en-yo/jw300-baseline-improve/en_yo_jw300_notebook_gdrive.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "name": "starter_notebook.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true,
      "include_colab_link": true
    },
    "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": "3.5.6"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/bamtak/masakhane/blob/master/en-yo/jw300-baseline-improve/en_yo_jw300_notebook_gdrive.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "Igc5itf-xMGj"
      },
      "source": [
        "# Masakhane - Machine Translation for African Languages (Using JoeyNMT)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "x4fXCKCf36IK"
      },
      "source": [
        "## Note before beginning:\n",
        "### - The idea is that you should be able to make minimal changes to this in order to get SOME result for your own translation corpus. \n",
        "\n",
        "### - The tl;dr: Go to the **\"TODO\"** comments which will tell you what to update to get up and running\n",
        "\n",
        "### - If you actually want to have a clue what you're doing, read the text and peek at the links\n",
        "\n",
        "### - With 100 epochs, it should take around 7 hours to run in Google Colab\n",
        "\n",
        "### - Once you've gotten a result for your language, please attach and email your notebook that generated it to masakhanetranslation@gmail.com\n",
        "\n",
        "### - If you care enough and get a chance, doing a brief background on your language would be amazing. See examples in  [(Martinus, 2019)](https://arxiv.org/abs/1906.05685)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "l929HimrxS0a"
      },
      "source": [
        "## Retrieve your data & make a parallel corpus\n",
        "\n",
        "If you are wanting to use the JW300 data referenced on the Masakhane website or in our GitHub repo, you can use `opus-tools` to convert the data into a convenient format. `opus_read` from that package provides a convenient tool for reading the native aligned XML files and to convert them to TMX format. The tool can also be used to fetch relevant files from OPUS on the fly and to filter the data as necessary. [Read the documentation](https://pypi.org/project/opustools-pkg/) for more details.\n",
        "\n",
        "Once you have your corpus files in TMX format (an xml structure which will include the sentences in your target language and your source language in a single file), we recommend reading them into a pandas dataframe. Thankfully, Jade wrote a silly `tmx2dataframe` package which converts your tmx file to a pandas dataframe. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "oGRmDELn7Az0",
        "outputId": "98551454-5471-46f5-81fc-fa483f28d350",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 127
        }
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Go to this URL in a browser: https://accounts.google.com/o/oauth2/auth?client_id=947318989803-6bn6qk8qdgf4n4g3pfee6491hc0brc4i.apps.googleusercontent.com&redirect_uri=urn%3aietf%3awg%3aoauth%3a2.0%3aoob&response_type=code&scope=email%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdocs.test%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdrive%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdrive.photos.readonly%20https%3a%2f%2fwww.googleapis.com%2fauth%2fpeopleapi.readonly\n",
            "\n",
            "Enter your authorization code:\n",
            "··········\n",
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "Cn3tgQLzUxwn",
        "colab": {}
      },
      "source": [
        "# TODO: Set your source and target languages. Keep in mind, these traditionally use language codes as found here:\n",
        "# These will also become the suffix's of all vocab and corpus files used throughout\n",
        "import os\n",
        "source_language = \"en\"\n",
        "target_language = \"yo\" \n",
        "lc = False  # If True, lowercase the data.\n",
        "seed = 42  # Random seed for shuffling.\n",
        "tag = \"baseline\" # Give a unique name to your folder - this is to ensure you don't rewrite any models you've already submitted\n",
        "\n",
        "os.environ[\"src\"] = source_language # Sets them in bash as well, since we often use bash scripts\n",
        "os.environ[\"tgt\"] = target_language\n",
        "os.environ[\"tag\"] = tag\n",
        "\n",
        "# This will save it to a folder in our gdrive instead! \n",
        "!mkdir -p \"/content/drive/My Drive/masakhane/$src-$tgt-$tag\"\n",
        "g_drive_path = \"/content/drive/My Drive/masakhane/%s-%s-%s\" % (source_language, target_language, tag)\n",
        "os.environ[\"gdrive_path\"] = g_drive_path\n",
        "models_path = '%s/models/%s%s_transformer'% (g_drive_path, source_language, target_language)\n",
        "# model temporary directory for training\n",
        "model_temp_dir = \"/content/drive/My Drive/masakhane/model-temp\"\n",
        "# model permanent storage on the drive\n",
        "!mkdir -p \"$gdrive_path/models/${src}${tgt}_transformer/\""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "kBSgJHEw7Nvx",
        "outputId": "0af3ab77-2c6a-431e-a299-9ea9f7904869",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "source": [
        "!echo $gdrive_path"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/content/drive/My Drive/masakhane/en-yo-baseline\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "gA75Fs9ys8Y9",
        "outputId": "835f0426-5e3e-4c70-8737-d1f1677ee041",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "source": [
        "#TODO: Skip for retrain\n",
        "# Install opus-tools\n",
        "! pip install opustools-pkg "
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: opustools-pkg in /usr/local/lib/python3.6/dist-packages (0.0.52)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "xq-tDZVks7ZD",
        "outputId": "96da3b64-d252-4747-95e9-38645b578431",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 215
        }
      },
      "source": [
        "#TODO: Skip for retrain\n",
        "# Downloading our corpus\n",
        "! opus_read -d JW300 -s $src -t $tgt -wm moses -w jw300.$src jw300.$tgt -q\n",
        "\n",
        "# extract the corpus file\n",
        "! gunzip JW300_latest_xml_$src-$tgt.xml.gz"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Alignment file /proj/nlpl/data/OPUS/JW300/latest/xml/en-yo.xml.gz not found. The following files are available for downloading:\n",
            "\n",
            "        ./JW300_latest_xml_en.zip already exists\n",
            "        ./JW300_latest_xml_yo.zip already exists\n",
            "   4 MB https://object.pouta.csc.fi/OPUS-JW300/v1/xml/en-yo.xml.gz\n",
            "\n",
            "   4 MB Total size\n",
            "./JW300_latest_xml_en-yo.xml.gz ... 100% of 4 MB\n",
            "gzip: JW300_latest_xml_en-yo.xml already exists; do you wish to overwrite (y or n)? n\n",
            "\tnot overwritten\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "n48GDRnP8y2G",
        "colab_type": "code",
        "outputId": "10dc77a6-1b2c-4db7-dea0-90648614af5b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 611
        }
      },
      "source": [
        "#TODO: Skip for retrain\n",
        "# Download the global test set.\n",
        "! wget https://raw.githubusercontent.com/juliakreutzer/masakhane/master/jw300_utils/test/test.en-any.en\n",
        "  \n",
        "# And the specific test set for this language pair.\n",
        "os.environ[\"trg\"] = target_language \n",
        "os.environ[\"src\"] = source_language \n",
        "\n",
        "! wget https://raw.githubusercontent.com/juliakreutzer/masakhane/master/jw300_utils/test/test.en-$trg.en \n",
        "! mv test.en-$trg.en test.en\n",
        "! wget https://raw.githubusercontent.com/juliakreutzer/masakhane/master/jw300_utils/test/test.en-$trg.$trg \n",
        "! mv test.en-$trg.$trg test.$trg"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2020-04-12 10:38:37--  https://raw.githubusercontent.com/juliakreutzer/masakhane/master/jw300_utils/test/test.en-any.en\n",
            "Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 151.101.0.133, 151.101.64.133, 151.101.128.133, ...\n",
            "Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|151.101.0.133|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 277791 (271K) [text/plain]\n",
            "Saving to: ‘test.en-any.en’\n",
            "\n",
            "\rtest.en-any.en        0%[                    ]       0  --.-KB/s               \rtest.en-any.en      100%[===================>] 271.28K  --.-KB/s    in 0.02s   \n",
            "\n",
            "2020-04-12 10:38:38 (17.4 MB/s) - ‘test.en-any.en’ saved [277791/277791]\n",
            "\n",
            "--2020-04-12 10:38:40--  https://raw.githubusercontent.com/juliakreutzer/masakhane/master/jw300_utils/test/test.en-yo.en\n",
            "Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 151.101.0.133, 151.101.64.133, 151.101.128.133, ...\n",
            "Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|151.101.0.133|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 201994 (197K) [text/plain]\n",
            "Saving to: ‘test.en-yo.en’\n",
            "\n",
            "test.en-yo.en       100%[===================>] 197.26K  --.-KB/s    in 0.02s   \n",
            "\n",
            "2020-04-12 10:38:40 (12.4 MB/s) - ‘test.en-yo.en’ saved [201994/201994]\n",
            "\n",
            "--2020-04-12 10:38:45--  https://raw.githubusercontent.com/juliakreutzer/masakhane/master/jw300_utils/test/test.en-yo.yo\n",
            "Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 151.101.0.133, 151.101.64.133, 151.101.128.133, ...\n",
            "Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|151.101.0.133|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 280073 (274K) [text/plain]\n",
            "Saving to: ‘test.en-yo.yo’\n",
            "\n",
            "test.en-yo.yo       100%[===================>] 273.51K  --.-KB/s    in 0.02s   \n",
            "\n",
            "2020-04-12 10:38:45 (13.8 MB/s) - ‘test.en-yo.yo’ saved [280073/280073]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NqDG-CI28y2L",
        "colab_type": "code",
        "outputId": "58edfffd-d1fe-47e4-aed8-bed2d882b389",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "source": [
        "#TODO: Skip for retrain\n",
        "# Read the test data to filter from train and dev splits.\n",
        "# Store english portion in set for quick filtering checks.\n",
        "en_test_sents = set()\n",
        "filter_test_sents = \"test.en-any.en\"\n",
        "j = 0\n",
        "with open(filter_test_sents) as f:\n",
        "  for line in f:\n",
        "    en_test_sents.add(line.strip())\n",
        "    j += 1\n",
        "print('Loaded {} global test sentences to filter from the training/dev data.'.format(j))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Loaded 3571 global test sentences to filter from the training/dev data.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "3CNdwLBCfSIl",
        "outputId": "7dd3286b-6348-4c6a-bdea-bd43a4c04dfb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 160
        }
      },
      "source": [
        "#TODO: Skip for retrain\n",
        "import pandas as pd\n",
        "\n",
        "# TMX file to dataframe\n",
        "source_file = 'jw300.' + source_language\n",
        "target_file = 'jw300.' + target_language\n",
        "\n",
        "source = []\n",
        "target = []\n",
        "skip_lines = []  # Collect the line numbers of the source portion to skip the same lines for the target portion.\n",
        "with open(source_file) as f:\n",
        "    for i, line in enumerate(f):\n",
        "        # Skip sentences that are contained in the test set.\n",
        "        if line.strip() not in en_test_sents:\n",
        "            source.append(line.strip())\n",
        "        else:\n",
        "            skip_lines.append(i)             \n",
        "with open(target_file) as f:\n",
        "    for j, line in enumerate(f):\n",
        "        # Only add to corpus if corresponding source was not skipped.\n",
        "        if j not in skip_lines:\n",
        "            target.append(line.strip())\n",
        "    \n",
        "print('Loaded data and skipped {}/{} lines since contained in test set.'.format(len(skip_lines), i))\n",
        "    \n",
        "df = pd.DataFrame(zip(source, target), columns=['source_sentence', 'target_sentence'])\n",
        "# if you get TypeError: data argument can't be an iterator is because of your zip version run this below\n",
        "#df = pd.DataFrame(list(zip(source, target)), columns=['source_sentence', 'target_sentence'])\n",
        "df.head(3)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Loaded data and skipped 5663/474986 lines since contained in test set.\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>source_sentence</th>\n",
              "      <th>target_sentence</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Using Ladders — Do You Make These Safety Checks ?</td>\n",
              "      <td>Lílo Àkàbà — Ǹjẹ́ O Máa Ń Ṣe Àyẹ̀wò Wọ̀nyí Tó...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>By Awake !</td>\n",
              "      <td>Látọwọ́ akọ̀ròyìn Jí !</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>correspondent in Ireland</td>\n",
              "      <td>ní Ireland</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                     source_sentence                                    target_sentence\n",
              "0  Using Ladders — Do You Make These Safety Checks ?  Lílo Àkàbà — Ǹjẹ́ O Máa Ń Ṣe Àyẹ̀wò Wọ̀nyí Tó...\n",
              "1                                         By Awake !                             Látọwọ́ akọ̀ròyìn Jí !\n",
              "2                           correspondent in Ireland                                         ní Ireland"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 25
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PvLN2qO1_NuA",
        "colab_type": "code",
        "outputId": "5f6e31c7-2d79-4ca5-d404-368b86bf261d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "source": [
        "print(df.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(469324, 2)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "YkuK3B4p2AkN"
      },
      "source": [
        "## Pre-processing and export\n",
        "\n",
        "It is generally a good idea to remove duplicate translations and conflicting translations from the corpus. In practice, these public corpora include some number of these that need to be cleaned.\n",
        "\n",
        "In addition we will split our data into dev/test/train and export to the filesystem."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "M_2ouEOH1_1q",
        "outputId": "29f8a1d0-a910-44a6-955a-40c6a3d4cb28",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 197
        }
      },
      "source": [
        "#TODO: Skip for retrain\n",
        "# drop duplicate translations\n",
        "df_pp = df.drop_duplicates()\n",
        "\n",
        "# drop conflicting translations\n",
        "# (this is optional and something that you might want to comment out \n",
        "# depending on the size of your corpus)\n",
        "df_pp.drop_duplicates(subset='source_sentence', inplace=True)\n",
        "df_pp.drop_duplicates(subset='target_sentence', inplace=True)\n",
        "\n",
        "# Shuffle the data to remove bias in dev set selection.\n",
        "df_pp = df_pp.sample(frac=1, random_state=seed).reset_index(drop=True)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:6: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  \n",
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:7: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  import sys\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Z_1BwAApEtMk",
        "colab_type": "code",
        "outputId": "9cb9e1c7-df14-4985-c08b-650dab4239c3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "#TODO: Skip for retrain\n",
        "# Install fuzzy wuzzy to remove \"almost duplicate\" sentences in the\n",
        "# test and training sets.\n",
        "! pip install fuzzywuzzy\n",
        "! pip install python-Levenshtein\n",
        "import time\n",
        "from fuzzywuzzy import process\n",
        "import numpy as np\n",
        "\n",
        "# reset the index of the training set after previous filtering\n",
        "df_pp.reset_index(drop=False, inplace=True)\n",
        "\n",
        "# Remove samples from the training data set if they \"almost overlap\" with the\n",
        "# samples in the test set.\n",
        "\n",
        "# Filtering function. Adjust pad to narrow down the candidate matches to\n",
        "# within a certain length of characters of the given sample.\n",
        "def fuzzfilter(sample, candidates, pad):\n",
        "  candidates = [x for x in candidates if len(x) <= len(sample)+pad and len(x) >= len(sample)-pad] \n",
        "  if len(candidates) > 0:\n",
        "    return process.extractOne(sample, candidates)[1]\n",
        "  else:\n",
        "    return np.nan\n",
        "\n",
        "# NOTE - This might run slow depending on the size of your training set. We are\n",
        "# printing some information to help you track how long it would take. \n",
        "scores = []\n",
        "start_time = time.time()\n",
        "for idx, row in df_pp.iterrows():\n",
        "  scores.append(fuzzfilter(row['source_sentence'], list(en_test_sents), 5))\n",
        "  if idx % 1000 == 0:\n",
        "    hours, rem = divmod(time.time() - start_time, 3600)\n",
        "    minutes, seconds = divmod(rem, 60)\n",
        "    print(\"{:0>2}:{:0>2}:{:05.2f}\".format(int(hours),int(minutes),seconds), \"%0.2f percent complete\" % (100.0*float(idx)/float(len(df_pp))))\n",
        "\n",
        "# Filter out \"almost overlapping samples\"\n",
        "df_pp['scores'] = scores\n",
        "df_pp = df_pp[df_pp['scores'] < 95]"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting fuzzywuzzy\n",
            "  Downloading https://files.pythonhosted.org/packages/43/ff/74f23998ad2f93b945c0309f825be92e04e0348e062026998b5eefef4c33/fuzzywuzzy-0.18.0-py2.py3-none-any.whl\n",
            "Installing collected packages: fuzzywuzzy\n",
            "Successfully installed fuzzywuzzy-0.18.0\n",
            "Collecting python-Levenshtein\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/42/a9/d1785c85ebf9b7dfacd08938dd028209c34a0ea3b1bcdb895208bd40a67d/python-Levenshtein-0.12.0.tar.gz (48kB)\n",
            "\u001b[K     |████████████████████████████████| 51kB 3.7MB/s \n",
            "\u001b[?25hRequirement already satisfied: setuptools in /usr/local/lib/python3.6/dist-packages (from python-Levenshtein) (46.1.3)\n",
            "Building wheels for collected packages: python-Levenshtein\n",
            "  Building wheel for python-Levenshtein (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for python-Levenshtein: filename=python_Levenshtein-0.12.0-cp36-cp36m-linux_x86_64.whl size=144791 sha256=a1c171e3c3fe2182059c6a83ce2ae60e74751a04db7685b0ea524689d1f62695\n",
            "  Stored in directory: /root/.cache/pip/wheels/de/c2/93/660fd5f7559049268ad2dc6d81c4e39e9e36518766eaf7e342\n",
            "Successfully built python-Levenshtein\n",
            "Installing collected packages: python-Levenshtein\n",
            "Successfully installed python-Levenshtein-0.12.0\n",
            "00:00:00.14 0.00 percent complete\n",
            "00:00:19.48 0.24 percent complete\n",
            "00:00:38.24 0.48 percent complete\n",
            "00:00:57.35 0.72 percent complete\n",
            "00:01:16.41 0.96 percent complete\n",
            "00:01:36.32 1.20 percent complete\n",
            "00:01:54.42 1.43 percent complete\n",
            "00:02:13.76 1.67 percent complete\n",
            "00:02:32.86 1.91 percent complete\n",
            "00:02:51.77 2.15 percent complete\n",
            "00:03:10.64 2.39 percent complete\n",
            "00:03:29.13 2.63 percent complete\n",
            "00:03:48.32 2.87 percent complete\n",
            "00:04:08.03 3.11 percent complete\n",
            "00:04:27.79 3.35 percent complete\n",
            "00:04:47.30 3.59 percent complete\n",
            "00:05:06.58 3.82 percent complete\n",
            "00:05:26.26 4.06 percent complete\n",
            "00:05:45.64 4.30 percent complete\n",
            "00:06:04.54 4.54 percent complete\n",
            "00:06:23.68 4.78 percent complete\n",
            "00:06:43.17 5.02 percent complete\n",
            "00:07:02.45 5.26 percent complete\n",
            "00:07:22.16 5.50 percent complete\n",
            "00:07:41.64 5.74 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '↓']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "00:08:01.48 5.98 percent complete\n",
            "00:08:21.53 6.22 percent complete\n",
            "00:08:40.78 6.45 percent complete\n",
            "00:09:00.17 6.69 percent complete\n",
            "00:09:19.45 6.93 percent complete\n",
            "00:09:39.29 7.17 percent complete\n",
            "00:09:59.40 7.41 percent complete\n",
            "00:10:18.60 7.65 percent complete\n",
            "00:10:38.11 7.89 percent complete\n",
            "00:10:57.05 8.13 percent complete\n",
            "00:11:16.07 8.37 percent complete\n",
            "00:11:34.85 8.61 percent complete\n",
            "00:11:54.55 8.84 percent complete\n",
            "00:12:13.53 9.08 percent complete\n",
            "00:12:33.17 9.32 percent complete\n",
            "00:12:52.62 9.56 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '” *']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "00:13:11.97 9.80 percent complete\n",
            "00:13:30.86 10.04 percent complete\n",
            "00:13:50.00 10.28 percent complete\n",
            "00:14:08.82 10.52 percent complete\n",
            "00:14:27.63 10.76 percent complete\n",
            "00:14:46.80 11.00 percent complete\n",
            "00:15:06.59 11.24 percent complete\n",
            "00:15:25.99 11.47 percent complete\n",
            "00:15:45.38 11.71 percent complete\n",
            "00:16:04.40 11.95 percent complete\n",
            "00:16:24.45 12.19 percent complete\n",
            "00:16:44.30 12.43 percent complete\n",
            "00:17:03.79 12.67 percent complete\n",
            "00:17:23.06 12.91 percent complete\n",
            "00:17:42.35 13.15 percent complete\n",
            "00:18:01.42 13.39 percent complete\n",
            "00:18:20.67 13.63 percent complete\n",
            "00:18:39.97 13.86 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '. .']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "00:18:59.55 14.10 percent complete\n",
            "00:19:18.33 14.34 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '․ ․']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "00:19:37.46 14.58 percent complete\n",
            "00:19:56.46 14.82 percent complete\n",
            "00:20:15.45 15.06 percent complete\n",
            "00:20:34.26 15.30 percent complete\n",
            "00:20:53.51 15.54 percent complete\n",
            "00:21:12.89 15.78 percent complete\n",
            "00:21:33.00 16.02 percent complete\n",
            "00:21:51.93 16.26 percent complete\n",
            "00:22:11.19 16.49 percent complete\n",
            "00:22:30.35 16.73 percent complete\n",
            "00:22:49.70 16.97 percent complete\n",
            "00:23:08.82 17.21 percent complete\n",
            "00:23:27.89 17.45 percent complete\n",
            "00:23:46.97 17.69 percent complete\n",
            "00:24:06.84 17.93 percent complete\n",
            "00:24:27.01 18.17 percent complete\n",
            "00:24:46.83 18.41 percent complete\n",
            "00:25:06.58 18.65 percent complete\n",
            "00:25:25.83 18.88 percent complete\n",
            "00:25:44.76 19.12 percent complete\n",
            "00:26:04.07 19.36 percent complete\n",
            "00:26:23.82 19.60 percent complete\n",
            "00:26:43.21 19.84 percent complete\n",
            "00:27:02.60 20.08 percent complete\n",
            "00:27:21.80 20.32 percent complete\n",
            "00:27:40.93 20.56 percent complete\n",
            "00:28:00.87 20.80 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '․ ․ ․']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "00:28:21.00 21.04 percent complete\n",
            "00:28:40.58 21.28 percent complete\n",
            "00:29:00.14 21.51 percent complete\n",
            "00:29:19.55 21.75 percent complete\n",
            "00:29:38.64 21.99 percent complete\n",
            "00:29:58.39 22.23 percent complete\n",
            "00:30:18.61 22.47 percent complete\n",
            "00:30:37.60 22.71 percent complete\n",
            "00:30:57.43 22.95 percent complete\n",
            "00:31:17.22 23.19 percent complete\n",
            "00:31:36.00 23.43 percent complete\n",
            "00:31:55.60 23.67 percent complete\n",
            "00:32:14.69 23.90 percent complete\n",
            "00:32:34.38 24.14 percent complete\n",
            "00:32:54.37 24.38 percent complete\n",
            "00:33:13.84 24.62 percent complete\n",
            "00:33:33.13 24.86 percent complete\n",
            "00:33:52.54 25.10 percent complete\n",
            "00:34:11.92 25.34 percent complete\n",
            "00:34:31.24 25.58 percent complete\n",
            "00:34:50.33 25.82 percent complete\n",
            "00:35:09.71 26.06 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '→ →']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "00:35:28.90 26.29 percent complete\n",
            "00:35:49.15 26.53 percent complete\n",
            "00:36:09.29 26.77 percent complete\n",
            "00:36:29.17 27.01 percent complete\n",
            "00:36:48.44 27.25 percent complete\n",
            "00:37:07.91 27.49 percent complete\n",
            "00:37:27.28 27.73 percent complete\n",
            "00:37:46.68 27.97 percent complete\n",
            "00:38:06.01 28.21 percent complete\n",
            "00:38:25.17 28.45 percent complete\n",
            "00:38:43.85 28.69 percent complete\n",
            "00:39:02.70 28.92 percent complete\n",
            "00:39:21.71 29.16 percent complete\n",
            "00:39:40.86 29.40 percent complete\n",
            "00:39:59.39 29.64 percent complete\n",
            "00:40:18.25 29.88 percent complete\n",
            "00:40:37.33 30.12 percent complete\n",
            "00:40:55.96 30.36 percent complete\n",
            "00:41:15.03 30.60 percent complete\n",
            "00:41:34.04 30.84 percent complete\n",
            "00:41:53.28 31.08 percent complete\n",
            "00:42:11.78 31.31 percent complete\n",
            "00:42:30.65 31.55 percent complete\n",
            "00:42:49.16 31.79 percent complete\n",
            "00:43:08.59 32.03 percent complete\n",
            "00:43:28.07 32.27 percent complete\n",
            "00:43:47.07 32.51 percent complete\n",
            "00:44:05.67 32.75 percent complete\n",
            "00:44:24.69 32.99 percent complete\n",
            "00:44:43.52 33.23 percent complete\n",
            "00:45:02.12 33.47 percent complete\n",
            "00:45:20.69 33.71 percent complete\n",
            "00:45:40.03 33.94 percent complete\n",
            "00:45:59.24 34.18 percent complete\n",
            "00:46:18.22 34.42 percent complete\n",
            "00:46:37.17 34.66 percent complete\n",
            "00:46:56.24 34.90 percent complete\n",
            "00:47:15.74 35.14 percent complete\n",
            "00:47:34.56 35.38 percent complete\n",
            "00:47:53.57 35.62 percent complete\n",
            "00:48:13.48 35.86 percent complete\n",
            "00:48:32.86 36.10 percent complete\n",
            "00:48:52.11 36.33 percent complete\n",
            "00:49:10.67 36.57 percent complete\n",
            "00:49:29.52 36.81 percent complete\n",
            "00:49:48.74 37.05 percent complete\n",
            "00:50:07.61 37.29 percent complete\n",
            "00:50:27.28 37.53 percent complete\n",
            "00:50:47.25 37.77 percent complete\n",
            "00:51:05.67 38.01 percent complete\n",
            "00:51:25.20 38.25 percent complete\n",
            "00:51:43.96 38.49 percent complete\n",
            "00:52:03.39 38.73 percent complete\n",
            "00:52:22.39 38.96 percent complete\n",
            "00:52:41.97 39.20 percent complete\n",
            "00:53:00.97 39.44 percent complete\n",
            "00:53:21.17 39.68 percent complete\n",
            "00:53:40.23 39.92 percent complete\n",
            "00:53:59.55 40.16 percent complete\n",
            "00:54:19.29 40.40 percent complete\n",
            "00:54:37.83 40.64 percent complete\n",
            "00:54:56.81 40.88 percent complete\n",
            "00:55:16.31 41.12 percent complete\n",
            "00:55:36.08 41.35 percent complete\n",
            "00:55:55.85 41.59 percent complete\n",
            "00:56:14.97 41.83 percent complete\n",
            "00:56:34.66 42.07 percent complete\n",
            "00:56:53.64 42.31 percent complete\n",
            "00:57:13.45 42.55 percent complete\n",
            "00:57:33.47 42.79 percent complete\n",
            "00:57:53.12 43.03 percent complete\n",
            "00:58:12.20 43.27 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '↓ ↓']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "00:58:32.08 43.51 percent complete\n",
            "00:58:51.55 43.75 percent complete\n",
            "00:59:10.78 43.98 percent complete\n",
            "00:59:29.87 44.22 percent complete\n",
            "00:59:49.41 44.46 percent complete\n",
            "01:00:08.94 44.70 percent complete\n",
            "01:00:28.17 44.94 percent complete\n",
            "01:00:47.95 45.18 percent complete\n",
            "01:01:07.58 45.42 percent complete\n",
            "01:01:26.95 45.66 percent complete\n",
            "01:01:46.50 45.90 percent complete\n",
            "01:02:05.59 46.14 percent complete\n",
            "01:02:25.13 46.37 percent complete\n",
            "01:02:44.58 46.61 percent complete\n",
            "01:03:04.08 46.85 percent complete\n",
            "01:03:23.16 47.09 percent complete\n",
            "01:03:43.64 47.33 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '*']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "01:04:03.31 47.57 percent complete\n",
            "01:04:23.01 47.81 percent complete\n",
            "01:04:42.57 48.05 percent complete\n",
            "01:05:02.17 48.29 percent complete\n",
            "01:05:21.78 48.53 percent complete\n",
            "01:05:41.50 48.77 percent complete\n",
            "01:06:01.03 49.00 percent complete\n",
            "01:06:20.08 49.24 percent complete\n",
            "01:06:39.39 49.48 percent complete\n",
            "01:06:59.05 49.72 percent complete\n",
            "01:07:18.55 49.96 percent complete\n",
            "01:07:37.54 50.20 percent complete\n",
            "01:07:56.74 50.44 percent complete\n",
            "01:08:15.77 50.68 percent complete\n",
            "01:08:35.36 50.92 percent complete\n",
            "01:08:55.00 51.16 percent complete\n",
            "01:09:14.66 51.39 percent complete\n",
            "01:09:34.11 51.63 percent complete\n",
            "01:09:52.83 51.87 percent complete\n",
            "01:10:12.29 52.11 percent complete\n",
            "01:10:32.19 52.35 percent complete\n",
            "01:10:52.16 52.59 percent complete\n",
            "01:11:11.94 52.83 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '⇩']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "01:11:31.35 53.07 percent complete\n",
            "01:11:50.96 53.31 percent complete\n",
            "01:12:10.04 53.55 percent complete\n",
            "01:12:29.43 53.79 percent complete\n",
            "01:12:49.31 54.02 percent complete\n",
            "01:13:08.41 54.26 percent complete\n",
            "01:13:27.83 54.50 percent complete\n",
            "01:13:47.41 54.74 percent complete\n",
            "01:14:06.99 54.98 percent complete\n",
            "01:14:26.57 55.22 percent complete\n",
            "01:14:45.85 55.46 percent complete\n",
            "01:15:05.62 55.70 percent complete\n",
            "01:15:25.60 55.94 percent complete\n",
            "01:15:45.42 56.18 percent complete\n",
            "01:16:05.41 56.41 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '”']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "01:16:25.79 56.65 percent complete\n",
            "01:16:45.75 56.89 percent complete\n",
            "01:17:05.58 57.13 percent complete\n",
            "01:17:25.10 57.37 percent complete\n",
            "01:17:45.24 57.61 percent complete\n",
            "01:18:05.41 57.85 percent complete\n",
            "01:18:24.90 58.09 percent complete\n",
            "01:18:44.64 58.33 percent complete\n",
            "01:19:04.44 58.57 percent complete\n",
            "01:19:25.12 58.81 percent complete\n",
            "01:19:45.56 59.04 percent complete\n",
            "01:20:05.63 59.28 percent complete\n",
            "01:20:25.81 59.52 percent complete\n",
            "01:20:45.91 59.76 percent complete\n",
            "01:21:05.27 60.00 percent complete\n",
            "01:21:25.29 60.24 percent complete\n",
            "01:21:44.55 60.48 percent complete\n",
            "01:22:04.45 60.72 percent complete\n",
            "01:22:24.81 60.96 percent complete\n",
            "01:22:44.36 61.20 percent complete\n",
            "01:23:04.16 61.43 percent complete\n",
            "01:23:23.03 61.67 percent complete\n",
            "01:23:42.99 61.91 percent complete\n",
            "01:24:02.77 62.15 percent complete\n",
            "01:24:23.07 62.39 percent complete\n",
            "01:24:42.63 62.63 percent complete\n",
            "01:25:02.01 62.87 percent complete\n",
            "01:25:21.57 63.11 percent complete\n",
            "01:25:40.93 63.35 percent complete\n",
            "01:26:00.62 63.59 percent complete\n",
            "01:26:20.30 63.83 percent complete\n",
            "01:26:39.92 64.06 percent complete\n",
            "01:26:59.41 64.30 percent complete\n",
            "01:27:18.80 64.54 percent complete\n",
            "01:27:38.64 64.78 percent complete\n",
            "01:27:58.23 65.02 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '↓ ↓ ↓ ↓']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "01:28:17.56 65.26 percent complete\n",
            "01:28:36.69 65.50 percent complete\n",
            "01:28:55.87 65.74 percent complete\n",
            "01:29:15.58 65.98 percent complete\n",
            "01:29:34.82 66.22 percent complete\n",
            "01:29:54.19 66.45 percent complete\n",
            "01:30:13.99 66.69 percent complete\n",
            "01:30:33.75 66.93 percent complete\n",
            "01:30:53.24 67.17 percent complete\n",
            "01:31:12.51 67.41 percent complete\n",
            "01:31:31.99 67.65 percent complete\n",
            "01:31:51.55 67.89 percent complete\n",
            "01:32:10.31 68.13 percent complete\n",
            "01:32:29.96 68.37 percent complete\n",
            "01:32:50.08 68.61 percent complete\n",
            "01:33:10.33 68.85 percent complete\n",
            "01:33:30.22 69.08 percent complete\n",
            "01:33:49.70 69.32 percent complete\n",
            "01:34:09.18 69.56 percent complete\n",
            "01:34:28.07 69.80 percent complete\n",
            "01:34:46.97 70.04 percent complete\n",
            "01:35:06.03 70.28 percent complete\n",
            "01:35:26.04 70.52 percent complete\n",
            "01:35:45.87 70.76 percent complete\n",
            "01:36:05.54 71.00 percent complete\n",
            "01:36:25.32 71.24 percent complete\n",
            "01:36:45.05 71.47 percent complete\n",
            "01:37:04.41 71.71 percent complete\n",
            "01:37:23.57 71.95 percent complete\n",
            "01:37:43.14 72.19 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '\\']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "01:38:03.03 72.43 percent complete\n",
            "01:38:22.67 72.67 percent complete\n",
            "01:38:42.48 72.91 percent complete\n",
            "01:39:02.12 73.15 percent complete\n",
            "01:39:21.48 73.39 percent complete\n",
            "01:39:41.34 73.63 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '●']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "01:40:01.28 73.86 percent complete\n",
            "01:40:20.93 74.10 percent complete\n",
            "01:40:40.14 74.34 percent complete\n",
            "01:41:00.03 74.58 percent complete\n",
            "01:41:20.40 74.82 percent complete\n",
            "01:41:39.97 75.06 percent complete\n",
            "01:41:59.30 75.30 percent complete\n",
            "01:42:18.86 75.54 percent complete\n",
            "01:42:38.21 75.78 percent complete\n",
            "01:42:56.99 76.02 percent complete\n",
            "01:43:15.90 76.26 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '․ ․ ․ ․ ․']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "01:43:35.48 76.49 percent complete\n",
            "01:43:55.55 76.73 percent complete\n",
            "01:44:14.82 76.97 percent complete\n",
            "01:44:34.10 77.21 percent complete\n",
            "01:44:53.94 77.45 percent complete\n",
            "01:45:13.17 77.69 percent complete\n",
            "01:45:32.40 77.93 percent complete\n",
            "01:45:52.10 78.17 percent complete\n",
            "01:46:12.18 78.41 percent complete\n",
            "01:46:32.08 78.65 percent complete\n",
            "01:46:51.92 78.88 percent complete\n",
            "01:47:10.54 79.12 percent complete\n",
            "01:47:30.26 79.36 percent complete\n",
            "01:47:50.22 79.60 percent complete\n",
            "01:48:09.37 79.84 percent complete\n",
            "01:48:28.54 80.08 percent complete\n",
            "01:48:47.98 80.32 percent complete\n",
            "01:49:07.93 80.56 percent complete\n",
            "01:49:27.37 80.80 percent complete\n",
            "01:49:46.70 81.04 percent complete\n",
            "01:50:05.55 81.28 percent complete\n",
            "01:50:25.02 81.51 percent complete\n",
            "01:50:43.80 81.75 percent complete\n",
            "01:51:03.04 81.99 percent complete\n",
            "01:51:22.42 82.23 percent complete\n",
            "01:51:41.95 82.47 percent complete\n",
            "01:52:01.26 82.71 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '□ ․ ․ ․ ․ ․']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "01:52:21.23 82.95 percent complete\n",
            "01:52:40.37 83.19 percent complete\n",
            "01:52:59.68 83.43 percent complete\n",
            "01:53:19.09 83.67 percent complete\n",
            "01:53:38.76 83.90 percent complete\n",
            "01:53:58.42 84.14 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '— ― ― ― ― ― ― ―']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "01:54:17.73 84.38 percent complete\n",
            "01:54:37.15 84.62 percent complete\n",
            "01:54:56.52 84.86 percent complete\n",
            "01:55:15.57 85.10 percent complete\n",
            "01:55:35.53 85.34 percent complete\n",
            "01:55:54.96 85.58 percent complete\n",
            "01:56:14.84 85.82 percent complete\n",
            "01:56:34.90 86.06 percent complete\n",
            "01:56:55.15 86.30 percent complete\n",
            "01:57:14.53 86.53 percent complete\n",
            "01:57:33.78 86.77 percent complete\n",
            "01:57:53.07 87.01 percent complete\n",
            "01:58:12.39 87.25 percent complete\n",
            "01:58:31.64 87.49 percent complete\n",
            "01:58:50.68 87.73 percent complete\n",
            "01:59:09.88 87.97 percent complete\n",
            "01:59:29.71 88.21 percent complete\n",
            "01:59:48.88 88.45 percent complete\n",
            "02:00:08.40 88.69 percent complete\n",
            "02:00:27.13 88.92 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '⇧']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "02:00:46.08 89.16 percent complete\n",
            "02:01:05.66 89.40 percent complete\n",
            "02:01:25.05 89.64 percent complete\n",
            "02:01:44.92 89.88 percent complete\n",
            "02:02:04.86 90.12 percent complete\n",
            "02:02:24.68 90.36 percent complete\n",
            "02:02:43.69 90.60 percent complete\n",
            "02:03:03.36 90.84 percent complete\n",
            "02:03:22.94 91.08 percent complete\n",
            "02:03:42.05 91.32 percent complete\n",
            "02:04:01.85 91.55 percent complete\n",
            "02:04:20.88 91.79 percent complete\n",
            "02:04:40.39 92.03 percent complete\n",
            "02:04:59.71 92.27 percent complete\n",
            "02:05:19.84 92.51 percent complete\n",
            "02:05:39.14 92.75 percent complete\n",
            "02:05:59.21 92.99 percent complete\n",
            "02:06:18.53 93.23 percent complete\n",
            "02:06:38.62 93.47 percent complete\n",
            "02:06:59.31 93.71 percent complete\n",
            "02:07:19.17 93.94 percent complete\n",
            "02:07:38.72 94.18 percent complete\n",
            "02:07:58.08 94.42 percent complete\n",
            "02:08:17.85 94.66 percent complete\n",
            "02:08:37.15 94.90 percent complete\n",
            "02:08:57.10 95.14 percent complete\n",
            "02:09:16.13 95.38 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '․ ․ ․ ․ ․ ․ ․ ․ ․ ․']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "02:09:36.09 95.62 percent complete\n",
            "02:09:55.25 95.86 percent complete\n",
            "02:10:14.43 96.10 percent complete\n",
            "02:10:34.19 96.34 percent complete\n",
            "02:10:53.38 96.57 percent complete\n",
            "02:11:12.67 96.81 percent complete\n",
            "02:11:32.19 97.05 percent complete\n",
            "02:11:51.43 97.29 percent complete\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:root:Applied processor reduces input query to empty string, all comparisons will have score 0. [Query: '']\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "02:12:11.22 97.53 percent complete\n",
            "02:12:31.26 97.77 percent complete\n",
            "02:12:50.80 98.01 percent complete\n",
            "02:13:10.16 98.25 percent complete\n",
            "02:13:29.41 98.49 percent complete\n",
            "02:13:48.66 98.73 percent complete\n",
            "02:14:08.34 98.96 percent complete\n",
            "02:14:27.76 99.20 percent complete\n",
            "02:14:48.28 99.44 percent complete\n",
            "02:15:08.40 99.68 percent complete\n",
            "02:15:28.13 99.92 percent complete\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "hxxBOCA-xXhy",
        "outputId": "d764afe7-2675-4fc3-80f0-bad7dcad720a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 865
        }
      },
      "source": [
        "#TODO: Skip for retrain\n",
        "# This section does the split between train/dev for the parallel corpora then saves them as separate files\n",
        "# We use 1000 dev test and the given test set.\n",
        "import csv\n",
        "\n",
        "# Do the split between dev/train and create parallel corpora\n",
        "num_dev_patterns = 1000\n",
        "\n",
        "# Optional: lower case the corpora - this will make it easier to generalize, but without proper casing.\n",
        "if lc:  # Julia: making lowercasing optional\n",
        "    df_pp[\"source_sentence\"] = df_pp[\"source_sentence\"].str.lower()\n",
        "    df_pp[\"target_sentence\"] = df_pp[\"target_sentence\"].str.lower()\n",
        "\n",
        "# Julia: test sets are already generated\n",
        "dev = df_pp.tail(num_dev_patterns) # Herman: Error in original\n",
        "stripped = df_pp.drop(df_pp.tail(num_dev_patterns).index)\n",
        "\n",
        "with open(\"train.\"+source_language, \"w\") as src_file, open(\"train.\"+target_language, \"w\") as trg_file:\n",
        "  for index, row in stripped.iterrows():\n",
        "    src_file.write(row[\"source_sentence\"]+\"\\n\")\n",
        "    trg_file.write(row[\"target_sentence\"]+\"\\n\")\n",
        "    \n",
        "with open(\"dev.\"+source_language, \"w\") as src_file, open(\"dev.\"+target_language, \"w\") as trg_file:\n",
        "  for index, row in dev.iterrows():\n",
        "    src_file.write(row[\"source_sentence\"]+\"\\n\")\n",
        "    trg_file.write(row[\"target_sentence\"]+\"\\n\")\n",
        "\n",
        "#stripped[[\"source_sentence\"]].to_csv(\"train.\"+source_language, header=False, index=False)  # Herman: Added `header=False` everywhere\n",
        "#stripped[[\"target_sentence\"]].to_csv(\"train.\"+target_language, header=False, index=False)  # Julia: Problematic handling of quotation marks.\n",
        "\n",
        "#dev[[\"source_sentence\"]].to_csv(\"dev.\"+source_language, header=False, index=False)\n",
        "#dev[[\"target_sentence\"]].to_csv(\"dev.\"+target_language, header=False, index=False)\n",
        "\n",
        "# Doublecheck the format below. There should be no extra quotation marks or weird characters.\n",
        "! head train.*\n",
        "! head dev.*"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "==> train.en <==\n",
            "TRAIN YOUR CHILDREN : “ I teach my children to check the expiration date of any packaged food items , such as snacks , before they buy them . ” ​ — Ruth , Nigeria\n",
            "When she replied that she was , he explained that he and his mother were trying to assist his sister with a school report on Canadians .\n",
            "Through the prophet Zephaniah , Jehovah answers : “ That day is a day of fury , a day of distress and of anguish , a day of storm and of desolation , a day of darkness and of gloominess , a day of clouds and of thick gloom . ”\n",
            "We do not require that people simply do as we tell them , but we give them convincing reasons to obey Christ’s command .\n",
            "Still , Jehovah can annihilate any rebel in the lake of fire , denying him any hope of a resurrection .\n",
            "Lucaris was arrested , and on July 27 , 1638 , he was taken on board a small boat as if for banishment .\n",
            "Yes , Jehovah remembered their faithful course .\n",
            "□ To appear tough\n",
            "During the ensuing confrontation , we Witnesses had to make our position clear to the agitated rebels and also explain our stand to the military guards .\n",
            "( See opening image . ) ( c ) Why should this Bible account about Samuel be of special interest to elders today ?\n",
            "\n",
            "==> train.yo <==\n",
            "KỌ́ ÀWỌN ỌMỌ RẸ : “ Mo kọ́ àwọn ọmọ mi pé kí wọ́n tó ra oúnjẹ bí ìpápánu , tó wà nínú agolo , ike , bébà , tàbí ọ̀rá , kí wọ́n máa yẹ ara oúnjẹ náà wò kí wọ́n lè mọ déètì tó máa bà jẹ́ . ” — Ruth , Nàìjíríà\n",
            "Nígbà tó sọ fún ọ̀dọ́kùnrin náà pé Kánádà lòun ti wá , ọ̀dọ́kùnrin náà sọ fún un pé òun àti màmá òun fẹ́ ran àbúrò òun obìnrin kan lọ́wọ́ láti kó ọ̀rọ̀ kan jọ nípa àwọn ará Kánádà , èyí tó fẹ́ mu lọ sílé ìwé .\n",
            "Jèhófà gbẹnu wòlíì Sefanáyà sọ ìdí rẹ̀ fún wa , ó ní : “ Ọjọ́ yẹn jẹ́ ọjọ́ ìbínú kíkan , ọjọ́ wàhálà àti làásìgbò , ọjọ́ ìjì àti ìsọdahoro , ọjọ́ òkùnkùn àti ìṣúdùdù , ọjọ́ àwọsánmà àti ìṣúdùdù tí ó nípọn . ”\n",
            "A ò fẹ́ káwọn èèyàn wulẹ̀ ṣe ohun tá a sọ fún wọn nìkan , àmọ́ à tún ń fún wọn ní ẹ̀rí tó dájú nípa ìdí tó fi yẹ kí wọ́n ṣègbọràn sí àṣẹ Kristi .\n",
            "Síbẹ̀ , Jèhófà lè pa ọlọ̀tẹ̀ èyíkéyìí run nípa sísọ ọ sínú adágún iná , tó túmọ̀ sí pé onítọ̀hún máa kú láìsí ìrètí kankan láti tún jíǹde .\n",
            "Ní wọ́n bá fi ọlọ́pàá mú Lucaris , nígbà tó sì di July 27 , 1638 , wọ́n fi ọkọ̀ ojú omi wà á lọ bí ẹni pé wọ́n fẹ́ gbé e lọ sí ilẹ̀ mìíràn .\n",
            "Bẹ́ẹ̀ ni o , Jèhófà kò gbàgbé ìṣòtítọ́ wọn .\n",
            "□ Kí wọ́n má bàa fojú ọ̀dẹ̀ wò mí\n",
            "Lákòókò tí àríyànjiyàn náà ń lọ lọ́wọ́ , àwa Ẹlẹ́rìí jẹ́ káwọn ọlọ̀tẹ̀ tínú ń bí yìí mọ̀ pé a ò lọ́wọ́ sí nǹkan tí wọ́n ń ṣe , a sì tún ṣàlàyé irú ẹni tá a jẹ́ fáwọn ológun tó ń ṣọ́ ọgbà ẹ̀wọ̀n náà .\n",
            "( Wo àwòrán tó wà níbẹ̀rẹ̀ àpilẹ̀kọ yìí . ) ( d ) Kí nìdí tó fi yẹ kí àwọn alàgbà lóde òní fún àkọsílẹ̀ Bíbélì yìí nípa Sámúẹ́lì láfiyèsí àrà ọ̀tọ̀ ?\n",
            "==> dev.en <==\n",
            "He is the Source of life , the One giving it as an undeserved gift through Christ .\n",
            "Now I had to find a legitimate line of work .\n",
            "Do I value material things more than my relationship with Jehovah and with people ?\n",
            "He has far more experience and stamina than you do , but he patiently walks near you .\n",
            "According to Harkavy’s Students ’ Hebrew and Chaldee Dictionary , ʽadh means “ duration , everlastingness , eternity , for ever . ”\n",
            "Why is rendering proper honor to elders a concern ?\n",
            "Jeremiah would rather be alone than be corrupted by bad companions .\n",
            "In years gone by , we believed that Jehovah became displeased with his people because they did not have a zealous share in the preaching work during World War I .\n",
            "Rather , they need to use a translation of the Bible in their own language .\n",
            "On a more personal level , showing honor to those to whom it is due keeps us from becoming self - centered .\n",
            "\n",
            "==> dev.yo <==\n",
            "Òun ni Orísun ìyè , Ẹni tí ń fi ìyè fúnni gẹ́gẹ́ bí ẹbùn tí a kò lẹ́tọ̀ọ́ sí nípasẹ̀ Kristi .\n",
            "Torí náà , mo ní láti wá iṣẹ́ gidi .\n",
            "Ṣé àwọn nǹkan tara ló jẹ mí lógún jù àbí àjọṣe mi pẹ̀lú Jèhófà àtàwọn èèyàn ?\n",
            "Ẹni tẹ́ ẹ jọ ń lọ yìí mọ ọ̀nà yẹn dáadáa .\n",
            "Gẹ́gẹ́ bí Harkavy’s Students ’ Hebrew and Chaldee Dictionary ṣe sọ , ʽadh túmọ̀ sí “ àkókò gígùn , àìnípẹ̀kun , títí gbére , títí láé . ”\n",
            "Kí nìdí tí kò fi yẹ ká máa gbé àwọn alàgbà gẹ̀gẹ̀ ju bó ṣe yẹ lọ ?\n",
            "Jeremáyà gbà kóun dá wà ju pé káwọn ọ̀rẹ́ burúkú wá kéèràn ran òun .\n",
            "Láwọn ọdún mélòó kan sẹ́yìn , a gbà pé inú Jèhófà ò dùn sáwọn èèyàn rẹ̀ torí pé wọn ò fìtara wàásù lásìkò Ogun Àgbáyé Kìíní .\n",
            "Àfi kí wọ́n ka Bíbélì tí wọ́n tú sí èdè wọn .\n",
            "Tó bá ti mọ́ wa lára láti máa bọlá fáwọn míì , a ò ní máa ro tara wa nìkan .\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "epeCydmCyS8X"
      },
      "source": [
        "\n",
        "\n",
        "---\n",
        "\n",
        "\n",
        "## Installation of JoeyNMT\n",
        "\n",
        "JoeyNMT is a simple, minimalist NMT package which is useful for learning and teaching. Check out the documentation for JoeyNMT [here](https://joeynmt.readthedocs.io)  "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "iBRMm4kMxZ8L",
        "outputId": "6cd58945-4935-4bfb-b392-c15b49e937ad",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "# Install JoeyNMT\n",
        "! git clone https://github.com/joeynmt/joeynmt.git\n",
        "! cd joeynmt; pip3 install ."
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Cloning into 'joeynmt'...\n",
            "remote: Enumerating objects: 3, done.\u001b[K\n",
            "remote: Counting objects: 100% (3/3), done.\u001b[K\n",
            "remote: Compressing objects: 100% (3/3), done.\u001b[K\n",
            "remote: Total 2380 (delta 0), reused 0 (delta 0), pack-reused 2377\u001b[K\n",
            "Receiving objects: 100% (2380/2380), 2.60 MiB | 4.24 MiB/s, done.\n",
            "Resolving deltas: 100% (1670/1670), done.\n",
            "Processing /content/joeynmt\n",
            "Requirement already satisfied: future in /usr/local/lib/python3.6/dist-packages (from joeynmt==0.0.1) (0.16.0)\n",
            "Requirement already satisfied: pillow in /usr/local/lib/python3.6/dist-packages (from joeynmt==0.0.1) (7.0.0)\n",
            "Requirement already satisfied: numpy<2.0,>=1.14.5 in /usr/local/lib/python3.6/dist-packages (from joeynmt==0.0.1) (1.18.2)\n",
            "Requirement already satisfied: setuptools>=41.0.0 in /usr/local/lib/python3.6/dist-packages (from joeynmt==0.0.1) (46.1.3)\n",
            "Requirement already satisfied: torch>=1.1 in /usr/local/lib/python3.6/dist-packages (from joeynmt==0.0.1) (1.4.0)\n",
            "Requirement already satisfied: tensorflow>=1.14 in /usr/local/lib/python3.6/dist-packages (from joeynmt==0.0.1) (2.2.0rc2)\n",
            "Requirement already satisfied: torchtext in /usr/local/lib/python3.6/dist-packages (from joeynmt==0.0.1) (0.3.1)\n",
            "Collecting sacrebleu>=1.3.6\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/f5/58/5c6cc352ea6271125325950715cf8b59b77abe5e93cf29f6e60b491a31d9/sacrebleu-1.4.6-py3-none-any.whl (59kB)\n",
            "\u001b[K     |████████████████████████████████| 61kB 4.3MB/s \n",
            "\u001b[?25hCollecting subword-nmt\n",
            "  Downloading https://files.pythonhosted.org/packages/74/60/6600a7bc09e7ab38bc53a48a20d8cae49b837f93f5842a41fe513a694912/subword_nmt-0.3.7-py2.py3-none-any.whl\n",
            "Requirement already satisfied: matplotlib in /usr/local/lib/python3.6/dist-packages (from joeynmt==0.0.1) (3.2.1)\n",
            "Requirement already satisfied: seaborn in /usr/local/lib/python3.6/dist-packages (from joeynmt==0.0.1) (0.10.0)\n",
            "Collecting pyyaml>=5.1\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/64/c2/b80047c7ac2478f9501676c988a5411ed5572f35d1beff9cae07d321512c/PyYAML-5.3.1.tar.gz (269kB)\n",
            "\u001b[K     |████████████████████████████████| 276kB 24.4MB/s \n",
            "\u001b[?25hCollecting pylint\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/e9/59/43fc36c5ee316bb9aeb7cf5329cdbdca89e5749c34d5602753827c0aa2dc/pylint-2.4.4-py3-none-any.whl (302kB)\n",
            "\u001b[K     |████████████████████████████████| 307kB 55.1MB/s \n",
            "\u001b[?25hRequirement already satisfied: six==1.12 in /usr/local/lib/python3.6/dist-packages (from joeynmt==0.0.1) (1.12.0)\n",
            "Collecting wrapt==1.11.1\n",
            "  Downloading https://files.pythonhosted.org/packages/67/b2/0f71ca90b0ade7fad27e3d20327c996c6252a2ffe88f50a95bba7434eda9/wrapt-1.11.1.tar.gz\n",
            "Requirement already satisfied: gast==0.3.3 in /usr/local/lib/python3.6/dist-packages (from tensorflow>=1.14->joeynmt==0.0.1) (0.3.3)\n",
            "Requirement already satisfied: tensorflow-estimator<2.3.0,>=2.2.0rc0 in /usr/local/lib/python3.6/dist-packages (from tensorflow>=1.14->joeynmt==0.0.1) (2.2.0rc0)\n",
            "Requirement already satisfied: keras-preprocessing>=1.1.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow>=1.14->joeynmt==0.0.1) (1.1.0)\n",
            "Requirement already satisfied: tensorboard<2.3.0,>=2.2.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow>=1.14->joeynmt==0.0.1) (2.2.0)\n",
            "Requirement already satisfied: scipy==1.4.1; python_version >= \"3\" in /usr/local/lib/python3.6/dist-packages (from tensorflow>=1.14->joeynmt==0.0.1) (1.4.1)\n",
            "Requirement already satisfied: protobuf>=3.8.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow>=1.14->joeynmt==0.0.1) (3.10.0)\n",
            "Requirement already satisfied: opt-einsum>=2.3.2 in /usr/local/lib/python3.6/dist-packages (from tensorflow>=1.14->joeynmt==0.0.1) (3.2.0)\n",
            "Requirement already satisfied: absl-py>=0.7.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow>=1.14->joeynmt==0.0.1) (0.9.0)\n",
            "Requirement already satisfied: wheel>=0.26; python_version >= \"3\" in /usr/local/lib/python3.6/dist-packages (from tensorflow>=1.14->joeynmt==0.0.1) (0.34.2)\n",
            "Requirement already satisfied: termcolor>=1.1.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow>=1.14->joeynmt==0.0.1) (1.1.0)\n",
            "Requirement already satisfied: astunparse==1.6.3 in /usr/local/lib/python3.6/dist-packages (from tensorflow>=1.14->joeynmt==0.0.1) (1.6.3)\n",
            "Requirement already satisfied: h5py<2.11.0,>=2.10.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow>=1.14->joeynmt==0.0.1) (2.10.0)\n",
            "Requirement already satisfied: google-pasta>=0.1.8 in /usr/local/lib/python3.6/dist-packages (from tensorflow>=1.14->joeynmt==0.0.1) (0.2.0)\n",
            "Requirement already satisfied: grpcio>=1.8.6 in /usr/local/lib/python3.6/dist-packages (from tensorflow>=1.14->joeynmt==0.0.1) (1.28.1)\n",
            "Requirement already satisfied: tqdm in /usr/local/lib/python3.6/dist-packages (from torchtext->joeynmt==0.0.1) (4.38.0)\n",
            "Requirement already satisfied: requests in /usr/local/lib/python3.6/dist-packages (from torchtext->joeynmt==0.0.1) (2.21.0)\n",
            "Collecting mecab-python3\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/18/49/b55a839a77189042960bf96490640c44816073f917d489acbc5d79fa5cc3/mecab_python3-0.996.5-cp36-cp36m-manylinux2010_x86_64.whl (17.1MB)\n",
            "\u001b[K     |████████████████████████████████| 17.1MB 201kB/s \n",
            "\u001b[?25hCollecting portalocker\n",
            "  Downloading https://files.pythonhosted.org/packages/64/03/9abfb3374d67838daf24f1a388528714bec1debb1d13749f0abd7fb07cfb/portalocker-1.6.0-py2.py3-none-any.whl\n",
            "Requirement already satisfied: typing in /usr/local/lib/python3.6/dist-packages (from sacrebleu>=1.3.6->joeynmt==0.0.1) (3.6.6)\n",
            "Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.6/dist-packages (from matplotlib->joeynmt==0.0.1) (0.10.0)\n",
            "Requirement already satisfied: python-dateutil>=2.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->joeynmt==0.0.1) (2.8.1)\n",
            "Requirement already satisfied: kiwisolver>=1.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->joeynmt==0.0.1) (1.2.0)\n",
            "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->joeynmt==0.0.1) (2.4.7)\n",
            "Requirement already satisfied: pandas>=0.22.0 in /usr/local/lib/python3.6/dist-packages (from seaborn->joeynmt==0.0.1) (1.0.3)\n",
            "Collecting isort<5,>=4.2.5\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/e5/b0/c121fd1fa3419ea9bfd55c7f9c4fedfec5143208d8c7ad3ce3db6c623c21/isort-4.3.21-py2.py3-none-any.whl (42kB)\n",
            "\u001b[K     |████████████████████████████████| 51kB 7.6MB/s \n",
            "\u001b[?25hCollecting mccabe<0.7,>=0.6\n",
            "  Downloading https://files.pythonhosted.org/packages/87/89/479dc97e18549e21354893e4ee4ef36db1d237534982482c3681ee6e7b57/mccabe-0.6.1-py2.py3-none-any.whl\n",
            "Collecting astroid<2.4,>=2.3.0\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/ad/ae/86734823047962e7b8c8529186a1ac4a7ca19aaf1aa0c7713c022ef593fd/astroid-2.3.3-py3-none-any.whl (205kB)\n",
            "\u001b[K     |████████████████████████████████| 215kB 54.1MB/s \n",
            "\u001b[?25hRequirement already satisfied: markdown>=2.6.8 in /usr/local/lib/python3.6/dist-packages (from tensorboard<2.3.0,>=2.2.0->tensorflow>=1.14->joeynmt==0.0.1) (3.2.1)\n",
            "Requirement already satisfied: werkzeug>=0.11.15 in /usr/local/lib/python3.6/dist-packages (from tensorboard<2.3.0,>=2.2.0->tensorflow>=1.14->joeynmt==0.0.1) (1.0.1)\n",
            "Requirement already satisfied: tensorboard-plugin-wit>=1.6.0 in /usr/local/lib/python3.6/dist-packages (from tensorboard<2.3.0,>=2.2.0->tensorflow>=1.14->joeynmt==0.0.1) (1.6.0.post3)\n",
            "Requirement already satisfied: google-auth-oauthlib<0.5,>=0.4.1 in /usr/local/lib/python3.6/dist-packages (from tensorboard<2.3.0,>=2.2.0->tensorflow>=1.14->joeynmt==0.0.1) (0.4.1)\n",
            "Requirement already satisfied: google-auth<2,>=1.6.3 in /usr/local/lib/python3.6/dist-packages (from tensorboard<2.3.0,>=2.2.0->tensorflow>=1.14->joeynmt==0.0.1) (1.7.2)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.6/dist-packages (from requests->torchtext->joeynmt==0.0.1) (2020.4.5.1)\n",
            "Requirement already satisfied: urllib3<1.25,>=1.21.1 in /usr/local/lib/python3.6/dist-packages (from requests->torchtext->joeynmt==0.0.1) (1.24.3)\n",
            "Requirement already satisfied: chardet<3.1.0,>=3.0.2 in /usr/local/lib/python3.6/dist-packages (from requests->torchtext->joeynmt==0.0.1) (3.0.4)\n",
            "Requirement already satisfied: idna<2.9,>=2.5 in /usr/local/lib/python3.6/dist-packages (from requests->torchtext->joeynmt==0.0.1) (2.8)\n",
            "Requirement already satisfied: pytz>=2017.2 in /usr/local/lib/python3.6/dist-packages (from pandas>=0.22.0->seaborn->joeynmt==0.0.1) (2018.9)\n",
            "Collecting lazy-object-proxy==1.4.*\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/0b/dd/b1e3407e9e6913cf178e506cd0dee818e58694d9a5cd1984e3f6a8b9a10f/lazy_object_proxy-1.4.3-cp36-cp36m-manylinux1_x86_64.whl (55kB)\n",
            "\u001b[K     |████████████████████████████████| 61kB 8.1MB/s \n",
            "\u001b[?25hCollecting typed-ast<1.5,>=1.4.0; implementation_name == \"cpython\" and python_version < \"3.8\"\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/90/ed/5459080d95eb87a02fe860d447197be63b6e2b5e9ff73c2b0a85622994f4/typed_ast-1.4.1-cp36-cp36m-manylinux1_x86_64.whl (737kB)\n",
            "\u001b[K     |████████████████████████████████| 747kB 54.4MB/s \n",
            "\u001b[?25hRequirement already satisfied: requests-oauthlib>=0.7.0 in /usr/local/lib/python3.6/dist-packages (from google-auth-oauthlib<0.5,>=0.4.1->tensorboard<2.3.0,>=2.2.0->tensorflow>=1.14->joeynmt==0.0.1) (1.3.0)\n",
            "Requirement already satisfied: pyasn1-modules>=0.2.1 in /usr/local/lib/python3.6/dist-packages (from google-auth<2,>=1.6.3->tensorboard<2.3.0,>=2.2.0->tensorflow>=1.14->joeynmt==0.0.1) (0.2.8)\n",
            "Requirement already satisfied: cachetools<3.2,>=2.0.0 in /usr/local/lib/python3.6/dist-packages (from google-auth<2,>=1.6.3->tensorboard<2.3.0,>=2.2.0->tensorflow>=1.14->joeynmt==0.0.1) (3.1.1)\n",
            "Requirement already satisfied: rsa<4.1,>=3.1.4 in /usr/local/lib/python3.6/dist-packages (from google-auth<2,>=1.6.3->tensorboard<2.3.0,>=2.2.0->tensorflow>=1.14->joeynmt==0.0.1) (4.0)\n",
            "Requirement already satisfied: oauthlib>=3.0.0 in /usr/local/lib/python3.6/dist-packages (from requests-oauthlib>=0.7.0->google-auth-oauthlib<0.5,>=0.4.1->tensorboard<2.3.0,>=2.2.0->tensorflow>=1.14->joeynmt==0.0.1) (3.1.0)\n",
            "Requirement already satisfied: pyasn1<0.5.0,>=0.4.6 in /usr/local/lib/python3.6/dist-packages (from pyasn1-modules>=0.2.1->google-auth<2,>=1.6.3->tensorboard<2.3.0,>=2.2.0->tensorflow>=1.14->joeynmt==0.0.1) (0.4.8)\n",
            "Building wheels for collected packages: joeynmt, pyyaml, wrapt\n",
            "  Building wheel for joeynmt (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for joeynmt: filename=joeynmt-0.0.1-cp36-none-any.whl size=73768 sha256=09642adb3c413596b594cde12da531b842bf2007c75f258d0eed07fa3b152a3d\n",
            "  Stored in directory: /tmp/pip-ephem-wheel-cache-5ss0uo8b/wheels/db/01/db/751cc9f3e7f6faec127c43644ba250a3ea7ad200594aeda70a\n",
            "  Building wheel for pyyaml (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for pyyaml: filename=PyYAML-5.3.1-cp36-cp36m-linux_x86_64.whl size=44621 sha256=326c10062d3be997e2721f4a455fa10fb95cb616ebf9d6957aac343e4fc9e7b7\n",
            "  Stored in directory: /root/.cache/pip/wheels/a7/c1/ea/cf5bd31012e735dc1dfea3131a2d5eae7978b251083d6247bd\n",
            "  Building wheel for wrapt (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for wrapt: filename=wrapt-1.11.1-cp36-cp36m-linux_x86_64.whl size=67435 sha256=ce4852f0ad71dc23caee5cfbf233c9da664c3e4716f81b23348701d716acff35\n",
            "  Stored in directory: /root/.cache/pip/wheels/89/67/41/63cbf0f6ac0a6156588b9587be4db5565f8c6d8ccef98202fc\n",
            "Successfully built joeynmt pyyaml wrapt\n",
            "Installing collected packages: mecab-python3, portalocker, sacrebleu, subword-nmt, pyyaml, isort, mccabe, wrapt, lazy-object-proxy, typed-ast, astroid, pylint, joeynmt\n",
            "  Found existing installation: PyYAML 3.13\n",
            "    Uninstalling PyYAML-3.13:\n",
            "      Successfully uninstalled PyYAML-3.13\n",
            "  Found existing installation: wrapt 1.12.1\n",
            "    Uninstalling wrapt-1.12.1:\n",
            "      Successfully uninstalled wrapt-1.12.1\n",
            "Successfully installed astroid-2.3.3 isort-4.3.21 joeynmt-0.0.1 lazy-object-proxy-1.4.3 mccabe-0.6.1 mecab-python3-0.996.5 portalocker-1.6.0 pylint-2.4.4 pyyaml-5.3.1 sacrebleu-1.4.6 subword-nmt-0.3.7 typed-ast-1.4.1 wrapt-1.11.1\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "AaE77Tcppex9"
      },
      "source": [
        "# Preprocessing the Data into Subword BPE Tokens\n",
        "\n",
        "- One of the most powerful improvements for agglutinative languages (a feature of most Bantu languages) is using BPE tokenization [ (Sennrich, 2015) ](https://arxiv.org/abs/1508.07909).\n",
        "\n",
        "- It was also shown that by optimizing the umber of BPE codes we significantly improve results for low-resourced languages [(Sennrich, 2019)](https://www.aclweb.org/anthology/P19-1021) [(Martinus, 2019)](https://arxiv.org/abs/1906.05685)\n",
        "\n",
        "- Below we have the scripts for doing BPE tokenization of our data. We use 4000 tokens as recommended by [(Sennrich, 2019)](https://www.aclweb.org/anthology/P19-1021). You do not need to change anything. Simply running the below will be suitable. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "H-TyjtmXB1mL",
        "outputId": "6a7b3ce9-239a-4a21-cc63-37107d521733",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 431
        }
      },
      "source": [
        "#TODO: Skip for retrain\n",
        "# One of the huge boosts in NMT performance was to use a different method of tokenizing. \n",
        "# Usually, NMT would tokenize by words. However, using a method called BPE gave amazing boosts to performance\n",
        "\n",
        "# Do subword NMT\n",
        "from os import path\n",
        "os.environ[\"src\"] = source_language # Sets them in bash as well, since we often use bash scripts\n",
        "os.environ[\"tgt\"] = target_language\n",
        "\n",
        "# Learn BPEs on the training data.\n",
        "os.environ[\"data_path\"] = path.join(\"joeynmt\", \"data\", source_language + target_language) # Herman! \n",
        "! subword-nmt learn-joint-bpe-and-vocab --input train.$src train.$tgt -s 4000 -o bpe.codes.4000 --write-vocabulary vocab.$src vocab.$tgt\n",
        "\n",
        "# Apply BPE splits to the development and test data.\n",
        "! subword-nmt apply-bpe -c bpe.codes.4000 --vocabulary vocab.$src < train.$src > train.bpe.$src\n",
        "! subword-nmt apply-bpe -c bpe.codes.4000 --vocabulary vocab.$tgt < train.$tgt > train.bpe.$tgt\n",
        "\n",
        "! subword-nmt apply-bpe -c bpe.codes.4000 --vocabulary vocab.$src < dev.$src > dev.bpe.$src\n",
        "! subword-nmt apply-bpe -c bpe.codes.4000 --vocabulary vocab.$tgt < dev.$tgt > dev.bpe.$tgt\n",
        "! subword-nmt apply-bpe -c bpe.codes.4000 --vocabulary vocab.$src < test.$src > test.bpe.$src\n",
        "! subword-nmt apply-bpe -c bpe.codes.4000 --vocabulary vocab.$tgt < test.$tgt > test.bpe.$tgt\n",
        "\n",
        "# Create directory, move everyone we care about to the correct location\n",
        "! mkdir -p $data_path\n",
        "! cp train.* $data_path\n",
        "! cp test.* $data_path\n",
        "! cp dev.* $data_path\n",
        "! cp bpe.codes.4000 $data_path\n",
        "! ls $data_path\n",
        "\n",
        "# Also move everything we care about to a mounted location in google drive (relevant if running in colab) at gdrive_path\n",
        "! cp train.* \"$gdrive_path\"\n",
        "! cp test.* \"$gdrive_path\"\n",
        "! cp dev.* \"$gdrive_path\"\n",
        "! cp bpe.codes.4000 \"$gdrive_path\"\n",
        "! ls \"$gdrive_path\"\n",
        "\n",
        "# Create that vocab using build_vocab\n",
        "! sudo chmod 777 joeynmt/scripts/build_vocab.py\n",
        "! joeynmt/scripts/build_vocab.py joeynmt/data/$src$tgt/train.bpe.$src joeynmt/data/$src$tgt/train.bpe.$tgt --output_path \"$gdrive_path/vocab.txt\"\n",
        "\n",
        "# Some output\n",
        "! echo \"BPE Xhosa Sentences\"\n",
        "! tail -n 5 test.bpe.$tgt\n",
        "! echo \"Combined BPE Vocab\"\n",
        "! tail -n 10 \"$gdrive_path/vocab.txt\"  # Herman"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "bpe.codes.4000\tdev.en\t     test.bpe.yo     test.yo\t   train.en\n",
            "dev.bpe.en\tdev.yo\t     test.en\t     train.bpe.en  train.yo\n",
            "dev.bpe.yo\ttest.bpe.en  test.en-any.en  train.bpe.yo\n",
            "bpe.codes.4000\tdev.en\ttest.bpe.en  test.en-any.en  train.bpe.yo  vocab.txt\n",
            "dev.bpe.en\tdev.yo\ttest.bpe.yo  test.yo\t     train.en\n",
            "dev.bpe.yo\tmodels\ttest.en      train.bpe.en    train.yo\n",
            "BPE Xhosa Sentences\n",
            "A@@ p@@ at@@ a ńlá ti ìgbàgbọ́ ( Wo ìpín@@ rọ̀ 12 - 14 )\n",
            "À@@ ṣí@@ borí ìgb@@ àlà ( Wo ìpín@@ rọ̀ 15 - 18 )\n",
            "Mo ti rí i pé àwọn èèyàn máa ń fẹ́ gb@@ ọ́@@ rọ̀ wa tí wọ́n bá rí i pé a lóye Bíbélì dáadáa , a sì fẹ́ ran àwọn lọ́wọ́ . ”\n",
            "I@@ dà ẹ̀mí ( Wo ìpín@@ rọ̀ 19 - 20 )\n",
            "Ó dájú pé lọ́@@ lá ìt@@ ì@@ lẹ́yìn Jèhófà , a máa dúró gb@@ ọ@@ in - in , Èṣù ò sì ní rí wa gbé ṣe .\n",
            "Combined BPE Vocab\n",
            "œ@@\n",
            "Ísír@@\n",
            "Isra@@\n",
            "̃\n",
            "×\n",
            "ô\n",
            "ʺ\n",
            "bítì\n",
            "Pété@@\n",
            "Jóò@@\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "Ixmzi60WsUZ8"
      },
      "source": [
        "# Creating the JoeyNMT Config\n",
        "\n",
        "JoeyNMT requires a yaml config. We provide a template below. We've also set a number of defaults with it, that you may play with!\n",
        "\n",
        "- We used Transformer architecture \n",
        "- We set our dropout to reasonably high: 0.3 (recommended in  [(Sennrich, 2019)](https://www.aclweb.org/anthology/P19-1021))\n",
        "\n",
        "Things worth playing with:\n",
        "- The batch size (also recommended to change for low-resourced languages)\n",
        "- The number of epochs (we've set it at 30 just so it runs in about an hour, for testing purposes)\n",
        "- The decoder options (beam_size, alpha)\n",
        "- Evaluation metrics (BLEU versus Crhf4)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Wc47fvWqyxbd",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def get_last_checkpoint(directory):\n",
        "  last_checkpoint = ''\n",
        "  try:\n",
        "    for filename in os.listdir(directory):\n",
        "      if 'best' in filename and filename.endswith(\".ckpt\"):\n",
        "        return filename\n",
        "      if not 'best' in filename and filename.endswith(\".ckpt\"):\n",
        "          if not last_checkpoint or int(filename.split('.')[0]) > int(last_checkpoint.split('.')[0]):\n",
        "            last_checkpoint = filename\n",
        "  except FileNotFoundError as e:\n",
        "    print('Error Occur ', e)\n",
        "  return last_checkpoint"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "x_ffEoFdy1Qo",
        "colab_type": "code",
        "outputId": "03eca8de-dd2b-4a95-d43e-4b6456f62294",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "source": [
        "# Copy the created models from the temporary storage to main storage on google drive for persistant storage \n",
        "# the content of te folder will be overwrite when you start trainin\n",
        "# !cp -r \"/content/drive/My Drive/masakhane/model-temp/\"* \"$gdrive_path/models/${src}${tgt}_transformer/\"\n",
        "last_checkpoint = get_last_checkpoint(models_path)\n",
        "print('Last checkpoint :',last_checkpoint)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Last checkpoint : best.ckpt\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "PIs1lY2hxMsl",
        "colab": {}
      },
      "source": [
        "# This creates the config file for our JoeyNMT system. It might seem overwhelming so we've provided a couple of useful parameters you'll need to update\n",
        "# (You can of course play with all the parameters if you'd like!)\n",
        "\n",
        "name = '%s%s' % (source_language, target_language)\n",
        "gdrive_path = os.environ[\"gdrive_path\"]\n",
        "\n",
        "# Create the config\n",
        "config = \"\"\"\n",
        "name: \"{name}_transformer\"\n",
        "\n",
        "data:\n",
        "    src: \"{source_language}\"\n",
        "    trg: \"{target_language}\"\n",
        "    train: \"{gdrive_path}/train.bpe\"\n",
        "    dev:   \"{gdrive_path}/dev.bpe\"\n",
        "    test:  \"{gdrive_path}/test.bpe\"\n",
        "    level: \"bpe\"\n",
        "    lowercase: False\n",
        "    max_sent_length: 100\n",
        "    src_vocab: \"{gdrive_path}/vocab.txt\"\n",
        "    trg_vocab: \"{gdrive_path}/vocab.txt\"\n",
        "\n",
        "testing:\n",
        "    beam_size: 5\n",
        "    alpha: 1.0\n",
        "\n",
        "training:\n",
        "    load_model: \"{gdrive_path}/models/{name}_transformer/{last_checkpoint}\" # uncommented to load a pre-trained model from last checkpoint\n",
        "    random_seed: 42\n",
        "    optimizer: \"adam\"\n",
        "    normalization: \"tokens\"\n",
        "    adam_betas: [0.9, 0.999] \n",
        "    scheduling: \"plateau\"           # TODO: try switching from plateau to Noam scheduling\n",
        "    patience: 5                     # For plateau: decrease learning rate by decrease_factor if validation score has not improved for this many validation rounds.\n",
        "    learning_rate_factor: 0.5       # factor for Noam scheduler (used with Transformer)\n",
        "    learning_rate_warmup: 1000      # warmup steps for Noam scheduler (used with Transformer)\n",
        "    decrease_factor: 0.7\n",
        "    loss: \"crossentropy\"\n",
        "    learning_rate: 0.0003\n",
        "    learning_rate_min: 0.00000001\n",
        "    weight_decay: 0.0\n",
        "    label_smoothing: 0.1\n",
        "    batch_size: 4096\n",
        "    batch_type: \"token\"\n",
        "    eval_batch_size: 3600\n",
        "    eval_batch_type: \"token\"\n",
        "    batch_multiplier: 1\n",
        "    early_stopping_metric: \"ppl\"\n",
        "    epochs: 2                     # TODO: Decrease for when playing around and checking of working. Around 30 is sufficient to check if its working at all\n",
        "    validation_freq: 1000          # TODO: Set to at least once per epoch.\n",
        "    logging_freq: 100\n",
        "    eval_metric: \"bleu\"\n",
        "    model_dir: \"{model_temp_dir}\"\n",
        "    overwrite: True               # TODO: Set to True if you want to overwrite possibly existing models. \n",
        "    shuffle: True\n",
        "    use_cuda: True\n",
        "    max_output_length: 100\n",
        "    print_valid_sents: [0, 1, 2, 3]\n",
        "    keep_last_ckpts: 3\n",
        "\n",
        "model:\n",
        "    initializer: \"xavier\"\n",
        "    bias_initializer: \"zeros\"\n",
        "    init_gain: 1.0\n",
        "    embed_initializer: \"xavier\"\n",
        "    embed_init_gain: 1.0\n",
        "    tied_embeddings: True\n",
        "    tied_softmax: True\n",
        "    encoder:\n",
        "        type: \"transformer\"\n",
        "        num_layers: 6\n",
        "        num_heads: 4             # TODO: Increase to 8 for larger data.\n",
        "        embeddings:\n",
        "            embedding_dim: 256   # TODO: Increase to 512 for larger data.\n",
        "            scale: True\n",
        "            dropout: 0.2\n",
        "        # typically ff_size = 4 x hidden_size\n",
        "        hidden_size: 256         # TODO: Increase to 512 for larger data.\n",
        "        ff_size: 1024            # TODO: Increase to 2048 for larger data.\n",
        "        dropout: 0.3\n",
        "    decoder:\n",
        "        type: \"transformer\"\n",
        "        num_layers: 6\n",
        "        num_heads: 4              # TODO: Increase to 8 for larger data.\n",
        "        embeddings:\n",
        "            embedding_dim: 256    # TODO: Increase to 512 for larger data.\n",
        "            scale: True\n",
        "            dropout: 0.2\n",
        "        # typically ff_size = 4 x hidden_size\n",
        "        hidden_size: 256         # TODO: Increase to 512 for larger data.\n",
        "        ff_size: 1024            # TODO: Increase to 2048 for larger data.\n",
        "        dropout: 0.3\n",
        "\"\"\".format(name=name, gdrive_path=os.environ[\"gdrive_path\"], source_language=source_language, target_language=target_language, model_temp_dir=model_temp_dir, last_checkpoint=last_checkpoint)\n",
        "with open(\"joeynmt/configs/transformer_{name}.yaml\".format(name=name),'w') as f:\n",
        "    f.write(config)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "pIifxE3Qzuvs"
      },
      "source": [
        "# Train the Model\n",
        "\n",
        "This single line of joeynmt runs the training using the config we made above"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "6ZBPFwT94WpI",
        "outputId": "842c925f-2f1a-4b4e-a6d1-f8439360c81f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "# Train the model\n",
        "# You can press Ctrl-C to stop. And then run the next cell to save your checkpoints! \n",
        "!cd joeynmt; python3 -m joeynmt train configs/transformer_$src$tgt.yaml"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2020-04-12 14:27:14,687 Hello! This is Joey-NMT.\n",
            "2020-04-12 14:27:14.845045: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1\n",
            "2020-04-12 14:27:16,195 Total params: 12188160\n",
            "2020-04-12 14:27:16,197 Trainable parameters: ['decoder.layer_norm.bias', 'decoder.layer_norm.weight', 'decoder.layers.0.dec_layer_norm.bias', 'decoder.layers.0.dec_layer_norm.weight', 'decoder.layers.0.feed_forward.layer_norm.bias', 'decoder.layers.0.feed_forward.layer_norm.weight', 'decoder.layers.0.feed_forward.pwff_layer.0.bias', 'decoder.layers.0.feed_forward.pwff_layer.0.weight', 'decoder.layers.0.feed_forward.pwff_layer.3.bias', 'decoder.layers.0.feed_forward.pwff_layer.3.weight', 'decoder.layers.0.src_trg_att.k_layer.bias', 'decoder.layers.0.src_trg_att.k_layer.weight', 'decoder.layers.0.src_trg_att.output_layer.bias', 'decoder.layers.0.src_trg_att.output_layer.weight', 'decoder.layers.0.src_trg_att.q_layer.bias', 'decoder.layers.0.src_trg_att.q_layer.weight', 'decoder.layers.0.src_trg_att.v_layer.bias', 'decoder.layers.0.src_trg_att.v_layer.weight', 'decoder.layers.0.trg_trg_att.k_layer.bias', 'decoder.layers.0.trg_trg_att.k_layer.weight', 'decoder.layers.0.trg_trg_att.output_layer.bias', 'decoder.layers.0.trg_trg_att.output_layer.weight', 'decoder.layers.0.trg_trg_att.q_layer.bias', 'decoder.layers.0.trg_trg_att.q_layer.weight', 'decoder.layers.0.trg_trg_att.v_layer.bias', 'decoder.layers.0.trg_trg_att.v_layer.weight', 'decoder.layers.0.x_layer_norm.bias', 'decoder.layers.0.x_layer_norm.weight', 'decoder.layers.1.dec_layer_norm.bias', 'decoder.layers.1.dec_layer_norm.weight', 'decoder.layers.1.feed_forward.layer_norm.bias', 'decoder.layers.1.feed_forward.layer_norm.weight', 'decoder.layers.1.feed_forward.pwff_layer.0.bias', 'decoder.layers.1.feed_forward.pwff_layer.0.weight', 'decoder.layers.1.feed_forward.pwff_layer.3.bias', 'decoder.layers.1.feed_forward.pwff_layer.3.weight', 'decoder.layers.1.src_trg_att.k_layer.bias', 'decoder.layers.1.src_trg_att.k_layer.weight', 'decoder.layers.1.src_trg_att.output_layer.bias', 'decoder.layers.1.src_trg_att.output_layer.weight', 'decoder.layers.1.src_trg_att.q_layer.bias', 'decoder.layers.1.src_trg_att.q_layer.weight', 'decoder.layers.1.src_trg_att.v_layer.bias', 'decoder.layers.1.src_trg_att.v_layer.weight', 'decoder.layers.1.trg_trg_att.k_layer.bias', 'decoder.layers.1.trg_trg_att.k_layer.weight', 'decoder.layers.1.trg_trg_att.output_layer.bias', 'decoder.layers.1.trg_trg_att.output_layer.weight', 'decoder.layers.1.trg_trg_att.q_layer.bias', 'decoder.layers.1.trg_trg_att.q_layer.weight', 'decoder.layers.1.trg_trg_att.v_layer.bias', 'decoder.layers.1.trg_trg_att.v_layer.weight', 'decoder.layers.1.x_layer_norm.bias', 'decoder.layers.1.x_layer_norm.weight', 'decoder.layers.2.dec_layer_norm.bias', 'decoder.layers.2.dec_layer_norm.weight', 'decoder.layers.2.feed_forward.layer_norm.bias', 'decoder.layers.2.feed_forward.layer_norm.weight', 'decoder.layers.2.feed_forward.pwff_layer.0.bias', 'decoder.layers.2.feed_forward.pwff_layer.0.weight', 'decoder.layers.2.feed_forward.pwff_layer.3.bias', 'decoder.layers.2.feed_forward.pwff_layer.3.weight', 'decoder.layers.2.src_trg_att.k_layer.bias', 'decoder.layers.2.src_trg_att.k_layer.weight', 'decoder.layers.2.src_trg_att.output_layer.bias', 'decoder.layers.2.src_trg_att.output_layer.weight', 'decoder.layers.2.src_trg_att.q_layer.bias', 'decoder.layers.2.src_trg_att.q_layer.weight', 'decoder.layers.2.src_trg_att.v_layer.bias', 'decoder.layers.2.src_trg_att.v_layer.weight', 'decoder.layers.2.trg_trg_att.k_layer.bias', 'decoder.layers.2.trg_trg_att.k_layer.weight', 'decoder.layers.2.trg_trg_att.output_layer.bias', 'decoder.layers.2.trg_trg_att.output_layer.weight', 'decoder.layers.2.trg_trg_att.q_layer.bias', 'decoder.layers.2.trg_trg_att.q_layer.weight', 'decoder.layers.2.trg_trg_att.v_layer.bias', 'decoder.layers.2.trg_trg_att.v_layer.weight', 'decoder.layers.2.x_layer_norm.bias', 'decoder.layers.2.x_layer_norm.weight', 'decoder.layers.3.dec_layer_norm.bias', 'decoder.layers.3.dec_layer_norm.weight', 'decoder.layers.3.feed_forward.layer_norm.bias', 'decoder.layers.3.feed_forward.layer_norm.weight', 'decoder.layers.3.feed_forward.pwff_layer.0.bias', 'decoder.layers.3.feed_forward.pwff_layer.0.weight', 'decoder.layers.3.feed_forward.pwff_layer.3.bias', 'decoder.layers.3.feed_forward.pwff_layer.3.weight', 'decoder.layers.3.src_trg_att.k_layer.bias', 'decoder.layers.3.src_trg_att.k_layer.weight', 'decoder.layers.3.src_trg_att.output_layer.bias', 'decoder.layers.3.src_trg_att.output_layer.weight', 'decoder.layers.3.src_trg_att.q_layer.bias', 'decoder.layers.3.src_trg_att.q_layer.weight', 'decoder.layers.3.src_trg_att.v_layer.bias', 'decoder.layers.3.src_trg_att.v_layer.weight', 'decoder.layers.3.trg_trg_att.k_layer.bias', 'decoder.layers.3.trg_trg_att.k_layer.weight', 'decoder.layers.3.trg_trg_att.output_layer.bias', 'decoder.layers.3.trg_trg_att.output_layer.weight', 'decoder.layers.3.trg_trg_att.q_layer.bias', 'decoder.layers.3.trg_trg_att.q_layer.weight', 'decoder.layers.3.trg_trg_att.v_layer.bias', 'decoder.layers.3.trg_trg_att.v_layer.weight', 'decoder.layers.3.x_layer_norm.bias', 'decoder.layers.3.x_layer_norm.weight', 'decoder.layers.4.dec_layer_norm.bias', 'decoder.layers.4.dec_layer_norm.weight', 'decoder.layers.4.feed_forward.layer_norm.bias', 'decoder.layers.4.feed_forward.layer_norm.weight', 'decoder.layers.4.feed_forward.pwff_layer.0.bias', 'decoder.layers.4.feed_forward.pwff_layer.0.weight', 'decoder.layers.4.feed_forward.pwff_layer.3.bias', 'decoder.layers.4.feed_forward.pwff_layer.3.weight', 'decoder.layers.4.src_trg_att.k_layer.bias', 'decoder.layers.4.src_trg_att.k_layer.weight', 'decoder.layers.4.src_trg_att.output_layer.bias', 'decoder.layers.4.src_trg_att.output_layer.weight', 'decoder.layers.4.src_trg_att.q_layer.bias', 'decoder.layers.4.src_trg_att.q_layer.weight', 'decoder.layers.4.src_trg_att.v_layer.bias', 'decoder.layers.4.src_trg_att.v_layer.weight', 'decoder.layers.4.trg_trg_att.k_layer.bias', 'decoder.layers.4.trg_trg_att.k_layer.weight', 'decoder.layers.4.trg_trg_att.output_layer.bias', 'decoder.layers.4.trg_trg_att.output_layer.weight', 'decoder.layers.4.trg_trg_att.q_layer.bias', 'decoder.layers.4.trg_trg_att.q_layer.weight', 'decoder.layers.4.trg_trg_att.v_layer.bias', 'decoder.layers.4.trg_trg_att.v_layer.weight', 'decoder.layers.4.x_layer_norm.bias', 'decoder.layers.4.x_layer_norm.weight', 'decoder.layers.5.dec_layer_norm.bias', 'decoder.layers.5.dec_layer_norm.weight', 'decoder.layers.5.feed_forward.layer_norm.bias', 'decoder.layers.5.feed_forward.layer_norm.weight', 'decoder.layers.5.feed_forward.pwff_layer.0.bias', 'decoder.layers.5.feed_forward.pwff_layer.0.weight', 'decoder.layers.5.feed_forward.pwff_layer.3.bias', 'decoder.layers.5.feed_forward.pwff_layer.3.weight', 'decoder.layers.5.src_trg_att.k_layer.bias', 'decoder.layers.5.src_trg_att.k_layer.weight', 'decoder.layers.5.src_trg_att.output_layer.bias', 'decoder.layers.5.src_trg_att.output_layer.weight', 'decoder.layers.5.src_trg_att.q_layer.bias', 'decoder.layers.5.src_trg_att.q_layer.weight', 'decoder.layers.5.src_trg_att.v_layer.bias', 'decoder.layers.5.src_trg_att.v_layer.weight', 'decoder.layers.5.trg_trg_att.k_layer.bias', 'decoder.layers.5.trg_trg_att.k_layer.weight', 'decoder.layers.5.trg_trg_att.output_layer.bias', 'decoder.layers.5.trg_trg_att.output_layer.weight', 'decoder.layers.5.trg_trg_att.q_layer.bias', 'decoder.layers.5.trg_trg_att.q_layer.weight', 'decoder.layers.5.trg_trg_att.v_layer.bias', 'decoder.layers.5.trg_trg_att.v_layer.weight', 'decoder.layers.5.x_layer_norm.bias', 'decoder.layers.5.x_layer_norm.weight', 'encoder.layer_norm.bias', 'encoder.layer_norm.weight', 'encoder.layers.0.feed_forward.layer_norm.bias', 'encoder.layers.0.feed_forward.layer_norm.weight', 'encoder.layers.0.feed_forward.pwff_layer.0.bias', 'encoder.layers.0.feed_forward.pwff_layer.0.weight', 'encoder.layers.0.feed_forward.pwff_layer.3.bias', 'encoder.layers.0.feed_forward.pwff_layer.3.weight', 'encoder.layers.0.layer_norm.bias', 'encoder.layers.0.layer_norm.weight', 'encoder.layers.0.src_src_att.k_layer.bias', 'encoder.layers.0.src_src_att.k_layer.weight', 'encoder.layers.0.src_src_att.output_layer.bias', 'encoder.layers.0.src_src_att.output_layer.weight', 'encoder.layers.0.src_src_att.q_layer.bias', 'encoder.layers.0.src_src_att.q_layer.weight', 'encoder.layers.0.src_src_att.v_layer.bias', 'encoder.layers.0.src_src_att.v_layer.weight', 'encoder.layers.1.feed_forward.layer_norm.bias', 'encoder.layers.1.feed_forward.layer_norm.weight', 'encoder.layers.1.feed_forward.pwff_layer.0.bias', 'encoder.layers.1.feed_forward.pwff_layer.0.weight', 'encoder.layers.1.feed_forward.pwff_layer.3.bias', 'encoder.layers.1.feed_forward.pwff_layer.3.weight', 'encoder.layers.1.layer_norm.bias', 'encoder.layers.1.layer_norm.weight', 'encoder.layers.1.src_src_att.k_layer.bias', 'encoder.layers.1.src_src_att.k_layer.weight', 'encoder.layers.1.src_src_att.output_layer.bias', 'encoder.layers.1.src_src_att.output_layer.weight', 'encoder.layers.1.src_src_att.q_layer.bias', 'encoder.layers.1.src_src_att.q_layer.weight', 'encoder.layers.1.src_src_att.v_layer.bias', 'encoder.layers.1.src_src_att.v_layer.weight', 'encoder.layers.2.feed_forward.layer_norm.bias', 'encoder.layers.2.feed_forward.layer_norm.weight', 'encoder.layers.2.feed_forward.pwff_layer.0.bias', 'encoder.layers.2.feed_forward.pwff_layer.0.weight', 'encoder.layers.2.feed_forward.pwff_layer.3.bias', 'encoder.layers.2.feed_forward.pwff_layer.3.weight', 'encoder.layers.2.layer_norm.bias', 'encoder.layers.2.layer_norm.weight', 'encoder.layers.2.src_src_att.k_layer.bias', 'encoder.layers.2.src_src_att.k_layer.weight', 'encoder.layers.2.src_src_att.output_layer.bias', 'encoder.layers.2.src_src_att.output_layer.weight', 'encoder.layers.2.src_src_att.q_layer.bias', 'encoder.layers.2.src_src_att.q_layer.weight', 'encoder.layers.2.src_src_att.v_layer.bias', 'encoder.layers.2.src_src_att.v_layer.weight', 'encoder.layers.3.feed_forward.layer_norm.bias', 'encoder.layers.3.feed_forward.layer_norm.weight', 'encoder.layers.3.feed_forward.pwff_layer.0.bias', 'encoder.layers.3.feed_forward.pwff_layer.0.weight', 'encoder.layers.3.feed_forward.pwff_layer.3.bias', 'encoder.layers.3.feed_forward.pwff_layer.3.weight', 'encoder.layers.3.layer_norm.bias', 'encoder.layers.3.layer_norm.weight', 'encoder.layers.3.src_src_att.k_layer.bias', 'encoder.layers.3.src_src_att.k_layer.weight', 'encoder.layers.3.src_src_att.output_layer.bias', 'encoder.layers.3.src_src_att.output_layer.weight', 'encoder.layers.3.src_src_att.q_layer.bias', 'encoder.layers.3.src_src_att.q_layer.weight', 'encoder.layers.3.src_src_att.v_layer.bias', 'encoder.layers.3.src_src_att.v_layer.weight', 'encoder.layers.4.feed_forward.layer_norm.bias', 'encoder.layers.4.feed_forward.layer_norm.weight', 'encoder.layers.4.feed_forward.pwff_layer.0.bias', 'encoder.layers.4.feed_forward.pwff_layer.0.weight', 'encoder.layers.4.feed_forward.pwff_layer.3.bias', 'encoder.layers.4.feed_forward.pwff_layer.3.weight', 'encoder.layers.4.layer_norm.bias', 'encoder.layers.4.layer_norm.weight', 'encoder.layers.4.src_src_att.k_layer.bias', 'encoder.layers.4.src_src_att.k_layer.weight', 'encoder.layers.4.src_src_att.output_layer.bias', 'encoder.layers.4.src_src_att.output_layer.weight', 'encoder.layers.4.src_src_att.q_layer.bias', 'encoder.layers.4.src_src_att.q_layer.weight', 'encoder.layers.4.src_src_att.v_layer.bias', 'encoder.layers.4.src_src_att.v_layer.weight', 'encoder.layers.5.feed_forward.layer_norm.bias', 'encoder.layers.5.feed_forward.layer_norm.weight', 'encoder.layers.5.feed_forward.pwff_layer.0.bias', 'encoder.layers.5.feed_forward.pwff_layer.0.weight', 'encoder.layers.5.feed_forward.pwff_layer.3.bias', 'encoder.layers.5.feed_forward.pwff_layer.3.weight', 'encoder.layers.5.layer_norm.bias', 'encoder.layers.5.layer_norm.weight', 'encoder.layers.5.src_src_att.k_layer.bias', 'encoder.layers.5.src_src_att.k_layer.weight', 'encoder.layers.5.src_src_att.output_layer.bias', 'encoder.layers.5.src_src_att.output_layer.weight', 'encoder.layers.5.src_src_att.q_layer.bias', 'encoder.layers.5.src_src_att.q_layer.weight', 'encoder.layers.5.src_src_att.v_layer.bias', 'encoder.layers.5.src_src_att.v_layer.weight', 'src_embed.lut.weight']\n",
            "2020-04-12 14:27:19,825 Loading model from /content/drive/My Drive/masakhane/en-yo-baseline/models/enyo_transformer/best.ckpt\n",
            "2020-04-12 14:27:20,177 cfg.name                           : enyo_transformer\n",
            "2020-04-12 14:27:20,178 cfg.data.src                       : en\n",
            "2020-04-12 14:27:20,178 cfg.data.trg                       : yo\n",
            "2020-04-12 14:27:20,178 cfg.data.train                     : /content/drive/My Drive/masakhane/en-yo-baseline/train.bpe\n",
            "2020-04-12 14:27:20,178 cfg.data.dev                       : /content/drive/My Drive/masakhane/en-yo-baseline/dev.bpe\n",
            "2020-04-12 14:27:20,178 cfg.data.test                      : /content/drive/My Drive/masakhane/en-yo-baseline/test.bpe\n",
            "2020-04-12 14:27:20,178 cfg.data.level                     : bpe\n",
            "2020-04-12 14:27:20,179 cfg.data.lowercase                 : False\n",
            "2020-04-12 14:27:20,179 cfg.data.max_sent_length           : 100\n",
            "2020-04-12 14:27:20,179 cfg.data.src_vocab                 : /content/drive/My Drive/masakhane/en-yo-baseline/vocab.txt\n",
            "2020-04-12 14:27:20,179 cfg.data.trg_vocab                 : /content/drive/My Drive/masakhane/en-yo-baseline/vocab.txt\n",
            "2020-04-12 14:27:20,179 cfg.testing.beam_size              : 5\n",
            "2020-04-12 14:27:20,179 cfg.testing.alpha                  : 1.0\n",
            "2020-04-12 14:27:20,179 cfg.training.load_model            : /content/drive/My Drive/masakhane/en-yo-baseline/models/enyo_transformer/best.ckpt\n",
            "2020-04-12 14:27:20,180 cfg.training.random_seed           : 42\n",
            "2020-04-12 14:27:20,180 cfg.training.optimizer             : adam\n",
            "2020-04-12 14:27:20,180 cfg.training.normalization         : tokens\n",
            "2020-04-12 14:27:20,180 cfg.training.adam_betas            : [0.9, 0.999]\n",
            "2020-04-12 14:27:20,180 cfg.training.scheduling            : plateau\n",
            "2020-04-12 14:27:20,180 cfg.training.patience              : 5\n",
            "2020-04-12 14:27:20,180 cfg.training.learning_rate_factor  : 0.5\n",
            "2020-04-12 14:27:20,180 cfg.training.learning_rate_warmup  : 1000\n",
            "2020-04-12 14:27:20,180 cfg.training.decrease_factor       : 0.7\n",
            "2020-04-12 14:27:20,181 cfg.training.loss                  : crossentropy\n",
            "2020-04-12 14:27:20,181 cfg.training.learning_rate         : 0.0003\n",
            "2020-04-12 14:27:20,181 cfg.training.learning_rate_min     : 1e-08\n",
            "2020-04-12 14:27:20,181 cfg.training.weight_decay          : 0.0\n",
            "2020-04-12 14:27:20,181 cfg.training.label_smoothing       : 0.1\n",
            "2020-04-12 14:27:20,181 cfg.training.batch_size            : 4096\n",
            "2020-04-12 14:27:20,181 cfg.training.batch_type            : token\n",
            "2020-04-12 14:27:20,181 cfg.training.eval_batch_size       : 3600\n",
            "2020-04-12 14:27:20,182 cfg.training.eval_batch_type       : token\n",
            "2020-04-12 14:27:20,182 cfg.training.batch_multiplier      : 1\n",
            "2020-04-12 14:27:20,182 cfg.training.early_stopping_metric : ppl\n",
            "2020-04-12 14:27:20,182 cfg.training.epochs                : 2\n",
            "2020-04-12 14:27:20,182 cfg.training.validation_freq       : 1000\n",
            "2020-04-12 14:27:20,182 cfg.training.logging_freq          : 100\n",
            "2020-04-12 14:27:20,182 cfg.training.eval_metric           : bleu\n",
            "2020-04-12 14:27:20,183 cfg.training.model_dir             : /content/drive/My Drive/masakhane/model-temp\n",
            "2020-04-12 14:27:20,183 cfg.training.overwrite             : True\n",
            "2020-04-12 14:27:20,183 cfg.training.shuffle               : True\n",
            "2020-04-12 14:27:20,183 cfg.training.use_cuda              : True\n",
            "2020-04-12 14:27:20,183 cfg.training.max_output_length     : 100\n",
            "2020-04-12 14:27:20,183 cfg.training.print_valid_sents     : [0, 1, 2, 3]\n",
            "2020-04-12 14:27:20,183 cfg.training.keep_last_ckpts       : 3\n",
            "2020-04-12 14:27:20,183 cfg.model.initializer              : xavier\n",
            "2020-04-12 14:27:20,184 cfg.model.bias_initializer         : zeros\n",
            "2020-04-12 14:27:20,184 cfg.model.init_gain                : 1.0\n",
            "2020-04-12 14:27:20,184 cfg.model.embed_initializer        : xavier\n",
            "2020-04-12 14:27:20,184 cfg.model.embed_init_gain          : 1.0\n",
            "2020-04-12 14:27:20,184 cfg.model.tied_embeddings          : True\n",
            "2020-04-12 14:27:20,184 cfg.model.tied_softmax             : True\n",
            "2020-04-12 14:27:20,184 cfg.model.encoder.type             : transformer\n",
            "2020-04-12 14:27:20,185 cfg.model.encoder.num_layers       : 6\n",
            "2020-04-12 14:27:20,185 cfg.model.encoder.num_heads        : 4\n",
            "2020-04-12 14:27:20,185 cfg.model.encoder.embeddings.embedding_dim : 256\n",
            "2020-04-12 14:27:20,185 cfg.model.encoder.embeddings.scale : True\n",
            "2020-04-12 14:27:20,185 cfg.model.encoder.embeddings.dropout : 0.2\n",
            "2020-04-12 14:27:20,185 cfg.model.encoder.hidden_size      : 256\n",
            "2020-04-12 14:27:20,185 cfg.model.encoder.ff_size          : 1024\n",
            "2020-04-12 14:27:20,185 cfg.model.encoder.dropout          : 0.3\n",
            "2020-04-12 14:27:20,186 cfg.model.decoder.type             : transformer\n",
            "2020-04-12 14:27:20,186 cfg.model.decoder.num_layers       : 6\n",
            "2020-04-12 14:27:20,186 cfg.model.decoder.num_heads        : 4\n",
            "2020-04-12 14:27:20,186 cfg.model.decoder.embeddings.embedding_dim : 256\n",
            "2020-04-12 14:27:20,186 cfg.model.decoder.embeddings.scale : True\n",
            "2020-04-12 14:27:20,186 cfg.model.decoder.embeddings.dropout : 0.2\n",
            "2020-04-12 14:27:20,186 cfg.model.decoder.hidden_size      : 256\n",
            "2020-04-12 14:27:20,186 cfg.model.decoder.ff_size          : 1024\n",
            "2020-04-12 14:27:20,187 cfg.model.decoder.dropout          : 0.3\n",
            "2020-04-12 14:27:20,187 Data set sizes: \n",
            "\ttrain 415100,\n",
            "\tvalid 1000,\n",
            "\ttest 2662\n",
            "2020-04-12 14:27:20,187 First training example:\n",
            "\t[SRC] T@@ R@@ A@@ IN Y@@ O@@ U@@ R C@@ H@@ I@@ L@@ D@@ R@@ E@@ N : “ I teach my children to ch@@ ec@@ k the exp@@ ir@@ ation d@@ ate of any p@@ ack@@ aged food it@@ em@@ s , such as s@@ n@@ ac@@ ks , before they bu@@ y them . ” ​ — Ru@@ th , N@@ ig@@ er@@ ia\n",
            "\t[TRG] K@@ Ọ́ ÀWỌN Ọ@@ M@@ Ọ R@@ Ẹ : “ Mo kọ́ àwọn ọmọ mi pé kí wọ́n tó ra oúnjẹ bí ìp@@ á@@ p@@ án@@ u , tó wà nínú ag@@ ol@@ o , i@@ ke , bé@@ bà , tàbí ọ̀r@@ á , kí wọ́n máa yẹ ara oúnjẹ náà wò kí wọ́n lè mọ dé@@ è@@ tì tó máa bà jẹ́ . ” — Ru@@ th , N@@ àì@@ jí@@ ríà\n",
            "2020-04-12 14:27:20,187 First 10 words (src): (0) <unk> (1) <pad> (2) <s> (3) </s> (4) , (5) . (6) the (7) tó (8) a (9) to\n",
            "2020-04-12 14:27:20,187 First 10 words (trg): (0) <unk> (1) <pad> (2) <s> (3) </s> (4) , (5) . (6) the (7) tó (8) a (9) to\n",
            "2020-04-12 14:27:20,187 Number of Src words (types): 4406\n",
            "2020-04-12 14:27:20,188 Number of Trg words (types): 4406\n",
            "2020-04-12 14:27:20,188 Model(\n",
            "\tencoder=TransformerEncoder(num_layers=6, num_heads=4),\n",
            "\tdecoder=TransformerDecoder(num_layers=6, num_heads=4),\n",
            "\tsrc_embed=Embeddings(embedding_dim=256, vocab_size=4406),\n",
            "\ttrg_embed=Embeddings(embedding_dim=256, vocab_size=4406))\n",
            "2020-04-12 14:27:20,198 EPOCH 1\n",
            "2020-04-12 14:27:31,916 Epoch   1 Step:   384100 Batch Loss:     2.240336 Tokens per Sec:    19764, Lr: 0.000004\n",
            "2020-04-12 14:27:42,982 Epoch   1 Step:   384200 Batch Loss:     1.883319 Tokens per Sec:    20576, Lr: 0.000004\n",
            "2020-04-12 14:27:54,021 Epoch   1 Step:   384300 Batch Loss:     1.927691 Tokens per Sec:    20939, Lr: 0.000004\n",
            "2020-04-12 14:28:05,130 Epoch   1 Step:   384400 Batch Loss:     1.918795 Tokens per Sec:    21147, Lr: 0.000004\n",
            "2020-04-12 14:28:16,247 Epoch   1 Step:   384500 Batch Loss:     2.029218 Tokens per Sec:    20929, Lr: 0.000004\n",
            "2020-04-12 14:28:27,333 Epoch   1 Step:   384600 Batch Loss:     1.894163 Tokens per Sec:    20946, Lr: 0.000004\n",
            "2020-04-12 14:28:38,420 Epoch   1 Step:   384700 Batch Loss:     1.838998 Tokens per Sec:    20906, Lr: 0.000004\n",
            "2020-04-12 14:28:49,432 Epoch   1 Step:   384800 Batch Loss:     1.793913 Tokens per Sec:    21056, Lr: 0.000004\n",
            "2020-04-12 14:29:00,482 Epoch   1 Step:   384900 Batch Loss:     1.827213 Tokens per Sec:    20679, Lr: 0.000004\n",
            "2020-04-12 14:29:11,456 Epoch   1 Step:   385000 Batch Loss:     1.768281 Tokens per Sec:    21078, Lr: 0.000004\n",
            "2020-04-12 14:29:28,421 Example #0\n",
            "2020-04-12 14:29:28,422 \tSource:     He is the Source of life , the One giving it as an undeserved gift through Christ .\n",
            "2020-04-12 14:29:28,422 \tReference:  Òun ni Orísun ìyè , Ẹni tí ń fi ìyè fúnni gẹ́gẹ́ bí ẹbùn tí a kò lẹ́tọ̀ọ́ sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:29:28,422 \tHypothesis: Òun ni Orísun ìyè , Ẹni tí ń fi í fúnni gẹ́gẹ́ bí ẹ̀bùn àìlẹ́tọ̀ọ́sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:29:28,422 Example #1\n",
            "2020-04-12 14:29:28,423 \tSource:     Now I had to find a legitimate line of work .\n",
            "2020-04-12 14:29:28,423 \tReference:  Torí náà , mo ní láti wá iṣẹ́ gidi .\n",
            "2020-04-12 14:29:28,423 \tHypothesis: Ní báyìí , mo ní láti wá iṣẹ́ tó yẹ kí n ṣe .\n",
            "2020-04-12 14:29:28,423 Example #2\n",
            "2020-04-12 14:29:28,424 \tSource:     Do I value material things more than my relationship with Jehovah and with people ?\n",
            "2020-04-12 14:29:28,424 \tReference:  Ṣé àwọn nǹkan tara ló jẹ mí lógún jù àbí àjọṣe mi pẹ̀lú Jèhófà àtàwọn èèyàn ?\n",
            "2020-04-12 14:29:28,424 \tHypothesis: Ǹjẹ́ mo mọyì àwọn nǹkan tara ju àjọṣe mi pẹ̀lú Jèhófà àti pẹ̀lú àwọn èèyàn lọ ?\n",
            "2020-04-12 14:29:28,424 Example #3\n",
            "2020-04-12 14:29:28,424 \tSource:     He has far more experience and stamina than you do , but he patiently walks near you .\n",
            "2020-04-12 14:29:28,424 \tReference:  Ẹni tẹ́ ẹ jọ ń lọ yìí mọ ọ̀nà yẹn dáadáa .\n",
            "2020-04-12 14:29:28,425 \tHypothesis: Ó ní ìrírí tó pọ̀ gan - an , ó sì tún ní ìrírí tó ju tìẹ lọ , àmọ́ ó fi sùúrù rìn nítòsí rẹ .\n",
            "2020-04-12 14:29:28,425 Validation result (greedy) at epoch   1, step   385000: bleu:  30.04, loss: 41967.2539, ppl:   4.0905, duration: 16.9679s\n",
            "2020-04-12 14:29:39,582 Epoch   1 Step:   385100 Batch Loss:     2.121287 Tokens per Sec:    21098, Lr: 0.000004\n",
            "2020-04-12 14:29:50,538 Epoch   1 Step:   385200 Batch Loss:     1.998006 Tokens per Sec:    20587, Lr: 0.000004\n",
            "2020-04-12 14:30:01,615 Epoch   1 Step:   385300 Batch Loss:     1.912782 Tokens per Sec:    21209, Lr: 0.000004\n",
            "2020-04-12 14:30:12,660 Epoch   1 Step:   385400 Batch Loss:     1.806685 Tokens per Sec:    20871, Lr: 0.000004\n",
            "2020-04-12 14:30:23,759 Epoch   1 Step:   385500 Batch Loss:     2.052587 Tokens per Sec:    21211, Lr: 0.000004\n",
            "2020-04-12 14:30:34,653 Epoch   1 Step:   385600 Batch Loss:     1.783153 Tokens per Sec:    20515, Lr: 0.000004\n",
            "2020-04-12 14:30:45,723 Epoch   1 Step:   385700 Batch Loss:     1.835905 Tokens per Sec:    21566, Lr: 0.000004\n",
            "2020-04-12 14:30:56,709 Epoch   1 Step:   385800 Batch Loss:     1.911717 Tokens per Sec:    20845, Lr: 0.000004\n",
            "2020-04-12 14:31:07,630 Epoch   1 Step:   385900 Batch Loss:     1.921858 Tokens per Sec:    21159, Lr: 0.000004\n",
            "2020-04-12 14:31:18,512 Epoch   1 Step:   386000 Batch Loss:     1.973545 Tokens per Sec:    20908, Lr: 0.000004\n",
            "2020-04-12 14:31:35,396 Example #0\n",
            "2020-04-12 14:31:35,397 \tSource:     He is the Source of life , the One giving it as an undeserved gift through Christ .\n",
            "2020-04-12 14:31:35,397 \tReference:  Òun ni Orísun ìyè , Ẹni tí ń fi ìyè fúnni gẹ́gẹ́ bí ẹbùn tí a kò lẹ́tọ̀ọ́ sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:31:35,397 \tHypothesis: Òun ni Orísun ìyè , Ẹni tí ń fi í fúnni gẹ́gẹ́ bí ẹ̀bùn àìlẹ́tọ̀ọ́sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:31:35,397 Example #1\n",
            "2020-04-12 14:31:35,398 \tSource:     Now I had to find a legitimate line of work .\n",
            "2020-04-12 14:31:35,398 \tReference:  Torí náà , mo ní láti wá iṣẹ́ gidi .\n",
            "2020-04-12 14:31:35,398 \tHypothesis: Ní báyìí , mo ní láti wá iṣẹ́ tó yẹ kí n ṣe .\n",
            "2020-04-12 14:31:35,398 Example #2\n",
            "2020-04-12 14:31:35,399 \tSource:     Do I value material things more than my relationship with Jehovah and with people ?\n",
            "2020-04-12 14:31:35,399 \tReference:  Ṣé àwọn nǹkan tara ló jẹ mí lógún jù àbí àjọṣe mi pẹ̀lú Jèhófà àtàwọn èèyàn ?\n",
            "2020-04-12 14:31:35,399 \tHypothesis: Ǹjẹ́ mo mọyì àwọn nǹkan tara ju àjọṣe mi pẹ̀lú Jèhófà àti pẹ̀lú àwọn èèyàn lọ ?\n",
            "2020-04-12 14:31:35,399 Example #3\n",
            "2020-04-12 14:31:35,400 \tSource:     He has far more experience and stamina than you do , but he patiently walks near you .\n",
            "2020-04-12 14:31:35,400 \tReference:  Ẹni tẹ́ ẹ jọ ń lọ yìí mọ ọ̀nà yẹn dáadáa .\n",
            "2020-04-12 14:31:35,400 \tHypothesis: Ó ní ìrírí tó pọ̀ gan - an , ó sì tún ní ìrírí tó ju tìẹ lọ , àmọ́ ó fi sùúrù rìn nítòsí rẹ .\n",
            "2020-04-12 14:31:35,400 Validation result (greedy) at epoch   1, step   386000: bleu:  29.92, loss: 42548.0859, ppl:   4.1711, duration: 16.8875s\n",
            "2020-04-12 14:31:46,453 Epoch   1 Step:   386100 Batch Loss:     2.060577 Tokens per Sec:    21051, Lr: 0.000004\n",
            "2020-04-12 14:31:57,360 Epoch   1 Step:   386200 Batch Loss:     1.789646 Tokens per Sec:    20830, Lr: 0.000004\n",
            "2020-04-12 14:32:08,332 Epoch   1 Step:   386300 Batch Loss:     1.944916 Tokens per Sec:    21407, Lr: 0.000004\n",
            "2020-04-12 14:32:19,253 Epoch   1 Step:   386400 Batch Loss:     1.973574 Tokens per Sec:    21195, Lr: 0.000004\n",
            "2020-04-12 14:32:30,182 Epoch   1 Step:   386500 Batch Loss:     1.965004 Tokens per Sec:    21150, Lr: 0.000004\n",
            "2020-04-12 14:32:41,084 Epoch   1 Step:   386600 Batch Loss:     1.814785 Tokens per Sec:    20966, Lr: 0.000004\n",
            "2020-04-12 14:32:52,090 Epoch   1 Step:   386700 Batch Loss:     2.025830 Tokens per Sec:    20638, Lr: 0.000004\n",
            "2020-04-12 14:33:03,165 Epoch   1 Step:   386800 Batch Loss:     1.706421 Tokens per Sec:    21197, Lr: 0.000004\n",
            "2020-04-12 14:33:14,067 Epoch   1 Step:   386900 Batch Loss:     1.815666 Tokens per Sec:    20733, Lr: 0.000004\n",
            "2020-04-12 14:33:24,980 Epoch   1 Step:   387000 Batch Loss:     1.893143 Tokens per Sec:    21026, Lr: 0.000004\n",
            "2020-04-12 14:33:41,969 Example #0\n",
            "2020-04-12 14:33:41,969 \tSource:     He is the Source of life , the One giving it as an undeserved gift through Christ .\n",
            "2020-04-12 14:33:41,969 \tReference:  Òun ni Orísun ìyè , Ẹni tí ń fi ìyè fúnni gẹ́gẹ́ bí ẹbùn tí a kò lẹ́tọ̀ọ́ sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:33:41,970 \tHypothesis: Òun ni Orísun ìyè , Ẹni tí ń fi í fúnni gẹ́gẹ́ bí ẹ̀bùn àìlẹ́tọ̀ọ́sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:33:41,970 Example #1\n",
            "2020-04-12 14:33:41,970 \tSource:     Now I had to find a legitimate line of work .\n",
            "2020-04-12 14:33:41,970 \tReference:  Torí náà , mo ní láti wá iṣẹ́ gidi .\n",
            "2020-04-12 14:33:41,971 \tHypothesis: Ní báyìí , mo ní láti wá iṣẹ́ tó dára .\n",
            "2020-04-12 14:33:41,971 Example #2\n",
            "2020-04-12 14:33:41,972 \tSource:     Do I value material things more than my relationship with Jehovah and with people ?\n",
            "2020-04-12 14:33:41,972 \tReference:  Ṣé àwọn nǹkan tara ló jẹ mí lógún jù àbí àjọṣe mi pẹ̀lú Jèhófà àtàwọn èèyàn ?\n",
            "2020-04-12 14:33:41,972 \tHypothesis: Ǹjẹ́ mo mọyì àwọn nǹkan tara ju àjọṣe mi pẹ̀lú Jèhófà àti pẹ̀lú àwọn èèyàn lọ ?\n",
            "2020-04-12 14:33:41,972 Example #3\n",
            "2020-04-12 14:33:41,973 \tSource:     He has far more experience and stamina than you do , but he patiently walks near you .\n",
            "2020-04-12 14:33:41,973 \tReference:  Ẹni tẹ́ ẹ jọ ń lọ yìí mọ ọ̀nà yẹn dáadáa .\n",
            "2020-04-12 14:33:41,973 \tHypothesis: Ó ní ìrírí tó pọ̀ gan - an , ó sì tún ní ìrírí tó ju tìẹ lọ , àmọ́ ó fi sùúrù rìn nítòsí rẹ .\n",
            "2020-04-12 14:33:41,973 Validation result (greedy) at epoch   1, step   387000: bleu:  29.87, loss: 42807.0391, ppl:   4.2075, duration: 16.9932s\n",
            "2020-04-12 14:33:52,943 Epoch   1 Step:   387100 Batch Loss:     1.831305 Tokens per Sec:    20895, Lr: 0.000004\n",
            "2020-04-12 14:34:03,745 Epoch   1 Step:   387200 Batch Loss:     1.805957 Tokens per Sec:    20377, Lr: 0.000004\n",
            "2020-04-12 14:34:14,840 Epoch   1 Step:   387300 Batch Loss:     1.742952 Tokens per Sec:    21086, Lr: 0.000004\n",
            "2020-04-12 14:34:25,757 Epoch   1 Step:   387400 Batch Loss:     2.395472 Tokens per Sec:    21045, Lr: 0.000004\n",
            "2020-04-12 14:34:36,705 Epoch   1 Step:   387500 Batch Loss:     1.808187 Tokens per Sec:    20949, Lr: 0.000004\n",
            "2020-04-12 14:34:47,693 Epoch   1 Step:   387600 Batch Loss:     1.924356 Tokens per Sec:    21305, Lr: 0.000004\n",
            "2020-04-12 14:34:58,665 Epoch   1 Step:   387700 Batch Loss:     1.883427 Tokens per Sec:    20812, Lr: 0.000004\n",
            "2020-04-12 14:35:09,662 Epoch   1 Step:   387800 Batch Loss:     1.921607 Tokens per Sec:    21186, Lr: 0.000004\n",
            "2020-04-12 14:35:20,718 Epoch   1 Step:   387900 Batch Loss:     1.909007 Tokens per Sec:    21307, Lr: 0.000004\n",
            "2020-04-12 14:35:31,647 Epoch   1 Step:   388000 Batch Loss:     1.892324 Tokens per Sec:    20464, Lr: 0.000004\n",
            "2020-04-12 14:35:48,532 Example #0\n",
            "2020-04-12 14:35:48,533 \tSource:     He is the Source of life , the One giving it as an undeserved gift through Christ .\n",
            "2020-04-12 14:35:48,533 \tReference:  Òun ni Orísun ìyè , Ẹni tí ń fi ìyè fúnni gẹ́gẹ́ bí ẹbùn tí a kò lẹ́tọ̀ọ́ sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:35:48,533 \tHypothesis: Òun ni Orísun ìyè , Ẹni tí ń fi í fúnni gẹ́gẹ́ bí ẹ̀bùn àìlẹ́tọ̀ọ́sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:35:48,533 Example #1\n",
            "2020-04-12 14:35:48,533 \tSource:     Now I had to find a legitimate line of work .\n",
            "2020-04-12 14:35:48,534 \tReference:  Torí náà , mo ní láti wá iṣẹ́ gidi .\n",
            "2020-04-12 14:35:48,534 \tHypothesis: Ní báyìí , mo ní láti wá iṣẹ́ tó dára .\n",
            "2020-04-12 14:35:48,534 Example #2\n",
            "2020-04-12 14:35:48,534 \tSource:     Do I value material things more than my relationship with Jehovah and with people ?\n",
            "2020-04-12 14:35:48,534 \tReference:  Ṣé àwọn nǹkan tara ló jẹ mí lógún jù àbí àjọṣe mi pẹ̀lú Jèhófà àtàwọn èèyàn ?\n",
            "2020-04-12 14:35:48,534 \tHypothesis: Ǹjẹ́ mo mọyì àwọn nǹkan tara ju àjọṣe mi pẹ̀lú Jèhófà àti pẹ̀lú àwọn èèyàn lọ ?\n",
            "2020-04-12 14:35:48,535 Example #3\n",
            "2020-04-12 14:35:48,535 \tSource:     He has far more experience and stamina than you do , but he patiently walks near you .\n",
            "2020-04-12 14:35:48,535 \tReference:  Ẹni tẹ́ ẹ jọ ń lọ yìí mọ ọ̀nà yẹn dáadáa .\n",
            "2020-04-12 14:35:48,535 \tHypothesis: Ó ní ìrírí tó pọ̀ gan - an , ó sì tún ní ìrírí tó ju tìẹ lọ , àmọ́ ó fi sùúrù rìn nítòsí rẹ .\n",
            "2020-04-12 14:35:48,535 Validation result (greedy) at epoch   1, step   388000: bleu:  29.81, loss: 42958.7578, ppl:   4.2290, duration: 16.8883s\n",
            "2020-04-12 14:35:59,503 Epoch   1 Step:   388100 Batch Loss:     1.835913 Tokens per Sec:    21177, Lr: 0.000004\n",
            "2020-04-12 14:36:10,367 Epoch   1 Step:   388200 Batch Loss:     1.863079 Tokens per Sec:    21249, Lr: 0.000004\n",
            "2020-04-12 14:36:21,399 Epoch   1 Step:   388300 Batch Loss:     1.825538 Tokens per Sec:    20914, Lr: 0.000004\n",
            "2020-04-12 14:36:32,421 Epoch   1 Step:   388400 Batch Loss:     1.788421 Tokens per Sec:    21343, Lr: 0.000004\n",
            "2020-04-12 14:36:43,501 Epoch   1 Step:   388500 Batch Loss:     2.025025 Tokens per Sec:    21474, Lr: 0.000004\n",
            "2020-04-12 14:36:54,499 Epoch   1 Step:   388600 Batch Loss:     1.974475 Tokens per Sec:    20873, Lr: 0.000004\n",
            "2020-04-12 14:37:05,496 Epoch   1 Step:   388700 Batch Loss:     1.740090 Tokens per Sec:    20847, Lr: 0.000004\n",
            "2020-04-12 14:37:16,508 Epoch   1 Step:   388800 Batch Loss:     1.982534 Tokens per Sec:    21475, Lr: 0.000004\n",
            "2020-04-12 14:37:27,535 Epoch   1 Step:   388900 Batch Loss:     1.769454 Tokens per Sec:    21121, Lr: 0.000004\n",
            "2020-04-12 14:37:38,498 Epoch   1 Step:   389000 Batch Loss:     2.170954 Tokens per Sec:    21243, Lr: 0.000004\n",
            "2020-04-12 14:37:55,188 Example #0\n",
            "2020-04-12 14:37:55,189 \tSource:     He is the Source of life , the One giving it as an undeserved gift through Christ .\n",
            "2020-04-12 14:37:55,189 \tReference:  Òun ni Orísun ìyè , Ẹni tí ń fi ìyè fúnni gẹ́gẹ́ bí ẹbùn tí a kò lẹ́tọ̀ọ́ sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:37:55,189 \tHypothesis: Òun ni Orísun ìyè , Ẹni tí ń fi í fúnni gẹ́gẹ́ bí ẹ̀bùn àìlẹ́tọ̀ọ́sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:37:55,189 Example #1\n",
            "2020-04-12 14:37:55,190 \tSource:     Now I had to find a legitimate line of work .\n",
            "2020-04-12 14:37:55,190 \tReference:  Torí náà , mo ní láti wá iṣẹ́ gidi .\n",
            "2020-04-12 14:37:55,190 \tHypothesis: Ní báyìí , mo ní láti wá iṣẹ́ tó dára .\n",
            "2020-04-12 14:37:55,190 Example #2\n",
            "2020-04-12 14:37:55,190 \tSource:     Do I value material things more than my relationship with Jehovah and with people ?\n",
            "2020-04-12 14:37:55,190 \tReference:  Ṣé àwọn nǹkan tara ló jẹ mí lógún jù àbí àjọṣe mi pẹ̀lú Jèhófà àtàwọn èèyàn ?\n",
            "2020-04-12 14:37:55,191 \tHypothesis: Ǹjẹ́ mo mọyì àwọn nǹkan tara ju àjọṣe mi pẹ̀lú Jèhófà àti pẹ̀lú àwọn èèyàn lọ ?\n",
            "2020-04-12 14:37:55,191 Example #3\n",
            "2020-04-12 14:37:55,191 \tSource:     He has far more experience and stamina than you do , but he patiently walks near you .\n",
            "2020-04-12 14:37:55,191 \tReference:  Ẹni tẹ́ ẹ jọ ń lọ yìí mọ ọ̀nà yẹn dáadáa .\n",
            "2020-04-12 14:37:55,191 \tHypothesis: Ó ní ìrírí tó pọ̀ gan - an , ó sì tún ní ìrírí tó ju tìẹ lọ , àmọ́ ó fi sùúrù rìn nítòsí rẹ .\n",
            "2020-04-12 14:37:55,192 Validation result (greedy) at epoch   1, step   389000: bleu:  29.81, loss: 43055.4961, ppl:   4.2427, duration: 16.6934s\n",
            "2020-04-12 14:38:06,145 Epoch   1 Step:   389100 Batch Loss:     2.074862 Tokens per Sec:    20912, Lr: 0.000004\n",
            "2020-04-12 14:38:17,133 Epoch   1 Step:   389200 Batch Loss:     1.827384 Tokens per Sec:    21191, Lr: 0.000004\n",
            "2020-04-12 14:38:22,901 Epoch   1: total training loss 9948.21\n",
            "2020-04-12 14:38:22,901 EPOCH 2\n",
            "2020-04-12 14:38:28,719 Epoch   2 Step:   389300 Batch Loss:     1.760182 Tokens per Sec:    18760, Lr: 0.000004\n",
            "2020-04-12 14:38:39,664 Epoch   2 Step:   389400 Batch Loss:     1.873315 Tokens per Sec:    21172, Lr: 0.000004\n",
            "2020-04-12 14:38:50,639 Epoch   2 Step:   389500 Batch Loss:     1.902953 Tokens per Sec:    20788, Lr: 0.000004\n",
            "2020-04-12 14:39:01,574 Epoch   2 Step:   389600 Batch Loss:     1.809665 Tokens per Sec:    21479, Lr: 0.000004\n",
            "2020-04-12 14:39:12,512 Epoch   2 Step:   389700 Batch Loss:     1.766662 Tokens per Sec:    20945, Lr: 0.000004\n",
            "2020-04-12 14:39:23,531 Epoch   2 Step:   389800 Batch Loss:     1.920965 Tokens per Sec:    21257, Lr: 0.000004\n",
            "2020-04-12 14:39:34,589 Epoch   2 Step:   389900 Batch Loss:     1.631354 Tokens per Sec:    20831, Lr: 0.000004\n",
            "2020-04-12 14:39:45,624 Epoch   2 Step:   390000 Batch Loss:     1.987037 Tokens per Sec:    21283, Lr: 0.000004\n",
            "2020-04-12 14:40:02,345 Example #0\n",
            "2020-04-12 14:40:02,346 \tSource:     He is the Source of life , the One giving it as an undeserved gift through Christ .\n",
            "2020-04-12 14:40:02,346 \tReference:  Òun ni Orísun ìyè , Ẹni tí ń fi ìyè fúnni gẹ́gẹ́ bí ẹbùn tí a kò lẹ́tọ̀ọ́ sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:40:02,346 \tHypothesis: Òun ni Orísun ìyè , Ẹni tí ń fi í fúnni gẹ́gẹ́ bí ẹ̀bùn àìlẹ́tọ̀ọ́sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:40:02,346 Example #1\n",
            "2020-04-12 14:40:02,347 \tSource:     Now I had to find a legitimate line of work .\n",
            "2020-04-12 14:40:02,347 \tReference:  Torí náà , mo ní láti wá iṣẹ́ gidi .\n",
            "2020-04-12 14:40:02,347 \tHypothesis: Ní báyìí , mo ní láti wá iṣẹ́ tó dára .\n",
            "2020-04-12 14:40:02,347 Example #2\n",
            "2020-04-12 14:40:02,348 \tSource:     Do I value material things more than my relationship with Jehovah and with people ?\n",
            "2020-04-12 14:40:02,348 \tReference:  Ṣé àwọn nǹkan tara ló jẹ mí lógún jù àbí àjọṣe mi pẹ̀lú Jèhófà àtàwọn èèyàn ?\n",
            "2020-04-12 14:40:02,348 \tHypothesis: Ǹjẹ́ mo mọyì àwọn nǹkan tara ju àjọṣe mi pẹ̀lú Jèhófà àti pẹ̀lú àwọn èèyàn lọ ?\n",
            "2020-04-12 14:40:02,348 Example #3\n",
            "2020-04-12 14:40:02,348 \tSource:     He has far more experience and stamina than you do , but he patiently walks near you .\n",
            "2020-04-12 14:40:02,348 \tReference:  Ẹni tẹ́ ẹ jọ ń lọ yìí mọ ọ̀nà yẹn dáadáa .\n",
            "2020-04-12 14:40:02,349 \tHypothesis: Ó ní ìrírí tó pọ̀ gan - an , ó sì tún ní ìrírí tó ju tìẹ lọ , àmọ́ ó fi sùúrù rìn nítòsí rẹ .\n",
            "2020-04-12 14:40:02,349 Validation result (greedy) at epoch   2, step   390000: bleu:  29.73, loss: 43157.8398, ppl:   4.2573, duration: 16.7243s\n",
            "2020-04-12 14:40:13,342 Epoch   2 Step:   390100 Batch Loss:     1.877792 Tokens per Sec:    21187, Lr: 0.000003\n",
            "2020-04-12 14:40:24,329 Epoch   2 Step:   390200 Batch Loss:     2.167672 Tokens per Sec:    21032, Lr: 0.000003\n",
            "2020-04-12 14:40:35,294 Epoch   2 Step:   390300 Batch Loss:     1.830189 Tokens per Sec:    21334, Lr: 0.000003\n",
            "2020-04-12 14:40:46,276 Epoch   2 Step:   390400 Batch Loss:     1.821566 Tokens per Sec:    21062, Lr: 0.000003\n",
            "2020-04-12 14:40:57,227 Epoch   2 Step:   390500 Batch Loss:     1.758496 Tokens per Sec:    20903, Lr: 0.000003\n",
            "2020-04-12 14:41:08,233 Epoch   2 Step:   390600 Batch Loss:     1.762770 Tokens per Sec:    21055, Lr: 0.000003\n",
            "2020-04-12 14:41:19,211 Epoch   2 Step:   390700 Batch Loss:     1.878187 Tokens per Sec:    21310, Lr: 0.000003\n",
            "2020-04-12 14:41:30,212 Epoch   2 Step:   390800 Batch Loss:     1.967181 Tokens per Sec:    21377, Lr: 0.000003\n",
            "2020-04-12 14:41:41,199 Epoch   2 Step:   390900 Batch Loss:     1.863368 Tokens per Sec:    21242, Lr: 0.000003\n",
            "2020-04-12 14:41:52,307 Epoch   2 Step:   391000 Batch Loss:     1.980855 Tokens per Sec:    20991, Lr: 0.000003\n",
            "2020-04-12 14:42:08,885 Example #0\n",
            "2020-04-12 14:42:08,885 \tSource:     He is the Source of life , the One giving it as an undeserved gift through Christ .\n",
            "2020-04-12 14:42:08,886 \tReference:  Òun ni Orísun ìyè , Ẹni tí ń fi ìyè fúnni gẹ́gẹ́ bí ẹbùn tí a kò lẹ́tọ̀ọ́ sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:42:08,886 \tHypothesis: Òun ni Orísun ìyè , Ẹni tí ń fi í fúnni gẹ́gẹ́ bí ẹ̀bùn àìlẹ́tọ̀ọ́sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:42:08,886 Example #1\n",
            "2020-04-12 14:42:08,886 \tSource:     Now I had to find a legitimate line of work .\n",
            "2020-04-12 14:42:08,886 \tReference:  Torí náà , mo ní láti wá iṣẹ́ gidi .\n",
            "2020-04-12 14:42:08,887 \tHypothesis: Ní báyìí , mo ní láti wá iṣẹ́ tó dára .\n",
            "2020-04-12 14:42:08,887 Example #2\n",
            "2020-04-12 14:42:08,887 \tSource:     Do I value material things more than my relationship with Jehovah and with people ?\n",
            "2020-04-12 14:42:08,887 \tReference:  Ṣé àwọn nǹkan tara ló jẹ mí lógún jù àbí àjọṣe mi pẹ̀lú Jèhófà àtàwọn èèyàn ?\n",
            "2020-04-12 14:42:08,887 \tHypothesis: Ǹjẹ́ mo mọyì àwọn nǹkan tara ju àjọṣe mi pẹ̀lú Jèhófà àti pẹ̀lú àwọn èèyàn lọ ?\n",
            "2020-04-12 14:42:08,887 Example #3\n",
            "2020-04-12 14:42:08,888 \tSource:     He has far more experience and stamina than you do , but he patiently walks near you .\n",
            "2020-04-12 14:42:08,888 \tReference:  Ẹni tẹ́ ẹ jọ ń lọ yìí mọ ọ̀nà yẹn dáadáa .\n",
            "2020-04-12 14:42:08,888 \tHypothesis: Ó ní ìrírí tó pọ̀ gan - an , ó sì tún ní ìrírí tó ju tìẹ lọ , àmọ́ ó fi sùúrù rìn nítòsí rẹ .\n",
            "2020-04-12 14:42:08,888 Validation result (greedy) at epoch   2, step   391000: bleu:  29.64, loss: 43197.9961, ppl:   4.2631, duration: 16.5808s\n",
            "2020-04-12 14:42:19,715 Epoch   2 Step:   391100 Batch Loss:     1.730690 Tokens per Sec:    20686, Lr: 0.000003\n",
            "2020-04-12 14:42:30,694 Epoch   2 Step:   391200 Batch Loss:     1.928249 Tokens per Sec:    21463, Lr: 0.000003\n",
            "2020-04-12 14:42:41,529 Epoch   2 Step:   391300 Batch Loss:     1.834248 Tokens per Sec:    21164, Lr: 0.000003\n",
            "2020-04-12 14:42:52,423 Epoch   2 Step:   391400 Batch Loss:     1.513258 Tokens per Sec:    21443, Lr: 0.000003\n",
            "2020-04-12 14:43:03,189 Epoch   2 Step:   391500 Batch Loss:     1.757843 Tokens per Sec:    21456, Lr: 0.000003\n",
            "2020-04-12 14:43:14,143 Epoch   2 Step:   391600 Batch Loss:     1.844177 Tokens per Sec:    21430, Lr: 0.000003\n",
            "2020-04-12 14:43:25,018 Epoch   2 Step:   391700 Batch Loss:     1.996000 Tokens per Sec:    21009, Lr: 0.000003\n",
            "2020-04-12 14:43:35,837 Epoch   2 Step:   391800 Batch Loss:     1.765452 Tokens per Sec:    21162, Lr: 0.000003\n",
            "2020-04-12 14:43:46,803 Epoch   2 Step:   391900 Batch Loss:     1.766898 Tokens per Sec:    21112, Lr: 0.000003\n",
            "2020-04-12 14:43:57,670 Epoch   2 Step:   392000 Batch Loss:     1.725068 Tokens per Sec:    21219, Lr: 0.000003\n",
            "2020-04-12 14:44:14,024 Example #0\n",
            "2020-04-12 14:44:14,025 \tSource:     He is the Source of life , the One giving it as an undeserved gift through Christ .\n",
            "2020-04-12 14:44:14,025 \tReference:  Òun ni Orísun ìyè , Ẹni tí ń fi ìyè fúnni gẹ́gẹ́ bí ẹbùn tí a kò lẹ́tọ̀ọ́ sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:44:14,025 \tHypothesis: Òun ni Orísun ìyè , Ẹni tí ń fi í fúnni gẹ́gẹ́ bí ẹ̀bùn àìlẹ́tọ̀ọ́sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:44:14,025 Example #1\n",
            "2020-04-12 14:44:14,026 \tSource:     Now I had to find a legitimate line of work .\n",
            "2020-04-12 14:44:14,026 \tReference:  Torí náà , mo ní láti wá iṣẹ́ gidi .\n",
            "2020-04-12 14:44:14,026 \tHypothesis: Ní báyìí , mo ní láti wá iṣẹ́ tó dára .\n",
            "2020-04-12 14:44:14,026 Example #2\n",
            "2020-04-12 14:44:14,026 \tSource:     Do I value material things more than my relationship with Jehovah and with people ?\n",
            "2020-04-12 14:44:14,026 \tReference:  Ṣé àwọn nǹkan tara ló jẹ mí lógún jù àbí àjọṣe mi pẹ̀lú Jèhófà àtàwọn èèyàn ?\n",
            "2020-04-12 14:44:14,027 \tHypothesis: Ǹjẹ́ mo mọyì àwọn nǹkan tara ju àjọṣe mi pẹ̀lú Jèhófà àti pẹ̀lú àwọn èèyàn lọ ?\n",
            "2020-04-12 14:44:14,027 Example #3\n",
            "2020-04-12 14:44:14,027 \tSource:     He has far more experience and stamina than you do , but he patiently walks near you .\n",
            "2020-04-12 14:44:14,027 \tReference:  Ẹni tẹ́ ẹ jọ ń lọ yìí mọ ọ̀nà yẹn dáadáa .\n",
            "2020-04-12 14:44:14,027 \tHypothesis: Ó ní ìrírí tó pọ̀ gan - an , ó sì tún ní ìrírí tó ju tìẹ lọ , àmọ́ ó fi sùúrù rìn nítòsí rẹ .\n",
            "2020-04-12 14:44:14,028 Validation result (greedy) at epoch   2, step   392000: bleu:  29.65, loss: 43272.7344, ppl:   4.2738, duration: 16.3569s\n",
            "2020-04-12 14:44:24,889 Epoch   2 Step:   392100 Batch Loss:     1.622569 Tokens per Sec:    21533, Lr: 0.000003\n",
            "2020-04-12 14:44:35,730 Epoch   2 Step:   392200 Batch Loss:     1.807818 Tokens per Sec:    21542, Lr: 0.000003\n",
            "2020-04-12 14:44:46,584 Epoch   2 Step:   392300 Batch Loss:     1.841716 Tokens per Sec:    21313, Lr: 0.000003\n",
            "2020-04-12 14:44:57,337 Epoch   2 Step:   392400 Batch Loss:     1.885677 Tokens per Sec:    20886, Lr: 0.000003\n",
            "2020-04-12 14:45:08,321 Epoch   2 Step:   392500 Batch Loss:     1.964448 Tokens per Sec:    21317, Lr: 0.000003\n",
            "2020-04-12 14:45:19,193 Epoch   2 Step:   392600 Batch Loss:     2.000385 Tokens per Sec:    21782, Lr: 0.000003\n",
            "2020-04-12 14:45:29,962 Epoch   2 Step:   392700 Batch Loss:     1.820094 Tokens per Sec:    21302, Lr: 0.000003\n",
            "2020-04-12 14:45:40,818 Epoch   2 Step:   392800 Batch Loss:     1.933245 Tokens per Sec:    21218, Lr: 0.000003\n",
            "2020-04-12 14:45:51,841 Epoch   2 Step:   392900 Batch Loss:     1.811371 Tokens per Sec:    21688, Lr: 0.000003\n",
            "2020-04-12 14:46:02,880 Epoch   2 Step:   393000 Batch Loss:     1.792009 Tokens per Sec:    20785, Lr: 0.000003\n",
            "2020-04-12 14:46:19,547 Example #0\n",
            "2020-04-12 14:46:19,548 \tSource:     He is the Source of life , the One giving it as an undeserved gift through Christ .\n",
            "2020-04-12 14:46:19,548 \tReference:  Òun ni Orísun ìyè , Ẹni tí ń fi ìyè fúnni gẹ́gẹ́ bí ẹbùn tí a kò lẹ́tọ̀ọ́ sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:46:19,548 \tHypothesis: Òun ni Orísun ìyè , Ẹni tí ń fi í fúnni gẹ́gẹ́ bí ẹ̀bùn àìlẹ́tọ̀ọ́sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:46:19,548 Example #1\n",
            "2020-04-12 14:46:19,548 \tSource:     Now I had to find a legitimate line of work .\n",
            "2020-04-12 14:46:19,549 \tReference:  Torí náà , mo ní láti wá iṣẹ́ gidi .\n",
            "2020-04-12 14:46:19,549 \tHypothesis: Ní báyìí , mo ní láti wá iṣẹ́ tó dára .\n",
            "2020-04-12 14:46:19,549 Example #2\n",
            "2020-04-12 14:46:19,549 \tSource:     Do I value material things more than my relationship with Jehovah and with people ?\n",
            "2020-04-12 14:46:19,550 \tReference:  Ṣé àwọn nǹkan tara ló jẹ mí lógún jù àbí àjọṣe mi pẹ̀lú Jèhófà àtàwọn èèyàn ?\n",
            "2020-04-12 14:46:19,550 \tHypothesis: Ǹjẹ́ mo mọyì àwọn nǹkan tara ju àjọṣe mi pẹ̀lú Jèhófà àti pẹ̀lú àwọn èèyàn lọ ?\n",
            "2020-04-12 14:46:19,550 Example #3\n",
            "2020-04-12 14:46:19,550 \tSource:     He has far more experience and stamina than you do , but he patiently walks near you .\n",
            "2020-04-12 14:46:19,550 \tReference:  Ẹni tẹ́ ẹ jọ ń lọ yìí mọ ọ̀nà yẹn dáadáa .\n",
            "2020-04-12 14:46:19,551 \tHypothesis: Ó ní ìrírí tó pọ̀ gan - an , ó sì tún ní ìrírí tó ju tìẹ lọ , àmọ́ ó fi sùúrù rìn nítòsí rẹ .\n",
            "2020-04-12 14:46:19,551 Validation result (greedy) at epoch   2, step   393000: bleu:  29.67, loss: 43318.9180, ppl:   4.2804, duration: 16.6700s\n",
            "2020-04-12 14:46:30,459 Epoch   2 Step:   393100 Batch Loss:     2.018415 Tokens per Sec:    21103, Lr: 0.000003\n",
            "2020-04-12 14:46:41,361 Epoch   2 Step:   393200 Batch Loss:     1.778643 Tokens per Sec:    21226, Lr: 0.000003\n",
            "2020-04-12 14:46:52,245 Epoch   2 Step:   393300 Batch Loss:     1.754664 Tokens per Sec:    21062, Lr: 0.000003\n",
            "2020-04-12 14:47:03,210 Epoch   2 Step:   393400 Batch Loss:     1.905242 Tokens per Sec:    20484, Lr: 0.000003\n",
            "2020-04-12 14:47:14,073 Epoch   2 Step:   393500 Batch Loss:     1.560444 Tokens per Sec:    21002, Lr: 0.000003\n",
            "2020-04-12 14:47:24,945 Epoch   2 Step:   393600 Batch Loss:     1.909894 Tokens per Sec:    21049, Lr: 0.000003\n",
            "2020-04-12 14:47:35,814 Epoch   2 Step:   393700 Batch Loss:     1.815693 Tokens per Sec:    21220, Lr: 0.000003\n",
            "2020-04-12 14:47:46,909 Epoch   2 Step:   393800 Batch Loss:     1.723883 Tokens per Sec:    20311, Lr: 0.000003\n",
            "2020-04-12 14:47:58,384 Epoch   2 Step:   393900 Batch Loss:     1.734728 Tokens per Sec:    21223, Lr: 0.000003\n",
            "2020-04-12 14:48:09,625 Epoch   2 Step:   394000 Batch Loss:     1.688648 Tokens per Sec:    20607, Lr: 0.000003\n",
            "2020-04-12 14:48:26,520 Example #0\n",
            "2020-04-12 14:48:26,521 \tSource:     He is the Source of life , the One giving it as an undeserved gift through Christ .\n",
            "2020-04-12 14:48:26,521 \tReference:  Òun ni Orísun ìyè , Ẹni tí ń fi ìyè fúnni gẹ́gẹ́ bí ẹbùn tí a kò lẹ́tọ̀ọ́ sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:48:26,521 \tHypothesis: Òun ni Orísun ìyè , Ẹni tí ń fi í fúnni gẹ́gẹ́ bí ẹ̀bùn àìlẹ́tọ̀ọ́sí nípasẹ̀ Kristi .\n",
            "2020-04-12 14:48:26,521 Example #1\n",
            "2020-04-12 14:48:26,522 \tSource:     Now I had to find a legitimate line of work .\n",
            "2020-04-12 14:48:26,522 \tReference:  Torí náà , mo ní láti wá iṣẹ́ gidi .\n",
            "2020-04-12 14:48:26,522 \tHypothesis: Ní báyìí , mo ní láti wá iṣẹ́ tó dára .\n",
            "2020-04-12 14:48:26,522 Example #2\n",
            "2020-04-12 14:48:26,522 \tSource:     Do I value material things more than my relationship with Jehovah and with people ?\n",
            "2020-04-12 14:48:26,523 \tReference:  Ṣé àwọn nǹkan tara ló jẹ mí lógún jù àbí àjọṣe mi pẹ̀lú Jèhófà àtàwọn èèyàn ?\n",
            "2020-04-12 14:48:26,523 \tHypothesis: Ǹjẹ́ mo mọyì àwọn nǹkan tara ju àjọṣe mi pẹ̀lú Jèhófà àti pẹ̀lú àwọn èèyàn lọ ?\n",
            "2020-04-12 14:48:26,523 Example #3\n",
            "2020-04-12 14:48:26,523 \tSource:     He has far more experience and stamina than you do , but he patiently walks near you .\n",
            "2020-04-12 14:48:26,523 \tReference:  Ẹni tẹ́ ẹ jọ ń lọ yìí mọ ọ̀nà yẹn dáadáa .\n",
            "2020-04-12 14:48:26,523 \tHypothesis: Ó ní ìrírí tó pọ̀ gan - an , ó sì tún ní ìrírí tó ju tìẹ lọ , àmọ́ ó fi sùúrù rìn nítòsí rẹ .\n",
            "2020-04-12 14:48:26,524 Validation result (greedy) at epoch   2, step   394000: bleu:  29.73, loss: 43349.5234, ppl:   4.2848, duration: 16.8980s\n",
            "2020-04-12 14:48:37,605 Epoch   2 Step:   394100 Batch Loss:     1.800462 Tokens per Sec:    20434, Lr: 0.000003\n",
            "2020-04-12 14:48:48,705 Epoch   2 Step:   394200 Batch Loss:     1.878787 Tokens per Sec:    20894, Lr: 0.000003\n",
            "2020-04-12 14:48:59,810 Epoch   2 Step:   394300 Batch Loss:     1.908489 Tokens per Sec:    20646, Lr: 0.000003\n",
            "2020-04-12 14:49:10,987 Epoch   2 Step:   394400 Batch Loss:     1.732751 Tokens per Sec:    20868, Lr: 0.000003\n",
            "2020-04-12 14:49:21,863 Epoch   2 Step:   394500 Batch Loss:     1.677493 Tokens per Sec:    20701, Lr: 0.000003\n",
            "2020-04-12 14:49:21,877 Epoch   2: total training loss 9712.73\n",
            "2020-04-12 14:49:21,878 Training ended after   2 epochs.\n",
            "2020-04-12 14:49:21,878 Best validation result (greedy) at step   384000:   3.72 ppl.\n",
            "2020-04-12 14:49:53,303  dev bleu:  31.03 [Beam search decoding with beam size = 5 and alpha = 1.0]\n",
            "2020-04-12 14:49:53,328 Translations saved to: /content/drive/My Drive/masakhane/model-temp/00384000.hyps.dev\n",
            "2020-04-12 14:50:31,869 test bleu:  38.62 [Beam search decoding with beam size = 5 and alpha = 1.0]\n",
            "2020-04-12 14:50:31,876 Translations saved to: /content/drive/My Drive/masakhane/model-temp/00384000.hyps.test\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "MBoDS09JM807",
        "colab": {}
      },
      "source": [
        "# Copy the created models from the temporary storage to main storage on google drive for persistant storage \n",
        "!cp -r \"/content/drive/My Drive/masakhane/model-temp/\"* \"$gdrive_path/models/${src}${tgt}_transformer/\""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "n94wlrCjVc17",
        "outputId": "90442631-4be7-4089-ac2d-18119a640d17",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 197
        }
      },
      "source": [
        "# Output our validation accuracy\n",
        "! cat \"$gdrive_path/models/${src}${tgt}_transformer/validations.txt\""
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Steps: 385000\tLoss: 41967.25391\tPPL: 4.09053\tbleu: 30.03572\tLR: 0.00000415\t\n",
            "Steps: 386000\tLoss: 42548.08594\tPPL: 4.17107\tbleu: 29.91872\tLR: 0.00000415\t\n",
            "Steps: 387000\tLoss: 42807.03906\tPPL: 4.20748\tbleu: 29.87333\tLR: 0.00000415\t\n",
            "Steps: 388000\tLoss: 42958.75781\tPPL: 4.22896\tbleu: 29.81204\tLR: 0.00000415\t\n",
            "Steps: 389000\tLoss: 43055.49609\tPPL: 4.24271\tbleu: 29.80630\tLR: 0.00000415\t\n",
            "Steps: 390000\tLoss: 43157.83984\tPPL: 4.25731\tbleu: 29.72661\tLR: 0.00000291\t\n",
            "Steps: 391000\tLoss: 43197.99609\tPPL: 4.26306\tbleu: 29.64012\tLR: 0.00000291\t\n",
            "Steps: 392000\tLoss: 43272.73438\tPPL: 4.27376\tbleu: 29.64561\tLR: 0.00000291\t\n",
            "Steps: 393000\tLoss: 43318.91797\tPPL: 4.28040\tbleu: 29.66845\tLR: 0.00000291\t\n",
            "Steps: 394000\tLoss: 43349.52344\tPPL: 4.28479\tbleu: 29.72774\tLR: 0.00000291\t\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "66WhRE9lIhoD",
        "outputId": "16aacfc5-552a-4b2c-c264-198ac5b1d7a7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 71
        }
      },
      "source": [
        "# Test our model\n",
        "! cd joeynmt; python3 -m joeynmt test \"$gdrive_path/models/${src}${tgt}_transformer/config.yaml\""
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2020-04-12 14:54:41,810 Hello! This is Joey-NMT.\n",
            "2020-04-12 14:55:15,182  dev bleu:  31.03 [Beam search decoding with beam size = 5 and alpha = 1.0]\n",
            "2020-04-12 14:55:53,503 test bleu:  38.62 [Beam search decoding with beam size = 5 and alpha = 1.0]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KaXDFfm-zgjK",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}