqnONem1EREQuScVNBXHiXNafvTa3NDE5jYiISPml4qaC+HDpfqw2g84NqtOxvnptRERELkfFTQVw8lwWczTWRkREpEhU3FQAHy4732tzbYNqXNugutlxREREyjUVN+XcyXNZzFl/8bk2GmsjIiJyNSpuyrkpyw6Qa7PTqX41OjdUr42IiMjVqLgpx06lZDF7/fmxNsO6q9dGRESkKFTclGMXe206qtdGRESkyFTclFPxKdl8ve5ir43ukBIRESkqFTfl1JRl+8/32tSrRmfdISUiIlJkKm7KofiUbL5a/2evjcViMTmRiIhIxaHiphyauvwAuXl2rqlXVWNtREREiknFTTmTkJrNrHVHgfN3SKnXRkREpHhU3JQzU5ad77WJrluVLuq1ERERKTYVN+VIYmo2X6nXRkREpERU3JQjU5YfICfPToe6VenaSL02IiIijlBxU04kpmYza+35Xpuht+gOKREREUepuCknpi4/SE6enfZ1grm+cYjZcURERCosFTflQGJaNl+uPQLAUI21ERERKREVN+XAtAu9Nu3qBNNNvTYiIiIlouLGZKfTcvjPxV4bjbUREREpMRU3Jpu24gDZVjtRkcHc0KSG2XFEREQqPBU3JjqdlsMXay6OtVGvjYiIiDOouDHRxysPkm210zYymBvVayMiIuIUKm5MkpSew8zVhwEYprE2IiIiTqPixiQfr7jQa1O7Cjc2Va+NiIiIs5SL4uaDDz6gXr16+Pn50alTJ9atW3fZbT/++GOuv/56qlatStWqVenevfsVty+PzvfaaKyNiIhIaTC9uJk9ezaxsbGMHTuWTZs20bZtW3r16kViYuIlt1+2bBkPPPAAS5cuZfXq1URGRtKzZ09OnDhRxskd9/HKg2RZbbSpXYWbmoaaHUdERMSleJkdYNKkSTz66KMMHjwYgKlTp/LTTz8xffp0Ro0aVWj7L7/8ssDrTz75hG+//Za4uDgGDRpUaPucnBxycnLyX6empgJgtVqxWq3ObEr+8a503DMZucz8/TAAT97YgLy8PKdmMFNR2u/q3P0cuHv7QedA7Xfv9kPpnYPiHM9iGIbh1HcvhtzcXPz9/Zk7dy59+/bNXx4TE8O5c+f44YcfrnqMtLQ0QkND+eabb/jb3/5WaP24ceMYP358oeWzZs3C39+/RPkdMf+IB3EnPYgMMBjR2oauSImIiFxdZmYm/fv3JyUlhaCgoCtua2rPTVJSEjabjbCwsALLw8LC2L17d5GOMXLkSCIiIujevfsl148ePZrY2Nj816mpqfmXsq52corLarWyePFievTogbe3d6H1yRm5jJ60ErDxQt923NLMtS5JXa397sDdz4G7tx90DtR+924/lN45uHjlpShMvyxVEm+88QZff/01y5Ytw8/P75Lb+Pr64uvrW2i5t7d3qX3wLnfsGWsOkJlro1WtIHq1inDZgcSleW4rCnc/B+7eftA5UPvdu/3g/HNQnGOZWtyEhITg6elJQkJCgeUJCQmEh4dfcd9//etfvPHGG/z666+0adOmNGM6RXJGLp9fGGsz9BbN/C0iIlJaTL1bysfHhw4dOhAXF5e/zG63ExcXR+fOnS+735tvvskrr7zCggULiI6OLouoJfbJyoNk5tpoGRFE9+audTlKRESkPDH9slRsbCwxMTFER0fTsWNHJk+eTEZGRv7dU4MGDaJWrVpMmDABgIkTJzJmzBhmzZpFvXr1iI+PB6By5cpUrlzZtHZcydkCvTZ6ro2IiEhpMr246devH6dPn2bMmDHEx8cTFRXFggUL8gcZHz16FA+PPzuYpkyZQm5uLvfcc0+B44wdO5Zx48aVZfQi+2TVQTJybbSoGUSPFmFX30FEREQcZnpxAzBkyBCGDBlyyXXLli0r8Prw4cOlH8iJzmbkMuO3w4CeRiwiIlIWTH9Csav7dNUhMnJtNK8ZRE/12oiIiJQ6FTel6FxmLjM01kZERKRMqbgpRZ+uOkR6Th7NwgPVayMiIlJGVNyUknOZf461Gda9MR4e6rUREREpCypuSsn0VYdIy++1ufIDCUVERMR5VNyUgpQsK59d6LV5+hb12oiIiJQlFTelYMbvR0jLyaNpWCC3tlSvjYiISFkqF8+5cSWZefD55qOAem1ERETMoJ4bJ1t+yoO07DyahFWmdyv12oiIiJQ1FTdOlJplZfmp8z016rURERExh4obJ/p89VGybBYa1QigT6uaZscRERFxSypunCQly8qM1UcAGHJTQ/XaiIiImEQDip3k5Lksqgf44G+xcmtLPY1YRETELOq5cZLmNYP45emu/F9zG57qtRERETGNihsn8vSwUM3X7BQiIiLuTcWNiIiIuBQVNyIiIuJSVNyIiIiIS1FxIyIiIi5FxY2IiIi4FBU3IiIi4lJU3IiIiIhLUXEjIiIiLkXFjYiIiLgUFTciIiLiUlTciIiIiEtRcSMiIiIuRcWNiIiIuBQvswOUNcMwAEhNTXX6sa1WK5mZmaSmpuLt7e3045d37t5+0Dlw9/aDzoHa797th9I7Bxf/3r749/iVuF1xk5aWBkBkZKTJSURERKS40tLSqFKlyhW3sRhFKYFciN1u5+TJkwQGBmKxWJx67NTUVCIjIzl27BhBQUFOPXZF4O7tB50Dd28/6Byo/e7dfii9c2AYBmlpaURERODhceVRNW7Xc+Ph4UHt2rVL9T2CgoLc9kMNaj/oHLh7+0HnQO137/ZD6ZyDq/XYXKQBxSIiIuJSVNyIiIiIS1Fx40S+vr6MHTsWX19fs6OYwt3bDzoH7t5+0DlQ+927/VA+zoHbDSgWERER16aeGxEREXEpKm5ERETEpai4EREREZei4kZERERcioobJ/nggw+oV68efn5+dOrUiXXr1pkdqcxMmDCBa665hsDAQEJDQ+nbty979uwxO5Zp3njjDSwWC8OGDTM7Spk6ceIEDz74INWrV6dSpUq0bt2aDRs2mB2rTNhsNl566SXq169PpUqVaNiwIa+88kqR5sCpqFasWMHtt99OREQEFouF77//vsB6wzAYM2YMNWvWpFKlSnTv3p19+/aZE7YUXKn9VquVkSNH0rp1awICAoiIiGDQoEGcPHnSvMCl4Gqfgb967LHHsFgsTJ48uUyyqbhxgtmzZxMbG8vYsWPZtGkTbdu2pVevXiQmJpodrUwsX76cJ598kjVr1rB48WKsVis9e/YkIyPD7Ghlbv369Xz00Ue0adPG7Chl6uzZs3Tt2hVvb29++eUX/vjjD95++22qVq1qdrQyMXHiRKZMmcL777/Prl27mDhxIm+++Sbvvfee2dFKTUZGBm3btuWDDz645Po333yTf//730ydOpW1a9cSEBBAr169yM7OLuOkpeNK7c/MzGTTpk289NJLbNq0iXnz5rFnzx7uuOMOE5KWnqt9Bi767rvvWLNmDREREWWUDDCkxDp27Gg8+eST+a9tNpsRERFhTJgwwcRU5klMTDQAY/ny5WZHKVNpaWlG48aNjcWLFxs33HCDMXToULMjlZmRI0ca1113ndkxTHPbbbcZDz30UIFld911lzFgwACTEpUtwPjuu+/yX9vtdiM8PNx466238pedO3fO8PX1Nb766isTEpau/23/paxbt84AjCNHjpRNqDJ2uXNw/Phxo1atWsaOHTuMunXrGu+8806Z5FHPTQnl5uayceNGunfvnr/Mw8OD7t27s3r1ahOTmSclJQWAatWqmZykbD355JPcdtttBT4L7mL+/PlER0dz7733EhoaSrt27fj444/NjlVmunTpQlxcHHv37gVg69atrFq1it69e5uczByHDh0iPj6+wP8LVapUoVOnTm79vWixWAgODjY7Spmx2+0MHDiQZ599lpYtW5bpe7vdxJnOlpSUhM1mIywsrMDysLAwdu/ebVIq89jtdoYNG0bXrl1p1aqV2XHKzNdff82mTZtYv3692VFMcfDgQaZMmUJsbCzPP/8869ev5+mnn8bHx4eYmBiz45W6UaNGkZqaSrNmzfD09MRms/Haa68xYMAAs6OZIj4+HuCS34sX17mT7OxsRo4cyQMPPOBWk2lOnDgRLy8vnn766TJ/bxU34lRPPvkkO3bsYNWqVWZHKTPHjh1j6NChLF68GD8/P7PjmMJutxMdHc3rr78OQLt27dixYwdTp051i+Jmzpw5fPnll8yaNYuWLVuyZcsWhg0bRkREhFu0Xy7ParVy3333YRgGU6ZMMTtOmdm4cSPvvvsumzZtwmKxlPn767JUCYWEhODp6UlCQkKB5QkJCYSHh5uUyhxDhgzhv//9L0uXLqV27dpmxykzGzduJDExkfbt2+Pl5YWXlxfLly/n3//+N15eXthsNrMjlrqaNWvSokWLAsuaN2/O0aNHTUpUtp599llGjRrF/fffT+vWrRk4cCDDhw9nwoQJZkczxcXvPnf/XrxY2Bw5coTFixe7Va/NypUrSUxMpE6dOvnfi0eOHGHEiBHUq1ev1N9fxU0J+fj40KFDB+Li4vKX2e124uLi6Ny5s4nJyo5hGAwZMoTvvvuOJUuWUL9+fbMjlalbbrmF7du3s2XLlvyf6OhoBgwYwJYtW/D09DQ7Yqnr2rVrodv/9+7dS926dU1KVLYyMzPx8Cj4derp6Yndbjcpkbnq169PeHh4ge/F1NRU1q5d6zbfixcLm3379vHrr79SvXp1syOVqYEDB7Jt27YC34sRERE8++yzLFy4sNTfX5elnCA2NpaYmBiio6Pp2LEjkydPJiMjg8GDB5sdrUw8+eSTzJo1ix9++IHAwMD8a+pVqlShUqVKJqcrfYGBgYXGFwUEBFC9enW3GXc0fPhwunTpwuuvv859993HunXrmDZtGtOmTTM7Wpm4/fbbee2116hTpw4tW7Zk8+bNTJo0iYceesjsaKUmPT2d/fv3578+dOgQW7ZsoVq1atSpU4dhw4bx6quv0rhxY+rXr89LL71EREQEffv2NS+0E12p/TVr1uSee+5h06ZN/Pe//8Vms+V/L1arVg0fHx+zYjvV1T4D/1vQeXt7Ex4eTtOmTUs/XJnck+UG3nvvPaNOnTqGj4+P0bFjR2PNmjVmRyozwCV/PvvsM7OjmcbdbgU3DMP48ccfjVatWhm+vr5Gs2bNjGnTppkdqcykpqYaQ4cONerUqWP4+fkZDRo0MF544QUjJyfH7GilZunSpZf8/z4mJsYwjPO3g7/00ktGWFiY4evra9xyyy3Gnj17zA3tRFdq/6FDhy77vbh06VKzozvN1T4D/6ssbwW3GIYLP0JTRERE3I7G3IiIiIhLUXEjIiIiLkXFjYiIiLgUFTciIiLiUlTciIiIiEtRcSMiIiIuRcWNiIiIuBQVNyIiIuJSVNyIiAAWi4Xvv//e7Bgi4gQqbkTEdP/4xz+wWCyFfm699Vazo4lIBaSJM0WkXLj11lv57LPPCizz9fU1KY2IVGTquRGRcsHX15fw8PACP1WrVgXOXzKaMmUKvXv3plKlSjRo0IC5c+cW2H/79u3cfPPNVKpUierVq/PPf/6T9PT0AttMnz6dli1b4uvrS82aNRkyZEiB9UlJSdx55534+/vTuHFj5s+fX7qNFpFSoeJGRCqEl156ibvvvputW7cyYMAA7r//fnbt2gVARkYGvXr1omrVqqxfv55vvvmGX3/9tUDxMmXKFJ588kn++c9/sn37dubPn0+jRo0KvMf48eO577772LZtG3369GHAgAEkJyeXaTtFxAnKZO5xEZEriImJMTw9PY2AgIACP6+99pphGIYBGI899liBfTp16mQ8/vjjhmEYxrRp04yqVasa6enp+et/+uknw8PDw4iPjzcMwzAiIiKMF1544bIZAOPFF1/Mf52enm4Axi+//OK0dopI2dCYGxEpF2666SamTJlSYFm1atXy/9y5c+cC6zp37syWLVsA2LVrF23btiUgICB/fdeuXbHb7ezZsweLxcLJkye55ZZbrpihTZs2+X8OCAggKCiIxMRER5skIiZRcSMi5UJAQEChy0TOUqlSpSJt5+3tXeC1xWLBbreXRiQRKUUacyMiFcKaNWsKvW7evDkAzZs3Z+vWrWRkZOSv/+233/Dw8KBp06YEBgZSr1494uLiyjSziJhDPTciUi7k5OQQHx9fYJmXlxchISEAfPPNN0RHR3Pdddfx5Zdfsm7dOj799FMABgwYwNixY4mJiWHcuHGcPn2ap556ioEDBxIWFgbAuHHjeOyxxwgNDaV3796kpaXx22+/8dRTT5VtQ0Wk1Km4EZFyYcGCBdSsWbPAsqZNm7J7927g/J1MX3/9NU888QQ1a9bkq6++okWLFgD4+/uzcOFChg4dyjXXXIO/vz933303kyZNyj9WTEwM2dnZvPPOOzzzzDOEhIRwzz33lF0DRaTMWAzDMMwOISJyJRaLhe+++46+ffuaHUVEKgCNuRERERGXouJGREREXIrG3IhIuaer5yJSHOq5EREREZei4kZERERcioobERERcSkqbkRERMSlqLgRERERl6LiRkRERFyKihsRERFxKSpuRERExKX8Pxx4HPtKOAyxAAAAAElFTkSuQmCC",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "test:\n",
            "{'jaccard_similarity': 0.7656637423081285, 'cosine_similarity': 0.8661194847755104, 'damerau-levenshtein_similarity': 0.6343659405293635, 'loss': 1.1603623446894855}\n"
          ]
        }
      ],
      "source": [
        "if __name__ == \"__main__\":\n",
        "    # not using absl (no need for command line), get config manually\n",
        "    config = get_config()\n",
        "    config.unlock() # update hyperparams before `config.lock()`\n",
        "    config.storage_folder = \"./storage/attention-attention\"\n",
        "    config.speech_to_text.decoder_type = DecoderType.TRANSFORMER\n",
        "    config.speech_to_text.decoder_kwargs = dict(\n",
        "        freq_dim=config.get_ref(\"n_vocab\"),\n",
        "        time_dim=config.get_ref(\"text_len\"),\n",
        "        hidden_dim=200,\n",
        "        n_heads=2,\n",
        "        feed_fwd_dim=400\n",
        "    )\n",
        "    config.lock()\n",
        "\n",
        "    # Manual seed\n",
        "    torch.manual_seed(config.seed)\n",
        "    np.random.seed(config.seed)\n",
        "    random.seed(config.seed)\n",
        "\n",
        "    # Device to use\n",
        "    DEVICE = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n",
        "    print(f\"Using {DEVICE}\")\n",
        "\n",
        "    # Training & testing\n",
        "    trainer = Trainer(config, DEVICE)\n",
        "    trainer.train()\n",
        "    trainer.test()\n",
        "\n",
        "    # Plotting - check './storage/attention-attention/{seed}/plots\n",
        "    plotting(config)"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "gpuType": "T4",
      "provenance": []
    },
    "kaggle": {
      "accelerator": "gpu",
      "dataSources": [],
      "dockerImageVersionId": 30627,
      "isGpuEnabled": true,
      "isInternetEnabled": true,
      "language": "python",
      "sourceType": "notebook"
    },
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.11.5"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}