{ "nbformat": 4, "nbformat_minor": 0, "metadata": { "colab": { "name": "Hausa_English_NMT.ipynb", "provenance": [], "collapsed_sections": [], "include_colab_link": true }, "kernelspec": { "name": "python3", "display_name": "Python 3" }, "accelerator": "GPU" }, "cells": [ { "cell_type": "markdown", "metadata": { "id": "view-in-github", "colab_type": "text" }, "source": [ "\"Open" ] }, { "cell_type": "code", "metadata": { "id": "rfqggSL3MOMG", "colab_type": "code", "outputId": "40facee6-726b-4699-c91a-446ee07b24d4", "colab": { "base_uri": "https://localhost:8080/", "height": 122 } }, "source": [ "\n", "from google.colab import drive\n", "drive.mount('/content/drive')" ], "execution_count": 1, "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": { "id": "KhgF50CRMU7l", "colab_type": "code", "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 = \"ha\"\n", "target_language = \"en\" \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-$tag\"\n", "# os.environ[\"gdrive_path\"] = \"/content/drive/My Drive/masakhane/%s-%s-%s\" % (source_language, target_language, tag)\n", "\n", "# This will save it to a folder in our gdrive instead! \n", "!mkdir -p \"/content/drive/My Drive/masakhane/$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": { "id": "EtjFUdP4MvnQ", "colab_type": "code", "outputId": "18cfb6cd-4c76-4659-8cf9-269df9e7d963", "colab": { "base_uri": "https://localhost:8080/", "height": 34 } }, "source": [ "!echo $gdrive_path" ], "execution_count": 3, "outputs": [ { "output_type": "stream", "text": [ "/content/drive/My Drive/masakhane/ha-en-baseline\n" ], "name": "stdout" } ] }, { "cell_type": "code", "metadata": { "id": "KMGnzvjIMzBe", "colab_type": "code", "outputId": "7af62cb6-935f-404e-d728-7942363e8346", "colab": { "base_uri": "https://localhost:8080/", "height": 102 } }, "source": [ "# Install opus-tools\n", "! pip install opustools-pkg" ], "execution_count": 4, "outputs": [ { "output_type": "stream", "text": [ "Collecting opustools-pkg\n", "\u001b[?25l Downloading https://files.pythonhosted.org/packages/6c/9f/e829a0cceccc603450cd18e1ff80807b6237a88d9a8df2c0bb320796e900/opustools_pkg-0.0.52-py3-none-any.whl (80kB)\n", "\r\u001b[K |████ | 10kB 25.0MB/s eta 0:00:01\r\u001b[K |████████ | 20kB 32.9MB/s eta 0:00:01\r\u001b[K |████████████▏ | 30kB 34.9MB/s eta 0:00:01\r\u001b[K |████████████████▏ | 40kB 38.7MB/s eta 0:00:01\r\u001b[K |████████████████████▎ | 51kB 40.1MB/s eta 0:00:01\r\u001b[K |████████████████████████▎ | 61kB 42.8MB/s eta 0:00:01\r\u001b[K |████████████████████████████▎ | 71kB 44.0MB/s eta 0:00:01\r\u001b[K |████████████████████████████████| 81kB 10.9MB/s \n", "\u001b[?25hInstalling collected packages: opustools-pkg\n", "Successfully installed opustools-pkg-0.0.52\n" ], "name": "stdout" } ] }, { "cell_type": "code", "metadata": { "id": "1oGBeknJM4SW", "colab_type": "code", "outputId": "cae7ae4b-dda3-47d1-e298-25a98c4ddea3", "colab": { "base_uri": "https://localhost:8080/", "height": 204 } }, "source": [ "\n", "# Downloading our corpus\n", "! opus_read -d JW300 -s $src -t $tgt -wm moses -w jw300.$src jw300.$tgt -q\n", "\n", "# extract the corpus file\n", "! gunzip JW300_latest_xml_$tgt-$src.xml.gz" ], "execution_count": 5, "outputs": [ { "output_type": "stream", "text": [ "\n", "Alignment file /proj/nlpl/data/OPUS/JW300/latest/xml/en-ha.xml.gz not found. The following files are available for downloading:\n", "\n", " 2 MB https://object.pouta.csc.fi/OPUS-JW300/v1/xml/en-ha.xml.gz\n", " 263 MB https://object.pouta.csc.fi/OPUS-JW300/v1/xml/en.zip\n", " 24 MB https://object.pouta.csc.fi/OPUS-JW300/v1/xml/ha.zip\n", "\n", " 289 MB Total size\n", "./JW300_latest_xml_en-ha.xml.gz ... 100% of 2 MB\n", "./JW300_latest_xml_en.zip ... 100% of 263 MB\n", "./JW300_latest_xml_ha.zip ... 100% of 24 MB\n" ], "name": "stdout" } ] }, { "cell_type": "code", "metadata": { "id": "jkL4ajrtNyvu", "colab_type": "code", "colab": { "base_uri": "https://localhost:8080/", "height": 176 }, "outputId": "d6f98c26-fcd7-49bf-cc17-d91a5ff4ffda" }, "source": [ "\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_file, target_file), 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", "df1= pd.DataFrame(list(zip(source, target)), columns=['source_sentence', 'target_sentence'])\n", "print(df1.shape)\n", "df1.head(3)\n" ], "execution_count": 6, "outputs": [ { "output_type": "stream", "text": [ "Loaded data and skipped 0/237064 lines since contained in test set.\n", "(237065, 2)\n" ], "name": "stdout" }, { "output_type": "execute_result", "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
source_sentencetarget_sentence
0Tambayoyi Daga Masu KaratuQuestions From Readers
1Yaya ya kamata Kiristoci su ɗauki riƙo ?How seriously should Christians view an engage...
2Riƙo dalili ne na farinciki , amma kuma muhimm...An engagement to marry is a cause for happines...
\n", "
" ], "text/plain": [ " source_sentence target_sentence\n", "0 Tambayoyi Daga Masu Karatu Questions From Readers\n", "1 Yaya ya kamata Kiristoci su ɗauki riƙo ? How seriously should Christians view an engage...\n", "2 Riƙo dalili ne na farinciki , amma kuma muhimm... An engagement to marry is a cause for happines..." ] }, "metadata": { "tags": [] }, "execution_count": 6 } ] }, { "cell_type": "code", "metadata": { "id": "4GfKzCP0u1d_", "colab_type": "code", "colab": { "base_uri": "https://localhost:8080/", "height": 204 }, "outputId": "7618557c-bda4-4eb4-933f-74586dc3b58d" }, "source": [ "\n", "# Downloading our corpus\n", "! opus_read -d Tatoeba -s $src -t $tgt -wm moses -w tatoeba.$src tatoeba.$tgt -q\n", "\n", "# extract the corpus file\n", "! gunzip Tatoeba_latest_xml_$tgt-$src.xml.gz" ], "execution_count": 7, "outputs": [ { "output_type": "stream", "text": [ "\n", "Alignment file /proj/nlpl/data/OPUS/Tatoeba/latest/xml/en-ha.xml.gz not found. The following files are available for downloading:\n", "\n", " 4 KB https://object.pouta.csc.fi/OPUS-Tatoeba/v20190709/xml/en-ha.xml.gz\n", " 90 MB https://object.pouta.csc.fi/OPUS-Tatoeba/v20190709/xml/en.zip\n", " 12 KB https://object.pouta.csc.fi/OPUS-Tatoeba/v20190709/xml/ha.zip\n", "\n", " 90 MB Total size\n", "./Tatoeba_latest_xml_en-ha.xml.gz ... 100% of 4 KB\n", "./Tatoeba_latest_xml_en.zip ... 100% of 90 MB\n", "./Tatoeba_latest_xml_ha.zip ... 100% of 12 KB\n" ], "name": "stdout" } ] }, { "cell_type": "code", "metadata": { "id": "A29X11AOvGKB", "colab_type": "code", "colab": { "base_uri": "https://localhost:8080/", "height": 176 }, "outputId": "fa886ac3-c985-4fa0-acce-6baa02fb02c3" }, "source": [ "import pandas as pd\n", "\n", "# TMX file to dataframe\n", "source_file = 'tatoeba.' + source_language\n", "target_file = 'tatoeba.' + 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_file, target_file), 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", "df2= pd.DataFrame(list(zip(source, target)), columns=['source_sentence', 'target_sentence'])\n", "print(df2.shape)\n", "df2.head(3)" ], "execution_count": 8, "outputs": [ { "output_type": "stream", "text": [ "Loaded data and skipped 0/55 lines since contained in test set.\n", "(56, 2)\n" ], "name": "stdout" }, { "output_type": "execute_result", "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
source_sentencetarget_sentence
0Kwarai kuwa zaka iya daukarsa idan kanaso .Of course you can take it if you want .
1Ka taɓa zuwa Paris ?Have you ever been to Paris ?
2Ina ne wajen bahaya ?Where is the toilet ?
\n", "
" ], "text/plain": [ " source_sentence target_sentence\n", "0 Kwarai kuwa zaka iya daukarsa idan kanaso . Of course you can take it if you want .\n", "1 Ka taɓa zuwa Paris ? Have you ever been to Paris ?\n", "2 Ina ne wajen bahaya ? Where is the toilet ?" ] }, "metadata": { "tags": [] }, "execution_count": 8 } ] }, { "cell_type": "code", "metadata": { "id": "bbyhNtcwzuZ8", "colab_type": "code", "colab": { "base_uri": "https://localhost:8080/", "height": 204 }, "outputId": "a7781b65-f341-4854-b346-392a73417d48" }, "source": [ "# Downloading our corpus\n", "! opus_read -d wikimedia -s $src -t $tgt -wm moses -w wikimedia.$src wikimedia.$tgt -q\n", "\n", "# extract the corpus file\n", "! gunzip wikimedia_latest_xml_$tgt-$src.xml.gz" ], "execution_count": 9, "outputs": [ { "output_type": "stream", "text": [ "\n", "Alignment file /proj/nlpl/data/OPUS/wikimedia/latest/xml/en-ha.xml.gz not found. The following files are available for downloading:\n", "\n", " 20 KB https://object.pouta.csc.fi/OPUS-wikimedia/v20190628/xml/en-ha.xml.gz\n", " 60 MB https://object.pouta.csc.fi/OPUS-wikimedia/v20190628/xml/en.zip\n", " 252 KB https://object.pouta.csc.fi/OPUS-wikimedia/v20190628/xml/ha.zip\n", "\n", " 60 MB Total size\n", "./wikimedia_latest_xml_en-ha.xml.gz ... 100% of 20 KB\n", "./wikimedia_latest_xml_en.zip ... 100% of 60 MB\n", "./wikimedia_latest_xml_ha.zip ... 100% of 252 KB\n" ], "name": "stdout" } ] }, { "cell_type": "code", "metadata": { "id": "u_bi-Pc7zulk", "colab_type": "code", "colab": { "base_uri": "https://localhost:8080/", "height": 176 }, "outputId": "65959478-c875-4ce3-b827-2baf30a21c48" }, "source": [ "\n", "# TMX file to dataframe\n", "source_file = 'wikimedia.' + source_language\n", "target_file = 'wikimedia.' + 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_file, target_file), 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", "df3= pd.DataFrame(list(zip(source, target)), columns=['source_sentence', 'target_sentence'])\n", "print(df3.shape)\n", "df3.head(3)" ], "execution_count": 10, "outputs": [ { "output_type": "stream", "text": [ "Loaded data and skipped 0/702 lines since contained in test set.\n", "(703, 2)\n" ], "name": "stdout" }, { "output_type": "execute_result", "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
source_sentencetarget_sentence
0ReferencesReferences
1↑ \" Will Smith Net Worth - atlantablackstar \" ...↑ \" Will Smith Net Worth - atlantablackstar \" ...
2Willard Carroll \" Will \" Smith Jr . [ 2 ] [ 3 ...Willard Carroll \" Will \" Smith Jr.[2][3][4 ] (...
\n", "
" ], "text/plain": [ " source_sentence target_sentence\n", "0 References References\n", "1 ↑ \" Will Smith Net Worth - atlantablackstar \" ... ↑ \" Will Smith Net Worth - atlantablackstar \" ...\n", "2 Willard Carroll \" Will \" Smith Jr . [ 2 ] [ 3 ... Willard Carroll \" Will \" Smith Jr.[2][3][4 ] (..." ] }, "metadata": { "tags": [] }, "execution_count": 10 } ] }, { "cell_type": "code", "metadata": { "id": "PRksk-iV0v-5", "colab_type": "code", "colab": { "base_uri": "https://localhost:8080/", "height": 204 }, "outputId": "3be17a57-38dd-4163-90ca-f1c6f88c01b5" }, "source": [ "# Downloading our corpus\n", "! opus_read -d Tanzil -s $src -t $tgt -wm moses -w tanzil.$src tanzil.$tgt -q\n", "\n", "# extract the corpus file\n", "! gunzip Tanzil_latest_xml_$tgt-$src.xml.gz" ], "execution_count": 11, "outputs": [ { "output_type": "stream", "text": [ "\n", "Alignment file /proj/nlpl/data/OPUS/Tanzil/latest/xml/en-ha.xml.gz not found. The following files are available for downloading:\n", "\n", " 1 MB https://object.pouta.csc.fi/OPUS-Tanzil/v1/xml/en-ha.xml.gz\n", " 34 MB https://object.pouta.csc.fi/OPUS-Tanzil/v1/xml/en.zip\n", " 1 MB https://object.pouta.csc.fi/OPUS-Tanzil/v1/xml/ha.zip\n", "\n", " 36 MB Total size\n", "./Tanzil_latest_xml_en-ha.xml.gz ... 100% of 1 MB\n", "./Tanzil_latest_xml_en.zip ... 100% of 34 MB\n", "./Tanzil_latest_xml_ha.zip ... 100% of 1 MB\n" ], "name": "stdout" } ] }, { "cell_type": "code", "metadata": { "id": "5q111T-n0wSr", "colab_type": "code", "colab": { "base_uri": "https://localhost:8080/", "height": 176 }, "outputId": "2d988d2e-e8ed-4723-a10c-b95d3d9dfbea" }, "source": [ "\n", "# TMX file to dataframe\n", "source_file = 'tanzil.' + source_language\n", "target_file = 'tanzil.' + 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_file, target_file), 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", "df4= pd.DataFrame(list(zip(source, target)), columns=['source_sentence', 'target_sentence'])\n", "print(df4.shape)\n", "df4.head(3)" ], "execution_count": 12, "outputs": [ { "output_type": "stream", "text": [ "Loaded data and skipped 0/132100 lines since contained in test set.\n", "(132101, 2)\n" ], "name": "stdout" }, { "output_type": "execute_result", "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
source_sentencetarget_sentence
0Da sũnan Allah , Mai rahama , Mai jin ƙai .In the name of Allah , most benevolent , ever-...
1Godiya ta tabbata ga Allah , Ubangijin halittu ;ALL PRAISE BE to Allah , Lord of all the worlds ,
2Mai rahama , Mai jin ƙai ;Most beneficent , ever-merciful ,
\n", "
" ], "text/plain": [ " source_sentence target_sentence\n", "0 Da sũnan Allah , Mai rahama , Mai jin ƙai . In the name of Allah , most benevolent , ever-...\n", "1 Godiya ta tabbata ga Allah , Ubangijin halittu ; ALL PRAISE BE to Allah , Lord of all the worlds ,\n", "2 Mai rahama , Mai jin ƙai ; Most beneficent , ever-merciful ," ] }, "metadata": { "tags": [] }, "execution_count": 12 } ] }, { "cell_type": "code", "metadata": { "id": "YkyK4ACmLfez", "colab_type": "code", "colab": {} }, "source": [ "List = [df1, df2, df3, df4] # List of your dataframes\n", "df = pd.concat(List)" ], "execution_count": 0, "outputs": [] }, { "cell_type": "code", "metadata": { "id": "Dw9SygdnN6S1", "colab_type": "code", "colab": { "base_uri": "https://localhost:8080/", "height": 34 }, "outputId": "73dd58d8-b5ef-43a1-bd49-d500d7aec783" }, "source": [ "df.shape" ], "execution_count": 14, "outputs": [ { "output_type": "execute_result", "data": { "text/plain": [ "(369925, 2)" ] }, "metadata": { "tags": [] }, "execution_count": 14 } ] }, { "cell_type": "code", "metadata": { "id": "agcS7za6YOcP", "colab_type": "code", "colab": {} }, "source": [ "\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": [] }, { "cell_type": "code", "metadata": { "id": "zHWZkn07ZXUd", "colab_type": "code", "colab": { "base_uri": "https://localhost:8080/", "height": 34 }, "outputId": "401ec8d1-070f-482b-afbb-f0d34c4a7338" }, "source": [ "df_pp.shape" ], "execution_count": 16, "outputs": [ { "output_type": "execute_result", "data": { "text/plain": [ "(351024, 2)" ] }, "metadata": { "tags": [] }, "execution_count": 16 } ] }, { "cell_type": "code", "metadata": { "id": "DC7zwjgPZa65", "colab_type": "code", "colab": {} }, "source": [ "# # 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": [] }, { "cell_type": "code", "metadata": { "id": "GSKrJmFAZkXh", "colab_type": "code", "colab": { "base_uri": "https://localhost:8080/", "height": 1000 }, "outputId": "28450261-2010-47d2-b64b-dc987a705b96" }, "source": [ "# 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", "import numpy as np\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", "#Divide df_pp into train and test sets\n", "\n", "msk = np.random.rand(len(df_pp)) < 0.98\n", "train = df_pp[msk]\n", "test_df = df_pp[~msk]\n", "\n", "# Further divide train into train and additional dev set which should be the last 1000\n", "dev_df = train.tail(num_dev_patterns) # Herman: Error in original\n", "stripped_df = train.drop(train.tail(num_dev_patterns).index)\n", "\n", "with open(\"test.\"+source_language, \"w\") as src_file, open(\"test.\"+target_language, \"w\") as trg_file:\n", " for index, row in test_df.iterrows():\n", " src_file.write(row[\"source_sentence\"]+\"\\n\")\n", " trg_file.write(row[\"target_sentence\"]+\"\\n\")\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_df.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_df.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.*\n", "! head test.*" ], "execution_count": 18, "outputs": [ { "output_type": "stream", "text": [ "==> train.en <==\n", "And We have placed in the earth firm mountains lest it should move away with them , and We placed therein Passages for paths , that haply they may be guided .\n", "“ YOU saw a lot of casual clothing , especially when it was hot , ” reported a Dutch newspaper regarding a meeting of church leaders .\n", "How will God relieve mankind of suffering ?\n", "Jesus felt compassion for the people he preached to , and we should likewise be moved by compassion to preach to others . ​ — Matthew 9 : 36 .\n", "Every soul shall have a taste of death : And only on the Day of Judgment shall you be paid your full recompense .\n", "And We had already tried before them the people of Pharaoh , and there came to them a noble messenger ,\n", "He was an instructor at the Bible School of Gilead , and later he served as a member of the Governing Body .\n", "Learning that the woman was Bath - sheba and that her husband , Uriah , was off to war , David sent for her and had relations with her .\n", "God had in mind for their children to have children , until the human family filled the whole earth .\n", "Reading the Bible daily and meditating on what we read can help to guard our minds against immoral thoughts .\n", "\n", "==> train.ha <==\n", "Kuma Mun sanya tabbatattun duwãtsu a cikin ƙasa dõmin kada ta karkata da su kuma Mun sanya ranguna , hanyõyi , a cikinsu ( duwãtsun ) , tsammãninsu sunã shiryuwa .\n", "WATA jarida da aka rubuta da yaren Dutch ta kwatanta irin shigar da shugabanan addinai suka yi a wani taronsu .\n", "Ta yaya Allah zai cire wahalar da ’ yan adam suke sha ?\n", "Yesu ya ji tausayin mutanen da yake yi wa wa’azi , saboda haka ya kamata mu ji tausayin waɗanda muke yi wa wa’azi . — Matta 9 : 36 .\n", "Kõwane rai mai ɗanɗanar mutuwa ne . Kuma ana cika muku ijãrõrinku kawai ne a Rãnar ¡ iyãma .\n", "Kuma lalle ne haƙĩƙa a gabaninsu , Mun fitini mutãnen Fir 'auna , kuma wani Manzo karĩmi ya jẽ musu .\n", "Shi malami ne a Makarantar Gilead kuma daga baya ya zama memban Hukumar da Ke Kula da Ayyukanmu .\n", "Da ya ji cewa Bath - sheba ce , kuma mijinta Uriah yana dagar yaƙi , Dauda ya aika a kira ta kuma ya kwana da ita .\n", "Ƙari ga haka , Allah yana so yaransu su haifi yara , har sai mutane sun cika duniya .\n", "( Yaƙ . 3 : 17 , Littafi Mai Tsarki ) Karanta Littafi Mai Tsarki da kuma yin bimbini a kan abin da muka karanta zai sa mu guji yin tunanin banza .\n", "==> dev.en <==\n", "Note how We explain the revelations , so that they may understand . ”\n", "If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "Whether thou ask for their forgiveness , or not , ( their sin is unforgivable ) : if thou ask seventy times for their forgiveness , Allah will not forgive them : because they have rejected Allah and His Messenger : and Allah guideth not those who are perversely rebellious .\n", "“ If possible , ” the Scriptures counsel us , “ as far as it depends upon you , be peaceable with all men . ”\n", "Farooq Kperogi Born March 30 1973 Baruten , Kwara State , Nigeria Nationality Nigerian Occupation Professor Academic background Alma mater Georgia State University ( Ph.D ) University of Louisiana at Lafayette ( M.Sc ) Bayero University ( B.A ) Thesis Webs of Resistance : The Citizen Online Journalism of the Nigerian Digital Diaspora  ( 2011 ) Doctoral advisor Michael L. Bruner Academic work Discipline English language , Communication , Social and Behavioral Sciences[1 ] Sub-discipline English usage , Media English , Nigerian English , World Englishes[1 ] Institutions Kennesaw State University Main interests New Media , English usage , Journalese , Political criticism[1 ] Notable works Glocal English : The Changing Face and Forms of Nigerian English in a Global World\n", "Our next article further considers the purpose of the resurrection .\n", "The fact that the Bible is available in all major languages on earth is a testimony to what ?\n", "But why were there not among the generations before you those possessing understanding , who should have forbidden the making of mischief in the earth , except a few of those whom We delivered from among them ?\n", "\n", "==> dev.ha <==\n", "\" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "Kõ kã nẽma musu gãfara ko ba ka nẽma musu ba , idan ka nẽma musu gãfara sau saba 'in , to , Allah bã zai gãfarta musu ba . Sabõda sũ , sun kãfirta da Allah da ManzonSa .\n", "Nassosi ya ba mu wannan shawarar : “ Idan ya yiwu , ku zama lafiya da dukan mutane , gwargwadon iyawarku . ” ( Rom .\n", "\n", "( Yahaya 5 : 28 , 29 ) Talifinmu na gaba zai ƙara ba da bayanin nufin tashin matattu .\n", "Wane tabbaci ne ake da shi cewa akwai Littafi Mai Tsarki cikin dukan manyan harsunan duniya ?\n", "To , don me mãsu hankali ba su kasance daga mutãnen ƙarnõnin da suke a gabãninku ba , sunã hani daga ɓarna a cikin ƙasa ? fãce kaɗan daga wanda Muka kuɓutar daga gare su ( sun yi hanin ) .\n", "==> test.en <==\n", "God did not do an injustice to them but they wronged themselves .\n", "Who whispers into the hearts of men ,\n", "Moreover , little children generally require the care of their mother .\n", "With that the master had him delivered to the jailers .\n", "Then after that your hearts hardened .\n", "We may have to endure some difficulties by relying solely on Jehovah .\n", "The prosecutor answered that she was not competent to comment on the Bible .\n", "Paul’s humble and friendly manner built up the congregations .\n", "Was she curious about what was happening ?\n", "Even if some elderly ones are unable to conduct a Bible study themselves , their upbuilding comments at congregation meetings strengthen the faith of all who hear them .\n", "\n", "==> test.ha <==\n", "Sabo da haka Allah ba Ya yiwuwa Ya zãlunce su , amma kansu suka kasance sunã zãlunta .\n", "\" Wanda ke sanya wasuwãsi a cikin ƙirãzan mutane . \"\n", "Bugu da ƙari , yara ƙanana suna bukatar kula daga uwarsu .\n", "Da haka ubangijin ya miƙa shi ga ma’aikatan gidan kurkuku .\n", "San 'nan kuma zukãtanku , suka ƙẽƙashe daga bãyan wancan .\n", "Saboda haka sai mu dogara ga Jehobah sa’ad da muke fuskantar wahaloli .\n", "Lauyan gwamnatin ta ce yadda take ji ba shi da muhimmanci tun da ba ta da yawan ilimi game da Littafi Mai Tsarki .\n", "Halin tawali’u da abokantaka na Bulus ya ƙarfafa ikilisiyoyi .\n", "Shin ta kalli baya ne don ta san abin da ke faruwa ?\n", "Ko da wasu cikin tsofaffin ba sa iya gudanar da nazarin Littafi Mai Tsarki da kansu , kalaminsu a taron ikilisiya yana ƙarfafa bangaskiyar dukan waɗanda suka saurare su .\n" ], "name": "stdout" } ] }, { "cell_type": "code", "metadata": { "id": "YP8MqOSrolWr", "colab_type": "code", "colab": {} }, "source": [ "stripped_df.to_csv('train.csv')\n", "!cp train.csv drive/My\\ Drive/masakhane/" ], "execution_count": 0, "outputs": [] }, { "cell_type": "code", "metadata": { "id": "-G_lTaVupEyG", "colab_type": "code", "colab": {} }, "source": [ "test_df.to_csv('test.csv')\n", "!cp test.csv drive/My\\ Drive/masakhane/" ], "execution_count": 0, "outputs": [] }, { "cell_type": "code", "metadata": { "id": "6QDAjsttpFIO", "colab_type": "code", "colab": {} }, "source": [ "dev_df.to_csv('dev.csv')\n", "!cp dev.csv drive/My\\ Drive/masakhane/" ], "execution_count": 0, "outputs": [] }, { "cell_type": "code", "metadata": { "id": "yG8VHg1_bTob", "colab_type": "code", "colab": { "base_uri": "https://localhost:8080/", "height": 1000 }, "outputId": "25d7d5dc-139b-40e8-866b-9f91077a028d" }, "source": [ "\n", "# Install JoeyNMT\n", "! git clone https://github.com/joeynmt/joeynmt.git\n", "! cd joeynmt; pip3 install ." ], "execution_count": 22, "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 | 10.95 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 10.1MB/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 46.3MB/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 52.7MB/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: 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: 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: 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: 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: grpcio>=1.8.6 in /usr/local/lib/python3.6/dist-packages (from tensorflow>=1.14->joeynmt==0.0.1) (1.27.2)\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: 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: 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: 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: 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: 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: 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: 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: 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: 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 211kB/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: 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) (2.4.6)\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: 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: pandas>=0.22.0 in /usr/local/lib/python3.6/dist-packages (from seaborn->joeynmt==0.0.1) (1.0.3)\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 63.3MB/s \n", "\u001b[?25hCollecting 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 9.2MB/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", "Requirement 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: 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.post2)\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: 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: 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: 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: 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: certifi>=2017.4.17 in /usr/local/lib/python3.6/dist-packages (from requests->torchtext->joeynmt==0.0.1) (2019.11.28)\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 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[?25hCollecting 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 10.4MB/s \n", "\u001b[?25hRequirement 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: 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: 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: 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>=0.1.3 in /usr/local/lib/python3.6/dist-packages (from rsa<4.1,>=3.1.4->google-auth<2,>=1.6.3->tensorboard<2.3.0,>=2.2.0->tensorflow>=1.14->joeynmt==0.0.1) (0.4.8)\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", "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=1867c4b093151a3b9a560e38117596da4a50dbd5242b617604a0677a3223006d\n", " Stored in directory: /tmp/pip-ephem-wheel-cache-hpeb8hxq/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=f9bedeaac806a6384a66f3dc63506581e0ce8781cf3d3b51d39ec935126d263b\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=67432 sha256=c994d70b003751d366dd49253087fe3fb924ef34dbec576e738ba57dcb0f08e8\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, typed-ast, wrapt, lazy-object-proxy, astroid, isort, mccabe, 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": "code", "metadata": { "id": "r8qyPXA2ZseA", "colab_type": "code", "colab": { "base_uri": "https://localhost:8080/", "height": 428 }, "outputId": "cd3af5ed-9c10-48b1-ba61-6c3c6c96c55a" }, "source": [ "\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 \"$gdrive_path/vocab.txt\"\n", "\n", "# Some output\n", "! echo \"BPE Hausa Sentences\"\n", "! tail -n 5 test.bpe.$tgt\n", "! echo \"Combined BPE Vocab\"\n", "! tail -n 10 joeynmt/data/$src$tgt/vocab.txt " ], "execution_count": 34, "outputs": [ { "output_type": "stream", "text": [ "bpe.codes.4000\tdev.csv test.bpe.en test.en\t train.bpe.ha train.ha\n", "dev.bpe.en\tdev.en\t test.bpe.ha test.ha\t train.csv\t vocab.txt\n", "dev.bpe.ha\tdev.ha\t test.csv train.bpe.en train.en\n", "bpe.codes.4000\tdev.csv models test.csv\ttrain.bpe.en train.en\n", "dev.bpe.en\tdev.en\t test.bpe.en test.en\ttrain.bpe.ha train.ha\n", "dev.bpe.ha\tdev.ha\t test.bpe.ha test.ha\ttrain.csv\n", "BPE Hausa Sentences\n", "So let man consider of what he was created ;\n", "They needed their bre@@ ad des@@ per@@ ately , just like we , because they were hun@@ g@@ ry .\n", "To be fa@@ ir , these ob@@ ser@@ v@@ ations would also be true of other coun@@ tri@@ es where so - called Christi@@ an@@ ity pre@@ va@@ ils .\n", "F@@ ur@@ ther@@ more , we thus show that we long to see the div@@ ine will done thr@@ ough@@ out the uni@@ verse .\n", "Ju@@ st as the woman in the illustr@@ ation us@@ es help@@ ful to@@ ol@@ s to ac@@ com@@ pl@@ ish her tas@@ k , so today we have to@@ ol@@ s that help us to ac@@ com@@ pl@@ ish our God - given responsib@@ ility to as@@ sis@@ t those who are wea@@ k .\n", "Combined BPE Vocab\n", "?@@\n", "ு@@\n", "ற@@\n", "ி@@\n", "்@@\n", "ோ@@\n", "ܨ@@\n", "̄\n", "ܘ@@\n", "“@@\n" ], "name": "stdout" } ] }, { "cell_type": "code", "metadata": { "id": "CWgRz1vNb_1c", "colab_type": "code", "colab": {} }, "source": [ "# # 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 joeynmt/data/$src$tgt/vocab.txt\n", "\n", "# # Some output\n", "# ! echo \"BPE Hausa Sentences\"\n", "# ! tail -n 5 test.bpe.$tgt\n", "# ! echo \"Combined BPE Vocab\"\n", "# ! tail -n 10 joeynmt/data/$src$tgt/vocab.txt # Herman" ], "execution_count": 0, "outputs": [] }, { "cell_type": "code", "metadata": { "id": "_TMh33L11_zJ", "colab_type": "code", "colab": {} }, "source": [ "def get_last_checkpoint(directory):\n", " last_checkpoint = ''\n", " try:\n", " for filename in os.listdir(directory):\n", " if 'best' in filename and filename.endswith(\".ckpt\"):\n", " return filename\n", " if not 'best' in filename and filename.endswith(\".ckpt\"):\n", " if not last_checkpoint or int(filename.split('.')[0]) > int(last_checkpoint.split('.')[0]):\n", " last_checkpoint = filename\n", " except FileNotFoundError as e:\n", " print('Error Occur ', e)\n", " return last_checkpoint" ], "execution_count": 0, "outputs": [] }, { "cell_type": "code", "metadata": { "id": "5GBr2T7O2Cyl", "colab_type": "code", "colab": { "base_uri": "https://localhost:8080/", "height": 34 }, "outputId": "dcfd9e93-256c-48bb-97a0-11c1b35fc408" }, "source": [ "# Copy the created models from the temporary storage to main storage on google drive for persistant storage \n", "# the content of te folder will be overwrite when you start trainin\n", "!cp -r \"/content/drive/My Drive/masakhane/model-temp/\"* \"$gdrive_path/models/${src}${tgt}_transformer/\"\n", "last_checkpoint = get_last_checkpoint(models_path)\n", "print('Last checkpoint :',last_checkpoint)" ], "execution_count": 63, "outputs": [ { "output_type": "stream", "text": [ "Last checkpoint : best.ckpt\n" ], "name": "stdout" } ] }, { "cell_type": "code", "metadata": { "id": "DS5Sf0XKV-e0", "colab_type": "code", "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: 10#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: 1 # 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": "code", "metadata": { "id": "kYy6pcfTdJDw", "colab_type": "code", "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: 30 # 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, last_checkpoint=last_checkpoint, model_temp_dir=model_temp_dir, target_language=target_language)\n", "with open(\"joeynmt/configs/transformer_{name}.yaml\".format(name=name),'w') as f:\n", " f.write(config)" ], "execution_count": 0, "outputs": [] }, { "cell_type": "code", "metadata": { "id": "tzS8C2gIdZwn", "colab_type": "code", "outputId": "8d250035-e40a-4f3e-a664-de6ddf74db2f", "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": 68, "outputs": [ { "output_type": "stream", "text": [ "2020-04-09 21:04:30,491 Hello! This is Joey-NMT.\n", "2020-04-09 21:04:30.708603: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1\n", "2020-04-09 21:04:33,075 Total params: 12222976\n", "2020-04-09 21:04:33,077 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", "2020-04-09 21:04:36,683 Loading model from /content/drive/My Drive/masakhane/ha-en-baseline/models/haen_transformer/best.ckpt\n", "2020-04-09 21:04:37,031 cfg.name : haen_transformer\n", "2020-04-09 21:04:37,031 cfg.data.src : ha\n", "2020-04-09 21:04:37,031 cfg.data.trg : en\n", "2020-04-09 21:04:37,031 cfg.data.train : /content/drive/My Drive/masakhane/ha-en-baseline/train.bpe\n", "2020-04-09 21:04:37,031 cfg.data.dev : /content/drive/My Drive/masakhane/ha-en-baseline/dev.bpe\n", "2020-04-09 21:04:37,031 cfg.data.test : /content/drive/My Drive/masakhane/ha-en-baseline/test.bpe\n", "2020-04-09 21:04:37,031 cfg.data.level : bpe\n", "2020-04-09 21:04:37,032 cfg.data.lowercase : False\n", "2020-04-09 21:04:37,032 cfg.data.max_sent_length : 100\n", "2020-04-09 21:04:37,032 cfg.data.src_vocab : /content/drive/My Drive/masakhane/ha-en-baseline/vocab.txt\n", "2020-04-09 21:04:37,032 cfg.data.trg_vocab : /content/drive/My Drive/masakhane/ha-en-baseline/vocab.txt\n", "2020-04-09 21:04:37,032 cfg.testing.beam_size : 5\n", "2020-04-09 21:04:37,032 cfg.testing.alpha : 1.0\n", "2020-04-09 21:04:37,032 cfg.training.load_model : /content/drive/My Drive/masakhane/ha-en-baseline/models/haen_transformer/best.ckpt\n", "2020-04-09 21:04:37,032 cfg.training.random_seed : 42\n", "2020-04-09 21:04:37,033 cfg.training.optimizer : adam\n", "2020-04-09 21:04:37,033 cfg.training.normalization : tokens\n", "2020-04-09 21:04:37,033 cfg.training.adam_betas : [0.9, 0.999]\n", "2020-04-09 21:04:37,033 cfg.training.scheduling : plateau\n", "2020-04-09 21:04:37,033 cfg.training.patience : 5\n", "2020-04-09 21:04:37,033 cfg.training.learning_rate_factor : 0.5\n", "2020-04-09 21:04:37,033 cfg.training.learning_rate_warmup : 1000\n", "2020-04-09 21:04:37,033 cfg.training.decrease_factor : 0.7\n", "2020-04-09 21:04:37,033 cfg.training.loss : crossentropy\n", "2020-04-09 21:04:37,034 cfg.training.learning_rate : 0.0003\n", "2020-04-09 21:04:37,034 cfg.training.learning_rate_min : 1e-08\n", "2020-04-09 21:04:37,034 cfg.training.weight_decay : 0.0\n", "2020-04-09 21:04:37,034 cfg.training.label_smoothing : 0.1\n", "2020-04-09 21:04:37,034 cfg.training.batch_size : 4096\n", "2020-04-09 21:04:37,034 cfg.training.batch_type : token\n", "2020-04-09 21:04:37,034 cfg.training.eval_batch_size : 3600\n", "2020-04-09 21:04:37,034 cfg.training.eval_batch_type : token\n", "2020-04-09 21:04:37,035 cfg.training.batch_multiplier : 1\n", "2020-04-09 21:04:37,035 cfg.training.early_stopping_metric : ppl\n", "2020-04-09 21:04:37,035 cfg.training.epochs : 30\n", "2020-04-09 21:04:37,035 cfg.training.validation_freq : 1000\n", "2020-04-09 21:04:37,035 cfg.training.logging_freq : 100\n", "2020-04-09 21:04:37,035 cfg.training.eval_metric : bleu\n", "2020-04-09 21:04:37,035 cfg.training.model_dir : /content/drive/My Drive/masakhane/model-temp\n", "2020-04-09 21:04:37,035 cfg.training.overwrite : True\n", "2020-04-09 21:04:37,035 cfg.training.shuffle : True\n", "2020-04-09 21:04:37,036 cfg.training.use_cuda : True\n", "2020-04-09 21:04:37,036 cfg.training.max_output_length : 100\n", "2020-04-09 21:04:37,036 cfg.training.print_valid_sents : [0, 1, 2, 3]\n", "2020-04-09 21:04:37,036 cfg.training.keep_last_ckpts : 3\n", "2020-04-09 21:04:37,036 cfg.model.initializer : xavier\n", "2020-04-09 21:04:37,036 cfg.model.bias_initializer : zeros\n", "2020-04-09 21:04:37,036 cfg.model.init_gain : 1.0\n", "2020-04-09 21:04:37,036 cfg.model.embed_initializer : xavier\n", "2020-04-09 21:04:37,037 cfg.model.embed_init_gain : 1.0\n", "2020-04-09 21:04:37,037 cfg.model.tied_embeddings : True\n", "2020-04-09 21:04:37,037 cfg.model.tied_softmax : True\n", "2020-04-09 21:04:37,037 cfg.model.encoder.type : transformer\n", "2020-04-09 21:04:37,037 cfg.model.encoder.num_layers : 6\n", "2020-04-09 21:04:37,037 cfg.model.encoder.num_heads : 4\n", "2020-04-09 21:04:37,037 cfg.model.encoder.embeddings.embedding_dim : 256\n", "2020-04-09 21:04:37,037 cfg.model.encoder.embeddings.scale : True\n", "2020-04-09 21:04:37,037 cfg.model.encoder.embeddings.dropout : 0.2\n", "2020-04-09 21:04:37,038 cfg.model.encoder.hidden_size : 256\n", "2020-04-09 21:04:37,038 cfg.model.encoder.ff_size : 1024\n", "2020-04-09 21:04:37,038 cfg.model.encoder.dropout : 0.3\n", "2020-04-09 21:04:37,038 cfg.model.decoder.type : transformer\n", "2020-04-09 21:04:37,038 cfg.model.decoder.num_layers : 6\n", "2020-04-09 21:04:37,038 cfg.model.decoder.num_heads : 4\n", "2020-04-09 21:04:37,038 cfg.model.decoder.embeddings.embedding_dim : 256\n", "2020-04-09 21:04:37,038 cfg.model.decoder.embeddings.scale : True\n", "2020-04-09 21:04:37,038 cfg.model.decoder.embeddings.dropout : 0.2\n", "2020-04-09 21:04:37,039 cfg.model.decoder.hidden_size : 256\n", "2020-04-09 21:04:37,039 cfg.model.decoder.ff_size : 1024\n", "2020-04-09 21:04:37,039 cfg.model.decoder.dropout : 0.3\n", "2020-04-09 21:04:37,039 Data set sizes: \n", "\ttrain 333888,\n", "\tvalid 966,\n", "\ttest 6898\n", "2020-04-09 21:04:37,039 First training example:\n", "\t[SRC] Kuma Mun sanya tabbat@@ at@@ tun du@@ wã@@ tsu a cikin ƙasa dõmin kada ta kark@@ ata da su kuma Mun sanya ran@@ g@@ una , han@@ y@@ õyi , a cikinsu ( du@@ wã@@ t@@ sun ) , tsammã@@ ninsu sunã shir@@ yuwa .\n", "\t[TRG] And We have pla@@ ced in the earth firm moun@@ tains l@@ est it should move away with them , and We pla@@ ced therein P@@ as@@ sa@@ ges for pa@@ ths , that ha@@ ply they may be guided .\n", "2020-04-09 21:04:37,039 First 10 words (src): (0) (1) (2) (3) (4) , (5) . (6) da (7) the (8) a (9) to\n", "2020-04-09 21:04:37,039 First 10 words (trg): (0) (1) (2) (3) (4) , (5) . (6) da (7) the (8) a (9) to\n", "2020-04-09 21:04:37,040 Number of Src words (types): 4542\n", "2020-04-09 21:04:37,040 Number of Trg words (types): 4542\n", "2020-04-09 21:04:37,040 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=4542),\n", "\ttrg_embed=Embeddings(embedding_dim=256, vocab_size=4542))\n", "2020-04-09 21:04:37,049 EPOCH 1\n", "2020-04-09 21:04:55,676 Epoch 1 Step: 8100 Batch Loss: 2.744179 Tokens per Sec: 10577, Lr: 0.000300\n", "2020-04-09 21:05:14,080 Epoch 1 Step: 8200 Batch Loss: 2.926254 Tokens per Sec: 11087, Lr: 0.000300\n", "2020-04-09 21:05:32,692 Epoch 1 Step: 8300 Batch Loss: 2.237095 Tokens per Sec: 10985, Lr: 0.000300\n", "2020-04-09 21:05:51,136 Epoch 1 Step: 8400 Batch Loss: 3.012698 Tokens per Sec: 10969, Lr: 0.000300\n", "2020-04-09 21:06:09,587 Epoch 1 Step: 8500 Batch Loss: 2.756017 Tokens per Sec: 11015, Lr: 0.000300\n", "2020-04-09 21:06:27,977 Epoch 1 Step: 8600 Batch Loss: 2.801084 Tokens per Sec: 10843, Lr: 0.000300\n", "2020-04-09 21:06:46,122 Epoch 1 Step: 8700 Batch Loss: 2.946009 Tokens per Sec: 10997, Lr: 0.000300\n", "2020-04-09 21:07:04,159 Epoch 1 Step: 8800 Batch Loss: 2.687814 Tokens per Sec: 10749, Lr: 0.000300\n", "2020-04-09 21:07:22,601 Epoch 1 Step: 8900 Batch Loss: 2.652180 Tokens per Sec: 10928, Lr: 0.000300\n", "2020-04-09 21:07:40,882 Epoch 1 Step: 9000 Batch Loss: 2.998913 Tokens per Sec: 10825, Lr: 0.000300\n", "2020-04-09 21:08:11,926 Hooray! New best validation result [ppl]!\n", "2020-04-09 21:08:11,926 Saving new checkpoint.\n", "2020-04-09 21:08:13,104 Example #0\n", "2020-04-09 21:08:13,105 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 21:08:13,105 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 21:08:13,105 \tHypothesis: And see how We have made Our signs for them that they may understand .\n", "2020-04-09 21:08:13,105 Example #1\n", "2020-04-09 21:08:13,106 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 21:08:13,106 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 21:08:13,106 \tHypothesis: If We willed , We would have made it a source of water , and do you not give thanks ?\n", "2020-04-09 21:08:13,106 Example #2\n", "2020-04-09 21:08:13,107 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 21:08:13,107 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 21:08:13,107 \tHypothesis: And those who disbelieve in Our signs are the people of the people of the jinn .\n", "2020-04-09 21:08:13,107 Example #3\n", "2020-04-09 21:08:13,107 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 21:08:13,107 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 21:08:13,108 \tHypothesis: And He will remove the source of the source of the poor , or after his people , and when they were scribed him .\n", "2020-04-09 21:08:13,108 Validation result (greedy) at epoch 1, step 9000: bleu: 11.57, loss: 73550.1016, ppl: 15.1917, duration: 32.2256s\n", "2020-04-09 21:08:31,351 Epoch 1 Step: 9100 Batch Loss: 2.904357 Tokens per Sec: 11015, Lr: 0.000300\n", "2020-04-09 21:08:49,591 Epoch 1 Step: 9200 Batch Loss: 2.769007 Tokens per Sec: 10919, Lr: 0.000300\n", "2020-04-09 21:09:07,932 Epoch 1 Step: 9300 Batch Loss: 2.760248 Tokens per Sec: 11067, Lr: 0.000300\n", "2020-04-09 21:09:26,221 Epoch 1 Step: 9400 Batch Loss: 2.594356 Tokens per Sec: 10948, Lr: 0.000300\n", "2020-04-09 21:09:44,235 Epoch 1 Step: 9500 Batch Loss: 2.590112 Tokens per Sec: 10724, Lr: 0.000300\n", "2020-04-09 21:10:02,444 Epoch 1 Step: 9600 Batch Loss: 2.314419 Tokens per Sec: 10905, Lr: 0.000300\n", "2020-04-09 21:10:20,861 Epoch 1 Step: 9700 Batch Loss: 2.841067 Tokens per Sec: 11078, Lr: 0.000300\n", "2020-04-09 21:10:39,139 Epoch 1 Step: 9800 Batch Loss: 3.299751 Tokens per Sec: 11075, Lr: 0.000300\n", "2020-04-09 21:10:57,649 Epoch 1 Step: 9900 Batch Loss: 2.662146 Tokens per Sec: 11178, Lr: 0.000300\n", "2020-04-09 21:11:16,118 Epoch 1 Step: 10000 Batch Loss: 2.978909 Tokens per Sec: 11149, Lr: 0.000300\n", "2020-04-09 21:11:45,745 Hooray! New best validation result [ppl]!\n", "2020-04-09 21:11:45,745 Saving new checkpoint.\n", "2020-04-09 21:11:46,932 Example #0\n", "2020-04-09 21:11:46,932 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 21:11:46,933 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 21:11:46,933 \tHypothesis: See how We have made clear signs for them that they may understand .\n", "2020-04-09 21:11:46,933 Example #1\n", "2020-04-09 21:11:46,933 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 21:11:46,934 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 21:11:46,934 \tHypothesis: If We willed , We would have made it a scattle , so do you not give thanks ?\n", "2020-04-09 21:11:46,934 Example #2\n", "2020-04-09 21:11:46,934 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 21:11:46,934 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 21:11:46,934 \tHypothesis: Those who disbelieve in Our signs are the people of the people .\n", "2020-04-09 21:11:46,935 Example #3\n", "2020-04-09 21:11:46,935 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 21:11:46,936 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 21:11:46,936 \tHypothesis: He will swear by the silver of the silver , or when the people of his people were killed , and they were swear by him .\n", "2020-04-09 21:11:46,936 Validation result (greedy) at epoch 1, step 10000: bleu: 12.26, loss: 71540.4219, ppl: 14.1033, duration: 30.8178s\n", "2020-04-09 21:12:05,194 Epoch 1 Step: 10100 Batch Loss: 2.599641 Tokens per Sec: 10607, Lr: 0.000300\n", "2020-04-09 21:12:23,330 Epoch 1 Step: 10200 Batch Loss: 1.996477 Tokens per Sec: 10879, Lr: 0.000300\n", "2020-04-09 21:12:41,565 Epoch 1 Step: 10300 Batch Loss: 2.676321 Tokens per Sec: 10993, Lr: 0.000300\n", "2020-04-09 21:12:59,882 Epoch 1 Step: 10400 Batch Loss: 2.508385 Tokens per Sec: 10888, Lr: 0.000300\n", "2020-04-09 21:13:18,372 Epoch 1 Step: 10500 Batch Loss: 2.875653 Tokens per Sec: 11222, Lr: 0.000300\n", "2020-04-09 21:13:36,948 Epoch 1 Step: 10600 Batch Loss: 3.037446 Tokens per Sec: 11150, Lr: 0.000300\n", "2020-04-09 21:13:55,316 Epoch 1 Step: 10700 Batch Loss: 2.795002 Tokens per Sec: 11155, Lr: 0.000300\n", "2020-04-09 21:14:13,620 Epoch 1 Step: 10800 Batch Loss: 2.607245 Tokens per Sec: 10853, Lr: 0.000300\n", "2020-04-09 21:14:31,894 Epoch 1 Step: 10900 Batch Loss: 2.554965 Tokens per Sec: 10907, Lr: 0.000300\n", "2020-04-09 21:14:50,062 Epoch 1 Step: 11000 Batch Loss: 2.012334 Tokens per Sec: 11039, Lr: 0.000300\n", "2020-04-09 21:15:12,871 Hooray! New best validation result [ppl]!\n", "2020-04-09 21:15:12,871 Saving new checkpoint.\n", "2020-04-09 21:15:14,063 Example #0\n", "2020-04-09 21:15:14,063 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 21:15:14,063 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 21:15:14,064 \tHypothesis: See how We have made Our signs for their understanding .\n", "2020-04-09 21:15:14,064 Example #1\n", "2020-04-09 21:15:14,064 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 21:15:14,064 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 21:15:14,064 \tHypothesis: If We willed , We would have made it a mountain , then do you not thank ?\n", "2020-04-09 21:15:14,064 Example #2\n", "2020-04-09 21:15:14,065 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 21:15:14,065 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 21:15:14,065 \tHypothesis: And those who disbelieve in Our signs are the people of the people of the righteous .\n", "2020-04-09 21:15:14,065 Example #3\n", "2020-04-09 21:15:14,065 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 21:15:14,066 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 21:15:14,066 \tHypothesis: And He made the source of the silver , or when some of his people were before him , they sleep him .\n", "2020-04-09 21:15:14,066 Validation result (greedy) at epoch 1, step 11000: bleu: 12.87, loss: 70733.1250, ppl: 13.6884, duration: 24.0037s\n", "2020-04-09 21:15:32,526 Epoch 1 Step: 11100 Batch Loss: 2.645327 Tokens per Sec: 10857, Lr: 0.000300\n", "2020-04-09 21:15:50,655 Epoch 1 Step: 11200 Batch Loss: 2.717508 Tokens per Sec: 10889, Lr: 0.000300\n", "2020-04-09 21:16:08,980 Epoch 1 Step: 11300 Batch Loss: 2.501031 Tokens per Sec: 11038, Lr: 0.000300\n", "2020-04-09 21:16:27,310 Epoch 1 Step: 11400 Batch Loss: 2.679977 Tokens per Sec: 11148, Lr: 0.000300\n", "2020-04-09 21:16:45,635 Epoch 1 Step: 11500 Batch Loss: 2.769779 Tokens per Sec: 11108, Lr: 0.000300\n", "2020-04-09 21:17:03,730 Epoch 1 Step: 11600 Batch Loss: 2.404546 Tokens per Sec: 10750, Lr: 0.000300\n", "2020-04-09 21:17:21,975 Epoch 1 Step: 11700 Batch Loss: 2.963691 Tokens per Sec: 10838, Lr: 0.000300\n", "2020-04-09 21:17:40,175 Epoch 1 Step: 11800 Batch Loss: 2.907357 Tokens per Sec: 10954, Lr: 0.000300\n", "2020-04-09 21:17:58,394 Epoch 1 Step: 11900 Batch Loss: 2.596608 Tokens per Sec: 11132, Lr: 0.000300\n", "2020-04-09 21:18:16,559 Epoch 1 Step: 12000 Batch Loss: 2.612819 Tokens per Sec: 10709, Lr: 0.000300\n", "2020-04-09 21:18:44,348 Hooray! New best validation result [ppl]!\n", "2020-04-09 21:18:44,348 Saving new checkpoint.\n", "2020-04-09 21:18:45,605 Example #0\n", "2020-04-09 21:18:45,606 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 21:18:45,606 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 21:18:45,606 \tHypothesis: See how We have placed Our signs , that they may understand .\n", "2020-04-09 21:18:45,606 Example #1\n", "2020-04-09 21:18:45,606 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 21:18:45,607 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 21:18:45,607 \tHypothesis: If We willed , We would have made it water , so that you may not be grateful ?\n", "2020-04-09 21:18:45,607 Example #2\n", "2020-04-09 21:18:45,607 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 21:18:45,607 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 21:18:45,607 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 21:18:45,608 Example #3\n", "2020-04-09 21:18:45,608 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 21:18:45,608 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 21:18:45,608 \tHypothesis: He will swear the poor , or when some of his people were destroyed before him , and they took him a moon .\n", "2020-04-09 21:18:45,608 Validation result (greedy) at epoch 1, step 12000: bleu: 13.12, loss: 69586.6797, ppl: 13.1200, duration: 29.0491s\n", "2020-04-09 21:19:04,078 Epoch 1 Step: 12100 Batch Loss: 2.710072 Tokens per Sec: 10835, Lr: 0.000300\n", "2020-04-09 21:19:22,535 Epoch 1 Step: 12200 Batch Loss: 2.457842 Tokens per Sec: 11304, Lr: 0.000300\n", "2020-04-09 21:19:40,941 Epoch 1 Step: 12300 Batch Loss: 2.591026 Tokens per Sec: 10885, Lr: 0.000300\n", "2020-04-09 21:19:59,216 Epoch 1 Step: 12400 Batch Loss: 2.461279 Tokens per Sec: 11037, Lr: 0.000300\n", "2020-04-09 21:20:02,395 Epoch 1: total training loss 12007.86\n", "2020-04-09 21:20:02,396 EPOCH 2\n", "2020-04-09 21:20:17,594 Epoch 2 Step: 12500 Batch Loss: 2.296296 Tokens per Sec: 10527, Lr: 0.000300\n", "2020-04-09 21:20:36,019 Epoch 2 Step: 12600 Batch Loss: 2.519243 Tokens per Sec: 11077, Lr: 0.000300\n", "2020-04-09 21:20:54,568 Epoch 2 Step: 12700 Batch Loss: 2.251203 Tokens per Sec: 11065, Lr: 0.000300\n", "2020-04-09 21:21:12,807 Epoch 2 Step: 12800 Batch Loss: 2.888603 Tokens per Sec: 10978, Lr: 0.000300\n", "2020-04-09 21:21:31,127 Epoch 2 Step: 12900 Batch Loss: 2.688028 Tokens per Sec: 11056, Lr: 0.000300\n", "2020-04-09 21:21:49,642 Epoch 2 Step: 13000 Batch Loss: 2.582552 Tokens per Sec: 11275, Lr: 0.000300\n", "2020-04-09 21:22:20,235 Hooray! New best validation result [ppl]!\n", "2020-04-09 21:22:20,236 Saving new checkpoint.\n", "2020-04-09 21:22:21,657 Example #0\n", "2020-04-09 21:22:21,658 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 21:22:21,658 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 21:22:21,658 \tHypothesis: See how We have made clear signs that they may understand .\n", "2020-04-09 21:22:21,658 Example #1\n", "2020-04-09 21:22:21,658 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 21:22:21,659 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 21:22:21,659 \tHypothesis: If We willed , We would have made it scatter , then why do you not give thanks ?\n", "2020-04-09 21:22:21,659 Example #2\n", "2020-04-09 21:22:21,659 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 21:22:21,659 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 21:22:21,659 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 21:22:21,660 Example #3\n", "2020-04-09 21:22:21,660 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 21:22:21,660 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 21:22:21,661 \tHypothesis: He will swear up the staff , and when some of his people were before him , and he will swear him .\n", "2020-04-09 21:22:21,661 Validation result (greedy) at epoch 2, step 13000: bleu: 14.33, loss: 67706.3047, ppl: 12.2384, duration: 32.0189s\n", "2020-04-09 21:22:40,075 Epoch 2 Step: 13100 Batch Loss: 2.782082 Tokens per Sec: 10770, Lr: 0.000300\n", "2020-04-09 21:22:58,381 Epoch 2 Step: 13200 Batch Loss: 2.683053 Tokens per Sec: 10995, Lr: 0.000300\n", "2020-04-09 21:23:16,823 Epoch 2 Step: 13300 Batch Loss: 2.662381 Tokens per Sec: 10983, Lr: 0.000300\n", "2020-04-09 21:23:35,281 Epoch 2 Step: 13400 Batch Loss: 2.565167 Tokens per Sec: 10933, Lr: 0.000300\n", "2020-04-09 21:23:53,636 Epoch 2 Step: 13500 Batch Loss: 2.549377 Tokens per Sec: 11101, Lr: 0.000300\n", "2020-04-09 21:24:11,864 Epoch 2 Step: 13600 Batch Loss: 2.523279 Tokens per Sec: 10937, Lr: 0.000300\n", "2020-04-09 21:24:29,995 Epoch 2 Step: 13700 Batch Loss: 2.452710 Tokens per Sec: 10782, Lr: 0.000300\n", "2020-04-09 21:24:48,253 Epoch 2 Step: 13800 Batch Loss: 2.861645 Tokens per Sec: 10842, Lr: 0.000300\n", "2020-04-09 21:25:06,543 Epoch 2 Step: 13900 Batch Loss: 2.296751 Tokens per Sec: 11120, Lr: 0.000300\n", "2020-04-09 21:25:24,965 Epoch 2 Step: 14000 Batch Loss: 2.505846 Tokens per Sec: 11093, Lr: 0.000300\n", "2020-04-09 21:25:52,942 Hooray! New best validation result [ppl]!\n", "2020-04-09 21:25:52,943 Saving new checkpoint.\n", "2020-04-09 21:25:54,207 Example #0\n", "2020-04-09 21:25:54,208 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 21:25:54,208 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 21:25:54,208 \tHypothesis: See how We have made Our signs , so that they may understand .\n", "2020-04-09 21:25:54,208 Example #1\n", "2020-04-09 21:25:54,209 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 21:25:54,209 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 21:25:54,209 \tHypothesis: If We willed , We would have brought it down , then why do you not give thanks ?\n", "2020-04-09 21:25:54,209 Example #2\n", "2020-04-09 21:25:54,210 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 21:25:54,210 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 21:25:54,210 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 21:25:54,210 Example #3\n", "2020-04-09 21:25:54,210 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 21:25:54,210 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 21:25:54,211 \tHypothesis: And He will send the mountains in the morning , and when some of his people were before him , they swear him .\n", "2020-04-09 21:25:54,211 Validation result (greedy) at epoch 2, step 14000: bleu: 14.99, loss: 66552.3047, ppl: 11.7270, duration: 29.2452s\n", "2020-04-09 21:26:12,786 Epoch 2 Step: 14100 Batch Loss: 2.731581 Tokens per Sec: 11040, Lr: 0.000300\n", "2020-04-09 21:26:30,992 Epoch 2 Step: 14200 Batch Loss: 2.455379 Tokens per Sec: 10890, Lr: 0.000300\n", "2020-04-09 21:26:49,293 Epoch 2 Step: 14300 Batch Loss: 1.839818 Tokens per Sec: 10964, Lr: 0.000300\n", "2020-04-09 21:27:07,394 Epoch 2 Step: 14400 Batch Loss: 2.703363 Tokens per Sec: 10900, Lr: 0.000300\n", "2020-04-09 21:27:25,958 Epoch 2 Step: 14500 Batch Loss: 2.679580 Tokens per Sec: 11299, Lr: 0.000300\n", "2020-04-09 21:27:44,247 Epoch 2 Step: 14600 Batch Loss: 2.673619 Tokens per Sec: 11061, Lr: 0.000300\n", "2020-04-09 21:28:02,547 Epoch 2 Step: 14700 Batch Loss: 2.739856 Tokens per Sec: 11102, Lr: 0.000300\n", "2020-04-09 21:28:20,885 Epoch 2 Step: 14800 Batch Loss: 2.997808 Tokens per Sec: 11188, Lr: 0.000300\n", "2020-04-09 21:28:38,692 Epoch 2 Step: 14900 Batch Loss: 2.572950 Tokens per Sec: 10473, Lr: 0.000300\n", "2020-04-09 21:28:57,147 Epoch 2 Step: 15000 Batch Loss: 2.505492 Tokens per Sec: 11118, Lr: 0.000300\n", "2020-04-09 21:29:29,053 Hooray! New best validation result [ppl]!\n", "2020-04-09 21:29:29,054 Saving new checkpoint.\n", "2020-04-09 21:29:30,250 Example #0\n", "2020-04-09 21:29:30,251 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 21:29:30,251 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 21:29:30,251 \tHypothesis: See how We have made clear signs that they may understand .\n", "2020-04-09 21:29:30,251 Example #1\n", "2020-04-09 21:29:30,252 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 21:29:30,252 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 21:29:30,252 \tHypothesis: If We willed , We would have brought it down water , so why do you not be grateful ?\n", "2020-04-09 21:29:30,252 Example #2\n", "2020-04-09 21:29:30,252 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 21:29:30,253 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 21:29:30,253 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 21:29:30,253 Example #3\n", "2020-04-09 21:29:30,253 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 21:29:30,253 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 21:29:30,253 \tHypothesis: He will send the mountains into the land , or when some of his people were before him , they mock him .\n", "2020-04-09 21:29:30,254 Validation result (greedy) at epoch 2, step 15000: bleu: 15.48, loss: 65527.4102, ppl: 11.2907, duration: 33.1061s\n", "2020-04-09 21:29:48,756 Epoch 2 Step: 15100 Batch Loss: 2.742007 Tokens per Sec: 10945, Lr: 0.000300\n", "2020-04-09 21:30:07,220 Epoch 2 Step: 15200 Batch Loss: 2.499738 Tokens per Sec: 11130, Lr: 0.000300\n", "2020-04-09 21:30:25,584 Epoch 2 Step: 15300 Batch Loss: 2.907837 Tokens per Sec: 11007, Lr: 0.000300\n", "2020-04-09 21:30:43,664 Epoch 2 Step: 15400 Batch Loss: 2.545413 Tokens per Sec: 10767, Lr: 0.000300\n", "2020-04-09 21:31:02,221 Epoch 2 Step: 15500 Batch Loss: 2.406178 Tokens per Sec: 11195, Lr: 0.000300\n", "2020-04-09 21:31:20,449 Epoch 2 Step: 15600 Batch Loss: 2.623208 Tokens per Sec: 10936, Lr: 0.000300\n", "2020-04-09 21:31:38,881 Epoch 2 Step: 15700 Batch Loss: 2.450566 Tokens per Sec: 11168, Lr: 0.000300\n", "2020-04-09 21:31:57,082 Epoch 2 Step: 15800 Batch Loss: 2.356850 Tokens per Sec: 10881, Lr: 0.000300\n", "2020-04-09 21:32:15,174 Epoch 2 Step: 15900 Batch Loss: 2.464229 Tokens per Sec: 11041, Lr: 0.000300\n", "2020-04-09 21:32:33,427 Epoch 2 Step: 16000 Batch Loss: 2.377052 Tokens per Sec: 10894, Lr: 0.000300\n", "2020-04-09 21:32:57,619 Hooray! New best validation result [ppl]!\n", "2020-04-09 21:32:57,619 Saving new checkpoint.\n", "2020-04-09 21:32:58,850 Example #0\n", "2020-04-09 21:32:58,850 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 21:32:58,850 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 21:32:58,851 \tHypothesis: See how We have made the signs for them that they may understand .\n", "2020-04-09 21:32:58,851 Example #1\n", "2020-04-09 21:32:58,851 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 21:32:58,851 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 21:32:58,851 \tHypothesis: If We willed , We would have brought it down , then why do you not give thanks ?\n", "2020-04-09 21:32:58,852 Example #2\n", "2020-04-09 21:32:58,852 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 21:32:58,852 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 21:32:58,852 \tHypothesis: Those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 21:32:58,852 Example #3\n", "2020-04-09 21:32:58,853 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 21:32:58,853 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 21:32:58,853 \tHypothesis: He will swear by the mountains , and when some of his people were before him , they mock him .\n", "2020-04-09 21:32:58,853 Validation result (greedy) at epoch 2, step 16000: bleu: 15.24, loss: 64649.4609, ppl: 10.9299, duration: 25.4259s\n", "2020-04-09 21:33:17,264 Epoch 2 Step: 16100 Batch Loss: 2.822345 Tokens per Sec: 10891, Lr: 0.000300\n", "2020-04-09 21:33:35,460 Epoch 2 Step: 16200 Batch Loss: 2.666233 Tokens per Sec: 10845, Lr: 0.000300\n", "2020-04-09 21:33:53,695 Epoch 2 Step: 16300 Batch Loss: 2.450842 Tokens per Sec: 10900, Lr: 0.000300\n", "2020-04-09 21:34:12,018 Epoch 2 Step: 16400 Batch Loss: 2.358483 Tokens per Sec: 11027, Lr: 0.000300\n", "2020-04-09 21:34:30,264 Epoch 2 Step: 16500 Batch Loss: 2.838802 Tokens per Sec: 10920, Lr: 0.000300\n", "2020-04-09 21:34:48,674 Epoch 2 Step: 16600 Batch Loss: 2.482951 Tokens per Sec: 11033, Lr: 0.000300\n", "2020-04-09 21:35:06,902 Epoch 2 Step: 16700 Batch Loss: 2.912965 Tokens per Sec: 11098, Lr: 0.000300\n", "2020-04-09 21:35:25,077 Epoch 2 Step: 16800 Batch Loss: 2.834556 Tokens per Sec: 10812, Lr: 0.000300\n", "2020-04-09 21:35:29,653 Epoch 2: total training loss 11432.02\n", "2020-04-09 21:35:29,654 EPOCH 3\n", "2020-04-09 21:35:43,783 Epoch 3 Step: 16900 Batch Loss: 2.626662 Tokens per Sec: 10704, Lr: 0.000300\n", "2020-04-09 21:36:02,224 Epoch 3 Step: 17000 Batch Loss: 2.497550 Tokens per Sec: 11078, Lr: 0.000300\n", "2020-04-09 21:36:28,097 Hooray! New best validation result [ppl]!\n", "2020-04-09 21:36:28,097 Saving new checkpoint.\n", "2020-04-09 21:36:29,361 Example #0\n", "2020-04-09 21:36:29,362 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 21:36:29,362 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 21:36:29,362 \tHypothesis: See how We have explained Our signs , so that they may understand . \"\n", "2020-04-09 21:36:29,362 Example #1\n", "2020-04-09 21:36:29,362 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 21:36:29,363 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 21:36:29,363 \tHypothesis: If We willed , We would have brought it down water , then why do you not give thanks ?\n", "2020-04-09 21:36:29,363 Example #2\n", "2020-04-09 21:36:29,363 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 21:36:29,363 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 21:36:29,364 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 21:36:29,364 Example #3\n", "2020-04-09 21:36:29,364 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 21:36:29,364 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 21:36:29,364 \tHypothesis: He will send the ships of the black , and when the leaders of his people came before him , they mock him .\n", "2020-04-09 21:36:29,364 Validation result (greedy) at epoch 3, step 17000: bleu: 15.91, loss: 63707.5977, ppl: 10.5556, duration: 27.1402s\n", "2020-04-09 21:36:47,752 Epoch 3 Step: 17100 Batch Loss: 2.498299 Tokens per Sec: 11019, Lr: 0.000300\n", "2020-04-09 21:37:05,941 Epoch 3 Step: 17200 Batch Loss: 2.555383 Tokens per Sec: 10963, Lr: 0.000300\n", "2020-04-09 21:37:24,346 Epoch 3 Step: 17300 Batch Loss: 2.544756 Tokens per Sec: 11054, Lr: 0.000300\n", "2020-04-09 21:37:42,391 Epoch 3 Step: 17400 Batch Loss: 2.525505 Tokens per Sec: 10804, Lr: 0.000300\n", "2020-04-09 21:38:00,719 Epoch 3 Step: 17500 Batch Loss: 2.694733 Tokens per Sec: 10785, Lr: 0.000300\n", "2020-04-09 21:38:18,893 Epoch 3 Step: 17600 Batch Loss: 2.502076 Tokens per Sec: 11040, Lr: 0.000300\n", "2020-04-09 21:38:37,061 Epoch 3 Step: 17700 Batch Loss: 2.207860 Tokens per Sec: 10898, Lr: 0.000300\n", "2020-04-09 21:38:55,406 Epoch 3 Step: 17800 Batch Loss: 2.588396 Tokens per Sec: 10934, Lr: 0.000300\n", "2020-04-09 21:39:13,520 Epoch 3 Step: 17900 Batch Loss: 2.916042 Tokens per Sec: 10910, Lr: 0.000300\n", "2020-04-09 21:39:31,972 Epoch 3 Step: 18000 Batch Loss: 2.472352 Tokens per Sec: 11151, Lr: 0.000300\n", "2020-04-09 21:40:02,573 Hooray! New best validation result [ppl]!\n", "2020-04-09 21:40:02,573 Saving new checkpoint.\n", "2020-04-09 21:40:03,771 Example #0\n", "2020-04-09 21:40:03,772 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 21:40:03,772 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 21:40:03,772 \tHypothesis: See how We explain Our signs that they may understand .\n", "2020-04-09 21:40:03,772 Example #1\n", "2020-04-09 21:40:03,773 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 21:40:03,773 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 21:40:03,773 \tHypothesis: Had We willed , We would have brought it down water , so why do you not give thanks ?\n", "2020-04-09 21:40:03,773 Example #2\n", "2020-04-09 21:40:03,773 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 21:40:03,774 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 21:40:03,774 \tHypothesis: Those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 21:40:03,774 Example #3\n", "2020-04-09 21:40:03,774 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 21:40:03,774 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 21:40:03,774 \tHypothesis: He will swear on the sea , and when some of his people passed before him , they mocked him .\n", "2020-04-09 21:40:03,775 Validation result (greedy) at epoch 3, step 18000: bleu: 16.89, loss: 62847.3516, ppl: 10.2250, duration: 31.8017s\n", "2020-04-09 21:40:22,187 Epoch 3 Step: 18100 Batch Loss: 2.439152 Tokens per Sec: 10900, Lr: 0.000300\n", "2020-04-09 21:40:40,570 Epoch 3 Step: 18200 Batch Loss: 2.490841 Tokens per Sec: 11166, Lr: 0.000300\n", "2020-04-09 21:40:58,607 Epoch 3 Step: 18300 Batch Loss: 2.704951 Tokens per Sec: 10795, Lr: 0.000300\n", "2020-04-09 21:41:16,892 Epoch 3 Step: 18400 Batch Loss: 2.389753 Tokens per Sec: 11100, Lr: 0.000300\n", "2020-04-09 21:41:34,997 Epoch 3 Step: 18500 Batch Loss: 2.442512 Tokens per Sec: 10845, Lr: 0.000300\n", "2020-04-09 21:41:53,454 Epoch 3 Step: 18600 Batch Loss: 2.791895 Tokens per Sec: 11346, Lr: 0.000300\n", "2020-04-09 21:42:11,485 Epoch 3 Step: 18700 Batch Loss: 2.511027 Tokens per Sec: 10891, Lr: 0.000300\n", "2020-04-09 21:42:29,621 Epoch 3 Step: 18800 Batch Loss: 2.446014 Tokens per Sec: 10755, Lr: 0.000300\n", "2020-04-09 21:42:48,116 Epoch 3 Step: 18900 Batch Loss: 2.595623 Tokens per Sec: 11038, Lr: 0.000300\n", "2020-04-09 21:43:06,373 Epoch 3 Step: 19000 Batch Loss: 2.462938 Tokens per Sec: 10967, Lr: 0.000300\n", "2020-04-09 21:43:36,078 Hooray! New best validation result [ppl]!\n", "2020-04-09 21:43:36,078 Saving new checkpoint.\n", "2020-04-09 21:43:37,307 Example #0\n", "2020-04-09 21:43:37,308 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 21:43:37,308 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 21:43:37,308 \tHypothesis: See how We have explained the signs that they may understand .\n", "2020-04-09 21:43:37,308 Example #1\n", "2020-04-09 21:43:37,308 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 21:43:37,308 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 21:43:37,309 \tHypothesis: If We willed , We would have made it a sleep , then why do you not give thanks ?\n", "2020-04-09 21:43:37,309 Example #2\n", "2020-04-09 21:43:37,309 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 21:43:37,309 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 21:43:37,309 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 21:43:37,310 Example #3\n", "2020-04-09 21:43:37,310 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 21:43:37,310 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 21:43:37,310 \tHypothesis: He is silver in the darkness , and when some of his people passed before him , they mock him .\n", "2020-04-09 21:43:37,311 Validation result (greedy) at epoch 3, step 19000: bleu: 16.97, loss: 62223.7656, ppl: 9.9919, duration: 30.9371s\n", "2020-04-09 21:43:55,781 Epoch 3 Step: 19100 Batch Loss: 2.580698 Tokens per Sec: 10817, Lr: 0.000300\n", "2020-04-09 21:44:13,917 Epoch 3 Step: 19200 Batch Loss: 2.149927 Tokens per Sec: 10983, Lr: 0.000300\n", "2020-04-09 21:44:32,033 Epoch 3 Step: 19300 Batch Loss: 2.509575 Tokens per Sec: 10843, Lr: 0.000300\n", "2020-04-09 21:44:50,407 Epoch 3 Step: 19400 Batch Loss: 2.393106 Tokens per Sec: 10953, Lr: 0.000300\n", "2020-04-09 21:45:08,619 Epoch 3 Step: 19500 Batch Loss: 2.572040 Tokens per Sec: 10951, Lr: 0.000300\n", "2020-04-09 21:45:26,741 Epoch 3 Step: 19600 Batch Loss: 2.254961 Tokens per Sec: 10983, Lr: 0.000300\n", "2020-04-09 21:45:44,854 Epoch 3 Step: 19700 Batch Loss: 2.653487 Tokens per Sec: 10913, Lr: 0.000300\n", "2020-04-09 21:46:03,151 Epoch 3 Step: 19800 Batch Loss: 2.419533 Tokens per Sec: 11079, Lr: 0.000300\n", "2020-04-09 21:46:21,491 Epoch 3 Step: 19900 Batch Loss: 2.411180 Tokens per Sec: 11165, Lr: 0.000300\n", "2020-04-09 21:46:39,763 Epoch 3 Step: 20000 Batch Loss: 2.163590 Tokens per Sec: 11044, Lr: 0.000300\n", "2020-04-09 21:47:09,992 Hooray! New best validation result [ppl]!\n", "2020-04-09 21:47:09,993 Saving new checkpoint.\n", "2020-04-09 21:47:11,178 Example #0\n", "2020-04-09 21:47:11,178 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 21:47:11,179 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 21:47:11,179 \tHypothesis: See how We explain Our signs that they may understand .\n", "2020-04-09 21:47:11,179 Example #1\n", "2020-04-09 21:47:11,179 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 21:47:11,179 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 21:47:11,179 \tHypothesis: If We willed , We would have brought him down water , so why do you not give thanks ?\n", "2020-04-09 21:47:11,180 Example #2\n", "2020-04-09 21:47:11,180 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 21:47:11,180 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 21:47:11,180 \tHypothesis: Those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 21:47:11,180 Example #3\n", "2020-04-09 21:47:11,181 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 21:47:11,181 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 21:47:11,181 \tHypothesis: He will send down the mountains , and when some of his people passed before him , they mocked him .\n", "2020-04-09 21:47:11,181 Validation result (greedy) at epoch 3, step 20000: bleu: 17.26, loss: 61801.8516, ppl: 9.8371, duration: 31.4179s\n", "2020-04-09 21:47:29,405 Epoch 3 Step: 20100 Batch Loss: 1.884119 Tokens per Sec: 10752, Lr: 0.000300\n", "2020-04-09 21:47:47,792 Epoch 3 Step: 20200 Batch Loss: 2.399179 Tokens per Sec: 11276, Lr: 0.000300\n", "2020-04-09 21:48:05,943 Epoch 3 Step: 20300 Batch Loss: 2.397899 Tokens per Sec: 11011, Lr: 0.000300\n", "2020-04-09 21:48:24,156 Epoch 3 Step: 20400 Batch Loss: 2.655838 Tokens per Sec: 10828, Lr: 0.000300\n", "2020-04-09 21:48:42,423 Epoch 3 Step: 20500 Batch Loss: 2.472438 Tokens per Sec: 10866, Lr: 0.000300\n", "2020-04-09 21:49:00,847 Epoch 3 Step: 20600 Batch Loss: 2.316028 Tokens per Sec: 11247, Lr: 0.000300\n", "2020-04-09 21:49:19,235 Epoch 3 Step: 20700 Batch Loss: 2.554988 Tokens per Sec: 11112, Lr: 0.000300\n", "2020-04-09 21:49:37,515 Epoch 3 Step: 20800 Batch Loss: 2.361322 Tokens per Sec: 11085, Lr: 0.000300\n", "2020-04-09 21:49:55,801 Epoch 3 Step: 20900 Batch Loss: 2.536736 Tokens per Sec: 11136, Lr: 0.000300\n", "2020-04-09 21:50:13,934 Epoch 3 Step: 21000 Batch Loss: 2.655164 Tokens per Sec: 10919, Lr: 0.000300\n", "2020-04-09 21:50:40,793 Hooray! New best validation result [ppl]!\n", "2020-04-09 21:50:40,793 Saving new checkpoint.\n", "2020-04-09 21:50:41,986 Example #0\n", "2020-04-09 21:50:41,988 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 21:50:41,988 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 21:50:41,988 \tHypothesis: See how We explain Our signs , so that they may understand .\n", "2020-04-09 21:50:41,988 Example #1\n", "2020-04-09 21:50:41,989 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 21:50:41,989 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 21:50:41,989 \tHypothesis: If We willed , We would have brought him down water , so why do you not give thanks ?\n", "2020-04-09 21:50:41,989 Example #2\n", "2020-04-09 21:50:41,989 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 21:50:41,990 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 21:50:41,990 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 21:50:41,990 Example #3\n", "2020-04-09 21:50:41,990 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 21:50:41,990 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 21:50:41,991 \tHypothesis: He will raise the silver in the scale , and when some of his people passed before him , they mock him .\n", "2020-04-09 21:50:41,991 Validation result (greedy) at epoch 3, step 21000: bleu: 17.68, loss: 61139.4961, ppl: 9.5990, duration: 28.0559s\n", "2020-04-09 21:51:00,436 Epoch 3 Step: 21100 Batch Loss: 2.190501 Tokens per Sec: 10938, Lr: 0.000300\n", "2020-04-09 21:51:18,681 Epoch 3 Step: 21200 Batch Loss: 2.153544 Tokens per Sec: 10899, Lr: 0.000300\n", "2020-04-09 21:51:26,509 Epoch 3: total training loss 10812.12\n", "2020-04-09 21:51:26,509 EPOCH 4\n", "2020-04-09 21:51:37,566 Epoch 4 Step: 21300 Batch Loss: 2.420489 Tokens per Sec: 10752, Lr: 0.000300\n", "2020-04-09 21:51:55,623 Epoch 4 Step: 21400 Batch Loss: 2.436672 Tokens per Sec: 10858, Lr: 0.000300\n", "2020-04-09 21:52:13,751 Epoch 4 Step: 21500 Batch Loss: 2.443198 Tokens per Sec: 10819, Lr: 0.000300\n", "2020-04-09 21:52:31,933 Epoch 4 Step: 21600 Batch Loss: 2.465627 Tokens per Sec: 10823, Lr: 0.000300\n", "2020-04-09 21:52:50,257 Epoch 4 Step: 21700 Batch Loss: 2.392741 Tokens per Sec: 10986, Lr: 0.000300\n", "2020-04-09 21:53:08,519 Epoch 4 Step: 21800 Batch Loss: 2.560066 Tokens per Sec: 10941, Lr: 0.000300\n", "2020-04-09 21:53:26,811 Epoch 4 Step: 21900 Batch Loss: 2.153594 Tokens per Sec: 11012, Lr: 0.000300\n", "2020-04-09 21:53:45,060 Epoch 4 Step: 22000 Batch Loss: 2.185071 Tokens per Sec: 11027, Lr: 0.000300\n", "2020-04-09 21:54:16,025 Hooray! New best validation result [ppl]!\n", "2020-04-09 21:54:16,026 Saving new checkpoint.\n", "2020-04-09 21:54:17,250 Example #0\n", "2020-04-09 21:54:17,251 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 21:54:17,251 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 21:54:17,251 \tHypothesis: See how We explain Our signs that they may understand . \"\n", "2020-04-09 21:54:17,251 Example #1\n", "2020-04-09 21:54:17,251 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 21:54:17,252 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 21:54:17,252 \tHypothesis: Had We willed , We would have brought him down water , so why do you not give thanks ?\n", "2020-04-09 21:54:17,252 Example #2\n", "2020-04-09 21:54:17,252 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 21:54:17,252 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 21:54:17,252 \tHypothesis: Those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 21:54:17,253 Example #3\n", "2020-04-09 21:54:17,253 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 21:54:17,253 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 21:54:17,253 \tHypothesis: He will swear in the land , and when the leaders of his people passed away before him , they mocked him .\n", "2020-04-09 21:54:17,253 Validation result (greedy) at epoch 4, step 22000: bleu: 17.86, loss: 60433.4375, ppl: 9.3516, duration: 32.1930s\n", "2020-04-09 21:54:35,541 Epoch 4 Step: 22100 Batch Loss: 2.235713 Tokens per Sec: 10651, Lr: 0.000300\n", "2020-04-09 21:54:53,925 Epoch 4 Step: 22200 Batch Loss: 2.199056 Tokens per Sec: 11215, Lr: 0.000300\n", "2020-04-09 21:55:12,079 Epoch 4 Step: 22300 Batch Loss: 2.569638 Tokens per Sec: 11194, Lr: 0.000300\n", "2020-04-09 21:55:30,442 Epoch 4 Step: 22400 Batch Loss: 2.184033 Tokens per Sec: 11143, Lr: 0.000300\n", "2020-04-09 21:55:48,582 Epoch 4 Step: 22500 Batch Loss: 2.249218 Tokens per Sec: 10884, Lr: 0.000300\n", "2020-04-09 21:56:06,831 Epoch 4 Step: 22600 Batch Loss: 2.020653 Tokens per Sec: 10945, Lr: 0.000300\n", "2020-04-09 21:56:25,028 Epoch 4 Step: 22700 Batch Loss: 2.322217 Tokens per Sec: 10945, Lr: 0.000300\n", "2020-04-09 21:56:43,174 Epoch 4 Step: 22800 Batch Loss: 2.227089 Tokens per Sec: 11120, Lr: 0.000300\n", "2020-04-09 21:57:01,558 Epoch 4 Step: 22900 Batch Loss: 2.167634 Tokens per Sec: 11034, Lr: 0.000300\n", "2020-04-09 21:57:19,619 Epoch 4 Step: 23000 Batch Loss: 2.206544 Tokens per Sec: 10910, Lr: 0.000300\n", "2020-04-09 21:57:45,138 Hooray! New best validation result [ppl]!\n", "2020-04-09 21:57:45,138 Saving new checkpoint.\n", "2020-04-09 21:57:46,358 Example #0\n", "2020-04-09 21:57:46,358 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 21:57:46,358 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 21:57:46,358 \tHypothesis: See how We explain Our signs that they may understand .\n", "2020-04-09 21:57:46,359 Example #1\n", "2020-04-09 21:57:46,359 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 21:57:46,359 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 21:57:46,359 \tHypothesis: Had We willed , We would have brought it down water , so why do you not give thanks ?\n", "2020-04-09 21:57:46,359 Example #2\n", "2020-04-09 21:57:46,360 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 21:57:46,360 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 21:57:46,360 \tHypothesis: And those who disbelieve Our revelations are the people of the nations .\n", "2020-04-09 21:57:46,360 Example #3\n", "2020-04-09 21:57:46,361 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 21:57:46,361 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 21:57:46,361 \tHypothesis: He will send the waters in the land , and when some of his people passed away before him , they mock him .\n", "2020-04-09 21:57:46,361 Validation result (greedy) at epoch 4, step 23000: bleu: 18.40, loss: 59807.6797, ppl: 9.1376, duration: 26.7420s\n", "2020-04-09 21:58:04,611 Epoch 4 Step: 23100 Batch Loss: 2.282868 Tokens per Sec: 10892, Lr: 0.000300\n", "2020-04-09 21:58:22,715 Epoch 4 Step: 23200 Batch Loss: 2.252471 Tokens per Sec: 10751, Lr: 0.000300\n", "2020-04-09 21:58:41,146 Epoch 4 Step: 23300 Batch Loss: 2.558650 Tokens per Sec: 11124, Lr: 0.000300\n", "2020-04-09 21:58:59,428 Epoch 4 Step: 23400 Batch Loss: 2.144667 Tokens per Sec: 10992, Lr: 0.000300\n", "2020-04-09 21:59:17,654 Epoch 4 Step: 23500 Batch Loss: 2.354735 Tokens per Sec: 11161, Lr: 0.000300\n", "2020-04-09 21:59:35,772 Epoch 4 Step: 23600 Batch Loss: 2.292526 Tokens per Sec: 10734, Lr: 0.000300\n", "2020-04-09 21:59:54,231 Epoch 4 Step: 23700 Batch Loss: 2.356517 Tokens per Sec: 11072, Lr: 0.000300\n", "2020-04-09 22:00:12,428 Epoch 4 Step: 23800 Batch Loss: 2.687940 Tokens per Sec: 10961, Lr: 0.000300\n", "2020-04-09 22:00:30,709 Epoch 4 Step: 23900 Batch Loss: 2.687732 Tokens per Sec: 10925, Lr: 0.000300\n", "2020-04-09 22:00:49,109 Epoch 4 Step: 24000 Batch Loss: 2.154740 Tokens per Sec: 11272, Lr: 0.000300\n", "2020-04-09 22:01:13,645 Hooray! New best validation result [ppl]!\n", "2020-04-09 22:01:13,646 Saving new checkpoint.\n", "2020-04-09 22:01:14,966 Example #0\n", "2020-04-09 22:01:14,966 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 22:01:14,967 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 22:01:14,967 \tHypothesis: See how We make clear signs that they may understand .\n", "2020-04-09 22:01:14,967 Example #1\n", "2020-04-09 22:01:14,967 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 22:01:14,967 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 22:01:14,968 \tHypothesis: If We willed , We would have brought it down water , so why do you not give thanks ?\n", "2020-04-09 22:01:14,968 Example #2\n", "2020-04-09 22:01:14,968 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 22:01:14,968 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 22:01:14,968 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 22:01:14,968 Example #3\n", "2020-04-09 22:01:14,969 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 22:01:14,969 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 22:01:14,969 \tHypothesis: He will send the ark in the crazing , and when some of his people passed away before him , they mocked him .\n", "2020-04-09 22:01:14,970 Validation result (greedy) at epoch 4, step 24000: bleu: 18.61, loss: 59333.5117, ppl: 8.9787, duration: 25.8604s\n", "2020-04-09 22:01:33,615 Epoch 4 Step: 24100 Batch Loss: 2.519805 Tokens per Sec: 10844, Lr: 0.000300\n", "2020-04-09 22:01:51,994 Epoch 4 Step: 24200 Batch Loss: 2.383247 Tokens per Sec: 11071, Lr: 0.000300\n", "2020-04-09 22:02:10,518 Epoch 4 Step: 24300 Batch Loss: 2.202720 Tokens per Sec: 11301, Lr: 0.000300\n", "2020-04-09 22:02:28,566 Epoch 4 Step: 24400 Batch Loss: 1.859843 Tokens per Sec: 10951, Lr: 0.000300\n", "2020-04-09 22:02:46,747 Epoch 4 Step: 24500 Batch Loss: 2.323406 Tokens per Sec: 10925, Lr: 0.000300\n", "2020-04-09 22:03:05,239 Epoch 4 Step: 24600 Batch Loss: 2.671881 Tokens per Sec: 11197, Lr: 0.000300\n", "2020-04-09 22:03:23,496 Epoch 4 Step: 24700 Batch Loss: 1.997586 Tokens per Sec: 10908, Lr: 0.000300\n", "2020-04-09 22:03:41,730 Epoch 4 Step: 24800 Batch Loss: 2.728327 Tokens per Sec: 11043, Lr: 0.000300\n", "2020-04-09 22:04:00,113 Epoch 4 Step: 24900 Batch Loss: 2.088190 Tokens per Sec: 11152, Lr: 0.000300\n", "2020-04-09 22:04:18,311 Epoch 4 Step: 25000 Batch Loss: 2.252918 Tokens per Sec: 10944, Lr: 0.000300\n", "2020-04-09 22:04:48,729 Hooray! New best validation result [ppl]!\n", "2020-04-09 22:04:48,730 Saving new checkpoint.\n", "2020-04-09 22:04:49,959 Example #0\n", "2020-04-09 22:04:49,960 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 22:04:49,960 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 22:04:49,960 \tHypothesis: See how We make the signs that they may understand .\n", "2020-04-09 22:04:49,960 Example #1\n", "2020-04-09 22:04:49,961 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 22:04:49,961 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 22:04:49,961 \tHypothesis: If We willed , We would have brought him down water , so why do you not give thanks ?\n", "2020-04-09 22:04:49,961 Example #2\n", "2020-04-09 22:04:49,961 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 22:04:49,962 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 22:04:49,962 \tHypothesis: Those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 22:04:49,962 Example #3\n", "2020-04-09 22:04:49,962 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 22:04:49,962 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 22:04:49,962 \tHypothesis: He will swear by the wild , and when the leaders of his people passed before him , they mocked him .\n", "2020-04-09 22:04:49,963 Validation result (greedy) at epoch 4, step 25000: bleu: 18.63, loss: 58789.0508, ppl: 8.7997, duration: 31.6513s\n", "2020-04-09 22:05:08,445 Epoch 4 Step: 25100 Batch Loss: 2.237121 Tokens per Sec: 10955, Lr: 0.000300\n", "2020-04-09 22:05:26,587 Epoch 4 Step: 25200 Batch Loss: 1.953338 Tokens per Sec: 10595, Lr: 0.000300\n", "2020-04-09 22:05:44,893 Epoch 4 Step: 25300 Batch Loss: 2.632457 Tokens per Sec: 11238, Lr: 0.000300\n", "2020-04-09 22:06:02,933 Epoch 4 Step: 25400 Batch Loss: 2.393584 Tokens per Sec: 10815, Lr: 0.000300\n", "2020-04-09 22:06:20,982 Epoch 4 Step: 25500 Batch Loss: 2.228080 Tokens per Sec: 10766, Lr: 0.000300\n", "2020-04-09 22:06:39,498 Epoch 4 Step: 25600 Batch Loss: 2.327147 Tokens per Sec: 11265, Lr: 0.000300\n", "2020-04-09 22:06:50,810 Epoch 4: total training loss 10366.13\n", "2020-04-09 22:06:50,810 EPOCH 5\n", "2020-04-09 22:06:57,872 Epoch 5 Step: 25700 Batch Loss: 2.271866 Tokens per Sec: 10315, Lr: 0.000300\n", "2020-04-09 22:07:16,046 Epoch 5 Step: 25800 Batch Loss: 2.448810 Tokens per Sec: 11142, Lr: 0.000300\n", "2020-04-09 22:07:34,188 Epoch 5 Step: 25900 Batch Loss: 2.329709 Tokens per Sec: 10571, Lr: 0.000300\n", "2020-04-09 22:07:52,443 Epoch 5 Step: 26000 Batch Loss: 2.480240 Tokens per Sec: 10962, Lr: 0.000300\n", "2020-04-09 22:08:20,088 Hooray! New best validation result [ppl]!\n", "2020-04-09 22:08:20,088 Saving new checkpoint.\n", "2020-04-09 22:08:21,293 Example #0\n", "2020-04-09 22:08:21,293 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 22:08:21,293 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 22:08:21,293 \tHypothesis: See how We make clear signs that they may understand .\n", "2020-04-09 22:08:21,294 Example #1\n", "2020-04-09 22:08:21,294 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 22:08:21,294 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 22:08:21,294 \tHypothesis: Had We willed , We would have brought him down water , so why do you not give thanks ?\n", "2020-04-09 22:08:21,295 Example #2\n", "2020-04-09 22:08:21,295 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 22:08:21,295 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 22:08:21,295 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 22:08:21,295 Example #3\n", "2020-04-09 22:08:21,296 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 22:08:21,296 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 22:08:21,296 \tHypothesis: He will send the ships in the land , and when the leaders of his people passed before him , they mocked him .\n", "2020-04-09 22:08:21,296 Validation result (greedy) at epoch 5, step 26000: bleu: 18.91, loss: 58786.1484, ppl: 8.7987, duration: 28.8523s\n", "2020-04-09 22:08:39,484 Epoch 5 Step: 26100 Batch Loss: 2.407242 Tokens per Sec: 10794, Lr: 0.000300\n", "2020-04-09 22:08:57,776 Epoch 5 Step: 26200 Batch Loss: 2.496454 Tokens per Sec: 11231, Lr: 0.000300\n", "2020-04-09 22:09:16,127 Epoch 5 Step: 26300 Batch Loss: 2.287539 Tokens per Sec: 11083, Lr: 0.000300\n", "2020-04-09 22:09:34,409 Epoch 5 Step: 26400 Batch Loss: 2.258843 Tokens per Sec: 11064, Lr: 0.000300\n", "2020-04-09 22:09:52,641 Epoch 5 Step: 26500 Batch Loss: 2.369582 Tokens per Sec: 10957, Lr: 0.000300\n", "2020-04-09 22:10:10,882 Epoch 5 Step: 26600 Batch Loss: 2.338110 Tokens per Sec: 11194, Lr: 0.000300\n", "2020-04-09 22:10:29,361 Epoch 5 Step: 26700 Batch Loss: 2.320955 Tokens per Sec: 11189, Lr: 0.000300\n", "2020-04-09 22:10:47,751 Epoch 5 Step: 26800 Batch Loss: 2.126230 Tokens per Sec: 11145, Lr: 0.000300\n", "2020-04-09 22:11:05,763 Epoch 5 Step: 26900 Batch Loss: 2.393279 Tokens per Sec: 10860, Lr: 0.000300\n", "2020-04-09 22:11:24,252 Epoch 5 Step: 27000 Batch Loss: 2.503567 Tokens per Sec: 11096, Lr: 0.000300\n", "2020-04-09 22:11:49,379 Hooray! New best validation result [ppl]!\n", "2020-04-09 22:11:49,379 Saving new checkpoint.\n", "2020-04-09 22:11:50,656 Example #0\n", "2020-04-09 22:11:50,656 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 22:11:50,657 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 22:11:50,657 \tHypothesis: See how We make clear signs that they may understand .\n", "2020-04-09 22:11:50,657 Example #1\n", "2020-04-09 22:11:50,658 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 22:11:50,658 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 22:11:50,659 \tHypothesis: Had We willed , We would have brought him a rain , so why do you not give thanks ?\n", "2020-04-09 22:11:50,659 Example #2\n", "2020-04-09 22:11:50,660 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 22:11:50,660 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 22:11:50,660 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 22:11:50,660 Example #3\n", "2020-04-09 22:11:50,660 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 22:11:50,660 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 22:11:50,661 \tHypothesis: He will send the ark in the black , and when some of his people passed away before him , they mocked him .\n", "2020-04-09 22:11:50,661 Validation result (greedy) at epoch 5, step 27000: bleu: 19.12, loss: 58117.3750, ppl: 8.5837, duration: 26.4078s\n", "2020-04-09 22:12:09,350 Epoch 5 Step: 27100 Batch Loss: 2.109674 Tokens per Sec: 10485, Lr: 0.000300\n", "2020-04-09 22:12:27,605 Epoch 5 Step: 27200 Batch Loss: 2.446875 Tokens per Sec: 10990, Lr: 0.000300\n", "2020-04-09 22:12:45,963 Epoch 5 Step: 27300 Batch Loss: 2.244982 Tokens per Sec: 11056, Lr: 0.000300\n", "2020-04-09 22:13:04,290 Epoch 5 Step: 27400 Batch Loss: 1.931579 Tokens per Sec: 11271, Lr: 0.000300\n", "2020-04-09 22:13:22,524 Epoch 5 Step: 27500 Batch Loss: 2.396963 Tokens per Sec: 10849, Lr: 0.000300\n", "2020-04-09 22:13:40,550 Epoch 5 Step: 27600 Batch Loss: 2.195024 Tokens per Sec: 10759, Lr: 0.000300\n", "2020-04-09 22:13:58,680 Epoch 5 Step: 27700 Batch Loss: 2.036140 Tokens per Sec: 10836, Lr: 0.000300\n", "2020-04-09 22:14:16,834 Epoch 5 Step: 27800 Batch Loss: 2.053679 Tokens per Sec: 10900, Lr: 0.000300\n", "2020-04-09 22:14:35,135 Epoch 5 Step: 27900 Batch Loss: 2.214127 Tokens per Sec: 11198, Lr: 0.000300\n", "2020-04-09 22:14:53,328 Epoch 5 Step: 28000 Batch Loss: 2.677039 Tokens per Sec: 10859, Lr: 0.000300\n", "2020-04-09 22:15:27,753 Hooray! New best validation result [ppl]!\n", "2020-04-09 22:15:27,753 Saving new checkpoint.\n", "2020-04-09 22:15:28,979 Example #0\n", "2020-04-09 22:15:28,980 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 22:15:28,980 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 22:15:28,980 \tHypothesis: See how We make the signs clear , so that they may understand .\n", "2020-04-09 22:15:28,980 Example #1\n", "2020-04-09 22:15:28,981 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 22:15:28,981 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 22:15:28,981 \tHypothesis: Had We willed , We would have brought him down water , so why do you not give thanks ?\n", "2020-04-09 22:15:28,981 Example #2\n", "2020-04-09 22:15:28,981 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 22:15:28,981 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 22:15:28,982 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 22:15:28,982 Example #3\n", "2020-04-09 22:15:28,982 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 22:15:28,982 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 22:15:28,982 \tHypothesis: He will send the ships in the grain , and when some of his people passed before him , they mocked him .\n", "2020-04-09 22:15:28,982 Validation result (greedy) at epoch 5, step 28000: bleu: 19.07, loss: 57682.4297, ppl: 8.4467, duration: 35.6536s\n", "2020-04-09 22:15:47,609 Epoch 5 Step: 28100 Batch Loss: 2.272454 Tokens per Sec: 11116, Lr: 0.000300\n", "2020-04-09 22:16:05,989 Epoch 5 Step: 28200 Batch Loss: 1.727880 Tokens per Sec: 10906, Lr: 0.000300\n", "2020-04-09 22:16:24,027 Epoch 5 Step: 28300 Batch Loss: 2.099026 Tokens per Sec: 10915, Lr: 0.000300\n", "2020-04-09 22:16:42,149 Epoch 5 Step: 28400 Batch Loss: 2.366949 Tokens per Sec: 10869, Lr: 0.000300\n", "2020-04-09 22:17:00,318 Epoch 5 Step: 28500 Batch Loss: 2.268248 Tokens per Sec: 11012, Lr: 0.000300\n", "2020-04-09 22:17:18,871 Epoch 5 Step: 28600 Batch Loss: 2.087840 Tokens per Sec: 11026, Lr: 0.000300\n", "2020-04-09 22:17:37,197 Epoch 5 Step: 28700 Batch Loss: 2.376420 Tokens per Sec: 11045, Lr: 0.000300\n", "2020-04-09 22:17:55,449 Epoch 5 Step: 28800 Batch Loss: 2.276690 Tokens per Sec: 11024, Lr: 0.000300\n", "2020-04-09 22:18:13,722 Epoch 5 Step: 28900 Batch Loss: 2.170886 Tokens per Sec: 11211, Lr: 0.000300\n", "2020-04-09 22:18:31,986 Epoch 5 Step: 29000 Batch Loss: 2.293583 Tokens per Sec: 10928, Lr: 0.000300\n", "2020-04-09 22:18:57,989 Hooray! New best validation result [ppl]!\n", "2020-04-09 22:18:57,989 Saving new checkpoint.\n", "2020-04-09 22:18:59,191 Example #0\n", "2020-04-09 22:18:59,191 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 22:18:59,191 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 22:18:59,191 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-09 22:18:59,192 Example #1\n", "2020-04-09 22:18:59,192 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 22:18:59,192 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 22:18:59,192 \tHypothesis: Had We willed , We would have brought it down to the rain , so why do you not give thanks ?\n", "2020-04-09 22:18:59,193 Example #2\n", "2020-04-09 22:18:59,193 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 22:18:59,193 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 22:18:59,193 \tHypothesis: Those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 22:18:59,193 Example #3\n", "2020-04-09 22:18:59,194 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 22:18:59,194 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 22:18:59,194 \tHypothesis: He will send the ship in the graves , and when some of his people passed before him , they mocked him .\n", "2020-04-09 22:18:59,194 Validation result (greedy) at epoch 5, step 29000: bleu: 19.28, loss: 57607.8828, ppl: 8.4235, duration: 27.2074s\n", "2020-04-09 22:19:17,392 Epoch 5 Step: 29100 Batch Loss: 2.160475 Tokens per Sec: 10751, Lr: 0.000300\n", "2020-04-09 22:19:35,660 Epoch 5 Step: 29200 Batch Loss: 2.120973 Tokens per Sec: 11135, Lr: 0.000300\n", "2020-04-09 22:19:53,992 Epoch 5 Step: 29300 Batch Loss: 2.308949 Tokens per Sec: 10901, Lr: 0.000300\n", "2020-04-09 22:20:12,098 Epoch 5 Step: 29400 Batch Loss: 2.203633 Tokens per Sec: 10806, Lr: 0.000300\n", "2020-04-09 22:20:30,189 Epoch 5 Step: 29500 Batch Loss: 2.124311 Tokens per Sec: 10898, Lr: 0.000300\n", "2020-04-09 22:20:48,549 Epoch 5 Step: 29600 Batch Loss: 2.330023 Tokens per Sec: 10780, Lr: 0.000300\n", "2020-04-09 22:21:06,604 Epoch 5 Step: 29700 Batch Loss: 2.142286 Tokens per Sec: 10872, Lr: 0.000300\n", "2020-04-09 22:21:24,948 Epoch 5 Step: 29800 Batch Loss: 2.385060 Tokens per Sec: 11036, Lr: 0.000300\n", "2020-04-09 22:21:43,180 Epoch 5 Step: 29900 Batch Loss: 2.357154 Tokens per Sec: 10891, Lr: 0.000300\n", "2020-04-09 22:22:01,539 Epoch 5 Step: 30000 Batch Loss: 1.879645 Tokens per Sec: 11107, Lr: 0.000300\n", "2020-04-09 22:22:30,442 Hooray! New best validation result [ppl]!\n", "2020-04-09 22:22:30,442 Saving new checkpoint.\n", "2020-04-09 22:22:31,679 Example #0\n", "2020-04-09 22:22:31,680 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 22:22:31,680 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 22:22:31,680 \tHypothesis: See how We make clear signs so that they may understand .\n", "2020-04-09 22:22:31,680 Example #1\n", "2020-04-09 22:22:31,681 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 22:22:31,681 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 22:22:31,681 \tHypothesis: Had We willed , We would have brought it down water , so why do you not give thanks ?\n", "2020-04-09 22:22:31,681 Example #2\n", "2020-04-09 22:22:31,682 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 22:22:31,682 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 22:22:31,682 \tHypothesis: And those who disbelieve in Our revelations are the people of the nations .\n", "2020-04-09 22:22:31,682 Example #3\n", "2020-04-09 22:22:31,683 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 22:22:31,683 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 22:22:31,683 \tHypothesis: And He will send the ark in the cup , and when the leaders of his people passed before him , they mock him .\n", "2020-04-09 22:22:31,683 Validation result (greedy) at epoch 5, step 30000: bleu: 19.90, loss: 57198.8086, ppl: 8.2970, duration: 30.1432s\n", "2020-04-09 22:22:48,080 Epoch 5: total training loss 10038.62\n", "2020-04-09 22:22:48,082 EPOCH 6\n", "2020-04-09 22:22:50,941 Epoch 6 Step: 30100 Batch Loss: 1.769536 Tokens per Sec: 10001, Lr: 0.000300\n", "2020-04-09 22:23:09,143 Epoch 6 Step: 30200 Batch Loss: 2.351325 Tokens per Sec: 11011, Lr: 0.000300\n", "2020-04-09 22:23:27,391 Epoch 6 Step: 30300 Batch Loss: 2.018995 Tokens per Sec: 10847, Lr: 0.000300\n", "2020-04-09 22:23:45,655 Epoch 6 Step: 30400 Batch Loss: 2.199352 Tokens per Sec: 11064, Lr: 0.000300\n", "2020-04-09 22:24:03,753 Epoch 6 Step: 30500 Batch Loss: 2.107277 Tokens per Sec: 10887, Lr: 0.000300\n", "2020-04-09 22:24:21,993 Epoch 6 Step: 30600 Batch Loss: 2.602820 Tokens per Sec: 11014, Lr: 0.000300\n", "2020-04-09 22:24:40,306 Epoch 6 Step: 30700 Batch Loss: 2.155402 Tokens per Sec: 10893, Lr: 0.000300\n", "2020-04-09 22:24:58,765 Epoch 6 Step: 30800 Batch Loss: 2.058830 Tokens per Sec: 11194, Lr: 0.000300\n", "2020-04-09 22:25:17,262 Epoch 6 Step: 30900 Batch Loss: 2.370862 Tokens per Sec: 11320, Lr: 0.000300\n", "2020-04-09 22:25:35,222 Epoch 6 Step: 31000 Batch Loss: 2.120316 Tokens per Sec: 10798, Lr: 0.000300\n", "2020-04-09 22:26:03,527 Hooray! New best validation result [ppl]!\n", "2020-04-09 22:26:03,528 Saving new checkpoint.\n", "2020-04-09 22:26:04,733 Example #0\n", "2020-04-09 22:26:04,734 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 22:26:04,734 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 22:26:04,734 \tHypothesis: See how We make clear signs that haply they may understand .\n", "2020-04-09 22:26:04,735 Example #1\n", "2020-04-09 22:26:04,735 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 22:26:04,735 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 22:26:04,735 \tHypothesis: If We willed , We would have brought him down to the water , so why do you not give thanks ?\n", "2020-04-09 22:26:04,735 Example #2\n", "2020-04-09 22:26:04,736 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 22:26:04,736 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 22:26:04,736 \tHypothesis: Those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 22:26:04,736 Example #3\n", "2020-04-09 22:26:04,736 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 22:26:04,737 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 22:26:04,737 \tHypothesis: He will send the ships in the bones , and when the leaders of his people passed away before him , they mock him .\n", "2020-04-09 22:26:04,737 Validation result (greedy) at epoch 6, step 31000: bleu: 19.62, loss: 57024.4297, ppl: 8.2436, duration: 29.5146s\n", "2020-04-09 22:26:23,161 Epoch 6 Step: 31100 Batch Loss: 2.066576 Tokens per Sec: 10862, Lr: 0.000300\n", "2020-04-09 22:26:41,247 Epoch 6 Step: 31200 Batch Loss: 2.277808 Tokens per Sec: 10708, Lr: 0.000300\n", "2020-04-09 22:26:59,359 Epoch 6 Step: 31300 Batch Loss: 2.233737 Tokens per Sec: 10886, Lr: 0.000300\n", "2020-04-09 22:27:17,552 Epoch 6 Step: 31400 Batch Loss: 2.222401 Tokens per Sec: 11055, Lr: 0.000300\n", "2020-04-09 22:27:35,679 Epoch 6 Step: 31500 Batch Loss: 1.836462 Tokens per Sec: 10891, Lr: 0.000300\n", "2020-04-09 22:27:54,048 Epoch 6 Step: 31600 Batch Loss: 2.044497 Tokens per Sec: 11083, Lr: 0.000300\n", "2020-04-09 22:28:12,145 Epoch 6 Step: 31700 Batch Loss: 2.310050 Tokens per Sec: 10698, Lr: 0.000300\n", "2020-04-09 22:28:30,507 Epoch 6 Step: 31800 Batch Loss: 2.326565 Tokens per Sec: 11117, Lr: 0.000300\n", "2020-04-09 22:28:48,677 Epoch 6 Step: 31900 Batch Loss: 2.002881 Tokens per Sec: 11191, Lr: 0.000300\n", "2020-04-09 22:29:06,987 Epoch 6 Step: 32000 Batch Loss: 2.135514 Tokens per Sec: 10778, Lr: 0.000300\n", "2020-04-09 22:29:36,780 Hooray! New best validation result [ppl]!\n", "2020-04-09 22:29:36,780 Saving new checkpoint.\n", "2020-04-09 22:29:38,145 Example #0\n", "2020-04-09 22:29:38,146 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 22:29:38,146 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 22:29:38,146 \tHypothesis: See how We make clear signs so that they may understand .\n", "2020-04-09 22:29:38,146 Example #1\n", "2020-04-09 22:29:38,147 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 22:29:38,147 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 22:29:38,147 \tHypothesis: If We willed , We would have removed him with water , so why do you not give thanks ?\n", "2020-04-09 22:29:38,147 Example #2\n", "2020-04-09 22:29:38,147 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 22:29:38,148 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 22:29:38,148 \tHypothesis: Those who disbelieve in Our signs are the people of the wicked .\n", "2020-04-09 22:29:38,148 Example #3\n", "2020-04-09 22:29:38,148 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 22:29:38,149 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 22:29:38,149 \tHypothesis: He will send the ships in the bones , and when the leaders of his people have passed before him , they mock him .\n", "2020-04-09 22:29:38,149 Validation result (greedy) at epoch 6, step 32000: bleu: 20.13, loss: 56367.0625, ppl: 8.0456, duration: 31.1619s\n", "2020-04-09 22:29:56,466 Epoch 6 Step: 32100 Batch Loss: 2.326706 Tokens per Sec: 10856, Lr: 0.000300\n", "2020-04-09 22:30:14,692 Epoch 6 Step: 32200 Batch Loss: 2.193179 Tokens per Sec: 10950, Lr: 0.000300\n", "2020-04-09 22:30:32,978 Epoch 6 Step: 32300 Batch Loss: 2.119092 Tokens per Sec: 11020, Lr: 0.000300\n", "2020-04-09 22:30:51,077 Epoch 6 Step: 32400 Batch Loss: 2.157750 Tokens per Sec: 10905, Lr: 0.000300\n", "2020-04-09 22:31:09,377 Epoch 6 Step: 32500 Batch Loss: 2.269197 Tokens per Sec: 10996, Lr: 0.000300\n", "2020-04-09 22:31:27,757 Epoch 6 Step: 32600 Batch Loss: 2.242722 Tokens per Sec: 11062, Lr: 0.000300\n", "2020-04-09 22:31:46,007 Epoch 6 Step: 32700 Batch Loss: 2.145940 Tokens per Sec: 11131, Lr: 0.000300\n", "2020-04-09 22:32:04,430 Epoch 6 Step: 32800 Batch Loss: 2.039839 Tokens per Sec: 10990, Lr: 0.000300\n", "2020-04-09 22:32:22,789 Epoch 6 Step: 32900 Batch Loss: 2.218368 Tokens per Sec: 11068, Lr: 0.000300\n", "2020-04-09 22:32:40,690 Epoch 6 Step: 33000 Batch Loss: 2.176511 Tokens per Sec: 10785, Lr: 0.000300\n", "2020-04-09 22:33:07,464 Example #0\n", "2020-04-09 22:33:07,464 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 22:33:07,465 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 22:33:07,465 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-09 22:33:07,465 Example #1\n", "2020-04-09 22:33:07,465 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 22:33:07,466 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 22:33:07,466 \tHypothesis: Had We willed , We would have brought it down , so why do you not give thanks ?\n", "2020-04-09 22:33:07,466 Example #2\n", "2020-04-09 22:33:07,466 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 22:33:07,467 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 22:33:07,467 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 22:33:07,467 Example #3\n", "2020-04-09 22:33:07,467 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 22:33:07,467 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 22:33:07,468 \tHypothesis: He will send the ark in the security , and when some of his people passed before him , they mock him .\n", "2020-04-09 22:33:07,468 Validation result (greedy) at epoch 6, step 33000: bleu: 19.51, loss: 56370.4180, ppl: 8.0466, duration: 26.7770s\n", "2020-04-09 22:33:25,567 Epoch 6 Step: 33100 Batch Loss: 2.295742 Tokens per Sec: 11089, Lr: 0.000300\n", "2020-04-09 22:33:43,892 Epoch 6 Step: 33200 Batch Loss: 2.323487 Tokens per Sec: 11003, Lr: 0.000300\n", "2020-04-09 22:34:02,365 Epoch 6 Step: 33300 Batch Loss: 2.269289 Tokens per Sec: 11382, Lr: 0.000300\n", "2020-04-09 22:34:20,551 Epoch 6 Step: 33400 Batch Loss: 2.187207 Tokens per Sec: 10966, Lr: 0.000300\n", "2020-04-09 22:34:39,013 Epoch 6 Step: 33500 Batch Loss: 2.300862 Tokens per Sec: 11352, Lr: 0.000300\n", "2020-04-09 22:34:57,270 Epoch 6 Step: 33600 Batch Loss: 2.210931 Tokens per Sec: 10861, Lr: 0.000300\n", "2020-04-09 22:35:15,724 Epoch 6 Step: 33700 Batch Loss: 2.247809 Tokens per Sec: 11097, Lr: 0.000300\n", "2020-04-09 22:35:34,026 Epoch 6 Step: 33800 Batch Loss: 2.042024 Tokens per Sec: 10956, Lr: 0.000300\n", "2020-04-09 22:35:52,328 Epoch 6 Step: 33900 Batch Loss: 2.444053 Tokens per Sec: 11090, Lr: 0.000300\n", "2020-04-09 22:36:10,561 Epoch 6 Step: 34000 Batch Loss: 2.443292 Tokens per Sec: 10999, Lr: 0.000300\n", "2020-04-09 22:36:37,376 Hooray! New best validation result [ppl]!\n", "2020-04-09 22:36:37,376 Saving new checkpoint.\n", "2020-04-09 22:36:38,557 Example #0\n", "2020-04-09 22:36:38,557 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 22:36:38,557 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 22:36:38,558 \tHypothesis: See how We make clear signs that they may understand .\n", "2020-04-09 22:36:38,558 Example #1\n", "2020-04-09 22:36:38,558 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 22:36:38,558 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 22:36:38,558 \tHypothesis: If We willed , We would have brought him down water , why do you not give thanks ?\n", "2020-04-09 22:36:38,558 Example #2\n", "2020-04-09 22:36:38,559 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 22:36:38,559 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 22:36:38,559 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 22:36:38,559 Example #3\n", "2020-04-09 22:36:38,560 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 22:36:38,560 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 22:36:38,560 \tHypothesis: He is going to swear in the security , and when the leaders of his people passed before him , they mock him .\n", "2020-04-09 22:36:38,560 Validation result (greedy) at epoch 6, step 34000: bleu: 20.57, loss: 56096.9102, ppl: 7.9656, duration: 27.9987s\n", "2020-04-09 22:36:56,940 Epoch 6 Step: 34100 Batch Loss: 2.276025 Tokens per Sec: 10775, Lr: 0.000300\n", "2020-04-09 22:37:15,428 Epoch 6 Step: 34200 Batch Loss: 2.242177 Tokens per Sec: 11203, Lr: 0.000300\n", "2020-04-09 22:37:33,790 Epoch 6 Step: 34300 Batch Loss: 2.284995 Tokens per Sec: 10965, Lr: 0.000300\n", "2020-04-09 22:37:51,981 Epoch 6 Step: 34400 Batch Loss: 2.292037 Tokens per Sec: 10958, Lr: 0.000300\n", "2020-04-09 22:38:10,219 Epoch 6 Step: 34500 Batch Loss: 2.542608 Tokens per Sec: 10880, Lr: 0.000300\n", "2020-04-09 22:38:10,441 Epoch 6: total training loss 9753.59\n", "2020-04-09 22:38:10,441 EPOCH 7\n", "2020-04-09 22:38:28,847 Epoch 7 Step: 34600 Batch Loss: 1.652930 Tokens per Sec: 10653, Lr: 0.000300\n", "2020-04-09 22:38:47,140 Epoch 7 Step: 34700 Batch Loss: 2.251304 Tokens per Sec: 11250, Lr: 0.000300\n", "2020-04-09 22:39:05,410 Epoch 7 Step: 34800 Batch Loss: 2.240629 Tokens per Sec: 11098, Lr: 0.000300\n", "2020-04-09 22:39:23,652 Epoch 7 Step: 34900 Batch Loss: 2.118052 Tokens per Sec: 10960, Lr: 0.000300\n", "2020-04-09 22:39:41,824 Epoch 7 Step: 35000 Batch Loss: 1.942208 Tokens per Sec: 10852, Lr: 0.000300\n", "2020-04-09 22:40:11,082 Hooray! New best validation result [ppl]!\n", "2020-04-09 22:40:11,082 Saving new checkpoint.\n", "2020-04-09 22:40:12,289 Example #0\n", "2020-04-09 22:40:12,289 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 22:40:12,289 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 22:40:12,289 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-09 22:40:12,290 Example #1\n", "2020-04-09 22:40:12,290 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 22:40:12,290 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 22:40:12,290 \tHypothesis: Had We willed , We would have brought him down water , so why do you not give thanks ?\n", "2020-04-09 22:40:12,291 Example #2\n", "2020-04-09 22:40:12,291 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 22:40:12,291 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 22:40:12,291 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 22:40:12,291 Example #3\n", "2020-04-09 22:40:12,292 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 22:40:12,292 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 22:40:12,292 \tHypothesis: And He sends the ships in the secure , and when some of his people passed before him , they mocked him .\n", "2020-04-09 22:40:12,292 Validation result (greedy) at epoch 7, step 35000: bleu: 20.94, loss: 55550.5625, ppl: 7.8062, duration: 30.4684s\n", "2020-04-09 22:40:30,635 Epoch 7 Step: 35100 Batch Loss: 2.243020 Tokens per Sec: 11049, Lr: 0.000300\n", "2020-04-09 22:40:48,800 Epoch 7 Step: 35200 Batch Loss: 1.575834 Tokens per Sec: 10774, Lr: 0.000300\n", "2020-04-09 22:41:06,906 Epoch 7 Step: 35300 Batch Loss: 1.838157 Tokens per Sec: 10849, Lr: 0.000300\n", "2020-04-09 22:41:25,271 Epoch 7 Step: 35400 Batch Loss: 2.306783 Tokens per Sec: 11141, Lr: 0.000300\n", "2020-04-09 22:41:43,542 Epoch 7 Step: 35500 Batch Loss: 2.136308 Tokens per Sec: 11001, Lr: 0.000300\n", "2020-04-09 22:42:01,729 Epoch 7 Step: 35600 Batch Loss: 1.842452 Tokens per Sec: 10960, Lr: 0.000300\n", "2020-04-09 22:42:20,041 Epoch 7 Step: 35700 Batch Loss: 2.477145 Tokens per Sec: 11092, Lr: 0.000300\n", "2020-04-09 22:42:38,243 Epoch 7 Step: 35800 Batch Loss: 2.298496 Tokens per Sec: 10961, Lr: 0.000300\n", "2020-04-09 22:42:56,591 Epoch 7 Step: 35900 Batch Loss: 2.072713 Tokens per Sec: 11077, Lr: 0.000300\n", "2020-04-09 22:43:14,722 Epoch 7 Step: 36000 Batch Loss: 2.111398 Tokens per Sec: 11030, Lr: 0.000300\n", "2020-04-09 22:43:44,824 Hooray! New best validation result [ppl]!\n", "2020-04-09 22:43:44,825 Saving new checkpoint.\n", "2020-04-09 22:43:46,064 Example #0\n", "2020-04-09 22:43:46,064 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 22:43:46,065 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 22:43:46,065 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-09 22:43:46,065 Example #1\n", "2020-04-09 22:43:46,065 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 22:43:46,065 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 22:43:46,066 \tHypothesis: If We willed , We would have brought him water water , so why do you not give thanks ?\n", "2020-04-09 22:43:46,066 Example #2\n", "2020-04-09 22:43:46,066 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 22:43:46,066 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 22:43:46,067 \tHypothesis: And those who disbelieve in Our revelations are the people of the communities .\n", "2020-04-09 22:43:46,067 Example #3\n", "2020-04-09 22:43:46,067 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 22:43:46,067 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 22:43:46,067 \tHypothesis: He sends the ships in the bones , and when some of his people passed away before him , they mocked him .\n", "2020-04-09 22:43:46,068 Validation result (greedy) at epoch 7, step 36000: bleu: 20.96, loss: 55508.8086, ppl: 7.7941, duration: 31.3451s\n", "2020-04-09 22:44:04,594 Epoch 7 Step: 36100 Batch Loss: 2.040615 Tokens per Sec: 10926, Lr: 0.000300\n", "2020-04-09 22:44:22,876 Epoch 7 Step: 36200 Batch Loss: 2.191253 Tokens per Sec: 11060, Lr: 0.000300\n", "2020-04-09 22:44:41,186 Epoch 7 Step: 36300 Batch Loss: 2.104839 Tokens per Sec: 10867, Lr: 0.000300\n", "2020-04-09 22:44:59,255 Epoch 7 Step: 36400 Batch Loss: 2.176226 Tokens per Sec: 10902, Lr: 0.000300\n", "2020-04-09 22:45:17,982 Epoch 7 Step: 36500 Batch Loss: 2.356600 Tokens per Sec: 11402, Lr: 0.000300\n", "2020-04-09 22:45:36,381 Epoch 7 Step: 36600 Batch Loss: 2.165979 Tokens per Sec: 11016, Lr: 0.000300\n", "2020-04-09 22:45:54,454 Epoch 7 Step: 36700 Batch Loss: 2.136996 Tokens per Sec: 10883, Lr: 0.000300\n", "2020-04-09 22:46:12,657 Epoch 7 Step: 36800 Batch Loss: 2.139226 Tokens per Sec: 10866, Lr: 0.000300\n", "2020-04-09 22:46:30,800 Epoch 7 Step: 36900 Batch Loss: 1.972817 Tokens per Sec: 10923, Lr: 0.000300\n", "2020-04-09 22:46:49,085 Epoch 7 Step: 37000 Batch Loss: 2.281073 Tokens per Sec: 11226, Lr: 0.000300\n", "2020-04-09 22:47:16,468 Hooray! New best validation result [ppl]!\n", "2020-04-09 22:47:16,469 Saving new checkpoint.\n", "2020-04-09 22:47:17,673 Example #0\n", "2020-04-09 22:47:17,674 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 22:47:17,674 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 22:47:17,674 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-09 22:47:17,674 Example #1\n", "2020-04-09 22:47:17,675 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 22:47:17,675 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 22:47:17,675 \tHypothesis: If We willed , We would have brought it down , so why do you not give thanks ?\n", "2020-04-09 22:47:17,675 Example #2\n", "2020-04-09 22:47:17,676 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 22:47:17,677 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 22:47:17,677 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 22:47:17,677 Example #3\n", "2020-04-09 22:47:17,677 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 22:47:17,678 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 22:47:17,678 \tHypothesis: He sends the ships in peace , and when some of his people passed before him , they mocked him .\n", "2020-04-09 22:47:17,678 Validation result (greedy) at epoch 7, step 37000: bleu: 20.84, loss: 55001.6914, ppl: 7.6493, duration: 28.5917s\n", "2020-04-09 22:47:36,144 Epoch 7 Step: 37100 Batch Loss: 2.136457 Tokens per Sec: 10952, Lr: 0.000300\n", "2020-04-09 22:47:54,450 Epoch 7 Step: 37200 Batch Loss: 2.298306 Tokens per Sec: 10887, Lr: 0.000300\n", "2020-04-09 22:48:12,901 Epoch 7 Step: 37300 Batch Loss: 2.054666 Tokens per Sec: 11155, Lr: 0.000300\n", "2020-04-09 22:48:31,129 Epoch 7 Step: 37400 Batch Loss: 2.227205 Tokens per Sec: 10853, Lr: 0.000300\n", "2020-04-09 22:48:49,341 Epoch 7 Step: 37500 Batch Loss: 2.201409 Tokens per Sec: 11144, Lr: 0.000300\n", "2020-04-09 22:49:07,735 Epoch 7 Step: 37600 Batch Loss: 2.358883 Tokens per Sec: 11230, Lr: 0.000300\n", "2020-04-09 22:49:25,839 Epoch 7 Step: 37700 Batch Loss: 2.324135 Tokens per Sec: 10885, Lr: 0.000300\n", "2020-04-09 22:49:44,130 Epoch 7 Step: 37800 Batch Loss: 2.272326 Tokens per Sec: 11006, Lr: 0.000300\n", "2020-04-09 22:50:02,450 Epoch 7 Step: 37900 Batch Loss: 2.176739 Tokens per Sec: 10953, Lr: 0.000300\n", "2020-04-09 22:50:20,704 Epoch 7 Step: 38000 Batch Loss: 2.219389 Tokens per Sec: 10926, Lr: 0.000300\n", "2020-04-09 22:50:49,197 Hooray! New best validation result [ppl]!\n", "2020-04-09 22:50:49,197 Saving new checkpoint.\n", "2020-04-09 22:50:50,459 Example #0\n", "2020-04-09 22:50:50,459 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 22:50:50,460 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 22:50:50,460 \tHypothesis: See how We explain the signs that they may understand . \"\n", "2020-04-09 22:50:50,460 Example #1\n", "2020-04-09 22:50:50,460 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 22:50:50,460 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 22:50:50,460 \tHypothesis: Had We willed , We could have brought it down to the rain , so why do you not give thanks ?\n", "2020-04-09 22:50:50,461 Example #2\n", "2020-04-09 22:50:50,461 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 22:50:50,461 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 22:50:50,461 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 22:50:50,461 Example #3\n", "2020-04-09 22:50:50,462 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 22:50:50,462 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 22:50:50,462 \tHypothesis: He sends the ships in security , and when the leaders of his people passed before him , they mocked him .\n", "2020-04-09 22:50:50,462 Validation result (greedy) at epoch 7, step 38000: bleu: 20.68, loss: 54860.3984, ppl: 7.6094, duration: 29.7582s\n", "2020-04-09 22:51:08,859 Epoch 7 Step: 38100 Batch Loss: 2.198525 Tokens per Sec: 10884, Lr: 0.000300\n", "2020-04-09 22:51:27,093 Epoch 7 Step: 38200 Batch Loss: 2.103558 Tokens per Sec: 10843, Lr: 0.000300\n", "2020-04-09 22:51:45,166 Epoch 7 Step: 38300 Batch Loss: 2.412749 Tokens per Sec: 10590, Lr: 0.000300\n", "2020-04-09 22:52:03,530 Epoch 7 Step: 38400 Batch Loss: 2.183279 Tokens per Sec: 11193, Lr: 0.000300\n", "2020-04-09 22:52:21,873 Epoch 7 Step: 38500 Batch Loss: 2.152642 Tokens per Sec: 11136, Lr: 0.000300\n", "2020-04-09 22:52:39,986 Epoch 7 Step: 38600 Batch Loss: 2.213567 Tokens per Sec: 10736, Lr: 0.000300\n", "2020-04-09 22:52:58,337 Epoch 7 Step: 38700 Batch Loss: 1.519967 Tokens per Sec: 11022, Lr: 0.000300\n", "2020-04-09 22:53:16,403 Epoch 7 Step: 38800 Batch Loss: 1.959008 Tokens per Sec: 10892, Lr: 0.000300\n", "2020-04-09 22:53:34,478 Epoch 7 Step: 38900 Batch Loss: 2.192921 Tokens per Sec: 10606, Lr: 0.000300\n", "2020-04-09 22:53:38,765 Epoch 7: total training loss 9552.65\n", "2020-04-09 22:53:38,765 EPOCH 8\n", "2020-04-09 22:53:52,788 Epoch 8 Step: 39000 Batch Loss: 2.122045 Tokens per Sec: 10672, Lr: 0.000300\n", "2020-04-09 22:54:20,487 Example #0\n", "2020-04-09 22:54:20,487 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 22:54:20,488 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 22:54:20,488 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-09 22:54:20,488 Example #1\n", "2020-04-09 22:54:20,488 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 22:54:20,489 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 22:54:20,489 \tHypothesis: If We willed , We would have brought it down water , so why do you not give thanks ?\n", "2020-04-09 22:54:20,489 Example #2\n", "2020-04-09 22:54:20,489 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 22:54:20,489 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 22:54:20,490 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 22:54:20,490 Example #3\n", "2020-04-09 22:54:20,490 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 22:54:20,490 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 22:54:20,491 \tHypothesis: And He sends the ships in the security , and when some of his people passed before him , they mocked him .\n", "2020-04-09 22:54:20,491 Validation result (greedy) at epoch 8, step 39000: bleu: 20.69, loss: 54868.7031, ppl: 7.6118, duration: 27.7021s\n", "2020-04-09 22:54:38,859 Epoch 8 Step: 39100 Batch Loss: 2.123837 Tokens per Sec: 11052, Lr: 0.000300\n", "2020-04-09 22:54:57,162 Epoch 8 Step: 39200 Batch Loss: 2.283272 Tokens per Sec: 10709, Lr: 0.000300\n", "2020-04-09 22:55:15,264 Epoch 8 Step: 39300 Batch Loss: 1.849268 Tokens per Sec: 11033, Lr: 0.000300\n", "2020-04-09 22:55:33,391 Epoch 8 Step: 39400 Batch Loss: 1.751283 Tokens per Sec: 11027, Lr: 0.000300\n", "2020-04-09 22:55:51,302 Epoch 8 Step: 39500 Batch Loss: 1.850424 Tokens per Sec: 10742, Lr: 0.000300\n", "2020-04-09 22:56:09,434 Epoch 8 Step: 39600 Batch Loss: 2.273306 Tokens per Sec: 11002, Lr: 0.000300\n", "2020-04-09 22:56:27,881 Epoch 8 Step: 39700 Batch Loss: 2.199118 Tokens per Sec: 11114, Lr: 0.000300\n", "2020-04-09 22:56:46,131 Epoch 8 Step: 39800 Batch Loss: 2.045624 Tokens per Sec: 11046, Lr: 0.000300\n", "2020-04-09 22:57:04,076 Epoch 8 Step: 39900 Batch Loss: 2.058761 Tokens per Sec: 10621, Lr: 0.000300\n", "2020-04-09 22:57:22,348 Epoch 8 Step: 40000 Batch Loss: 2.125954 Tokens per Sec: 10939, Lr: 0.000300\n", "2020-04-09 22:57:55,923 Hooray! New best validation result [ppl]!\n", "2020-04-09 22:57:55,924 Saving new checkpoint.\n", "2020-04-09 22:57:57,569 Example #0\n", "2020-04-09 22:57:57,570 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 22:57:57,570 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 22:57:57,570 \tHypothesis: See how We make clear signs that they may understand .\n", "2020-04-09 22:57:57,570 Example #1\n", "2020-04-09 22:57:57,571 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 22:57:57,571 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 22:57:57,571 \tHypothesis: If We willed , We would have brought it down , so why do you not give thanks ?\n", "2020-04-09 22:57:57,571 Example #2\n", "2020-04-09 22:57:57,571 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 22:57:57,571 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 22:57:57,572 \tHypothesis: And those who disbelieve in Our signs are the people of the people of the nations .\n", "2020-04-09 22:57:57,572 Example #3\n", "2020-04-09 22:57:57,572 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 22:57:57,572 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 22:57:57,572 \tHypothesis: He sends the ships in peace , and when some of his people passed before him , they mock him .\n", "2020-04-09 22:57:57,573 Validation result (greedy) at epoch 8, step 40000: bleu: 21.26, loss: 54804.7461, ppl: 7.5938, duration: 35.2244s\n", "2020-04-09 22:58:15,970 Epoch 8 Step: 40100 Batch Loss: 2.264594 Tokens per Sec: 10762, Lr: 0.000300\n", "2020-04-09 22:58:34,118 Epoch 8 Step: 40200 Batch Loss: 2.029600 Tokens per Sec: 10978, Lr: 0.000300\n", "2020-04-09 22:58:52,301 Epoch 8 Step: 40300 Batch Loss: 2.173855 Tokens per Sec: 10935, Lr: 0.000300\n", "2020-04-09 22:59:10,695 Epoch 8 Step: 40400 Batch Loss: 2.229180 Tokens per Sec: 10910, Lr: 0.000300\n", "2020-04-09 22:59:28,773 Epoch 8 Step: 40500 Batch Loss: 2.336877 Tokens per Sec: 10854, Lr: 0.000300\n", "2020-04-09 22:59:47,018 Epoch 8 Step: 40600 Batch Loss: 2.316767 Tokens per Sec: 11122, Lr: 0.000300\n", "2020-04-09 23:00:05,212 Epoch 8 Step: 40700 Batch Loss: 2.323744 Tokens per Sec: 11096, Lr: 0.000300\n", "2020-04-09 23:00:23,715 Epoch 8 Step: 40800 Batch Loss: 2.254515 Tokens per Sec: 10863, Lr: 0.000300\n", "2020-04-09 23:00:42,131 Epoch 8 Step: 40900 Batch Loss: 2.295140 Tokens per Sec: 11250, Lr: 0.000300\n", "2020-04-09 23:01:00,190 Epoch 8 Step: 41000 Batch Loss: 2.179672 Tokens per Sec: 10769, Lr: 0.000300\n", "2020-04-09 23:01:26,690 Hooray! New best validation result [ppl]!\n", "2020-04-09 23:01:26,691 Saving new checkpoint.\n", "2020-04-09 23:01:27,932 Example #0\n", "2020-04-09 23:01:27,933 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 23:01:27,933 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 23:01:27,933 \tHypothesis: See how We make clear signs that haply they may understand .\n", "2020-04-09 23:01:27,933 Example #1\n", "2020-04-09 23:01:27,933 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 23:01:27,934 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 23:01:27,934 \tHypothesis: If We willed , We would have made it a rain , so why do you not give thanks ?\n", "2020-04-09 23:01:27,934 Example #2\n", "2020-04-09 23:01:27,934 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 23:01:27,934 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 23:01:27,934 \tHypothesis: And those who disbelieve in Our signs are the people of the people of the community .\n", "2020-04-09 23:01:27,935 Example #3\n", "2020-04-09 23:01:27,935 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 23:01:27,935 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 23:01:27,935 \tHypothesis: He will send the ships in security , and when some of his people passed before him , they mocked him .\n", "2020-04-09 23:01:27,935 Validation result (greedy) at epoch 8, step 41000: bleu: 21.22, loss: 54244.3281, ppl: 7.4380, duration: 27.7449s\n", "2020-04-09 23:01:46,494 Epoch 8 Step: 41100 Batch Loss: 2.098592 Tokens per Sec: 11043, Lr: 0.000300\n", "2020-04-09 23:02:05,028 Epoch 8 Step: 41200 Batch Loss: 2.218289 Tokens per Sec: 11263, Lr: 0.000300\n", "2020-04-09 23:02:23,280 Epoch 8 Step: 41300 Batch Loss: 2.110460 Tokens per Sec: 10980, Lr: 0.000300\n", "2020-04-09 23:02:41,466 Epoch 8 Step: 41400 Batch Loss: 1.964119 Tokens per Sec: 10941, Lr: 0.000300\n", "2020-04-09 23:02:59,899 Epoch 8 Step: 41500 Batch Loss: 2.156255 Tokens per Sec: 10941, Lr: 0.000300\n", "2020-04-09 23:03:17,864 Epoch 8 Step: 41600 Batch Loss: 2.160904 Tokens per Sec: 10882, Lr: 0.000300\n", "2020-04-09 23:03:36,036 Epoch 8 Step: 41700 Batch Loss: 2.071249 Tokens per Sec: 11101, Lr: 0.000300\n", "2020-04-09 23:03:54,431 Epoch 8 Step: 41800 Batch Loss: 1.970897 Tokens per Sec: 10872, Lr: 0.000300\n", "2020-04-09 23:04:12,745 Epoch 8 Step: 41900 Batch Loss: 2.100407 Tokens per Sec: 11188, Lr: 0.000300\n", "2020-04-09 23:04:31,043 Epoch 8 Step: 42000 Batch Loss: 2.062652 Tokens per Sec: 10870, Lr: 0.000300\n", "2020-04-09 23:05:01,873 Hooray! New best validation result [ppl]!\n", "2020-04-09 23:05:01,873 Saving new checkpoint.\n", "2020-04-09 23:05:03,093 Example #0\n", "2020-04-09 23:05:03,096 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 23:05:03,096 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 23:05:03,096 \tHypothesis: See how We explain the signs that haply they may understand .\n", "2020-04-09 23:05:03,096 Example #1\n", "2020-04-09 23:05:03,096 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 23:05:03,097 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 23:05:03,097 \tHypothesis: Had We willed , We would have brought him down the rain , so why do you not give thanks ?\n", "2020-04-09 23:05:03,097 Example #2\n", "2020-04-09 23:05:03,097 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 23:05:03,097 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 23:05:03,098 \tHypothesis: And those who disbelieve in Our signs are the companions of the people of the Right Hand .\n", "2020-04-09 23:05:03,098 Example #3\n", "2020-04-09 23:05:03,098 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 23:05:03,098 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 23:05:03,098 \tHypothesis: He is moving the ships in the security , and when some of his people passed away before him , they mock him .\n", "2020-04-09 23:05:03,099 Validation result (greedy) at epoch 8, step 42000: bleu: 21.43, loss: 54089.2109, ppl: 7.3954, duration: 32.0546s\n", "2020-04-09 23:05:21,344 Epoch 8 Step: 42100 Batch Loss: 2.304792 Tokens per Sec: 10888, Lr: 0.000300\n", "2020-04-09 23:05:39,565 Epoch 8 Step: 42200 Batch Loss: 2.192781 Tokens per Sec: 10942, Lr: 0.000300\n", "2020-04-09 23:05:57,956 Epoch 8 Step: 42300 Batch Loss: 2.130342 Tokens per Sec: 11164, Lr: 0.000300\n", "2020-04-09 23:06:16,132 Epoch 8 Step: 42400 Batch Loss: 2.363187 Tokens per Sec: 10706, Lr: 0.000300\n", "2020-04-09 23:06:34,416 Epoch 8 Step: 42500 Batch Loss: 2.291660 Tokens per Sec: 10884, Lr: 0.000300\n", "2020-04-09 23:06:52,730 Epoch 8 Step: 42600 Batch Loss: 2.206863 Tokens per Sec: 11100, Lr: 0.000300\n", "2020-04-09 23:07:11,039 Epoch 8 Step: 42700 Batch Loss: 2.141305 Tokens per Sec: 10933, Lr: 0.000300\n", "2020-04-09 23:07:29,269 Epoch 8 Step: 42800 Batch Loss: 2.108359 Tokens per Sec: 11036, Lr: 0.000300\n", "2020-04-09 23:07:47,615 Epoch 8 Step: 42900 Batch Loss: 2.179616 Tokens per Sec: 11016, Lr: 0.000300\n", "2020-04-09 23:08:05,988 Epoch 8 Step: 43000 Batch Loss: 2.228909 Tokens per Sec: 11192, Lr: 0.000300\n", "2020-04-09 23:08:34,568 Hooray! New best validation result [ppl]!\n", "2020-04-09 23:08:34,568 Saving new checkpoint.\n", "2020-04-09 23:08:36,186 Example #0\n", "2020-04-09 23:08:36,186 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 23:08:36,186 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 23:08:36,186 \tHypothesis: See how We explain Our signs that they may understand .\n", "2020-04-09 23:08:36,187 Example #1\n", "2020-04-09 23:08:36,187 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 23:08:36,187 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 23:08:36,187 \tHypothesis: If We willed , We could have made it a scale , so why do you not give thanks ?\n", "2020-04-09 23:08:36,187 Example #2\n", "2020-04-09 23:08:36,188 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 23:08:36,188 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 23:08:36,188 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 23:08:36,189 Example #3\n", "2020-04-09 23:08:36,189 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 23:08:36,189 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 23:08:36,189 \tHypothesis: And He sends the ship in the security , and when the leaders of his people passed before him , they mocked him .\n", "2020-04-09 23:08:36,189 Validation result (greedy) at epoch 8, step 43000: bleu: 21.52, loss: 53769.9023, ppl: 7.3086, duration: 30.2008s\n", "2020-04-09 23:08:54,741 Epoch 8 Step: 43100 Batch Loss: 2.133504 Tokens per Sec: 10998, Lr: 0.000300\n", "2020-04-09 23:09:13,143 Epoch 8 Step: 43200 Batch Loss: 2.271922 Tokens per Sec: 11031, Lr: 0.000300\n", "2020-04-09 23:09:31,292 Epoch 8 Step: 43300 Batch Loss: 2.036513 Tokens per Sec: 11015, Lr: 0.000300\n", "2020-04-09 23:09:40,067 Epoch 8: total training loss 9379.64\n", "2020-04-09 23:09:40,067 EPOCH 9\n", "2020-04-09 23:09:49,969 Epoch 9 Step: 43400 Batch Loss: 1.958349 Tokens per Sec: 10582, Lr: 0.000300\n", "2020-04-09 23:10:08,249 Epoch 9 Step: 43500 Batch Loss: 2.125014 Tokens per Sec: 11156, Lr: 0.000300\n", "2020-04-09 23:10:26,251 Epoch 9 Step: 43600 Batch Loss: 2.180873 Tokens per Sec: 10753, Lr: 0.000300\n", "2020-04-09 23:10:44,631 Epoch 9 Step: 43700 Batch Loss: 2.179083 Tokens per Sec: 11328, Lr: 0.000300\n", "2020-04-09 23:11:02,926 Epoch 9 Step: 43800 Batch Loss: 1.994536 Tokens per Sec: 11272, Lr: 0.000300\n", "2020-04-09 23:11:21,247 Epoch 9 Step: 43900 Batch Loss: 2.165617 Tokens per Sec: 10896, Lr: 0.000300\n", "2020-04-09 23:11:39,641 Epoch 9 Step: 44000 Batch Loss: 2.075650 Tokens per Sec: 11018, Lr: 0.000300\n", "2020-04-09 23:12:08,383 Example #0\n", "2020-04-09 23:12:08,384 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 23:12:08,384 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 23:12:08,384 \tHypothesis: See how We explain the signs that haply they may understand .\n", "2020-04-09 23:12:08,384 Example #1\n", "2020-04-09 23:12:08,384 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 23:12:08,384 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 23:12:08,385 \tHypothesis: If We willed , We would have brought it down , so why do you not give thanks ?\n", "2020-04-09 23:12:08,385 Example #2\n", "2020-04-09 23:12:08,385 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 23:12:08,385 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 23:12:08,385 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 23:12:08,386 Example #3\n", "2020-04-09 23:12:08,386 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 23:12:08,386 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 23:12:08,386 \tHypothesis: And He sends the ships in the security , and when some of his people passed before him , they mocked him .\n", "2020-04-09 23:12:08,386 Validation result (greedy) at epoch 9, step 44000: bleu: 21.30, loss: 53794.7305, ppl: 7.3153, duration: 28.7451s\n", "2020-04-09 23:12:26,752 Epoch 9 Step: 44100 Batch Loss: 1.944477 Tokens per Sec: 11027, Lr: 0.000300\n", "2020-04-09 23:12:44,933 Epoch 9 Step: 44200 Batch Loss: 1.634747 Tokens per Sec: 10848, Lr: 0.000300\n", "2020-04-09 23:13:02,909 Epoch 9 Step: 44300 Batch Loss: 2.026499 Tokens per Sec: 10784, Lr: 0.000300\n", "2020-04-09 23:13:21,309 Epoch 9 Step: 44400 Batch Loss: 2.118422 Tokens per Sec: 10905, Lr: 0.000300\n", "2020-04-09 23:13:39,601 Epoch 9 Step: 44500 Batch Loss: 2.234929 Tokens per Sec: 11042, Lr: 0.000300\n", "2020-04-09 23:13:57,968 Epoch 9 Step: 44600 Batch Loss: 2.032033 Tokens per Sec: 10945, Lr: 0.000300\n", "2020-04-09 23:14:16,306 Epoch 9 Step: 44700 Batch Loss: 2.207828 Tokens per Sec: 11037, Lr: 0.000300\n", "2020-04-09 23:14:34,529 Epoch 9 Step: 44800 Batch Loss: 2.059973 Tokens per Sec: 11064, Lr: 0.000300\n", "2020-04-09 23:14:52,794 Epoch 9 Step: 44900 Batch Loss: 2.120345 Tokens per Sec: 11178, Lr: 0.000300\n", "2020-04-09 23:15:11,096 Epoch 9 Step: 45000 Batch Loss: 2.246343 Tokens per Sec: 11074, Lr: 0.000300\n", "2020-04-09 23:15:41,218 Hooray! New best validation result [ppl]!\n", "2020-04-09 23:15:41,218 Saving new checkpoint.\n", "2020-04-09 23:15:42,453 Example #0\n", "2020-04-09 23:15:42,453 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 23:15:42,454 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 23:15:42,454 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-09 23:15:42,454 Example #1\n", "2020-04-09 23:15:42,454 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 23:15:42,454 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 23:15:42,454 \tHypothesis: If We willed , We would have brought it down , so why do you not give thanks ?\n", "2020-04-09 23:15:42,455 Example #2\n", "2020-04-09 23:15:42,455 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 23:15:42,455 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 23:15:42,455 \tHypothesis: Those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 23:15:42,455 Example #3\n", "2020-04-09 23:15:42,456 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 23:15:42,456 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 23:15:42,456 \tHypothesis: And He sends the ships in security , and when the leaders of his people passed before him , they mock him .\n", "2020-04-09 23:15:42,456 Validation result (greedy) at epoch 9, step 45000: bleu: 21.60, loss: 53501.6523, ppl: 7.2364, duration: 31.3597s\n", "2020-04-09 23:16:00,965 Epoch 9 Step: 45100 Batch Loss: 2.204039 Tokens per Sec: 10888, Lr: 0.000300\n", "2020-04-09 23:16:19,209 Epoch 9 Step: 45200 Batch Loss: 2.258697 Tokens per Sec: 10921, Lr: 0.000300\n", "2020-04-09 23:16:37,367 Epoch 9 Step: 45300 Batch Loss: 2.276276 Tokens per Sec: 11040, Lr: 0.000300\n", "2020-04-09 23:16:55,604 Epoch 9 Step: 45400 Batch Loss: 2.063883 Tokens per Sec: 10911, Lr: 0.000300\n", "2020-04-09 23:17:13,563 Epoch 9 Step: 45500 Batch Loss: 2.302160 Tokens per Sec: 10632, Lr: 0.000300\n", "2020-04-09 23:17:31,834 Epoch 9 Step: 45600 Batch Loss: 2.138355 Tokens per Sec: 10942, Lr: 0.000300\n", "2020-04-09 23:17:49,977 Epoch 9 Step: 45700 Batch Loss: 2.278868 Tokens per Sec: 10823, Lr: 0.000300\n", "2020-04-09 23:18:08,288 Epoch 9 Step: 45800 Batch Loss: 2.133148 Tokens per Sec: 11149, Lr: 0.000300\n", "2020-04-09 23:18:26,367 Epoch 9 Step: 45900 Batch Loss: 2.073554 Tokens per Sec: 10843, Lr: 0.000300\n", "2020-04-09 23:18:44,788 Epoch 9 Step: 46000 Batch Loss: 2.244796 Tokens per Sec: 11108, Lr: 0.000300\n", "2020-04-09 23:19:10,593 Hooray! New best validation result [ppl]!\n", "2020-04-09 23:19:10,594 Saving new checkpoint.\n", "2020-04-09 23:19:11,793 Example #0\n", "2020-04-09 23:19:11,798 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 23:19:11,798 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 23:19:11,799 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-09 23:19:11,799 Example #1\n", "2020-04-09 23:19:11,799 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 23:19:11,800 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 23:19:11,800 \tHypothesis: If We willed , We would have made it a rain , so why do you not give thanks ?\n", "2020-04-09 23:19:11,800 Example #2\n", "2020-04-09 23:19:11,800 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 23:19:11,800 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 23:19:11,800 \tHypothesis: And those who disbelieve in Our signs are the people of the nation .\n", "2020-04-09 23:19:11,801 Example #3\n", "2020-04-09 23:19:11,801 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 23:19:11,801 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 23:19:11,801 \tHypothesis: And He sends the ships in the secure , and when the chiefs of his people passed before him , they mocked him .\n", "2020-04-09 23:19:11,801 Validation result (greedy) at epoch 9, step 46000: bleu: 21.44, loss: 53400.7969, ppl: 7.2095, duration: 27.0126s\n", "2020-04-09 23:19:29,880 Epoch 9 Step: 46100 Batch Loss: 2.119853 Tokens per Sec: 10630, Lr: 0.000300\n", "2020-04-09 23:19:48,204 Epoch 9 Step: 46200 Batch Loss: 2.053347 Tokens per Sec: 11177, Lr: 0.000300\n", "2020-04-09 23:20:06,455 Epoch 9 Step: 46300 Batch Loss: 2.076783 Tokens per Sec: 11029, Lr: 0.000300\n", "2020-04-09 23:20:24,734 Epoch 9 Step: 46400 Batch Loss: 1.912648 Tokens per Sec: 11139, Lr: 0.000300\n", "2020-04-09 23:20:42,961 Epoch 9 Step: 46500 Batch Loss: 1.998083 Tokens per Sec: 10916, Lr: 0.000300\n", "2020-04-09 23:21:01,238 Epoch 9 Step: 46600 Batch Loss: 2.216117 Tokens per Sec: 10895, Lr: 0.000300\n", "2020-04-09 23:21:19,443 Epoch 9 Step: 46700 Batch Loss: 2.486369 Tokens per Sec: 10981, Lr: 0.000300\n", "2020-04-09 23:21:37,847 Epoch 9 Step: 46800 Batch Loss: 2.164602 Tokens per Sec: 10895, Lr: 0.000300\n", "2020-04-09 23:21:55,967 Epoch 9 Step: 46900 Batch Loss: 2.107483 Tokens per Sec: 10743, Lr: 0.000300\n", "2020-04-09 23:22:14,273 Epoch 9 Step: 47000 Batch Loss: 2.304420 Tokens per Sec: 11072, Lr: 0.000300\n", "2020-04-09 23:22:43,000 Hooray! New best validation result [ppl]!\n", "2020-04-09 23:22:43,000 Saving new checkpoint.\n", "2020-04-09 23:22:44,227 Example #0\n", "2020-04-09 23:22:44,228 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 23:22:44,228 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 23:22:44,229 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-09 23:22:44,229 Example #1\n", "2020-04-09 23:22:44,229 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 23:22:44,229 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 23:22:44,229 \tHypothesis: If We willed , We would have brought it down a rain , so why do you not give thanks ?\n", "2020-04-09 23:22:44,229 Example #2\n", "2020-04-09 23:22:44,230 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 23:22:44,230 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 23:22:44,230 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 23:22:44,230 Example #3\n", "2020-04-09 23:22:44,231 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 23:22:44,231 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 23:22:44,231 \tHypothesis: He will send the ark in the secure , and when the chiefs of his people passed away before him , they mock him .\n", "2020-04-09 23:22:44,231 Validation result (greedy) at epoch 9, step 47000: bleu: 21.57, loss: 53217.5508, ppl: 7.1607, duration: 29.9575s\n", "2020-04-09 23:23:02,658 Epoch 9 Step: 47100 Batch Loss: 2.094570 Tokens per Sec: 10824, Lr: 0.000300\n", "2020-04-09 23:23:20,878 Epoch 9 Step: 47200 Batch Loss: 2.053883 Tokens per Sec: 11009, Lr: 0.000300\n", "2020-04-09 23:23:39,195 Epoch 9 Step: 47300 Batch Loss: 2.054940 Tokens per Sec: 10956, Lr: 0.000300\n", "2020-04-09 23:23:57,624 Epoch 9 Step: 47400 Batch Loss: 2.270289 Tokens per Sec: 11255, Lr: 0.000300\n", "2020-04-09 23:24:15,710 Epoch 9 Step: 47500 Batch Loss: 2.118424 Tokens per Sec: 10833, Lr: 0.000300\n", "2020-04-09 23:24:34,236 Epoch 9 Step: 47600 Batch Loss: 2.299340 Tokens per Sec: 10992, Lr: 0.000300\n", "2020-04-09 23:24:52,520 Epoch 9 Step: 47700 Batch Loss: 2.464327 Tokens per Sec: 10875, Lr: 0.000300\n", "2020-04-09 23:25:05,572 Epoch 9: total training loss 9243.69\n", "2020-04-09 23:25:05,573 EPOCH 10\n", "2020-04-09 23:25:11,228 Epoch 10 Step: 47800 Batch Loss: 2.010685 Tokens per Sec: 10531, Lr: 0.000300\n", "2020-04-09 23:25:29,308 Epoch 10 Step: 47900 Batch Loss: 2.320683 Tokens per Sec: 10700, Lr: 0.000300\n", "2020-04-09 23:25:47,415 Epoch 10 Step: 48000 Batch Loss: 1.773332 Tokens per Sec: 10907, Lr: 0.000300\n", "2020-04-09 23:26:15,043 Example #0\n", "2020-04-09 23:26:15,044 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 23:26:15,044 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 23:26:15,044 \tHypothesis: See how We make clear signs so that they may understand .\n", "2020-04-09 23:26:15,044 Example #1\n", "2020-04-09 23:26:15,045 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 23:26:15,045 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 23:26:15,045 \tHypothesis: Had We willed , We would have brought it down , so why do you not give thanks ?\n", "2020-04-09 23:26:15,045 Example #2\n", "2020-04-09 23:26:15,045 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 23:26:15,046 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 23:26:15,046 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 23:26:15,046 Example #3\n", "2020-04-09 23:26:15,046 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 23:26:15,046 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 23:26:15,046 \tHypothesis: And He sends the ships in the secure , and when the chiefs of his people passed before him , they mock him .\n", "2020-04-09 23:26:15,047 Validation result (greedy) at epoch 10, step 48000: bleu: 21.67, loss: 53295.6172, ppl: 7.1815, duration: 27.6309s\n", "2020-04-09 23:26:33,282 Epoch 10 Step: 48100 Batch Loss: 2.043831 Tokens per Sec: 10691, Lr: 0.000300\n", "2020-04-09 23:26:51,725 Epoch 10 Step: 48200 Batch Loss: 2.287551 Tokens per Sec: 11132, Lr: 0.000300\n", "2020-04-09 23:27:10,047 Epoch 10 Step: 48300 Batch Loss: 1.951195 Tokens per Sec: 11285, Lr: 0.000300\n", "2020-04-09 23:27:28,175 Epoch 10 Step: 48400 Batch Loss: 2.161692 Tokens per Sec: 10855, Lr: 0.000300\n", "2020-04-09 23:27:46,473 Epoch 10 Step: 48500 Batch Loss: 1.970119 Tokens per Sec: 10936, Lr: 0.000300\n", "2020-04-09 23:28:04,605 Epoch 10 Step: 48600 Batch Loss: 2.080714 Tokens per Sec: 11076, Lr: 0.000300\n", "2020-04-09 23:28:22,905 Epoch 10 Step: 48700 Batch Loss: 1.907820 Tokens per Sec: 11007, Lr: 0.000300\n", "2020-04-09 23:28:41,381 Epoch 10 Step: 48800 Batch Loss: 2.079331 Tokens per Sec: 11266, Lr: 0.000300\n", "2020-04-09 23:28:59,875 Epoch 10 Step: 48900 Batch Loss: 1.969763 Tokens per Sec: 11227, Lr: 0.000300\n", "2020-04-09 23:29:18,062 Epoch 10 Step: 49000 Batch Loss: 2.113390 Tokens per Sec: 11051, Lr: 0.000300\n", "2020-04-09 23:29:47,530 Hooray! New best validation result [ppl]!\n", "2020-04-09 23:29:47,530 Saving new checkpoint.\n", "2020-04-09 23:29:48,821 Example #0\n", "2020-04-09 23:29:48,821 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 23:29:48,821 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 23:29:48,822 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-09 23:29:48,822 Example #1\n", "2020-04-09 23:29:48,822 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 23:29:48,822 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 23:29:48,822 \tHypothesis: If We willed , We would have made it a scalf , so why do you not give thanks ?\n", "2020-04-09 23:29:48,822 Example #2\n", "2020-04-09 23:29:48,823 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 23:29:48,823 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 23:29:48,823 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 23:29:48,823 Example #3\n", "2020-04-09 23:29:48,823 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 23:29:48,824 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 23:29:48,824 \tHypothesis: He sends the ships in security , and when the leaders of his people passed before him , they mock him .\n", "2020-04-09 23:29:48,824 Validation result (greedy) at epoch 10, step 49000: bleu: 22.09, loss: 52955.1719, ppl: 7.0916, duration: 30.7611s\n", "2020-04-09 23:30:07,061 Epoch 10 Step: 49100 Batch Loss: 2.022794 Tokens per Sec: 10783, Lr: 0.000300\n", "2020-04-09 23:30:25,348 Epoch 10 Step: 49200 Batch Loss: 2.064187 Tokens per Sec: 11006, Lr: 0.000300\n", "2020-04-09 23:30:43,760 Epoch 10 Step: 49300 Batch Loss: 1.730834 Tokens per Sec: 11082, Lr: 0.000300\n", "2020-04-09 23:31:02,166 Epoch 10 Step: 49400 Batch Loss: 1.957734 Tokens per Sec: 11233, Lr: 0.000300\n", "2020-04-09 23:31:20,618 Epoch 10 Step: 49500 Batch Loss: 2.026736 Tokens per Sec: 10924, Lr: 0.000300\n", "2020-04-09 23:31:38,861 Epoch 10 Step: 49600 Batch Loss: 1.908111 Tokens per Sec: 11005, Lr: 0.000300\n", "2020-04-09 23:31:57,164 Epoch 10 Step: 49700 Batch Loss: 2.311419 Tokens per Sec: 11104, Lr: 0.000300\n", "2020-04-09 23:32:15,372 Epoch 10 Step: 49800 Batch Loss: 2.055347 Tokens per Sec: 11056, Lr: 0.000300\n", "2020-04-09 23:32:33,719 Epoch 10 Step: 49900 Batch Loss: 2.064936 Tokens per Sec: 10927, Lr: 0.000300\n", "2020-04-09 23:32:51,739 Epoch 10 Step: 50000 Batch Loss: 2.213390 Tokens per Sec: 10789, Lr: 0.000300\n", "2020-04-09 23:33:19,845 Hooray! New best validation result [ppl]!\n", "2020-04-09 23:33:19,845 Saving new checkpoint.\n", "2020-04-09 23:33:21,087 Example #0\n", "2020-04-09 23:33:21,088 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 23:33:21,088 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 23:33:21,088 \tHypothesis: See how We explain the verses that they may understand .\n", "2020-04-09 23:33:21,088 Example #1\n", "2020-04-09 23:33:21,089 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 23:33:21,089 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 23:33:21,089 \tHypothesis: Had We willed , We could have made it a rain , so why do you not give thanks ?\n", "2020-04-09 23:33:21,089 Example #2\n", "2020-04-09 23:33:21,090 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 23:33:21,090 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 23:33:21,090 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 23:33:21,090 Example #3\n", "2020-04-09 23:33:21,090 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 23:33:21,090 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 23:33:21,091 \tHypothesis: And He sends the ships in the security , and when some of his people passed away before him , they mocked at him .\n", "2020-04-09 23:33:21,091 Validation result (greedy) at epoch 10, step 50000: bleu: 22.02, loss: 52565.1875, ppl: 6.9900, duration: 29.3516s\n", "2020-04-09 23:33:39,367 Epoch 10 Step: 50100 Batch Loss: 2.326136 Tokens per Sec: 10819, Lr: 0.000300\n", "2020-04-09 23:33:57,697 Epoch 10 Step: 50200 Batch Loss: 2.143932 Tokens per Sec: 10889, Lr: 0.000300\n", "2020-04-09 23:34:16,249 Epoch 10 Step: 50300 Batch Loss: 1.877056 Tokens per Sec: 11303, Lr: 0.000300\n", "2020-04-09 23:34:34,733 Epoch 10 Step: 50400 Batch Loss: 1.809703 Tokens per Sec: 11128, Lr: 0.000300\n", "2020-04-09 23:34:53,172 Epoch 10 Step: 50500 Batch Loss: 2.160360 Tokens per Sec: 11267, Lr: 0.000300\n", "2020-04-09 23:35:11,401 Epoch 10 Step: 50600 Batch Loss: 2.017428 Tokens per Sec: 10817, Lr: 0.000300\n", "2020-04-09 23:35:29,563 Epoch 10 Step: 50700 Batch Loss: 2.144158 Tokens per Sec: 10885, Lr: 0.000300\n", "2020-04-09 23:35:47,895 Epoch 10 Step: 50800 Batch Loss: 1.944361 Tokens per Sec: 10989, Lr: 0.000300\n", "2020-04-09 23:36:06,267 Epoch 10 Step: 50900 Batch Loss: 2.223985 Tokens per Sec: 11217, Lr: 0.000300\n", "2020-04-09 23:36:24,529 Epoch 10 Step: 51000 Batch Loss: 2.218631 Tokens per Sec: 10962, Lr: 0.000300\n", "2020-04-09 23:36:53,553 Example #0\n", "2020-04-09 23:36:53,554 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 23:36:53,554 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 23:36:53,554 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-09 23:36:53,554 Example #1\n", "2020-04-09 23:36:53,555 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 23:36:53,555 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 23:36:53,555 \tHypothesis: If We willed , We could have removed it down a rain , so why do you not give thanks ?\n", "2020-04-09 23:36:53,555 Example #2\n", "2020-04-09 23:36:53,556 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 23:36:53,556 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 23:36:53,561 \tHypothesis: Those who disbelieve in Our signs are the people of the nations .\n", "2020-04-09 23:36:53,562 Example #3\n", "2020-04-09 23:36:53,563 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 23:36:53,563 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 23:36:53,563 \tHypothesis: He sends the ships in security , and when the leaders of his people passed before him , they ridicule him .\n", "2020-04-09 23:36:53,563 Validation result (greedy) at epoch 10, step 51000: bleu: 22.09, loss: 52727.7344, ppl: 7.0322, duration: 29.0338s\n", "2020-04-09 23:37:11,740 Epoch 10 Step: 51100 Batch Loss: 2.201875 Tokens per Sec: 10864, Lr: 0.000300\n", "2020-04-09 23:37:30,052 Epoch 10 Step: 51200 Batch Loss: 1.671625 Tokens per Sec: 11095, Lr: 0.000300\n", "2020-04-09 23:37:48,360 Epoch 10 Step: 51300 Batch Loss: 1.745461 Tokens per Sec: 10932, Lr: 0.000300\n", "2020-04-09 23:38:06,512 Epoch 10 Step: 51400 Batch Loss: 1.872573 Tokens per Sec: 10805, Lr: 0.000300\n", "2020-04-09 23:38:24,922 Epoch 10 Step: 51500 Batch Loss: 2.234218 Tokens per Sec: 10954, Lr: 0.000300\n", "2020-04-09 23:38:43,110 Epoch 10 Step: 51600 Batch Loss: 2.124661 Tokens per Sec: 10807, Lr: 0.000300\n", "2020-04-09 23:39:01,495 Epoch 10 Step: 51700 Batch Loss: 2.211341 Tokens per Sec: 11116, Lr: 0.000300\n", "2020-04-09 23:39:19,551 Epoch 10 Step: 51800 Batch Loss: 2.008241 Tokens per Sec: 10791, Lr: 0.000300\n", "2020-04-09 23:39:37,710 Epoch 10 Step: 51900 Batch Loss: 2.126257 Tokens per Sec: 10938, Lr: 0.000300\n", "2020-04-09 23:39:56,049 Epoch 10 Step: 52000 Batch Loss: 2.031232 Tokens per Sec: 11124, Lr: 0.000300\n", "2020-04-09 23:40:29,242 Hooray! New best validation result [ppl]!\n", "2020-04-09 23:40:29,242 Saving new checkpoint.\n", "2020-04-09 23:40:30,523 Example #0\n", "2020-04-09 23:40:30,524 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 23:40:30,524 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 23:40:30,524 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-09 23:40:30,524 Example #1\n", "2020-04-09 23:40:30,525 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 23:40:30,525 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 23:40:30,525 \tHypothesis: If We willed , We would have brought it down , so why do you not give thanks ?\n", "2020-04-09 23:40:30,525 Example #2\n", "2020-04-09 23:40:30,526 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 23:40:30,526 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 23:40:30,526 \tHypothesis: Those who disbelieve in Our signs are the companions of the people of the wicked .\n", "2020-04-09 23:40:30,526 Example #3\n", "2020-04-09 23:40:30,527 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 23:40:30,527 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 23:40:30,527 \tHypothesis: He sends the ark in security , and when the leaders of his people passed before him , they mocked him .\n", "2020-04-09 23:40:30,527 Validation result (greedy) at epoch 10, step 52000: bleu: 22.09, loss: 52320.3359, ppl: 6.9270, duration: 34.4780s\n", "2020-04-09 23:40:49,346 Epoch 10 Step: 52100 Batch Loss: 1.941027 Tokens per Sec: 10930, Lr: 0.000300\n", "2020-04-09 23:41:04,114 Epoch 10: total training loss 9078.04\n", "2020-04-09 23:41:04,114 EPOCH 11\n", "2020-04-09 23:41:07,914 Epoch 11 Step: 52200 Batch Loss: 2.191005 Tokens per Sec: 9272, Lr: 0.000300\n", "2020-04-09 23:41:26,332 Epoch 11 Step: 52300 Batch Loss: 1.820207 Tokens per Sec: 11043, Lr: 0.000300\n", "2020-04-09 23:41:44,597 Epoch 11 Step: 52400 Batch Loss: 2.626487 Tokens per Sec: 11154, Lr: 0.000300\n", "2020-04-09 23:42:03,116 Epoch 11 Step: 52500 Batch Loss: 2.159490 Tokens per Sec: 11041, Lr: 0.000300\n", "2020-04-09 23:42:21,529 Epoch 11 Step: 52600 Batch Loss: 1.945245 Tokens per Sec: 11211, Lr: 0.000300\n", "2020-04-09 23:42:40,108 Epoch 11 Step: 52700 Batch Loss: 2.105855 Tokens per Sec: 11207, Lr: 0.000300\n", "2020-04-09 23:42:58,393 Epoch 11 Step: 52800 Batch Loss: 2.098129 Tokens per Sec: 11094, Lr: 0.000300\n", "2020-04-09 23:43:16,804 Epoch 11 Step: 52900 Batch Loss: 2.024961 Tokens per Sec: 11018, Lr: 0.000300\n", "2020-04-09 23:43:35,150 Epoch 11 Step: 53000 Batch Loss: 1.960102 Tokens per Sec: 11021, Lr: 0.000300\n", "2020-04-09 23:44:05,004 Hooray! New best validation result [ppl]!\n", "2020-04-09 23:44:05,005 Saving new checkpoint.\n", "2020-04-09 23:44:06,645 Example #0\n", "2020-04-09 23:44:06,646 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 23:44:06,646 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 23:44:06,646 \tHypothesis: See how We explain Our signs that haply they may understand .\n", "2020-04-09 23:44:06,646 Example #1\n", "2020-04-09 23:44:06,647 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 23:44:06,647 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 23:44:06,647 \tHypothesis: If We willed , We could have removed him with a scorn , so why do you not give thanks ?\n", "2020-04-09 23:44:06,647 Example #2\n", "2020-04-09 23:44:06,647 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 23:44:06,648 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 23:44:06,648 \tHypothesis: And those who disbelieve in Our signs are the people of the Left .\n", "2020-04-09 23:44:06,648 Example #3\n", "2020-04-09 23:44:06,648 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 23:44:06,648 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 23:44:06,649 \tHypothesis: He sends the ark in security , and when some of his people passed away before him , they mocked him .\n", "2020-04-09 23:44:06,649 Validation result (greedy) at epoch 11, step 53000: bleu: 21.96, loss: 52218.3359, ppl: 6.9009, duration: 31.4982s\n", "2020-04-09 23:44:24,986 Epoch 11 Step: 53100 Batch Loss: 2.118927 Tokens per Sec: 10999, Lr: 0.000300\n", "2020-04-09 23:44:43,224 Epoch 11 Step: 53200 Batch Loss: 2.021477 Tokens per Sec: 10886, Lr: 0.000300\n", "2020-04-09 23:45:01,491 Epoch 11 Step: 53300 Batch Loss: 2.073700 Tokens per Sec: 11041, Lr: 0.000300\n", "2020-04-09 23:45:19,735 Epoch 11 Step: 53400 Batch Loss: 2.189031 Tokens per Sec: 10978, Lr: 0.000300\n", "2020-04-09 23:45:37,797 Epoch 11 Step: 53500 Batch Loss: 1.868712 Tokens per Sec: 10739, Lr: 0.000300\n", "2020-04-09 23:45:56,212 Epoch 11 Step: 53600 Batch Loss: 2.167693 Tokens per Sec: 10942, Lr: 0.000300\n", "2020-04-09 23:46:14,387 Epoch 11 Step: 53700 Batch Loss: 1.953278 Tokens per Sec: 10964, Lr: 0.000300\n", "2020-04-09 23:46:32,577 Epoch 11 Step: 53800 Batch Loss: 2.179255 Tokens per Sec: 11098, Lr: 0.000300\n", "2020-04-09 23:46:50,942 Epoch 11 Step: 53900 Batch Loss: 2.124580 Tokens per Sec: 11133, Lr: 0.000300\n", "2020-04-09 23:47:09,046 Epoch 11 Step: 54000 Batch Loss: 1.801505 Tokens per Sec: 10726, Lr: 0.000300\n", "2020-04-09 23:47:37,176 Example #0\n", "2020-04-09 23:47:37,177 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 23:47:37,177 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 23:47:37,177 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-09 23:47:37,177 Example #1\n", "2020-04-09 23:47:37,178 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 23:47:37,178 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 23:47:37,178 \tHypothesis: If We willed , We could have brought it down a rain , so why do you not give thanks ?\n", "2020-04-09 23:47:37,178 Example #2\n", "2020-04-09 23:47:37,179 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 23:47:37,179 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 23:47:37,179 \tHypothesis: And those who disbelieve in Our signs are the dwellers of the wicked .\n", "2020-04-09 23:47:37,179 Example #3\n", "2020-04-09 23:47:37,180 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 23:47:37,180 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 23:47:37,180 \tHypothesis: He sends the ships in security , and when the leaders of his people passed before him , they mock him .\n", "2020-04-09 23:47:37,180 Validation result (greedy) at epoch 11, step 54000: bleu: 21.81, loss: 52261.8867, ppl: 6.9120, duration: 28.1339s\n", "2020-04-09 23:47:55,572 Epoch 11 Step: 54100 Batch Loss: 1.916748 Tokens per Sec: 11051, Lr: 0.000300\n", "2020-04-09 23:48:13,660 Epoch 11 Step: 54200 Batch Loss: 2.093851 Tokens per Sec: 11008, Lr: 0.000300\n", "2020-04-09 23:48:31,893 Epoch 11 Step: 54300 Batch Loss: 2.006753 Tokens per Sec: 10905, Lr: 0.000300\n", "2020-04-09 23:48:50,048 Epoch 11 Step: 54400 Batch Loss: 1.814026 Tokens per Sec: 11115, Lr: 0.000300\n", "2020-04-09 23:49:08,516 Epoch 11 Step: 54500 Batch Loss: 2.147173 Tokens per Sec: 10900, Lr: 0.000300\n", "2020-04-09 23:49:26,734 Epoch 11 Step: 54600 Batch Loss: 1.892052 Tokens per Sec: 10938, Lr: 0.000300\n", "2020-04-09 23:49:45,164 Epoch 11 Step: 54700 Batch Loss: 2.088869 Tokens per Sec: 11029, Lr: 0.000300\n", "2020-04-09 23:50:03,664 Epoch 11 Step: 54800 Batch Loss: 2.336154 Tokens per Sec: 11186, Lr: 0.000300\n", "2020-04-09 23:50:21,897 Epoch 11 Step: 54900 Batch Loss: 1.933826 Tokens per Sec: 11004, Lr: 0.000300\n", "2020-04-09 23:50:40,138 Epoch 11 Step: 55000 Batch Loss: 1.969119 Tokens per Sec: 10980, Lr: 0.000300\n", "2020-04-09 23:51:11,557 Hooray! New best validation result [ppl]!\n", "2020-04-09 23:51:11,557 Saving new checkpoint.\n", "2020-04-09 23:51:12,782 Example #0\n", "2020-04-09 23:51:12,783 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 23:51:12,783 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 23:51:12,783 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-09 23:51:12,784 Example #1\n", "2020-04-09 23:51:12,784 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 23:51:12,784 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 23:51:12,784 \tHypothesis: If We willed , We would have made it a rain , so why do you not give thanks ?\n", "2020-04-09 23:51:12,784 Example #2\n", "2020-04-09 23:51:12,785 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 23:51:12,785 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 23:51:12,785 \tHypothesis: Those who disbelieve in Our signs are the people of the wicked .\n", "2020-04-09 23:51:12,786 Example #3\n", "2020-04-09 23:51:12,786 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 23:51:12,786 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 23:51:12,786 \tHypothesis: He sends the ships in security , and whenever some of his people passed before him , they mocked him .\n", "2020-04-09 23:51:12,787 Validation result (greedy) at epoch 11, step 55000: bleu: 22.27, loss: 52147.5391, ppl: 6.8828, duration: 32.6479s\n", "2020-04-09 23:51:31,245 Epoch 11 Step: 55100 Batch Loss: 2.169886 Tokens per Sec: 10609, Lr: 0.000300\n", "2020-04-09 23:51:49,326 Epoch 11 Step: 55200 Batch Loss: 2.055598 Tokens per Sec: 11001, Lr: 0.000300\n", "2020-04-09 23:52:07,752 Epoch 11 Step: 55300 Batch Loss: 2.228842 Tokens per Sec: 11247, Lr: 0.000300\n", "2020-04-09 23:52:26,017 Epoch 11 Step: 55400 Batch Loss: 2.130361 Tokens per Sec: 11080, Lr: 0.000300\n", "2020-04-09 23:52:44,376 Epoch 11 Step: 55500 Batch Loss: 2.160210 Tokens per Sec: 10861, Lr: 0.000300\n", "2020-04-09 23:53:02,676 Epoch 11 Step: 55600 Batch Loss: 2.061281 Tokens per Sec: 11019, Lr: 0.000300\n", "2020-04-09 23:53:21,018 Epoch 11 Step: 55700 Batch Loss: 2.197393 Tokens per Sec: 10931, Lr: 0.000300\n", "2020-04-09 23:53:39,331 Epoch 11 Step: 55800 Batch Loss: 1.915841 Tokens per Sec: 10810, Lr: 0.000300\n", "2020-04-09 23:53:57,523 Epoch 11 Step: 55900 Batch Loss: 2.184733 Tokens per Sec: 11202, Lr: 0.000300\n", "2020-04-09 23:54:15,571 Epoch 11 Step: 56000 Batch Loss: 1.980049 Tokens per Sec: 10742, Lr: 0.000300\n", "2020-04-09 23:54:45,633 Hooray! New best validation result [ppl]!\n", "2020-04-09 23:54:45,633 Saving new checkpoint.\n", "2020-04-09 23:54:46,853 Example #0\n", "2020-04-09 23:54:46,854 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 23:54:46,854 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 23:54:46,854 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-09 23:54:46,854 Example #1\n", "2020-04-09 23:54:46,855 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 23:54:46,855 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 23:54:46,855 \tHypothesis: If We willed , We would have brought it down a scalf , so why do you not give thanks ?\n", "2020-04-09 23:54:46,855 Example #2\n", "2020-04-09 23:54:46,856 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 23:54:46,856 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 23:54:46,856 \tHypothesis: And those who disbelieve in Our signs are the people of the people of the Last .\n", "2020-04-09 23:54:46,856 Example #3\n", "2020-04-09 23:54:46,857 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 23:54:46,857 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 23:54:46,857 \tHypothesis: He sends the ark in security , and when the leaders of his people passed before him , they mock him .\n", "2020-04-09 23:54:46,857 Validation result (greedy) at epoch 11, step 56000: bleu: 22.20, loss: 52079.8203, ppl: 6.8656, duration: 31.2855s\n", "2020-04-09 23:55:05,151 Epoch 11 Step: 56100 Batch Loss: 1.872254 Tokens per Sec: 10440, Lr: 0.000300\n", "2020-04-09 23:55:23,312 Epoch 11 Step: 56200 Batch Loss: 2.045106 Tokens per Sec: 10745, Lr: 0.000300\n", "2020-04-09 23:55:41,690 Epoch 11 Step: 56300 Batch Loss: 1.987326 Tokens per Sec: 11257, Lr: 0.000300\n", "2020-04-09 23:55:59,961 Epoch 11 Step: 56400 Batch Loss: 2.013717 Tokens per Sec: 10962, Lr: 0.000300\n", "2020-04-09 23:56:18,099 Epoch 11 Step: 56500 Batch Loss: 1.970747 Tokens per Sec: 11011, Lr: 0.000300\n", "2020-04-09 23:56:35,016 Epoch 11: total training loss 8971.71\n", "2020-04-09 23:56:35,017 EPOCH 12\n", "2020-04-09 23:56:36,883 Epoch 12 Step: 56600 Batch Loss: 2.049063 Tokens per Sec: 8153, Lr: 0.000300\n", "2020-04-09 23:56:54,879 Epoch 12 Step: 56700 Batch Loss: 2.050419 Tokens per Sec: 10864, Lr: 0.000300\n", "2020-04-09 23:57:13,296 Epoch 12 Step: 56800 Batch Loss: 2.177431 Tokens per Sec: 11007, Lr: 0.000300\n", "2020-04-09 23:57:31,165 Epoch 12 Step: 56900 Batch Loss: 2.072167 Tokens per Sec: 10669, Lr: 0.000300\n", "2020-04-09 23:57:49,671 Epoch 12 Step: 57000 Batch Loss: 2.118921 Tokens per Sec: 11042, Lr: 0.000300\n", "2020-04-09 23:58:19,120 Hooray! New best validation result [ppl]!\n", "2020-04-09 23:58:19,120 Saving new checkpoint.\n", "2020-04-09 23:58:20,380 Example #0\n", "2020-04-09 23:58:20,381 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-09 23:58:20,381 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-09 23:58:20,381 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-09 23:58:20,381 Example #1\n", "2020-04-09 23:58:20,382 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-09 23:58:20,382 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-09 23:58:20,382 \tHypothesis: If We willed , We would have brought him down a blind , so why do you not give thanks ?\n", "2020-04-09 23:58:20,382 Example #2\n", "2020-04-09 23:58:20,382 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-09 23:58:20,383 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-09 23:58:20,383 \tHypothesis: And those who disbelieve in Our signs are the people of the nation of the nation .\n", "2020-04-09 23:58:20,383 Example #3\n", "2020-04-09 23:58:20,383 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-09 23:58:20,383 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-09 23:58:20,384 \tHypothesis: And He sends the ships in security , and when the leaders of his people passed before him , they mock him .\n", "2020-04-09 23:58:20,384 Validation result (greedy) at epoch 12, step 57000: bleu: 22.22, loss: 51867.7734, ppl: 6.8120, duration: 30.7125s\n", "2020-04-09 23:58:39,026 Epoch 12 Step: 57100 Batch Loss: 2.083498 Tokens per Sec: 11167, Lr: 0.000300\n", "2020-04-09 23:58:57,330 Epoch 12 Step: 57200 Batch Loss: 2.018097 Tokens per Sec: 11126, Lr: 0.000300\n", "2020-04-09 23:59:15,602 Epoch 12 Step: 57300 Batch Loss: 2.214853 Tokens per Sec: 11063, Lr: 0.000300\n", "2020-04-09 23:59:33,837 Epoch 12 Step: 57400 Batch Loss: 1.777696 Tokens per Sec: 10919, Lr: 0.000300\n", "2020-04-09 23:59:52,145 Epoch 12 Step: 57500 Batch Loss: 1.950492 Tokens per Sec: 10913, Lr: 0.000300\n", "2020-04-10 00:00:10,497 Epoch 12 Step: 57600 Batch Loss: 1.759585 Tokens per Sec: 10966, Lr: 0.000300\n", "2020-04-10 00:00:28,797 Epoch 12 Step: 57700 Batch Loss: 2.525295 Tokens per Sec: 10881, Lr: 0.000300\n", "2020-04-10 00:00:47,215 Epoch 12 Step: 57800 Batch Loss: 1.982021 Tokens per Sec: 11169, Lr: 0.000300\n", "2020-04-10 00:01:05,602 Epoch 12 Step: 57900 Batch Loss: 1.834756 Tokens per Sec: 11052, Lr: 0.000300\n", "2020-04-10 00:01:24,025 Epoch 12 Step: 58000 Batch Loss: 2.025802 Tokens per Sec: 11147, Lr: 0.000300\n", "2020-04-10 00:01:53,617 Example #0\n", "2020-04-10 00:01:53,618 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 00:01:53,618 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 00:01:53,618 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-10 00:01:53,618 Example #1\n", "2020-04-10 00:01:53,619 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 00:01:53,619 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 00:01:53,619 \tHypothesis: Had We willed , We would have made it a scalf , so why do you not give thanks ?\n", "2020-04-10 00:01:53,619 Example #2\n", "2020-04-10 00:01:53,619 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 00:01:53,620 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 00:01:53,620 \tHypothesis: And those who disbelieve in Our signs are the people of the wicked .\n", "2020-04-10 00:01:53,620 Example #3\n", "2020-04-10 00:01:53,620 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 00:01:53,620 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 00:01:53,620 \tHypothesis: And He sends the ark in security , and when some of his people passed before him , they mocked him .\n", "2020-04-10 00:01:53,621 Validation result (greedy) at epoch 12, step 58000: bleu: 22.15, loss: 51899.0625, ppl: 6.8199, duration: 29.5948s\n", "2020-04-10 00:02:11,740 Epoch 12 Step: 58100 Batch Loss: 1.922659 Tokens per Sec: 10778, Lr: 0.000300\n", "2020-04-10 00:02:30,084 Epoch 12 Step: 58200 Batch Loss: 2.083060 Tokens per Sec: 11018, Lr: 0.000300\n", "2020-04-10 00:02:48,501 Epoch 12 Step: 58300 Batch Loss: 1.356486 Tokens per Sec: 10978, Lr: 0.000300\n", "2020-04-10 00:03:06,553 Epoch 12 Step: 58400 Batch Loss: 1.794775 Tokens per Sec: 10615, Lr: 0.000300\n", "2020-04-10 00:03:24,946 Epoch 12 Step: 58500 Batch Loss: 1.955103 Tokens per Sec: 10933, Lr: 0.000300\n", "2020-04-10 00:03:43,398 Epoch 12 Step: 58600 Batch Loss: 2.099446 Tokens per Sec: 10987, Lr: 0.000300\n", "2020-04-10 00:04:01,714 Epoch 12 Step: 58700 Batch Loss: 2.077677 Tokens per Sec: 10923, Lr: 0.000300\n", "2020-04-10 00:04:20,038 Epoch 12 Step: 58800 Batch Loss: 1.990346 Tokens per Sec: 10962, Lr: 0.000300\n", "2020-04-10 00:04:38,573 Epoch 12 Step: 58900 Batch Loss: 2.216553 Tokens per Sec: 10947, Lr: 0.000300\n", "2020-04-10 00:04:57,001 Epoch 12 Step: 59000 Batch Loss: 2.114666 Tokens per Sec: 11175, Lr: 0.000300\n", "2020-04-10 00:05:27,331 Hooray! New best validation result [ppl]!\n", "2020-04-10 00:05:27,331 Saving new checkpoint.\n", "2020-04-10 00:05:28,556 Example #0\n", "2020-04-10 00:05:28,557 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 00:05:28,557 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 00:05:28,557 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-10 00:05:28,557 Example #1\n", "2020-04-10 00:05:28,558 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 00:05:28,558 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 00:05:28,558 \tHypothesis: If We willed , We would have made it a scalf , so why do you not give thanks ?\n", "2020-04-10 00:05:28,559 Example #2\n", "2020-04-10 00:05:28,559 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 00:05:28,559 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 00:05:28,559 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-10 00:05:28,559 Example #3\n", "2020-04-10 00:05:28,560 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 00:05:28,560 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 00:05:28,560 \tHypothesis: He sends the ark in security , and when the leaders of his people passed before him , they mocked him .\n", "2020-04-10 00:05:28,560 Validation result (greedy) at epoch 12, step 59000: bleu: 22.63, loss: 51732.4727, ppl: 6.7780, duration: 31.5592s\n", "2020-04-10 00:05:46,742 Epoch 12 Step: 59100 Batch Loss: 1.985844 Tokens per Sec: 10742, Lr: 0.000300\n", "2020-04-10 00:06:04,941 Epoch 12 Step: 59200 Batch Loss: 2.322878 Tokens per Sec: 11029, Lr: 0.000300\n", "2020-04-10 00:06:22,888 Epoch 12 Step: 59300 Batch Loss: 1.939017 Tokens per Sec: 10749, Lr: 0.000300\n", "2020-04-10 00:06:41,259 Epoch 12 Step: 59400 Batch Loss: 2.131201 Tokens per Sec: 10872, Lr: 0.000300\n", "2020-04-10 00:06:59,063 Epoch 12 Step: 59500 Batch Loss: 1.773757 Tokens per Sec: 10521, Lr: 0.000300\n", "2020-04-10 00:07:17,191 Epoch 12 Step: 59600 Batch Loss: 1.899326 Tokens per Sec: 10818, Lr: 0.000300\n", "2020-04-10 00:07:35,363 Epoch 12 Step: 59700 Batch Loss: 1.969768 Tokens per Sec: 10890, Lr: 0.000300\n", "2020-04-10 00:07:53,639 Epoch 12 Step: 59800 Batch Loss: 2.106566 Tokens per Sec: 10788, Lr: 0.000300\n", "2020-04-10 00:08:12,149 Epoch 12 Step: 59900 Batch Loss: 1.954037 Tokens per Sec: 11133, Lr: 0.000300\n", "2020-04-10 00:08:30,434 Epoch 12 Step: 60000 Batch Loss: 2.156806 Tokens per Sec: 10947, Lr: 0.000300\n", "2020-04-10 00:09:02,015 Hooray! New best validation result [ppl]!\n", "2020-04-10 00:09:02,016 Saving new checkpoint.\n", "2020-04-10 00:09:03,512 Example #0\n", "2020-04-10 00:09:03,513 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 00:09:03,513 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 00:09:03,513 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 00:09:03,513 Example #1\n", "2020-04-10 00:09:03,514 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 00:09:03,514 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 00:09:03,514 \tHypothesis: If We willed , We would have brought it down a blazing water , so why do you not give thanks ?\n", "2020-04-10 00:09:03,514 Example #2\n", "2020-04-10 00:09:03,514 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 00:09:03,514 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 00:09:03,515 \tHypothesis: And those who disbelieve in Our signs are the companions of the people of the Last .\n", "2020-04-10 00:09:03,515 Example #3\n", "2020-04-10 00:09:03,515 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 00:09:03,515 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 00:09:03,515 \tHypothesis: And He sends the ships in security , and whenever there were the chiefs of his people passed away before him , they mocked at him .\n", "2020-04-10 00:09:03,516 Validation result (greedy) at epoch 12, step 60000: bleu: 22.11, loss: 51570.8320, ppl: 6.7376, duration: 33.0813s\n", "2020-04-10 00:09:21,891 Epoch 12 Step: 60100 Batch Loss: 2.108980 Tokens per Sec: 10939, Lr: 0.000300\n", "2020-04-10 00:09:40,426 Epoch 12 Step: 60200 Batch Loss: 2.014201 Tokens per Sec: 11255, Lr: 0.000300\n", "2020-04-10 00:09:58,719 Epoch 12 Step: 60300 Batch Loss: 1.751254 Tokens per Sec: 11030, Lr: 0.000300\n", "2020-04-10 00:10:16,874 Epoch 12 Step: 60400 Batch Loss: 2.079721 Tokens per Sec: 10953, Lr: 0.000300\n", "2020-04-10 00:10:35,374 Epoch 12 Step: 60500 Batch Loss: 2.099009 Tokens per Sec: 11084, Lr: 0.000300\n", "2020-04-10 00:10:53,709 Epoch 12 Step: 60600 Batch Loss: 2.114868 Tokens per Sec: 11041, Lr: 0.000300\n", "2020-04-10 00:11:12,108 Epoch 12 Step: 60700 Batch Loss: 2.079616 Tokens per Sec: 11057, Lr: 0.000300\n", "2020-04-10 00:11:29,959 Epoch 12 Step: 60800 Batch Loss: 1.737789 Tokens per Sec: 10644, Lr: 0.000300\n", "2020-04-10 00:11:48,485 Epoch 12 Step: 60900 Batch Loss: 2.157420 Tokens per Sec: 11195, Lr: 0.000300\n", "2020-04-10 00:12:06,829 Epoch 12 Step: 61000 Batch Loss: 1.606453 Tokens per Sec: 11110, Lr: 0.000300\n", "2020-04-10 00:12:36,332 Example #0\n", "2020-04-10 00:12:36,333 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 00:12:36,333 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 00:12:36,333 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-10 00:12:36,333 Example #1\n", "2020-04-10 00:12:36,333 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 00:12:36,334 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 00:12:36,334 \tHypothesis: Had We willed , We would have brought it down a scalf , so why do you not give thanks ?\n", "2020-04-10 00:12:36,334 Example #2\n", "2020-04-10 00:12:36,334 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 00:12:36,334 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 00:12:36,334 \tHypothesis: And those who disbelieve in Our signs are the companions of the nations .\n", "2020-04-10 00:12:36,335 Example #3\n", "2020-04-10 00:12:36,335 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 00:12:36,335 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 00:12:36,335 \tHypothesis: He sends the ships in security , and whenever the leaders of his people passed before him , they mock him .\n", "2020-04-10 00:12:36,335 Validation result (greedy) at epoch 12, step 61000: bleu: 22.47, loss: 51693.0586, ppl: 6.7681, duration: 29.5062s\n", "2020-04-10 00:12:39,317 Epoch 12: total training loss 8902.33\n", "2020-04-10 00:12:39,317 EPOCH 13\n", "2020-04-10 00:12:55,185 Epoch 13 Step: 61100 Batch Loss: 1.951422 Tokens per Sec: 10935, Lr: 0.000300\n", "2020-04-10 00:13:13,209 Epoch 13 Step: 61200 Batch Loss: 1.997106 Tokens per Sec: 10617, Lr: 0.000300\n", "2020-04-10 00:13:31,416 Epoch 13 Step: 61300 Batch Loss: 1.927300 Tokens per Sec: 11108, Lr: 0.000300\n", "2020-04-10 00:13:49,750 Epoch 13 Step: 61400 Batch Loss: 2.027156 Tokens per Sec: 11053, Lr: 0.000300\n", "2020-04-10 00:14:08,021 Epoch 13 Step: 61500 Batch Loss: 2.197399 Tokens per Sec: 10913, Lr: 0.000300\n", "2020-04-10 00:14:26,156 Epoch 13 Step: 61600 Batch Loss: 2.138607 Tokens per Sec: 10914, Lr: 0.000300\n", "2020-04-10 00:14:44,557 Epoch 13 Step: 61700 Batch Loss: 1.713070 Tokens per Sec: 11063, Lr: 0.000300\n", "2020-04-10 00:15:02,641 Epoch 13 Step: 61800 Batch Loss: 1.938872 Tokens per Sec: 11019, Lr: 0.000300\n", "2020-04-10 00:15:20,942 Epoch 13 Step: 61900 Batch Loss: 2.002184 Tokens per Sec: 10954, Lr: 0.000300\n", "2020-04-10 00:15:38,912 Epoch 13 Step: 62000 Batch Loss: 2.074891 Tokens per Sec: 10647, Lr: 0.000300\n", "2020-04-10 00:16:10,043 Hooray! New best validation result [ppl]!\n", "2020-04-10 00:16:10,043 Saving new checkpoint.\n", "2020-04-10 00:16:11,252 Example #0\n", "2020-04-10 00:16:11,252 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 00:16:11,253 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 00:16:11,253 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 00:16:11,253 Example #1\n", "2020-04-10 00:16:11,254 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 00:16:11,254 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 00:16:11,254 \tHypothesis: If We willed , We would have made it rain , so why do you not give thanks ?\n", "2020-04-10 00:16:11,254 Example #2\n", "2020-04-10 00:16:11,254 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 00:16:11,255 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 00:16:11,255 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-10 00:16:11,255 Example #3\n", "2020-04-10 00:16:11,255 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 00:16:11,256 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 00:16:11,256 \tHypothesis: He sends the ships in security , and when the leaders of his people passed before him , they mock him .\n", "2020-04-10 00:16:11,256 Validation result (greedy) at epoch 13, step 62000: bleu: 22.43, loss: 51519.8945, ppl: 6.7249, duration: 32.3433s\n", "2020-04-10 00:16:29,887 Epoch 13 Step: 62100 Batch Loss: 1.980540 Tokens per Sec: 11123, Lr: 0.000300\n", "2020-04-10 00:16:48,161 Epoch 13 Step: 62200 Batch Loss: 1.906483 Tokens per Sec: 10577, Lr: 0.000300\n", "2020-04-10 00:17:06,550 Epoch 13 Step: 62300 Batch Loss: 1.815236 Tokens per Sec: 11157, Lr: 0.000300\n", "2020-04-10 00:17:24,844 Epoch 13 Step: 62400 Batch Loss: 2.022269 Tokens per Sec: 10855, Lr: 0.000300\n", "2020-04-10 00:17:43,090 Epoch 13 Step: 62500 Batch Loss: 2.156144 Tokens per Sec: 11041, Lr: 0.000300\n", "2020-04-10 00:18:01,475 Epoch 13 Step: 62600 Batch Loss: 2.223251 Tokens per Sec: 11235, Lr: 0.000300\n", "2020-04-10 00:18:19,869 Epoch 13 Step: 62700 Batch Loss: 2.332296 Tokens per Sec: 11066, Lr: 0.000300\n", "2020-04-10 00:18:38,172 Epoch 13 Step: 62800 Batch Loss: 1.843963 Tokens per Sec: 10835, Lr: 0.000300\n", "2020-04-10 00:18:56,469 Epoch 13 Step: 62900 Batch Loss: 2.225714 Tokens per Sec: 11078, Lr: 0.000300\n", "2020-04-10 00:19:14,694 Epoch 13 Step: 63000 Batch Loss: 2.248154 Tokens per Sec: 10898, Lr: 0.000300\n", "2020-04-10 00:19:40,515 Hooray! New best validation result [ppl]!\n", "2020-04-10 00:19:40,516 Saving new checkpoint.\n", "2020-04-10 00:19:41,853 Example #0\n", "2020-04-10 00:19:41,854 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 00:19:41,854 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 00:19:41,854 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 00:19:41,854 Example #1\n", "2020-04-10 00:19:41,855 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 00:19:41,855 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 00:19:41,855 \tHypothesis: Had We willed , We could have made it a blazing water , so why do you not give thanks ?\n", "2020-04-10 00:19:41,855 Example #2\n", "2020-04-10 00:19:41,856 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 00:19:41,856 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 00:19:41,856 \tHypothesis: And those who disbelieve in Our signs are the people of the Last Day .\n", "2020-04-10 00:19:41,856 Example #3\n", "2020-04-10 00:19:41,856 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 00:19:41,857 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 00:19:41,857 \tHypothesis: He sends the ship in security , and whenever the leaders of his people passed before him , they mock him .\n", "2020-04-10 00:19:41,857 Validation result (greedy) at epoch 13, step 63000: bleu: 22.70, loss: 51195.3477, ppl: 6.6446, duration: 27.1621s\n", "2020-04-10 00:20:00,347 Epoch 13 Step: 63100 Batch Loss: 2.162723 Tokens per Sec: 10738, Lr: 0.000300\n", "2020-04-10 00:20:18,665 Epoch 13 Step: 63200 Batch Loss: 1.952012 Tokens per Sec: 11072, Lr: 0.000300\n", "2020-04-10 00:20:36,776 Epoch 13 Step: 63300 Batch Loss: 2.208684 Tokens per Sec: 10958, Lr: 0.000300\n", "2020-04-10 00:20:54,966 Epoch 13 Step: 63400 Batch Loss: 2.102542 Tokens per Sec: 10761, Lr: 0.000300\n", "2020-04-10 00:21:13,222 Epoch 13 Step: 63500 Batch Loss: 2.032974 Tokens per Sec: 11026, Lr: 0.000300\n", "2020-04-10 00:21:31,374 Epoch 13 Step: 63600 Batch Loss: 2.063340 Tokens per Sec: 10885, Lr: 0.000300\n", "2020-04-10 00:21:49,661 Epoch 13 Step: 63700 Batch Loss: 1.845389 Tokens per Sec: 10951, Lr: 0.000300\n", "2020-04-10 00:22:07,840 Epoch 13 Step: 63800 Batch Loss: 1.700786 Tokens per Sec: 11037, Lr: 0.000300\n", "2020-04-10 00:22:26,110 Epoch 13 Step: 63900 Batch Loss: 1.989260 Tokens per Sec: 10939, Lr: 0.000300\n", "2020-04-10 00:22:44,290 Epoch 13 Step: 64000 Batch Loss: 2.045196 Tokens per Sec: 10915, Lr: 0.000300\n", "2020-04-10 00:23:14,312 Example #0\n", "2020-04-10 00:23:14,312 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 00:23:14,312 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 00:23:14,312 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 00:23:14,313 Example #1\n", "2020-04-10 00:23:14,313 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 00:23:14,313 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 00:23:14,313 \tHypothesis: Had We willed , We would have brought it down a rain ; so why do you not give thanks ?\n", "2020-04-10 00:23:14,313 Example #2\n", "2020-04-10 00:23:14,314 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 00:23:14,314 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 00:23:14,314 \tHypothesis: Those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 00:23:14,314 Example #3\n", "2020-04-10 00:23:14,314 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 00:23:14,315 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 00:23:14,315 \tHypothesis: He sends the ark in security , and whenever some of his people passed away before him they mocked him .\n", "2020-04-10 00:23:14,315 Validation result (greedy) at epoch 13, step 64000: bleu: 22.50, loss: 51292.9297, ppl: 6.6687, duration: 30.0252s\n", "2020-04-10 00:23:32,632 Epoch 13 Step: 64100 Batch Loss: 2.019084 Tokens per Sec: 10977, Lr: 0.000300\n", "2020-04-10 00:23:50,851 Epoch 13 Step: 64200 Batch Loss: 2.140482 Tokens per Sec: 10747, Lr: 0.000300\n", "2020-04-10 00:24:09,093 Epoch 13 Step: 64300 Batch Loss: 2.019103 Tokens per Sec: 11197, Lr: 0.000300\n", "2020-04-10 00:24:27,472 Epoch 13 Step: 64400 Batch Loss: 2.012290 Tokens per Sec: 10918, Lr: 0.000300\n", "2020-04-10 00:24:45,683 Epoch 13 Step: 64500 Batch Loss: 2.073479 Tokens per Sec: 11111, Lr: 0.000300\n", "2020-04-10 00:25:04,032 Epoch 13 Step: 64600 Batch Loss: 2.106380 Tokens per Sec: 10909, Lr: 0.000300\n", "2020-04-10 00:25:22,572 Epoch 13 Step: 64700 Batch Loss: 1.755913 Tokens per Sec: 11325, Lr: 0.000300\n", "2020-04-10 00:25:40,963 Epoch 13 Step: 64800 Batch Loss: 1.878993 Tokens per Sec: 11128, Lr: 0.000300\n", "2020-04-10 00:25:59,453 Epoch 13 Step: 64900 Batch Loss: 2.018988 Tokens per Sec: 11167, Lr: 0.000300\n", "2020-04-10 00:26:17,675 Epoch 13 Step: 65000 Batch Loss: 1.831195 Tokens per Sec: 10894, Lr: 0.000300\n", "2020-04-10 00:26:49,174 Example #0\n", "2020-04-10 00:26:49,175 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 00:26:49,175 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 00:26:49,175 \tHypothesis: See how We explain Our signs that they may understand .\n", "2020-04-10 00:26:49,175 Example #1\n", "2020-04-10 00:26:49,176 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 00:26:49,176 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 00:26:49,176 \tHypothesis: If We willed , We could have made it a rain , so why do you not give thanks ?\n", "2020-04-10 00:26:49,176 Example #2\n", "2020-04-10 00:26:49,177 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 00:26:49,177 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 00:26:49,177 \tHypothesis: And those who disbelieve in Our signs are the companions of the people of the nations .\n", "2020-04-10 00:26:49,177 Example #3\n", "2020-04-10 00:26:49,177 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 00:26:49,177 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 00:26:49,178 \tHypothesis: And He sends the ark in security , and whenever some of his people passed away before him they mocked him .\n", "2020-04-10 00:26:49,178 Validation result (greedy) at epoch 13, step 65000: bleu: 22.38, loss: 51368.1445, ppl: 6.6872, duration: 31.5021s\n", "2020-04-10 00:27:07,476 Epoch 13 Step: 65100 Batch Loss: 1.389902 Tokens per Sec: 11012, Lr: 0.000300\n", "2020-04-10 00:27:26,008 Epoch 13 Step: 65200 Batch Loss: 1.927866 Tokens per Sec: 11095, Lr: 0.000300\n", "2020-04-10 00:27:44,315 Epoch 13 Step: 65300 Batch Loss: 2.159589 Tokens per Sec: 10989, Lr: 0.000300\n", "2020-04-10 00:28:02,422 Epoch 13 Step: 65400 Batch Loss: 2.090349 Tokens per Sec: 10725, Lr: 0.000300\n", "2020-04-10 00:28:08,724 Epoch 13: total training loss 8806.44\n", "2020-04-10 00:28:08,725 EPOCH 14\n", "2020-04-10 00:28:21,127 Epoch 14 Step: 65500 Batch Loss: 1.866094 Tokens per Sec: 10715, Lr: 0.000300\n", "2020-04-10 00:28:39,596 Epoch 14 Step: 65600 Batch Loss: 1.818756 Tokens per Sec: 11154, Lr: 0.000300\n", "2020-04-10 00:28:57,990 Epoch 14 Step: 65700 Batch Loss: 1.968616 Tokens per Sec: 10891, Lr: 0.000300\n", "2020-04-10 00:29:16,419 Epoch 14 Step: 65800 Batch Loss: 1.723485 Tokens per Sec: 11220, Lr: 0.000300\n", "2020-04-10 00:29:34,506 Epoch 14 Step: 65900 Batch Loss: 2.196171 Tokens per Sec: 10943, Lr: 0.000300\n", "2020-04-10 00:29:52,634 Epoch 14 Step: 66000 Batch Loss: 2.055669 Tokens per Sec: 10906, Lr: 0.000300\n", "2020-04-10 00:30:24,258 Hooray! New best validation result [ppl]!\n", "2020-04-10 00:30:24,258 Saving new checkpoint.\n", "2020-04-10 00:30:25,471 Example #0\n", "2020-04-10 00:30:25,471 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 00:30:25,472 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 00:30:25,472 \tHypothesis: Look how We explain the signs that they may understand .\n", "2020-04-10 00:30:25,472 Example #1\n", "2020-04-10 00:30:25,472 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 00:30:25,472 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 00:30:25,473 \tHypothesis: Had We willed , We would have brought it down to a blind , so why do you not give thanks ?\n", "2020-04-10 00:30:25,473 Example #2\n", "2020-04-10 00:30:25,473 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 00:30:25,473 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 00:30:25,473 \tHypothesis: And those who disbelieve in Our signs are the dwellers of the nations .\n", "2020-04-10 00:30:25,473 Example #3\n", "2020-04-10 00:30:25,474 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 00:30:25,474 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 00:30:25,474 \tHypothesis: He sends the ark in security , and whenever the leaders of his people passed away before him , they mock him .\n", "2020-04-10 00:30:25,474 Validation result (greedy) at epoch 14, step 66000: bleu: 23.04, loss: 51037.9961, ppl: 6.6061, duration: 32.8401s\n", "2020-04-10 00:30:43,917 Epoch 14 Step: 66100 Batch Loss: 1.962536 Tokens per Sec: 10964, Lr: 0.000300\n", "2020-04-10 00:31:02,414 Epoch 14 Step: 66200 Batch Loss: 2.044894 Tokens per Sec: 11117, Lr: 0.000300\n", "2020-04-10 00:31:20,748 Epoch 14 Step: 66300 Batch Loss: 2.042278 Tokens per Sec: 10860, Lr: 0.000300\n", "2020-04-10 00:31:38,957 Epoch 14 Step: 66400 Batch Loss: 1.522688 Tokens per Sec: 10964, Lr: 0.000300\n", "2020-04-10 00:31:57,261 Epoch 14 Step: 66500 Batch Loss: 1.536105 Tokens per Sec: 11237, Lr: 0.000300\n", "2020-04-10 00:32:15,511 Epoch 14 Step: 66600 Batch Loss: 1.975232 Tokens per Sec: 10916, Lr: 0.000300\n", "2020-04-10 00:32:33,803 Epoch 14 Step: 66700 Batch Loss: 2.235552 Tokens per Sec: 11153, Lr: 0.000300\n", "2020-04-10 00:32:52,225 Epoch 14 Step: 66800 Batch Loss: 2.013879 Tokens per Sec: 10928, Lr: 0.000300\n", "2020-04-10 00:33:10,528 Epoch 14 Step: 66900 Batch Loss: 1.920696 Tokens per Sec: 10999, Lr: 0.000300\n", "2020-04-10 00:33:28,645 Epoch 14 Step: 67000 Batch Loss: 2.038061 Tokens per Sec: 10952, Lr: 0.000300\n", "2020-04-10 00:33:56,036 Example #0\n", "2020-04-10 00:33:56,036 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 00:33:56,037 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 00:33:56,037 \tHypothesis: Look ! how We explain Our signs that haply they may understand .\n", "2020-04-10 00:33:56,037 Example #1\n", "2020-04-10 00:33:56,037 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 00:33:56,040 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 00:33:56,041 \tHypothesis: If We willed , We could have made it a severe rain , so why do you not give thanks ?\n", "2020-04-10 00:33:56,041 Example #2\n", "2020-04-10 00:33:56,042 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 00:33:56,042 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 00:33:56,042 \tHypothesis: Those who disbelieve in Our signs are the people of the nations .\n", "2020-04-10 00:33:56,042 Example #3\n", "2020-04-10 00:33:56,043 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 00:33:56,043 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 00:33:56,043 \tHypothesis: And He sends the ships in security , and whenever some of his people passed away before him , they mocked him .\n", "2020-04-10 00:33:56,043 Validation result (greedy) at epoch 14, step 67000: bleu: 22.19, loss: 51285.6094, ppl: 6.6669, duration: 27.3973s\n", "2020-04-10 00:34:14,405 Epoch 14 Step: 67100 Batch Loss: 1.963850 Tokens per Sec: 11036, Lr: 0.000300\n", "2020-04-10 00:34:32,733 Epoch 14 Step: 67200 Batch Loss: 1.781273 Tokens per Sec: 10921, Lr: 0.000300\n", "2020-04-10 00:34:50,959 Epoch 14 Step: 67300 Batch Loss: 1.952495 Tokens per Sec: 11138, Lr: 0.000300\n", "2020-04-10 00:35:09,205 Epoch 14 Step: 67400 Batch Loss: 2.078780 Tokens per Sec: 11023, Lr: 0.000300\n", "2020-04-10 00:35:27,332 Epoch 14 Step: 67500 Batch Loss: 1.957240 Tokens per Sec: 10857, Lr: 0.000300\n", "2020-04-10 00:35:45,598 Epoch 14 Step: 67600 Batch Loss: 2.166666 Tokens per Sec: 10945, Lr: 0.000300\n", "2020-04-10 00:36:03,915 Epoch 14 Step: 67700 Batch Loss: 1.957902 Tokens per Sec: 11080, Lr: 0.000300\n", "2020-04-10 00:36:22,112 Epoch 14 Step: 67800 Batch Loss: 2.148292 Tokens per Sec: 10938, Lr: 0.000300\n", "2020-04-10 00:36:40,453 Epoch 14 Step: 67900 Batch Loss: 2.139485 Tokens per Sec: 10945, Lr: 0.000300\n", "2020-04-10 00:36:58,673 Epoch 14 Step: 68000 Batch Loss: 1.913926 Tokens per Sec: 10661, Lr: 0.000300\n", "2020-04-10 00:37:30,464 Hooray! New best validation result [ppl]!\n", "2020-04-10 00:37:30,464 Saving new checkpoint.\n", "2020-04-10 00:37:31,679 Example #0\n", "2020-04-10 00:37:31,680 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 00:37:31,680 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 00:37:31,680 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 00:37:31,680 Example #1\n", "2020-04-10 00:37:31,681 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 00:37:31,681 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 00:37:31,681 \tHypothesis: Had We willed , We would have made it a scale , so why do you not give thanks ?\n", "2020-04-10 00:37:31,681 Example #2\n", "2020-04-10 00:37:31,682 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 00:37:31,682 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 00:37:31,682 \tHypothesis: And those who disbelieve in Our signs are the People of the Left .\n", "2020-04-10 00:37:31,682 Example #3\n", "2020-04-10 00:37:31,683 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 00:37:31,683 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 00:37:31,683 \tHypothesis: And He sends the ship in a cup , and whenever some of his people passed away before him , they mock at him .\n", "2020-04-10 00:37:31,683 Validation result (greedy) at epoch 14, step 68000: bleu: 22.96, loss: 50936.9648, ppl: 6.5814, duration: 33.0094s\n", "2020-04-10 00:37:50,358 Epoch 14 Step: 68100 Batch Loss: 2.041265 Tokens per Sec: 11040, Lr: 0.000300\n", "2020-04-10 00:38:08,683 Epoch 14 Step: 68200 Batch Loss: 1.951928 Tokens per Sec: 10823, Lr: 0.000300\n", "2020-04-10 00:38:26,731 Epoch 14 Step: 68300 Batch Loss: 1.918482 Tokens per Sec: 10885, Lr: 0.000300\n", "2020-04-10 00:38:44,865 Epoch 14 Step: 68400 Batch Loss: 2.056036 Tokens per Sec: 10927, Lr: 0.000300\n", "2020-04-10 00:39:03,282 Epoch 14 Step: 68500 Batch Loss: 2.259863 Tokens per Sec: 11112, Lr: 0.000300\n", "2020-04-10 00:39:21,502 Epoch 14 Step: 68600 Batch Loss: 1.837015 Tokens per Sec: 10906, Lr: 0.000300\n", "2020-04-10 00:39:40,009 Epoch 14 Step: 68700 Batch Loss: 1.521235 Tokens per Sec: 10976, Lr: 0.000300\n", "2020-04-10 00:39:58,378 Epoch 14 Step: 68800 Batch Loss: 2.082549 Tokens per Sec: 11017, Lr: 0.000300\n", "2020-04-10 00:40:16,861 Epoch 14 Step: 68900 Batch Loss: 2.020568 Tokens per Sec: 11272, Lr: 0.000300\n", "2020-04-10 00:40:34,787 Epoch 14 Step: 69000 Batch Loss: 2.058884 Tokens per Sec: 10630, Lr: 0.000300\n", "2020-04-10 00:41:06,065 Hooray! New best validation result [ppl]!\n", "2020-04-10 00:41:06,066 Saving new checkpoint.\n", "2020-04-10 00:41:07,259 Example #0\n", "2020-04-10 00:41:07,261 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 00:41:07,261 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 00:41:07,261 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 00:41:07,262 Example #1\n", "2020-04-10 00:41:07,262 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 00:41:07,262 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 00:41:07,262 \tHypothesis: Had We willed , We would have made it a rain , so why do you not give thanks ?\n", "2020-04-10 00:41:07,262 Example #2\n", "2020-04-10 00:41:07,263 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 00:41:07,263 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 00:41:07,263 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-10 00:41:07,263 Example #3\n", "2020-04-10 00:41:07,264 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 00:41:07,264 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 00:41:07,264 \tHypothesis: He sends the ship in security , and whenever some of his people passed away before him , they mock him .\n", "2020-04-10 00:41:07,264 Validation result (greedy) at epoch 14, step 69000: bleu: 22.95, loss: 50715.9688, ppl: 6.5278, duration: 32.4772s\n", "2020-04-10 00:41:25,920 Epoch 14 Step: 69100 Batch Loss: 2.107017 Tokens per Sec: 10939, Lr: 0.000300\n", "2020-04-10 00:41:44,042 Epoch 14 Step: 69200 Batch Loss: 1.982954 Tokens per Sec: 10828, Lr: 0.000300\n", "2020-04-10 00:42:02,197 Epoch 14 Step: 69300 Batch Loss: 2.180037 Tokens per Sec: 10940, Lr: 0.000300\n", "2020-04-10 00:42:20,529 Epoch 14 Step: 69400 Batch Loss: 1.816124 Tokens per Sec: 10996, Lr: 0.000300\n", "2020-04-10 00:42:38,705 Epoch 14 Step: 69500 Batch Loss: 1.984770 Tokens per Sec: 10859, Lr: 0.000300\n", "2020-04-10 00:42:57,074 Epoch 14 Step: 69600 Batch Loss: 2.038921 Tokens per Sec: 10899, Lr: 0.000300\n", "2020-04-10 00:43:15,397 Epoch 14 Step: 69700 Batch Loss: 1.957741 Tokens per Sec: 11028, Lr: 0.000300\n", "2020-04-10 00:43:33,739 Epoch 14 Step: 69800 Batch Loss: 1.958010 Tokens per Sec: 11089, Lr: 0.000300\n", "2020-04-10 00:43:42,729 Epoch 14: total training loss 8720.33\n", "2020-04-10 00:43:42,729 EPOCH 15\n", "2020-04-10 00:43:52,554 Epoch 15 Step: 69900 Batch Loss: 2.078084 Tokens per Sec: 10680, Lr: 0.000300\n", "2020-04-10 00:44:10,952 Epoch 15 Step: 70000 Batch Loss: 2.210284 Tokens per Sec: 11313, Lr: 0.000300\n", "2020-04-10 00:44:38,850 Hooray! New best validation result [ppl]!\n", "2020-04-10 00:44:38,851 Saving new checkpoint.\n", "2020-04-10 00:44:40,475 Example #0\n", "2020-04-10 00:44:40,476 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 00:44:40,476 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 00:44:40,476 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 00:44:40,476 Example #1\n", "2020-04-10 00:44:40,477 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 00:44:40,477 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 00:44:40,477 \tHypothesis: Had We willed , We would have brought it down to a shining shining , so why do you not give thanks ?\n", "2020-04-10 00:44:40,477 Example #2\n", "2020-04-10 00:44:40,477 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 00:44:40,478 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 00:44:40,478 \tHypothesis: Those who disbelieve in Our signs are the people of the Last Day .\n", "2020-04-10 00:44:40,478 Example #3\n", "2020-04-10 00:44:40,478 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 00:44:40,478 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 00:44:40,478 \tHypothesis: He sends the ark in security , and whenever some of his people passed away before him , they mocked at him .\n", "2020-04-10 00:44:40,479 Validation result (greedy) at epoch 15, step 70000: bleu: 22.91, loss: 50636.3086, ppl: 6.5086, duration: 29.5266s\n", "2020-04-10 00:44:58,692 Epoch 15 Step: 70100 Batch Loss: 1.982051 Tokens per Sec: 10838, Lr: 0.000300\n", "2020-04-10 00:45:16,765 Epoch 15 Step: 70200 Batch Loss: 2.024493 Tokens per Sec: 10972, Lr: 0.000300\n", "2020-04-10 00:45:35,226 Epoch 15 Step: 70300 Batch Loss: 2.110310 Tokens per Sec: 11172, Lr: 0.000300\n", "2020-04-10 00:45:53,374 Epoch 15 Step: 70400 Batch Loss: 2.027920 Tokens per Sec: 10889, Lr: 0.000300\n", "2020-04-10 00:46:11,570 Epoch 15 Step: 70500 Batch Loss: 1.965156 Tokens per Sec: 10938, Lr: 0.000300\n", "2020-04-10 00:46:29,995 Epoch 15 Step: 70600 Batch Loss: 1.913314 Tokens per Sec: 11052, Lr: 0.000300\n", "2020-04-10 00:46:48,351 Epoch 15 Step: 70700 Batch Loss: 1.959119 Tokens per Sec: 11169, Lr: 0.000300\n", "2020-04-10 00:47:06,747 Epoch 15 Step: 70800 Batch Loss: 2.076412 Tokens per Sec: 10890, Lr: 0.000300\n", "2020-04-10 00:47:25,176 Epoch 15 Step: 70900 Batch Loss: 1.971274 Tokens per Sec: 11116, Lr: 0.000300\n", "2020-04-10 00:47:43,550 Epoch 15 Step: 71000 Batch Loss: 1.877722 Tokens per Sec: 10732, Lr: 0.000300\n", "2020-04-10 00:48:14,086 Example #0\n", "2020-04-10 00:48:14,087 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 00:48:14,087 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 00:48:14,087 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 00:48:14,087 Example #1\n", "2020-04-10 00:48:14,088 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 00:48:14,088 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 00:48:14,088 \tHypothesis: If We willed , We would have brought it down to a rain , so why do you not give thanks ?\n", "2020-04-10 00:48:14,088 Example #2\n", "2020-04-10 00:48:14,089 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 00:48:14,089 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 00:48:14,089 \tHypothesis: And those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 00:48:14,089 Example #3\n", "2020-04-10 00:48:14,090 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 00:48:14,090 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 00:48:14,090 \tHypothesis: And he smoke the ark in the crafty , and when there came to his people , they mocked at him .\n", "2020-04-10 00:48:14,090 Validation result (greedy) at epoch 15, step 71000: bleu: 23.23, loss: 50686.0820, ppl: 6.5206, duration: 30.5393s\n", "2020-04-10 00:48:32,436 Epoch 15 Step: 71100 Batch Loss: 1.638475 Tokens per Sec: 10919, Lr: 0.000300\n", "2020-04-10 00:48:50,667 Epoch 15 Step: 71200 Batch Loss: 1.844209 Tokens per Sec: 10812, Lr: 0.000300\n", "2020-04-10 00:49:08,708 Epoch 15 Step: 71300 Batch Loss: 1.948277 Tokens per Sec: 10989, Lr: 0.000300\n", "2020-04-10 00:49:26,872 Epoch 15 Step: 71400 Batch Loss: 2.060926 Tokens per Sec: 10877, Lr: 0.000300\n", "2020-04-10 00:49:45,224 Epoch 15 Step: 71500 Batch Loss: 1.657561 Tokens per Sec: 11019, Lr: 0.000300\n", "2020-04-10 00:50:03,482 Epoch 15 Step: 71600 Batch Loss: 1.743101 Tokens per Sec: 10998, Lr: 0.000300\n", "2020-04-10 00:50:21,614 Epoch 15 Step: 71700 Batch Loss: 1.961392 Tokens per Sec: 10933, Lr: 0.000300\n", "2020-04-10 00:50:39,680 Epoch 15 Step: 71800 Batch Loss: 2.056450 Tokens per Sec: 10752, Lr: 0.000300\n", "2020-04-10 00:50:57,977 Epoch 15 Step: 71900 Batch Loss: 2.014215 Tokens per Sec: 10894, Lr: 0.000300\n", "2020-04-10 00:51:16,209 Epoch 15 Step: 72000 Batch Loss: 1.957076 Tokens per Sec: 11077, Lr: 0.000300\n", "2020-04-10 00:51:48,331 Hooray! New best validation result [ppl]!\n", "2020-04-10 00:51:48,332 Saving new checkpoint.\n", "2020-04-10 00:51:49,585 Example #0\n", "2020-04-10 00:51:49,586 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 00:51:49,586 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 00:51:49,586 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-10 00:51:49,586 Example #1\n", "2020-04-10 00:51:49,586 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 00:51:49,587 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 00:51:49,587 \tHypothesis: If We willed , We would have brought it down to a rain , so why do you not give thanks ?\n", "2020-04-10 00:51:49,587 Example #2\n", "2020-04-10 00:51:49,587 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 00:51:49,587 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 00:51:49,587 \tHypothesis: And those who disbelieve in Our signs are the People of the Last Day .\n", "2020-04-10 00:51:49,588 Example #3\n", "2020-04-10 00:51:49,588 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 00:51:49,588 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 00:51:49,588 \tHypothesis: He sends the ship in security , and whenever some of his people passed away before him , they mock him .\n", "2020-04-10 00:51:49,588 Validation result (greedy) at epoch 15, step 72000: bleu: 22.84, loss: 50441.1875, ppl: 6.4618, duration: 33.3787s\n", "2020-04-10 00:52:08,159 Epoch 15 Step: 72100 Batch Loss: 1.985393 Tokens per Sec: 11029, Lr: 0.000300\n", "2020-04-10 00:52:26,511 Epoch 15 Step: 72200 Batch Loss: 1.587585 Tokens per Sec: 10916, Lr: 0.000300\n", "2020-04-10 00:52:44,982 Epoch 15 Step: 72300 Batch Loss: 2.111837 Tokens per Sec: 10978, Lr: 0.000300\n", "2020-04-10 00:53:03,413 Epoch 15 Step: 72400 Batch Loss: 1.961910 Tokens per Sec: 11003, Lr: 0.000300\n", "2020-04-10 00:53:21,903 Epoch 15 Step: 72500 Batch Loss: 1.986619 Tokens per Sec: 11026, Lr: 0.000300\n", "2020-04-10 00:53:40,092 Epoch 15 Step: 72600 Batch Loss: 1.602548 Tokens per Sec: 10918, Lr: 0.000300\n", "2020-04-10 00:53:58,291 Epoch 15 Step: 72700 Batch Loss: 1.945370 Tokens per Sec: 10873, Lr: 0.000300\n", "2020-04-10 00:54:16,315 Epoch 15 Step: 72800 Batch Loss: 2.180970 Tokens per Sec: 10939, Lr: 0.000300\n", "2020-04-10 00:54:34,688 Epoch 15 Step: 72900 Batch Loss: 2.246078 Tokens per Sec: 10961, Lr: 0.000300\n", "2020-04-10 00:54:52,725 Epoch 15 Step: 73000 Batch Loss: 1.940509 Tokens per Sec: 10795, Lr: 0.000300\n", "2020-04-10 00:55:23,365 Example #0\n", "2020-04-10 00:55:23,366 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 00:55:23,366 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 00:55:23,366 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 00:55:23,366 Example #1\n", "2020-04-10 00:55:23,367 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 00:55:23,367 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 00:55:23,367 \tHypothesis: If We had willed , We would have brought it down , so why do you not give thanks ?\n", "2020-04-10 00:55:23,367 Example #2\n", "2020-04-10 00:55:23,367 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 00:55:23,367 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 00:55:23,368 \tHypothesis: And those who disbelieve in Our signs are the people of the Last Day .\n", "2020-04-10 00:55:23,368 Example #3\n", "2020-04-10 00:55:23,368 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 00:55:23,368 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 00:55:23,368 \tHypothesis: He causes the ship to be smoke , and whenever there came to the leaders of his people passed away before him , they mocked him .\n", "2020-04-10 00:55:23,369 Validation result (greedy) at epoch 15, step 73000: bleu: 23.01, loss: 50537.3047, ppl: 6.4848, duration: 30.6426s\n", "2020-04-10 00:55:41,620 Epoch 15 Step: 73100 Batch Loss: 2.040757 Tokens per Sec: 10969, Lr: 0.000300\n", "2020-04-10 00:55:59,839 Epoch 15 Step: 73200 Batch Loss: 2.063497 Tokens per Sec: 11030, Lr: 0.000300\n", "2020-04-10 00:56:17,948 Epoch 15 Step: 73300 Batch Loss: 1.944863 Tokens per Sec: 10834, Lr: 0.000300\n", "2020-04-10 00:56:36,273 Epoch 15 Step: 73400 Batch Loss: 1.852272 Tokens per Sec: 10816, Lr: 0.000300\n", "2020-04-10 00:56:54,514 Epoch 15 Step: 73500 Batch Loss: 1.706273 Tokens per Sec: 10993, Lr: 0.000300\n", "2020-04-10 00:57:12,807 Epoch 15 Step: 73600 Batch Loss: 1.724387 Tokens per Sec: 10920, Lr: 0.000300\n", "2020-04-10 00:57:30,971 Epoch 15 Step: 73700 Batch Loss: 2.101259 Tokens per Sec: 10870, Lr: 0.000300\n", "2020-04-10 00:57:49,237 Epoch 15 Step: 73800 Batch Loss: 1.808204 Tokens per Sec: 10838, Lr: 0.000300\n", "2020-04-10 00:58:07,388 Epoch 15 Step: 73900 Batch Loss: 2.035917 Tokens per Sec: 11033, Lr: 0.000300\n", "2020-04-10 00:58:25,722 Epoch 15 Step: 74000 Batch Loss: 1.899449 Tokens per Sec: 11129, Lr: 0.000300\n", "2020-04-10 00:58:55,180 Hooray! New best validation result [ppl]!\n", "2020-04-10 00:58:55,180 Saving new checkpoint.\n", "2020-04-10 00:58:56,383 Example #0\n", "2020-04-10 00:58:56,384 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 00:58:56,385 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 00:58:56,385 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-10 00:58:56,386 Example #1\n", "2020-04-10 00:58:56,386 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 00:58:56,387 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 00:58:56,387 \tHypothesis: If We willed , We would have made it a rain , so why do you not give thanks ?\n", "2020-04-10 00:58:56,387 Example #2\n", "2020-04-10 00:58:56,388 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 00:58:56,388 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 00:58:56,388 \tHypothesis: And those who disbelieve in Our signs are the people of the nations .\n", "2020-04-10 00:58:56,388 Example #3\n", "2020-04-10 00:58:56,389 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 00:58:56,389 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 00:58:56,389 \tHypothesis: He sends the ship in security , and whenever there came to his people passed away before him , they mocked at him .\n", "2020-04-10 00:58:56,389 Validation result (greedy) at epoch 15, step 74000: bleu: 23.24, loss: 50347.6016, ppl: 6.4395, duration: 30.6663s\n", "2020-04-10 00:59:14,989 Epoch 15 Step: 74100 Batch Loss: 1.977336 Tokens per Sec: 11037, Lr: 0.000300\n", "2020-04-10 00:59:33,558 Epoch 15 Step: 74200 Batch Loss: 2.133296 Tokens per Sec: 11227, Lr: 0.000300\n", "2020-04-10 00:59:46,468 Epoch 15: total training loss 8661.53\n", "2020-04-10 00:59:46,469 EPOCH 16\n", "2020-04-10 00:59:52,416 Epoch 16 Step: 74300 Batch Loss: 2.146594 Tokens per Sec: 10334, Lr: 0.000300\n", "2020-04-10 01:00:10,828 Epoch 16 Step: 74400 Batch Loss: 1.970908 Tokens per Sec: 10928, Lr: 0.000300\n", "2020-04-10 01:00:28,925 Epoch 16 Step: 74500 Batch Loss: 1.899306 Tokens per Sec: 11161, Lr: 0.000300\n", "2020-04-10 01:00:47,234 Epoch 16 Step: 74600 Batch Loss: 1.958111 Tokens per Sec: 11092, Lr: 0.000300\n", "2020-04-10 01:01:05,459 Epoch 16 Step: 74700 Batch Loss: 2.015564 Tokens per Sec: 10848, Lr: 0.000300\n", "2020-04-10 01:01:23,401 Epoch 16 Step: 74800 Batch Loss: 1.955702 Tokens per Sec: 10858, Lr: 0.000300\n", "2020-04-10 01:01:41,857 Epoch 16 Step: 74900 Batch Loss: 1.948092 Tokens per Sec: 11016, Lr: 0.000300\n", "2020-04-10 01:02:00,028 Epoch 16 Step: 75000 Batch Loss: 1.675818 Tokens per Sec: 10857, Lr: 0.000300\n", "2020-04-10 01:02:30,715 Hooray! New best validation result [ppl]!\n", "2020-04-10 01:02:30,716 Saving new checkpoint.\n", "2020-04-10 01:02:31,952 Example #0\n", "2020-04-10 01:02:31,952 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 01:02:31,952 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 01:02:31,952 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 01:02:31,953 Example #1\n", "2020-04-10 01:02:31,953 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 01:02:31,953 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 01:02:31,953 \tHypothesis: If We had willed , We would have made it a rain , so why do you not give thanks ?\n", "2020-04-10 01:02:31,953 Example #2\n", "2020-04-10 01:02:31,954 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 01:02:31,954 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 01:02:31,954 \tHypothesis: And those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 01:02:31,954 Example #3\n", "2020-04-10 01:02:31,955 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 01:02:31,955 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 01:02:31,955 \tHypothesis: And He causes the ship to be safe , and whenever there came to his people passed before him , they mocked him .\n", "2020-04-10 01:02:31,955 Validation result (greedy) at epoch 16, step 75000: bleu: 23.20, loss: 50295.0117, ppl: 6.4270, duration: 31.9266s\n", "2020-04-10 01:02:50,443 Epoch 16 Step: 75100 Batch Loss: 1.945771 Tokens per Sec: 10808, Lr: 0.000300\n", "2020-04-10 01:03:08,944 Epoch 16 Step: 75200 Batch Loss: 1.939548 Tokens per Sec: 11133, Lr: 0.000300\n", "2020-04-10 01:03:27,113 Epoch 16 Step: 75300 Batch Loss: 2.183349 Tokens per Sec: 11011, Lr: 0.000300\n", "2020-04-10 01:03:45,332 Epoch 16 Step: 75400 Batch Loss: 1.680499 Tokens per Sec: 10929, Lr: 0.000300\n", "2020-04-10 01:04:03,704 Epoch 16 Step: 75500 Batch Loss: 2.030691 Tokens per Sec: 11091, Lr: 0.000300\n", "2020-04-10 01:04:21,973 Epoch 16 Step: 75600 Batch Loss: 1.664226 Tokens per Sec: 11032, Lr: 0.000300\n", "2020-04-10 01:04:40,315 Epoch 16 Step: 75700 Batch Loss: 1.897389 Tokens per Sec: 10926, Lr: 0.000300\n", "2020-04-10 01:04:58,764 Epoch 16 Step: 75800 Batch Loss: 1.984963 Tokens per Sec: 11059, Lr: 0.000300\n", "2020-04-10 01:05:17,093 Epoch 16 Step: 75900 Batch Loss: 1.761254 Tokens per Sec: 10725, Lr: 0.000300\n", "2020-04-10 01:05:35,336 Epoch 16 Step: 76000 Batch Loss: 2.017479 Tokens per Sec: 11090, Lr: 0.000300\n", "2020-04-10 01:06:04,950 Example #0\n", "2020-04-10 01:06:04,951 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 01:06:04,951 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 01:06:04,951 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 01:06:04,951 Example #1\n", "2020-04-10 01:06:04,952 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 01:06:04,952 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 01:06:04,952 \tHypothesis: If We willed , We would have made it rain , so why do you not give thanks ?\n", "2020-04-10 01:06:04,952 Example #2\n", "2020-04-10 01:06:04,952 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 01:06:04,953 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 01:06:04,953 \tHypothesis: And those who disbelieve in Our signs are the people of the Left ( Paradise ) .\n", "2020-04-10 01:06:04,953 Example #3\n", "2020-04-10 01:06:04,953 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 01:06:04,953 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 01:06:04,954 \tHypothesis: He sends the ships in security , and whenever some of his people passed away before him , they mocked him .\n", "2020-04-10 01:06:04,954 Validation result (greedy) at epoch 16, step 76000: bleu: 23.29, loss: 50363.2852, ppl: 6.4432, duration: 29.6176s\n", "2020-04-10 01:06:23,330 Epoch 16 Step: 76100 Batch Loss: 1.958177 Tokens per Sec: 11029, Lr: 0.000300\n", "2020-04-10 01:06:41,615 Epoch 16 Step: 76200 Batch Loss: 1.960025 Tokens per Sec: 10908, Lr: 0.000300\n", "2020-04-10 01:07:00,059 Epoch 16 Step: 76300 Batch Loss: 2.178654 Tokens per Sec: 11165, Lr: 0.000300\n", "2020-04-10 01:07:18,350 Epoch 16 Step: 76400 Batch Loss: 2.035555 Tokens per Sec: 10741, Lr: 0.000300\n", "2020-04-10 01:07:36,928 Epoch 16 Step: 76500 Batch Loss: 1.936760 Tokens per Sec: 10972, Lr: 0.000300\n", "2020-04-10 01:07:55,216 Epoch 16 Step: 76600 Batch Loss: 2.012488 Tokens per Sec: 10751, Lr: 0.000300\n", "2020-04-10 01:08:13,729 Epoch 16 Step: 76700 Batch Loss: 2.082892 Tokens per Sec: 10756, Lr: 0.000300\n", "2020-04-10 01:08:31,977 Epoch 16 Step: 76800 Batch Loss: 2.066797 Tokens per Sec: 11012, Lr: 0.000300\n", "2020-04-10 01:08:50,130 Epoch 16 Step: 76900 Batch Loss: 2.083507 Tokens per Sec: 10876, Lr: 0.000300\n", "2020-04-10 01:09:08,237 Epoch 16 Step: 77000 Batch Loss: 1.940081 Tokens per Sec: 10774, Lr: 0.000300\n", "2020-04-10 01:09:36,892 Hooray! New best validation result [ppl]!\n", "2020-04-10 01:09:36,893 Saving new checkpoint.\n", "2020-04-10 01:09:38,097 Example #0\n", "2020-04-10 01:09:38,098 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 01:09:38,098 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 01:09:38,098 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 01:09:38,098 Example #1\n", "2020-04-10 01:09:38,099 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 01:09:38,099 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 01:09:38,099 \tHypothesis: If We had willed , We would have brought it down rain , so why do you not give thanks ?\n", "2020-04-10 01:09:38,099 Example #2\n", "2020-04-10 01:09:38,099 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 01:09:38,100 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 01:09:38,100 \tHypothesis: And those who disbelieve in Our signs are the companions of the Left\n", "2020-04-10 01:09:38,100 Example #3\n", "2020-04-10 01:09:38,100 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 01:09:38,101 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 01:09:38,101 \tHypothesis: He causes the ships in security , and whenever some of his people passed away before him , they mocked him .\n", "2020-04-10 01:09:38,101 Validation result (greedy) at epoch 16, step 77000: bleu: 22.80, loss: 50225.0586, ppl: 6.4104, duration: 29.8631s\n", "2020-04-10 01:09:56,578 Epoch 16 Step: 77100 Batch Loss: 1.948406 Tokens per Sec: 10874, Lr: 0.000300\n", "2020-04-10 01:10:15,086 Epoch 16 Step: 77200 Batch Loss: 2.118070 Tokens per Sec: 10902, Lr: 0.000300\n", "2020-04-10 01:10:33,351 Epoch 16 Step: 77300 Batch Loss: 2.119480 Tokens per Sec: 10667, Lr: 0.000300\n", "2020-04-10 01:10:51,645 Epoch 16 Step: 77400 Batch Loss: 1.999082 Tokens per Sec: 11056, Lr: 0.000300\n", "2020-04-10 01:11:09,902 Epoch 16 Step: 77500 Batch Loss: 2.055047 Tokens per Sec: 10795, Lr: 0.000300\n", "2020-04-10 01:11:28,431 Epoch 16 Step: 77600 Batch Loss: 1.981309 Tokens per Sec: 11260, Lr: 0.000300\n", "2020-04-10 01:11:46,849 Epoch 16 Step: 77700 Batch Loss: 1.975920 Tokens per Sec: 10933, Lr: 0.000300\n", "2020-04-10 01:12:05,080 Epoch 16 Step: 77800 Batch Loss: 1.664971 Tokens per Sec: 10712, Lr: 0.000300\n", "2020-04-10 01:12:23,527 Epoch 16 Step: 77900 Batch Loss: 1.979631 Tokens per Sec: 11161, Lr: 0.000300\n", "2020-04-10 01:12:41,753 Epoch 16 Step: 78000 Batch Loss: 1.965306 Tokens per Sec: 10979, Lr: 0.000300\n", "2020-04-10 01:13:12,193 Example #0\n", "2020-04-10 01:13:12,193 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 01:13:12,194 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 01:13:12,194 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 01:13:12,194 Example #1\n", "2020-04-10 01:13:12,194 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 01:13:12,194 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 01:13:12,194 \tHypothesis: Had We willed , We would have made it a rain , so why do you not give thanks ?\n", "2020-04-10 01:13:12,195 Example #2\n", "2020-04-10 01:13:12,195 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 01:13:12,195 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 01:13:12,195 \tHypothesis: And those who disbelieve in Our signs are the people of the Right Hand .\n", "2020-04-10 01:13:12,195 Example #3\n", "2020-04-10 01:13:12,196 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 01:13:12,196 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 01:13:12,196 \tHypothesis: He sends the ship in security , and whenever there are some of his people passed away before him , they mocked him .\n", "2020-04-10 01:13:12,196 Validation result (greedy) at epoch 16, step 78000: bleu: 23.40, loss: 50386.0898, ppl: 6.4487, duration: 30.4428s\n", "2020-04-10 01:13:30,633 Epoch 16 Step: 78100 Batch Loss: 1.746766 Tokens per Sec: 10993, Lr: 0.000300\n", "2020-04-10 01:13:48,808 Epoch 16 Step: 78200 Batch Loss: 1.984667 Tokens per Sec: 10950, Lr: 0.000300\n", "2020-04-10 01:14:07,179 Epoch 16 Step: 78300 Batch Loss: 2.289903 Tokens per Sec: 11132, Lr: 0.000300\n", "2020-04-10 01:14:25,704 Epoch 16 Step: 78400 Batch Loss: 1.984331 Tokens per Sec: 11178, Lr: 0.000300\n", "2020-04-10 01:14:43,938 Epoch 16 Step: 78500 Batch Loss: 2.064981 Tokens per Sec: 11043, Lr: 0.000300\n", "2020-04-10 01:15:02,306 Epoch 16 Step: 78600 Batch Loss: 1.840224 Tokens per Sec: 11219, Lr: 0.000300\n", "2020-04-10 01:15:17,266 Epoch 16: total training loss 8578.26\n", "2020-04-10 01:15:17,266 EPOCH 17\n", "2020-04-10 01:15:21,084 Epoch 17 Step: 78700 Batch Loss: 1.791155 Tokens per Sec: 9421, Lr: 0.000300\n", "2020-04-10 01:15:39,553 Epoch 17 Step: 78800 Batch Loss: 2.138428 Tokens per Sec: 11121, Lr: 0.000300\n", "2020-04-10 01:15:57,627 Epoch 17 Step: 78900 Batch Loss: 1.785225 Tokens per Sec: 10748, Lr: 0.000300\n", "2020-04-10 01:16:15,702 Epoch 17 Step: 79000 Batch Loss: 2.019567 Tokens per Sec: 10911, Lr: 0.000300\n", "2020-04-10 01:16:44,098 Hooray! New best validation result [ppl]!\n", "2020-04-10 01:16:44,098 Saving new checkpoint.\n", "2020-04-10 01:16:45,311 Example #0\n", "2020-04-10 01:16:45,312 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 01:16:45,312 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 01:16:45,312 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-10 01:16:45,312 Example #1\n", "2020-04-10 01:16:45,313 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 01:16:45,313 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 01:16:45,313 \tHypothesis: If We willed , We would have made it a blazing water , so why do you not give thanks ?\n", "2020-04-10 01:16:45,313 Example #2\n", "2020-04-10 01:16:45,313 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 01:16:45,313 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 01:16:45,314 \tHypothesis: Those who disbelieve in Our signs are the people of the nations .\n", "2020-04-10 01:16:45,314 Example #3\n", "2020-04-10 01:16:45,314 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 01:16:45,314 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 01:16:45,314 \tHypothesis: He sends the ship in security , and whenever the chiefs of his people passed away before him , they mock him .\n", "2020-04-10 01:16:45,315 Validation result (greedy) at epoch 17, step 79000: bleu: 23.23, loss: 50002.6797, ppl: 6.3579, duration: 29.6119s\n", "2020-04-10 01:17:03,883 Epoch 17 Step: 79100 Batch Loss: 1.968331 Tokens per Sec: 10941, Lr: 0.000300\n", "2020-04-10 01:17:22,140 Epoch 17 Step: 79200 Batch Loss: 2.244795 Tokens per Sec: 10686, Lr: 0.000300\n", "2020-04-10 01:17:40,516 Epoch 17 Step: 79300 Batch Loss: 2.105882 Tokens per Sec: 11298, Lr: 0.000300\n", "2020-04-10 01:17:58,609 Epoch 17 Step: 79400 Batch Loss: 2.352234 Tokens per Sec: 10733, Lr: 0.000300\n", "2020-04-10 01:18:17,044 Epoch 17 Step: 79500 Batch Loss: 1.919507 Tokens per Sec: 11111, Lr: 0.000300\n", "2020-04-10 01:18:35,444 Epoch 17 Step: 79600 Batch Loss: 1.791587 Tokens per Sec: 11049, Lr: 0.000300\n", "2020-04-10 01:18:53,680 Epoch 17 Step: 79700 Batch Loss: 2.108922 Tokens per Sec: 11106, Lr: 0.000300\n", "2020-04-10 01:19:11,980 Epoch 17 Step: 79800 Batch Loss: 2.052773 Tokens per Sec: 11016, Lr: 0.000300\n", "2020-04-10 01:19:30,242 Epoch 17 Step: 79900 Batch Loss: 2.059338 Tokens per Sec: 11029, Lr: 0.000300\n", "2020-04-10 01:19:48,607 Epoch 17 Step: 80000 Batch Loss: 1.894915 Tokens per Sec: 10818, Lr: 0.000300\n", "2020-04-10 01:20:19,005 Hooray! New best validation result [ppl]!\n", "2020-04-10 01:20:19,005 Saving new checkpoint.\n", "2020-04-10 01:20:20,184 Example #0\n", "2020-04-10 01:20:20,185 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 01:20:20,185 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 01:20:20,185 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-10 01:20:20,185 Example #1\n", "2020-04-10 01:20:20,186 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 01:20:20,186 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 01:20:20,186 \tHypothesis: If We willed , We would have made it a blazing water , why do you not give thanks ?\n", "2020-04-10 01:20:20,186 Example #2\n", "2020-04-10 01:20:20,186 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 01:20:20,186 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 01:20:20,187 \tHypothesis: Those who disbelieve in Our signs are the companions of the Left .\n", "2020-04-10 01:20:20,187 Example #3\n", "2020-04-10 01:20:20,187 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 01:20:20,187 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 01:20:20,187 \tHypothesis: He causes the ship to be safe , and whenever the leaders of his people passed away before him , they mocked him .\n", "2020-04-10 01:20:20,188 Validation result (greedy) at epoch 17, step 80000: bleu: 23.38, loss: 49990.4805, ppl: 6.3550, duration: 31.5797s\n", "2020-04-10 01:20:38,367 Epoch 17 Step: 80100 Batch Loss: 1.763065 Tokens per Sec: 10718, Lr: 0.000300\n", "2020-04-10 01:20:56,602 Epoch 17 Step: 80200 Batch Loss: 2.106579 Tokens per Sec: 10843, Lr: 0.000300\n", "2020-04-10 01:21:14,928 Epoch 17 Step: 80300 Batch Loss: 2.089027 Tokens per Sec: 10960, Lr: 0.000300\n", "2020-04-10 01:21:33,178 Epoch 17 Step: 80400 Batch Loss: 1.656638 Tokens per Sec: 11098, Lr: 0.000300\n", "2020-04-10 01:21:51,463 Epoch 17 Step: 80500 Batch Loss: 2.070889 Tokens per Sec: 10955, Lr: 0.000300\n", "2020-04-10 01:22:09,830 Epoch 17 Step: 80600 Batch Loss: 2.043451 Tokens per Sec: 11123, Lr: 0.000300\n", "2020-04-10 01:22:28,140 Epoch 17 Step: 80700 Batch Loss: 2.110359 Tokens per Sec: 10959, Lr: 0.000300\n", "2020-04-10 01:22:46,350 Epoch 17 Step: 80800 Batch Loss: 1.977466 Tokens per Sec: 10927, Lr: 0.000300\n", "2020-04-10 01:23:04,767 Epoch 17 Step: 80900 Batch Loss: 1.983456 Tokens per Sec: 11009, Lr: 0.000300\n", "2020-04-10 01:23:23,207 Epoch 17 Step: 81000 Batch Loss: 1.797660 Tokens per Sec: 11110, Lr: 0.000300\n", "2020-04-10 01:23:54,813 Example #0\n", "2020-04-10 01:23:54,814 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 01:23:54,814 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 01:23:54,814 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 01:23:54,814 Example #1\n", "2020-04-10 01:23:54,815 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 01:23:54,815 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 01:23:54,815 \tHypothesis: If We willed , We would have made it a scorching , so why do you not give thanks ?\n", "2020-04-10 01:23:54,815 Example #2\n", "2020-04-10 01:23:54,816 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 01:23:54,816 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 01:23:54,816 \tHypothesis: And those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 01:23:54,816 Example #3\n", "2020-04-10 01:23:54,816 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 01:23:54,816 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 01:23:54,817 \tHypothesis: He sends the ship in security , and whenever the leaders of his people passed away before him , they mocked him .\n", "2020-04-10 01:23:54,817 Validation result (greedy) at epoch 17, step 81000: bleu: 23.49, loss: 49995.0977, ppl: 6.3561, duration: 31.6094s\n", "2020-04-10 01:24:13,145 Epoch 17 Step: 81100 Batch Loss: 1.966325 Tokens per Sec: 10922, Lr: 0.000300\n", "2020-04-10 01:24:31,454 Epoch 17 Step: 81200 Batch Loss: 2.208890 Tokens per Sec: 11047, Lr: 0.000300\n", "2020-04-10 01:24:49,787 Epoch 17 Step: 81300 Batch Loss: 2.079608 Tokens per Sec: 10834, Lr: 0.000300\n", "2020-04-10 01:25:07,999 Epoch 17 Step: 81400 Batch Loss: 1.858481 Tokens per Sec: 10836, Lr: 0.000300\n", "2020-04-10 01:25:26,310 Epoch 17 Step: 81500 Batch Loss: 1.903145 Tokens per Sec: 10879, Lr: 0.000300\n", "2020-04-10 01:25:44,673 Epoch 17 Step: 81600 Batch Loss: 2.181380 Tokens per Sec: 10934, Lr: 0.000300\n", "2020-04-10 01:26:02,770 Epoch 17 Step: 81700 Batch Loss: 1.951004 Tokens per Sec: 10948, Lr: 0.000300\n", "2020-04-10 01:26:21,114 Epoch 17 Step: 81800 Batch Loss: 2.032413 Tokens per Sec: 10962, Lr: 0.000300\n", "2020-04-10 01:26:39,122 Epoch 17 Step: 81900 Batch Loss: 1.909287 Tokens per Sec: 10671, Lr: 0.000300\n", "2020-04-10 01:26:57,491 Epoch 17 Step: 82000 Batch Loss: 1.856526 Tokens per Sec: 11184, Lr: 0.000300\n", "2020-04-10 01:27:25,357 Hooray! New best validation result [ppl]!\n", "2020-04-10 01:27:25,358 Saving new checkpoint.\n", "2020-04-10 01:27:26,562 Example #0\n", "2020-04-10 01:27:26,563 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 01:27:26,563 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 01:27:26,563 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 01:27:26,563 Example #1\n", "2020-04-10 01:27:26,563 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 01:27:26,564 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 01:27:26,564 \tHypothesis: If We willed , We could have made it blazing , so why do you not give thanks ?\n", "2020-04-10 01:27:26,564 Example #2\n", "2020-04-10 01:27:26,564 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 01:27:26,564 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 01:27:26,564 \tHypothesis: And those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 01:27:26,565 Example #3\n", "2020-04-10 01:27:26,565 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 01:27:26,565 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 01:27:26,565 \tHypothesis: He sends the ark in security , and whenever there came to his people , they mock him .\n", "2020-04-10 01:27:26,565 Validation result (greedy) at epoch 17, step 82000: bleu: 23.11, loss: 49730.7930, ppl: 6.2942, duration: 29.0743s\n", "2020-04-10 01:27:45,216 Epoch 17 Step: 82100 Batch Loss: 2.088522 Tokens per Sec: 11041, Lr: 0.000300\n", "2020-04-10 01:28:03,631 Epoch 17 Step: 82200 Batch Loss: 1.947450 Tokens per Sec: 11000, Lr: 0.000300\n", "2020-04-10 01:28:22,232 Epoch 17 Step: 82300 Batch Loss: 1.758387 Tokens per Sec: 10816, Lr: 0.000300\n", "2020-04-10 01:28:40,489 Epoch 17 Step: 82400 Batch Loss: 2.132944 Tokens per Sec: 10979, Lr: 0.000300\n", "2020-04-10 01:28:58,910 Epoch 17 Step: 82500 Batch Loss: 1.880717 Tokens per Sec: 10926, Lr: 0.000300\n", "2020-04-10 01:29:17,134 Epoch 17 Step: 82600 Batch Loss: 2.056727 Tokens per Sec: 10765, Lr: 0.000300\n", "2020-04-10 01:29:35,465 Epoch 17 Step: 82700 Batch Loss: 1.954727 Tokens per Sec: 11108, Lr: 0.000300\n", "2020-04-10 01:29:53,875 Epoch 17 Step: 82800 Batch Loss: 1.922287 Tokens per Sec: 11117, Lr: 0.000300\n", "2020-04-10 01:30:12,086 Epoch 17 Step: 82900 Batch Loss: 1.667030 Tokens per Sec: 10985, Lr: 0.000300\n", "2020-04-10 01:30:30,226 Epoch 17 Step: 83000 Batch Loss: 1.848236 Tokens per Sec: 10827, Lr: 0.000300\n", "2020-04-10 01:31:02,280 Example #0\n", "2020-04-10 01:31:02,280 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 01:31:02,280 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 01:31:02,281 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 01:31:02,281 Example #1\n", "2020-04-10 01:31:02,281 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 01:31:02,281 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 01:31:02,282 \tHypothesis: If We willed , We would have brought it to a shining shine , so why do you not give thanks ?\n", "2020-04-10 01:31:02,282 Example #2\n", "2020-04-10 01:31:02,282 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 01:31:02,282 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 01:31:02,283 \tHypothesis: And those who disbelieve in Our signs are the companions of the Left .\n", "2020-04-10 01:31:02,283 Example #3\n", "2020-04-10 01:31:02,283 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 01:31:02,283 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 01:31:02,284 \tHypothesis: And He causes the ship to be safe , and whenever some of his people passed away before him , they mocked him .\n", "2020-04-10 01:31:02,284 Validation result (greedy) at epoch 17, step 83000: bleu: 23.37, loss: 49861.5898, ppl: 6.3248, duration: 32.0573s\n", "2020-04-10 01:31:20,697 Epoch 17 Step: 83100 Batch Loss: 1.994778 Tokens per Sec: 11212, Lr: 0.000300\n", "2020-04-10 01:31:20,711 Epoch 17: total training loss 8536.55\n", "2020-04-10 01:31:20,711 EPOCH 18\n", "2020-04-10 01:31:39,553 Epoch 18 Step: 83200 Batch Loss: 1.944407 Tokens per Sec: 10897, Lr: 0.000300\n", "2020-04-10 01:31:58,041 Epoch 18 Step: 83300 Batch Loss: 1.957616 Tokens per Sec: 11156, Lr: 0.000300\n", "2020-04-10 01:32:16,338 Epoch 18 Step: 83400 Batch Loss: 1.893644 Tokens per Sec: 10903, Lr: 0.000300\n", "2020-04-10 01:32:34,758 Epoch 18 Step: 83500 Batch Loss: 2.090825 Tokens per Sec: 11260, Lr: 0.000300\n", "2020-04-10 01:32:53,084 Epoch 18 Step: 83600 Batch Loss: 2.034973 Tokens per Sec: 10964, Lr: 0.000300\n", "2020-04-10 01:33:11,558 Epoch 18 Step: 83700 Batch Loss: 2.109057 Tokens per Sec: 10935, Lr: 0.000300\n", "2020-04-10 01:33:29,761 Epoch 18 Step: 83800 Batch Loss: 1.974979 Tokens per Sec: 10952, Lr: 0.000300\n", "2020-04-10 01:33:48,151 Epoch 18 Step: 83900 Batch Loss: 1.822299 Tokens per Sec: 11117, Lr: 0.000300\n", "2020-04-10 01:34:06,221 Epoch 18 Step: 84000 Batch Loss: 1.943298 Tokens per Sec: 10763, Lr: 0.000300\n", "2020-04-10 01:34:39,208 Example #0\n", "2020-04-10 01:34:39,209 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 01:34:39,209 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 01:34:39,209 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 01:34:39,210 Example #1\n", "2020-04-10 01:34:39,210 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 01:34:39,210 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 01:34:39,210 \tHypothesis: Had We willed , We would have made it a blazing water , so why do you not give thanks ?\n", "2020-04-10 01:34:39,210 Example #2\n", "2020-04-10 01:34:39,211 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 01:34:39,211 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 01:34:39,211 \tHypothesis: And those who disbelieve in Our signs are the companions of the Left .\n", "2020-04-10 01:34:39,211 Example #3\n", "2020-04-10 01:34:39,211 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 01:34:39,212 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 01:34:39,212 \tHypothesis: He causes the ship to be safe , and when the leaders of his people passed away before him , they mocked him .\n", "2020-04-10 01:34:39,212 Validation result (greedy) at epoch 18, step 84000: bleu: 23.25, loss: 49811.2617, ppl: 6.3130, duration: 32.9909s\n", "2020-04-10 01:34:57,707 Epoch 18 Step: 84100 Batch Loss: 1.885644 Tokens per Sec: 11156, Lr: 0.000300\n", "2020-04-10 01:35:15,868 Epoch 18 Step: 84200 Batch Loss: 2.058369 Tokens per Sec: 11036, Lr: 0.000300\n", "2020-04-10 01:35:33,969 Epoch 18 Step: 84300 Batch Loss: 2.176130 Tokens per Sec: 10898, Lr: 0.000300\n", "2020-04-10 01:35:52,307 Epoch 18 Step: 84400 Batch Loss: 2.058196 Tokens per Sec: 10990, Lr: 0.000300\n", "2020-04-10 01:36:10,508 Epoch 18 Step: 84500 Batch Loss: 1.762303 Tokens per Sec: 10982, Lr: 0.000300\n", "2020-04-10 01:36:28,905 Epoch 18 Step: 84600 Batch Loss: 2.004754 Tokens per Sec: 11114, Lr: 0.000300\n", "2020-04-10 01:36:47,093 Epoch 18 Step: 84700 Batch Loss: 1.658600 Tokens per Sec: 10655, Lr: 0.000300\n", "2020-04-10 01:37:05,390 Epoch 18 Step: 84800 Batch Loss: 2.141808 Tokens per Sec: 11060, Lr: 0.000300\n", "2020-04-10 01:37:23,488 Epoch 18 Step: 84900 Batch Loss: 1.842981 Tokens per Sec: 10730, Lr: 0.000300\n", "2020-04-10 01:37:41,924 Epoch 18 Step: 85000 Batch Loss: 2.145619 Tokens per Sec: 11009, Lr: 0.000300\n", "2020-04-10 01:38:12,650 Example #0\n", "2020-04-10 01:38:12,650 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 01:38:12,651 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 01:38:12,651 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 01:38:12,651 Example #1\n", "2020-04-10 01:38:12,651 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 01:38:12,651 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 01:38:12,652 \tHypothesis: Had We willed , We could have brought it down to a blazing blazing , so why do you not give thanks ?\n", "2020-04-10 01:38:12,652 Example #2\n", "2020-04-10 01:38:12,652 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 01:38:12,652 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 01:38:12,652 \tHypothesis: Those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 01:38:12,652 Example #3\n", "2020-04-10 01:38:12,653 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 01:38:12,653 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 01:38:12,653 \tHypothesis: And He sends the ship in a secure , and whenever some of his people passed away before him , they mock him .\n", "2020-04-10 01:38:12,653 Validation result (greedy) at epoch 18, step 85000: bleu: 23.41, loss: 50014.1016, ppl: 6.3605, duration: 30.7291s\n", "2020-04-10 01:38:30,876 Epoch 18 Step: 85100 Batch Loss: 1.949485 Tokens per Sec: 10939, Lr: 0.000300\n", "2020-04-10 01:38:49,218 Epoch 18 Step: 85200 Batch Loss: 1.798659 Tokens per Sec: 10868, Lr: 0.000300\n", "2020-04-10 01:39:07,587 Epoch 18 Step: 85300 Batch Loss: 1.736262 Tokens per Sec: 10911, Lr: 0.000300\n", "2020-04-10 01:39:25,906 Epoch 18 Step: 85400 Batch Loss: 1.856417 Tokens per Sec: 10847, Lr: 0.000300\n", "2020-04-10 01:39:44,305 Epoch 18 Step: 85500 Batch Loss: 1.955724 Tokens per Sec: 11151, Lr: 0.000300\n", "2020-04-10 01:40:02,432 Epoch 18 Step: 85600 Batch Loss: 1.918325 Tokens per Sec: 10846, Lr: 0.000300\n", "2020-04-10 01:40:20,973 Epoch 18 Step: 85700 Batch Loss: 2.064615 Tokens per Sec: 11318, Lr: 0.000300\n", "2020-04-10 01:40:39,163 Epoch 18 Step: 85800 Batch Loss: 2.122475 Tokens per Sec: 11111, Lr: 0.000300\n", "2020-04-10 01:40:57,360 Epoch 18 Step: 85900 Batch Loss: 1.792347 Tokens per Sec: 10885, Lr: 0.000300\n", "2020-04-10 01:41:15,586 Epoch 18 Step: 86000 Batch Loss: 1.974522 Tokens per Sec: 10866, Lr: 0.000300\n", "2020-04-10 01:41:46,514 Example #0\n", "2020-04-10 01:41:46,514 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 01:41:46,514 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 01:41:46,515 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 01:41:46,515 Example #1\n", "2020-04-10 01:41:46,515 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 01:41:46,515 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 01:41:46,515 \tHypothesis: If We willed , We would have made it a blazing blazing , so why do you not give thanks ?\n", "2020-04-10 01:41:46,515 Example #2\n", "2020-04-10 01:41:46,516 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 01:41:46,516 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 01:41:46,516 \tHypothesis: Those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 01:41:46,516 Example #3\n", "2020-04-10 01:41:46,517 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 01:41:46,517 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 01:41:46,517 \tHypothesis: He sends the ark in security , and whenever some of his people passed away before him , they mocked him .\n", "2020-04-10 01:41:46,517 Validation result (greedy) at epoch 18, step 86000: bleu: 23.39, loss: 49752.8867, ppl: 6.2994, duration: 30.9302s\n", "2020-04-10 01:42:05,013 Epoch 18 Step: 86100 Batch Loss: 1.441408 Tokens per Sec: 11154, Lr: 0.000300\n", "2020-04-10 01:42:23,152 Epoch 18 Step: 86200 Batch Loss: 1.868245 Tokens per Sec: 10792, Lr: 0.000300\n", "2020-04-10 01:42:41,339 Epoch 18 Step: 86300 Batch Loss: 1.986871 Tokens per Sec: 10957, Lr: 0.000300\n", "2020-04-10 01:42:59,665 Epoch 18 Step: 86400 Batch Loss: 1.945903 Tokens per Sec: 10963, Lr: 0.000300\n", "2020-04-10 01:43:18,003 Epoch 18 Step: 86500 Batch Loss: 1.862186 Tokens per Sec: 11251, Lr: 0.000300\n", "2020-04-10 01:43:36,071 Epoch 18 Step: 86600 Batch Loss: 2.094389 Tokens per Sec: 10922, Lr: 0.000300\n", "2020-04-10 01:43:54,571 Epoch 18 Step: 86700 Batch Loss: 1.962776 Tokens per Sec: 10759, Lr: 0.000300\n", "2020-04-10 01:44:12,930 Epoch 18 Step: 86800 Batch Loss: 2.135858 Tokens per Sec: 10906, Lr: 0.000300\n", "2020-04-10 01:44:31,018 Epoch 18 Step: 86900 Batch Loss: 2.082438 Tokens per Sec: 10881, Lr: 0.000300\n", "2020-04-10 01:44:49,515 Epoch 18 Step: 87000 Batch Loss: 2.256247 Tokens per Sec: 11072, Lr: 0.000300\n", "2020-04-10 01:45:19,659 Hooray! New best validation result [ppl]!\n", "2020-04-10 01:45:19,660 Saving new checkpoint.\n", "2020-04-10 01:45:20,952 Example #0\n", "2020-04-10 01:45:20,953 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 01:45:20,953 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 01:45:20,954 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 01:45:20,954 Example #1\n", "2020-04-10 01:45:20,954 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 01:45:20,954 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 01:45:20,954 \tHypothesis: If We willed , We would have made it blazing water , so why do you not give thanks ?\n", "2020-04-10 01:45:20,954 Example #2\n", "2020-04-10 01:45:20,955 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 01:45:20,955 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 01:45:20,955 \tHypothesis: And those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 01:45:20,955 Example #3\n", "2020-04-10 01:45:20,956 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 01:45:20,956 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 01:45:20,956 \tHypothesis: He sends the ark in security , and when some of his people passed away before him , they mocked him .\n", "2020-04-10 01:45:20,956 Validation result (greedy) at epoch 18, step 87000: bleu: 23.50, loss: 49512.0352, ppl: 6.2435, duration: 31.4405s\n", "2020-04-10 01:45:39,443 Epoch 18 Step: 87100 Batch Loss: 2.077125 Tokens per Sec: 10650, Lr: 0.000300\n", "2020-04-10 01:45:57,836 Epoch 18 Step: 87200 Batch Loss: 2.074671 Tokens per Sec: 10848, Lr: 0.000300\n", "2020-04-10 01:46:16,058 Epoch 18 Step: 87300 Batch Loss: 2.051377 Tokens per Sec: 10780, Lr: 0.000300\n", "2020-04-10 01:46:34,367 Epoch 18 Step: 87400 Batch Loss: 1.893348 Tokens per Sec: 10795, Lr: 0.000300\n", "2020-04-10 01:46:52,522 Epoch 18 Step: 87500 Batch Loss: 1.922227 Tokens per Sec: 10824, Lr: 0.000300\n", "2020-04-10 01:46:56,564 Epoch 18: total training loss 8481.83\n", "2020-04-10 01:46:56,564 EPOCH 19\n", "2020-04-10 01:47:11,308 Epoch 19 Step: 87600 Batch Loss: 1.993919 Tokens per Sec: 10609, Lr: 0.000300\n", "2020-04-10 01:47:29,675 Epoch 19 Step: 87700 Batch Loss: 1.674491 Tokens per Sec: 11071, Lr: 0.000300\n", "2020-04-10 01:47:47,997 Epoch 19 Step: 87800 Batch Loss: 1.904007 Tokens per Sec: 10906, Lr: 0.000300\n", "2020-04-10 01:48:06,423 Epoch 19 Step: 87900 Batch Loss: 2.008139 Tokens per Sec: 10850, Lr: 0.000300\n", "2020-04-10 01:48:24,861 Epoch 19 Step: 88000 Batch Loss: 1.904158 Tokens per Sec: 11142, Lr: 0.000300\n", "2020-04-10 01:48:53,488 Example #0\n", "2020-04-10 01:48:53,489 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 01:48:53,489 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 01:48:53,489 \tHypothesis: See how We explain Our signs that they may understand .\n", "2020-04-10 01:48:53,489 Example #1\n", "2020-04-10 01:48:53,490 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 01:48:53,490 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 01:48:53,490 \tHypothesis: If We willed , We would have made it blazing , so why do you not give thanks ?\n", "2020-04-10 01:48:53,490 Example #2\n", "2020-04-10 01:48:53,490 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 01:48:53,491 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 01:48:53,491 \tHypothesis: Those who disbelieve Our signs are the people of the Left .\n", "2020-04-10 01:48:53,491 Example #3\n", "2020-04-10 01:48:53,491 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 01:48:53,491 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 01:48:53,492 \tHypothesis: He sends the ark in security , and whenever some of his people passed away before him , they mocked him .\n", "2020-04-10 01:48:53,492 Validation result (greedy) at epoch 19, step 88000: bleu: 23.48, loss: 49672.2227, ppl: 6.2806, duration: 28.6302s\n", "2020-04-10 01:49:11,961 Epoch 19 Step: 88100 Batch Loss: 1.951505 Tokens per Sec: 11001, Lr: 0.000300\n", "2020-04-10 01:49:30,503 Epoch 19 Step: 88200 Batch Loss: 1.848330 Tokens per Sec: 10909, Lr: 0.000300\n", "2020-04-10 01:49:48,751 Epoch 19 Step: 88300 Batch Loss: 1.891136 Tokens per Sec: 11017, Lr: 0.000300\n", "2020-04-10 01:50:06,918 Epoch 19 Step: 88400 Batch Loss: 1.820073 Tokens per Sec: 10710, Lr: 0.000300\n", "2020-04-10 01:50:25,425 Epoch 19 Step: 88500 Batch Loss: 1.878468 Tokens per Sec: 10991, Lr: 0.000300\n", "2020-04-10 01:50:43,751 Epoch 19 Step: 88600 Batch Loss: 1.577677 Tokens per Sec: 10873, Lr: 0.000300\n", "2020-04-10 01:51:02,078 Epoch 19 Step: 88700 Batch Loss: 1.965250 Tokens per Sec: 10931, Lr: 0.000300\n", "2020-04-10 01:51:20,475 Epoch 19 Step: 88800 Batch Loss: 2.070602 Tokens per Sec: 10868, Lr: 0.000300\n", "2020-04-10 01:51:38,994 Epoch 19 Step: 88900 Batch Loss: 1.955767 Tokens per Sec: 10763, Lr: 0.000300\n", "2020-04-10 01:51:57,436 Epoch 19 Step: 89000 Batch Loss: 2.033603 Tokens per Sec: 11180, Lr: 0.000300\n", "2020-04-10 01:52:26,506 Hooray! New best validation result [ppl]!\n", "2020-04-10 01:52:26,507 Saving new checkpoint.\n", "2020-04-10 01:52:27,769 Example #0\n", "2020-04-10 01:52:27,770 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 01:52:27,770 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 01:52:27,770 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 01:52:27,771 Example #1\n", "2020-04-10 01:52:27,771 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 01:52:27,771 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 01:52:27,771 \tHypothesis: Had We willed , We would have made it blazing , so why do you not give thanks ?\n", "2020-04-10 01:52:27,772 Example #2\n", "2020-04-10 01:52:27,772 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 01:52:27,772 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 01:52:27,772 \tHypothesis: Those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 01:52:27,772 Example #3\n", "2020-04-10 01:52:27,773 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 01:52:27,773 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 01:52:27,773 \tHypothesis: He sends the ark in security , and whenever some of his people passed away before him , they mocked at him .\n", "2020-04-10 01:52:27,773 Validation result (greedy) at epoch 19, step 89000: bleu: 23.39, loss: 49346.0508, ppl: 6.2053, duration: 30.3371s\n", "2020-04-10 01:52:46,364 Epoch 19 Step: 89100 Batch Loss: 1.851570 Tokens per Sec: 10671, Lr: 0.000300\n", "2020-04-10 01:53:04,653 Epoch 19 Step: 89200 Batch Loss: 2.036609 Tokens per Sec: 10746, Lr: 0.000300\n", "2020-04-10 01:53:22,968 Epoch 19 Step: 89300 Batch Loss: 1.959805 Tokens per Sec: 10681, Lr: 0.000300\n", "2020-04-10 01:53:41,390 Epoch 19 Step: 89400 Batch Loss: 1.864794 Tokens per Sec: 11069, Lr: 0.000300\n", "2020-04-10 01:53:59,709 Epoch 19 Step: 89500 Batch Loss: 1.987709 Tokens per Sec: 10810, Lr: 0.000300\n", "2020-04-10 01:54:18,093 Epoch 19 Step: 89600 Batch Loss: 1.976975 Tokens per Sec: 10939, Lr: 0.000300\n", "2020-04-10 01:54:36,624 Epoch 19 Step: 89700 Batch Loss: 2.133315 Tokens per Sec: 11260, Lr: 0.000300\n", "2020-04-10 01:54:55,122 Epoch 19 Step: 89800 Batch Loss: 2.212672 Tokens per Sec: 11078, Lr: 0.000300\n", "2020-04-10 01:55:13,460 Epoch 19 Step: 89900 Batch Loss: 1.987521 Tokens per Sec: 10819, Lr: 0.000300\n", "2020-04-10 01:55:31,815 Epoch 19 Step: 90000 Batch Loss: 1.882458 Tokens per Sec: 10865, Lr: 0.000300\n", "2020-04-10 01:56:01,238 Hooray! New best validation result [ppl]!\n", "2020-04-10 01:56:01,238 Saving new checkpoint.\n", "2020-04-10 01:56:02,869 Example #0\n", "2020-04-10 01:56:02,870 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 01:56:02,870 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 01:56:02,870 \tHypothesis: See how We explain Our signs , that they may understand .\n", "2020-04-10 01:56:02,871 Example #1\n", "2020-04-10 01:56:02,871 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 01:56:02,871 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 01:56:02,871 \tHypothesis: If We willed , We would have made it a blazing black , so why do you not give thanks ?\n", "2020-04-10 01:56:02,871 Example #2\n", "2020-04-10 01:56:02,872 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 01:56:02,872 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 01:56:02,872 \tHypothesis: And those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 01:56:02,872 Example #3\n", "2020-04-10 01:56:02,873 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 01:56:02,873 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 01:56:02,873 \tHypothesis: He causes the ark in security , and whenever there came to him some of his people , they mocked him .\n", "2020-04-10 01:56:02,873 Validation result (greedy) at epoch 19, step 90000: bleu: 23.58, loss: 49344.7539, ppl: 6.2050, duration: 31.0579s\n", "2020-04-10 01:56:21,511 Epoch 19 Step: 90100 Batch Loss: 2.055625 Tokens per Sec: 10960, Lr: 0.000300\n", "2020-04-10 01:56:39,777 Epoch 19 Step: 90200 Batch Loss: 2.168376 Tokens per Sec: 10834, Lr: 0.000300\n", "2020-04-10 01:56:58,015 Epoch 19 Step: 90300 Batch Loss: 2.059725 Tokens per Sec: 10905, Lr: 0.000300\n", "2020-04-10 01:57:16,511 Epoch 19 Step: 90400 Batch Loss: 1.731590 Tokens per Sec: 10915, Lr: 0.000300\n", "2020-04-10 01:57:34,975 Epoch 19 Step: 90500 Batch Loss: 1.906534 Tokens per Sec: 10899, Lr: 0.000300\n", "2020-04-10 01:57:53,154 Epoch 19 Step: 90600 Batch Loss: 1.896634 Tokens per Sec: 10803, Lr: 0.000300\n", "2020-04-10 01:58:11,331 Epoch 19 Step: 90700 Batch Loss: 2.076320 Tokens per Sec: 10894, Lr: 0.000300\n", "2020-04-10 01:58:29,636 Epoch 19 Step: 90800 Batch Loss: 1.846678 Tokens per Sec: 11015, Lr: 0.000300\n", "2020-04-10 01:58:47,835 Epoch 19 Step: 90900 Batch Loss: 1.934550 Tokens per Sec: 10732, Lr: 0.000300\n", "2020-04-10 01:59:06,211 Epoch 19 Step: 91000 Batch Loss: 1.667401 Tokens per Sec: 11127, Lr: 0.000300\n", "2020-04-10 01:59:37,921 Hooray! New best validation result [ppl]!\n", "2020-04-10 01:59:37,921 Saving new checkpoint.\n", "2020-04-10 01:59:39,115 Example #0\n", "2020-04-10 01:59:39,116 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 01:59:39,116 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 01:59:39,116 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 01:59:39,116 Example #1\n", "2020-04-10 01:59:39,117 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 01:59:39,117 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 01:59:39,117 \tHypothesis: If We willed , We could have made it rain , so why do you not give thanks ?\n", "2020-04-10 01:59:39,117 Example #2\n", "2020-04-10 01:59:39,117 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 01:59:39,118 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 01:59:39,118 \tHypothesis: Those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 01:59:39,118 Example #3\n", "2020-04-10 01:59:39,119 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 01:59:39,119 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 01:59:39,119 \tHypothesis: He sends the ark in security , and whenever there came to him some of his people passed away before him , they mocked him .\n", "2020-04-10 01:59:39,119 Validation result (greedy) at epoch 19, step 91000: bleu: 23.66, loss: 49236.4922, ppl: 6.1802, duration: 32.9079s\n", "2020-04-10 01:59:57,725 Epoch 19 Step: 91100 Batch Loss: 1.917826 Tokens per Sec: 11088, Lr: 0.000300\n", "2020-04-10 02:00:16,073 Epoch 19 Step: 91200 Batch Loss: 1.980175 Tokens per Sec: 11101, Lr: 0.000300\n", "2020-04-10 02:00:34,269 Epoch 19 Step: 91300 Batch Loss: 1.592009 Tokens per Sec: 10770, Lr: 0.000300\n", "2020-04-10 02:00:52,443 Epoch 19 Step: 91400 Batch Loss: 1.768723 Tokens per Sec: 10831, Lr: 0.000300\n", "2020-04-10 02:01:10,919 Epoch 19 Step: 91500 Batch Loss: 1.772911 Tokens per Sec: 11196, Lr: 0.000300\n", "2020-04-10 02:01:29,295 Epoch 19 Step: 91600 Batch Loss: 1.939131 Tokens per Sec: 10988, Lr: 0.000300\n", "2020-04-10 02:01:47,579 Epoch 19 Step: 91700 Batch Loss: 2.032266 Tokens per Sec: 11071, Lr: 0.000300\n", "2020-04-10 02:02:05,620 Epoch 19 Step: 91800 Batch Loss: 1.939207 Tokens per Sec: 10889, Lr: 0.000300\n", "2020-04-10 02:02:24,067 Epoch 19 Step: 91900 Batch Loss: 1.901432 Tokens per Sec: 11056, Lr: 0.000300\n", "2020-04-10 02:02:30,815 Epoch 19: total training loss 8423.95\n", "2020-04-10 02:02:30,815 EPOCH 20\n", "2020-04-10 02:02:42,706 Epoch 20 Step: 92000 Batch Loss: 1.734878 Tokens per Sec: 10730, Lr: 0.000300\n", "2020-04-10 02:03:13,908 Example #0\n", "2020-04-10 02:03:13,909 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 02:03:13,909 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 02:03:13,909 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 02:03:13,909 Example #1\n", "2020-04-10 02:03:13,910 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 02:03:13,910 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 02:03:13,910 \tHypothesis: If We willed , We would have made it rain , so why do you not give thanks ?\n", "2020-04-10 02:03:13,910 Example #2\n", "2020-04-10 02:03:13,911 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 02:03:13,911 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 02:03:13,911 \tHypothesis: And those who disbelieve in Our signs are the People of the Left .\n", "2020-04-10 02:03:13,911 Example #3\n", "2020-04-10 02:03:13,911 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 02:03:13,912 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 02:03:13,912 \tHypothesis: And He causes the ships to be safe , and whenever some of his people passed away before him they mock at him .\n", "2020-04-10 02:03:13,912 Validation result (greedy) at epoch 20, step 92000: bleu: 23.52, loss: 49448.0117, ppl: 6.2287, duration: 31.2052s\n", "2020-04-10 02:03:32,215 Epoch 20 Step: 92100 Batch Loss: 1.764992 Tokens per Sec: 10981, Lr: 0.000300\n", "2020-04-10 02:03:50,468 Epoch 20 Step: 92200 Batch Loss: 1.606150 Tokens per Sec: 10818, Lr: 0.000300\n", "2020-04-10 02:04:08,785 Epoch 20 Step: 92300 Batch Loss: 2.158339 Tokens per Sec: 10737, Lr: 0.000300\n", "2020-04-10 02:04:27,049 Epoch 20 Step: 92400 Batch Loss: 1.830504 Tokens per Sec: 10722, Lr: 0.000300\n", "2020-04-10 02:04:45,720 Epoch 20 Step: 92500 Batch Loss: 1.962649 Tokens per Sec: 11394, Lr: 0.000300\n", "2020-04-10 02:05:04,036 Epoch 20 Step: 92600 Batch Loss: 1.995850 Tokens per Sec: 10883, Lr: 0.000300\n", "2020-04-10 02:05:22,455 Epoch 20 Step: 92700 Batch Loss: 1.986660 Tokens per Sec: 11080, Lr: 0.000300\n", "2020-04-10 02:05:40,876 Epoch 20 Step: 92800 Batch Loss: 1.884072 Tokens per Sec: 10870, Lr: 0.000300\n", "2020-04-10 02:05:59,082 Epoch 20 Step: 92900 Batch Loss: 1.954565 Tokens per Sec: 10920, Lr: 0.000300\n", "2020-04-10 02:06:17,333 Epoch 20 Step: 93000 Batch Loss: 1.967526 Tokens per Sec: 10898, Lr: 0.000300\n", "2020-04-10 02:06:45,295 Example #0\n", "2020-04-10 02:06:45,296 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 02:06:45,296 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 02:06:45,296 \tHypothesis: See how We explain Our signs , that they may understand .\n", "2020-04-10 02:06:45,296 Example #1\n", "2020-04-10 02:06:45,297 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 02:06:45,297 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 02:06:45,297 \tHypothesis: If We willed , We could have made it rain , so why do you not give thanks ?\n", "2020-04-10 02:06:45,297 Example #2\n", "2020-04-10 02:06:45,297 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 02:06:45,298 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 02:06:45,298 \tHypothesis: Those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 02:06:45,298 Example #3\n", "2020-04-10 02:06:45,298 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 02:06:45,298 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 02:06:45,299 \tHypothesis: He sends the ark in peace , and whenever there came to his people , they mock him .\n", "2020-04-10 02:06:45,299 Validation result (greedy) at epoch 20, step 93000: bleu: 23.38, loss: 49437.2148, ppl: 6.2262, duration: 27.9652s\n", "2020-04-10 02:07:03,694 Epoch 20 Step: 93100 Batch Loss: 2.001143 Tokens per Sec: 10985, Lr: 0.000300\n", "2020-04-10 02:07:22,011 Epoch 20 Step: 93200 Batch Loss: 1.891721 Tokens per Sec: 10922, Lr: 0.000300\n", "2020-04-10 02:07:40,348 Epoch 20 Step: 93300 Batch Loss: 1.740505 Tokens per Sec: 11100, Lr: 0.000300\n", "2020-04-10 02:07:58,563 Epoch 20 Step: 93400 Batch Loss: 1.578363 Tokens per Sec: 11007, Lr: 0.000300\n", "2020-04-10 02:08:16,861 Epoch 20 Step: 93500 Batch Loss: 1.920019 Tokens per Sec: 10997, Lr: 0.000300\n", "2020-04-10 02:08:35,236 Epoch 20 Step: 93600 Batch Loss: 1.862135 Tokens per Sec: 10880, Lr: 0.000300\n", "2020-04-10 02:08:53,628 Epoch 20 Step: 93700 Batch Loss: 1.929412 Tokens per Sec: 11251, Lr: 0.000300\n", "2020-04-10 02:09:11,985 Epoch 20 Step: 93800 Batch Loss: 1.881828 Tokens per Sec: 11173, Lr: 0.000300\n", "2020-04-10 02:09:30,504 Epoch 20 Step: 93900 Batch Loss: 1.938331 Tokens per Sec: 10979, Lr: 0.000300\n", "2020-04-10 02:09:48,738 Epoch 20 Step: 94000 Batch Loss: 2.181306 Tokens per Sec: 10854, Lr: 0.000300\n", "2020-04-10 02:10:21,264 Hooray! New best validation result [ppl]!\n", "2020-04-10 02:10:21,264 Saving new checkpoint.\n", "2020-04-10 02:10:22,521 Example #0\n", "2020-04-10 02:10:22,521 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 02:10:22,522 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 02:10:22,522 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 02:10:22,522 Example #1\n", "2020-04-10 02:10:22,522 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 02:10:22,522 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 02:10:22,523 \tHypothesis: Had We willed , We would have made it rain , so why do you not give thanks ?\n", "2020-04-10 02:10:22,523 Example #2\n", "2020-04-10 02:10:22,523 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 02:10:22,523 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 02:10:22,523 \tHypothesis: And those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 02:10:22,524 Example #3\n", "2020-04-10 02:10:22,524 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 02:10:22,524 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 02:10:22,524 \tHypothesis: And He causes the ship to be safe , and whenever there came to his people , they mocked him .\n", "2020-04-10 02:10:22,524 Validation result (greedy) at epoch 20, step 94000: bleu: 23.89, loss: 49196.5977, ppl: 6.1711, duration: 33.7856s\n", "2020-04-10 02:10:41,205 Epoch 20 Step: 94100 Batch Loss: 2.035070 Tokens per Sec: 11163, Lr: 0.000300\n", "2020-04-10 02:10:59,459 Epoch 20 Step: 94200 Batch Loss: 2.080710 Tokens per Sec: 11246, Lr: 0.000300\n", "2020-04-10 02:11:17,663 Epoch 20 Step: 94300 Batch Loss: 2.031420 Tokens per Sec: 10904, Lr: 0.000300\n", "2020-04-10 02:11:35,983 Epoch 20 Step: 94400 Batch Loss: 1.974053 Tokens per Sec: 11047, Lr: 0.000300\n", "2020-04-10 02:11:54,165 Epoch 20 Step: 94500 Batch Loss: 1.926105 Tokens per Sec: 11219, Lr: 0.000300\n", "2020-04-10 02:12:12,411 Epoch 20 Step: 94600 Batch Loss: 2.197823 Tokens per Sec: 11160, Lr: 0.000300\n", "2020-04-10 02:12:30,713 Epoch 20 Step: 94700 Batch Loss: 1.628919 Tokens per Sec: 10784, Lr: 0.000300\n", "2020-04-10 02:12:49,054 Epoch 20 Step: 94800 Batch Loss: 1.814257 Tokens per Sec: 11232, Lr: 0.000300\n", "2020-04-10 02:13:07,439 Epoch 20 Step: 94900 Batch Loss: 1.899114 Tokens per Sec: 11142, Lr: 0.000300\n", "2020-04-10 02:13:25,520 Epoch 20 Step: 95000 Batch Loss: 1.796813 Tokens per Sec: 10910, Lr: 0.000300\n", "2020-04-10 02:13:54,463 Example #0\n", "2020-04-10 02:13:54,464 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 02:13:54,464 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 02:13:54,464 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 02:13:54,465 Example #1\n", "2020-04-10 02:13:54,465 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 02:13:54,465 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 02:13:54,465 \tHypothesis: If We willed , We would have made it a black of black , so why do you not give thanks ?\n", "2020-04-10 02:13:54,465 Example #2\n", "2020-04-10 02:13:54,466 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 02:13:54,466 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 02:13:54,466 \tHypothesis: And those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 02:13:54,466 Example #3\n", "2020-04-10 02:13:54,467 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 02:13:54,467 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 02:13:54,467 \tHypothesis: And He causes the ship in security , and whenever there came to his people passed away before him , they mocked him .\n", "2020-04-10 02:13:54,467 Validation result (greedy) at epoch 20, step 95000: bleu: 23.79, loss: 49235.1055, ppl: 6.1799, duration: 28.9466s\n", "2020-04-10 02:14:12,683 Epoch 20 Step: 95100 Batch Loss: 1.921516 Tokens per Sec: 10959, Lr: 0.000300\n", "2020-04-10 02:14:31,097 Epoch 20 Step: 95200 Batch Loss: 1.924880 Tokens per Sec: 11314, Lr: 0.000300\n", "2020-04-10 02:14:49,425 Epoch 20 Step: 95300 Batch Loss: 2.041879 Tokens per Sec: 11030, Lr: 0.000300\n", "2020-04-10 02:15:07,691 Epoch 20 Step: 95400 Batch Loss: 2.142836 Tokens per Sec: 10972, Lr: 0.000300\n", "2020-04-10 02:15:25,900 Epoch 20 Step: 95500 Batch Loss: 1.973491 Tokens per Sec: 11013, Lr: 0.000300\n", "2020-04-10 02:15:44,251 Epoch 20 Step: 95600 Batch Loss: 1.976262 Tokens per Sec: 10967, Lr: 0.000300\n", "2020-04-10 02:16:02,446 Epoch 20 Step: 95700 Batch Loss: 2.034365 Tokens per Sec: 10710, Lr: 0.000300\n", "2020-04-10 02:16:20,728 Epoch 20 Step: 95800 Batch Loss: 1.902493 Tokens per Sec: 11092, Lr: 0.000300\n", "2020-04-10 02:16:38,959 Epoch 20 Step: 95900 Batch Loss: 1.509552 Tokens per Sec: 11014, Lr: 0.000300\n", "2020-04-10 02:16:57,146 Epoch 20 Step: 96000 Batch Loss: 1.875306 Tokens per Sec: 10856, Lr: 0.000300\n", "2020-04-10 02:17:26,693 Hooray! New best validation result [ppl]!\n", "2020-04-10 02:17:26,693 Saving new checkpoint.\n", "2020-04-10 02:17:28,134 Example #0\n", "2020-04-10 02:17:28,135 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 02:17:28,135 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 02:17:28,136 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-10 02:17:28,136 Example #1\n", "2020-04-10 02:17:28,136 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 02:17:28,136 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 02:17:28,137 \tHypothesis: If We willed , We would have made it a rain , so why do you not give thanks ?\n", "2020-04-10 02:17:28,137 Example #2\n", "2020-04-10 02:17:28,137 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 02:17:28,138 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 02:17:28,138 \tHypothesis: And those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 02:17:28,138 Example #3\n", "2020-04-10 02:17:28,139 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 02:17:28,139 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 02:17:28,139 \tHypothesis: He sends the ark in peace , and whenever some of his people passed away before him , they mocked him .\n", "2020-04-10 02:17:28,140 Validation result (greedy) at epoch 20, step 96000: bleu: 24.11, loss: 49117.6211, ppl: 6.1531, duration: 30.9928s\n", "2020-04-10 02:17:46,560 Epoch 20 Step: 96100 Batch Loss: 2.128038 Tokens per Sec: 10779, Lr: 0.000300\n", "2020-04-10 02:18:04,838 Epoch 20 Step: 96200 Batch Loss: 2.188960 Tokens per Sec: 11171, Lr: 0.000300\n", "2020-04-10 02:18:23,079 Epoch 20 Step: 96300 Batch Loss: 1.803448 Tokens per Sec: 11099, Lr: 0.000300\n", "2020-04-10 02:18:29,685 Epoch 20: total training loss 8342.08\n", "2020-04-10 02:18:29,686 EPOCH 21\n", "2020-04-10 02:18:41,721 Epoch 21 Step: 96400 Batch Loss: 1.790403 Tokens per Sec: 10214, Lr: 0.000300\n", "2020-04-10 02:19:00,034 Epoch 21 Step: 96500 Batch Loss: 1.811777 Tokens per Sec: 10922, Lr: 0.000300\n", "2020-04-10 02:19:18,185 Epoch 21 Step: 96600 Batch Loss: 1.796744 Tokens per Sec: 10990, Lr: 0.000300\n", "2020-04-10 02:19:36,478 Epoch 21 Step: 96700 Batch Loss: 1.944327 Tokens per Sec: 11202, Lr: 0.000300\n", "2020-04-10 02:19:54,610 Epoch 21 Step: 96800 Batch Loss: 1.864659 Tokens per Sec: 10801, Lr: 0.000300\n", "2020-04-10 02:20:13,007 Epoch 21 Step: 96900 Batch Loss: 1.892391 Tokens per Sec: 11000, Lr: 0.000300\n", "2020-04-10 02:20:31,155 Epoch 21 Step: 97000 Batch Loss: 1.959144 Tokens per Sec: 10871, Lr: 0.000300\n", "2020-04-10 02:21:00,460 Hooray! New best validation result [ppl]!\n", "2020-04-10 02:21:00,461 Saving new checkpoint.\n", "2020-04-10 02:21:02,053 Example #0\n", "2020-04-10 02:21:02,054 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 02:21:02,054 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 02:21:02,054 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 02:21:02,054 Example #1\n", "2020-04-10 02:21:02,054 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 02:21:02,055 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 02:21:02,055 \tHypothesis: If We willed , We could have made it rain , so why do you not give thanks ?\n", "2020-04-10 02:21:02,055 Example #2\n", "2020-04-10 02:21:02,055 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 02:21:02,055 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 02:21:02,056 \tHypothesis: And those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 02:21:02,056 Example #3\n", "2020-04-10 02:21:02,056 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 02:21:02,057 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 02:21:02,057 \tHypothesis: And He causes the ship to be safe , and whenever there came to his people passed away before him , they ridicule him .\n", "2020-04-10 02:21:02,057 Validation result (greedy) at epoch 21, step 97000: bleu: 24.04, loss: 48887.0234, ppl: 6.1008, duration: 30.9013s\n", "2020-04-10 02:21:20,289 Epoch 21 Step: 97100 Batch Loss: 1.713458 Tokens per Sec: 10908, Lr: 0.000300\n", "2020-04-10 02:21:38,695 Epoch 21 Step: 97200 Batch Loss: 2.069952 Tokens per Sec: 10813, Lr: 0.000300\n", "2020-04-10 02:21:56,911 Epoch 21 Step: 97300 Batch Loss: 1.941131 Tokens per Sec: 10866, Lr: 0.000300\n", "2020-04-10 02:22:15,274 Epoch 21 Step: 97400 Batch Loss: 2.034267 Tokens per Sec: 11290, Lr: 0.000300\n", "2020-04-10 02:22:33,729 Epoch 21 Step: 97500 Batch Loss: 1.815179 Tokens per Sec: 10745, Lr: 0.000300\n", "2020-04-10 02:22:52,216 Epoch 21 Step: 97600 Batch Loss: 1.807081 Tokens per Sec: 11005, Lr: 0.000300\n", "2020-04-10 02:23:10,372 Epoch 21 Step: 97700 Batch Loss: 1.821677 Tokens per Sec: 11088, Lr: 0.000300\n", "2020-04-10 02:23:28,478 Epoch 21 Step: 97800 Batch Loss: 1.579910 Tokens per Sec: 10926, Lr: 0.000300\n", "2020-04-10 02:23:46,912 Epoch 21 Step: 97900 Batch Loss: 1.775105 Tokens per Sec: 11067, Lr: 0.000300\n", "2020-04-10 02:24:05,296 Epoch 21 Step: 98000 Batch Loss: 1.904306 Tokens per Sec: 10995, Lr: 0.000300\n", "2020-04-10 02:24:35,971 Example #0\n", "2020-04-10 02:24:35,972 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 02:24:35,972 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 02:24:35,972 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 02:24:35,972 Example #1\n", "2020-04-10 02:24:35,973 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 02:24:35,973 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 02:24:35,973 \tHypothesis: If We willed , We could have made it blind , so why do you not give thanks ?\n", "2020-04-10 02:24:35,973 Example #2\n", "2020-04-10 02:24:35,974 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 02:24:35,974 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 02:24:35,974 \tHypothesis: And those who disbelieve in Our signs , they are the people of the Left Hand .\n", "2020-04-10 02:24:35,975 Example #3\n", "2020-04-10 02:24:35,975 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 02:24:35,975 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 02:24:35,975 \tHypothesis: And He causes the ship to be safe , and when there came to pass away from his people , they mocked him .\n", "2020-04-10 02:24:35,976 Validation result (greedy) at epoch 21, step 98000: bleu: 23.84, loss: 49047.7109, ppl: 6.1372, duration: 30.6787s\n", "2020-04-10 02:24:54,280 Epoch 21 Step: 98100 Batch Loss: 2.045422 Tokens per Sec: 10701, Lr: 0.000300\n", "2020-04-10 02:25:12,718 Epoch 21 Step: 98200 Batch Loss: 2.055113 Tokens per Sec: 11047, Lr: 0.000300\n", "2020-04-10 02:25:31,192 Epoch 21 Step: 98300 Batch Loss: 1.965874 Tokens per Sec: 11114, Lr: 0.000300\n", "2020-04-10 02:25:49,683 Epoch 21 Step: 98400 Batch Loss: 1.915390 Tokens per Sec: 10887, Lr: 0.000300\n", "2020-04-10 02:26:08,073 Epoch 21 Step: 98500 Batch Loss: 1.830495 Tokens per Sec: 10958, Lr: 0.000300\n", "2020-04-10 02:26:26,069 Epoch 21 Step: 98600 Batch Loss: 2.197199 Tokens per Sec: 10933, Lr: 0.000300\n", "2020-04-10 02:26:44,380 Epoch 21 Step: 98700 Batch Loss: 1.972440 Tokens per Sec: 10743, Lr: 0.000300\n", "2020-04-10 02:27:02,980 Epoch 21 Step: 98800 Batch Loss: 1.882878 Tokens per Sec: 11085, Lr: 0.000300\n", "2020-04-10 02:27:21,339 Epoch 21 Step: 98900 Batch Loss: 1.772267 Tokens per Sec: 11340, Lr: 0.000300\n", "2020-04-10 02:27:39,510 Epoch 21 Step: 99000 Batch Loss: 1.776191 Tokens per Sec: 10974, Lr: 0.000300\n", "2020-04-10 02:28:10,026 Hooray! New best validation result [ppl]!\n", "2020-04-10 02:28:10,027 Saving new checkpoint.\n", "2020-04-10 02:28:11,319 Example #0\n", "2020-04-10 02:28:11,320 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 02:28:11,320 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 02:28:11,320 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 02:28:11,320 Example #1\n", "2020-04-10 02:28:11,321 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 02:28:11,321 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 02:28:11,321 \tHypothesis: If We willed , We could have made it a rain , so why do you not give thanks ?\n", "2020-04-10 02:28:11,321 Example #2\n", "2020-04-10 02:28:11,321 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 02:28:11,322 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 02:28:11,322 \tHypothesis: And those who disbelieve in Our revelations are the people of the Left Hand .\n", "2020-04-10 02:28:11,322 Example #3\n", "2020-04-10 02:28:11,322 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 02:28:11,322 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 02:28:11,322 \tHypothesis: And He causes the ship to be safe , and whenever there came to his people passed away before him , they mocked him .\n", "2020-04-10 02:28:11,323 Validation result (greedy) at epoch 21, step 99000: bleu: 23.90, loss: 48865.9961, ppl: 6.0961, duration: 31.8122s\n", "2020-04-10 02:28:29,813 Epoch 21 Step: 99100 Batch Loss: 1.568060 Tokens per Sec: 10683, Lr: 0.000300\n", "2020-04-10 02:28:48,305 Epoch 21 Step: 99200 Batch Loss: 2.030564 Tokens per Sec: 11189, Lr: 0.000300\n", "2020-04-10 02:29:06,517 Epoch 21 Step: 99300 Batch Loss: 1.808883 Tokens per Sec: 10897, Lr: 0.000300\n", "2020-04-10 02:29:24,724 Epoch 21 Step: 99400 Batch Loss: 1.563194 Tokens per Sec: 11070, Lr: 0.000300\n", "2020-04-10 02:29:43,040 Epoch 21 Step: 99500 Batch Loss: 1.968221 Tokens per Sec: 11088, Lr: 0.000300\n", "2020-04-10 02:30:01,311 Epoch 21 Step: 99600 Batch Loss: 2.059013 Tokens per Sec: 11115, Lr: 0.000300\n", "2020-04-10 02:30:19,530 Epoch 21 Step: 99700 Batch Loss: 2.015598 Tokens per Sec: 10867, Lr: 0.000300\n", "2020-04-10 02:30:37,956 Epoch 21 Step: 99800 Batch Loss: 1.677501 Tokens per Sec: 11071, Lr: 0.000300\n", "2020-04-10 02:30:56,137 Epoch 21 Step: 99900 Batch Loss: 1.702487 Tokens per Sec: 10909, Lr: 0.000300\n", "2020-04-10 02:31:14,491 Epoch 21 Step: 100000 Batch Loss: 2.048004 Tokens per Sec: 11049, Lr: 0.000300\n", "2020-04-10 02:31:44,540 Example #0\n", "2020-04-10 02:31:44,541 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 02:31:44,541 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 02:31:44,542 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 02:31:44,542 Example #1\n", "2020-04-10 02:31:44,542 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 02:31:44,542 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 02:31:44,542 \tHypothesis: If We willed , We could have made it blazing , so why do you not give thanks ?\n", "2020-04-10 02:31:44,543 Example #2\n", "2020-04-10 02:31:44,543 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 02:31:44,543 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 02:31:44,543 \tHypothesis: And those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 02:31:44,543 Example #3\n", "2020-04-10 02:31:44,544 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 02:31:44,544 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 02:31:44,544 \tHypothesis: He sends the ark in security , and whenever some of his people passed away before him , they mock him .\n", "2020-04-10 02:31:44,544 Validation result (greedy) at epoch 21, step 100000: bleu: 23.64, loss: 48984.0898, ppl: 6.1227, duration: 30.0528s\n", "2020-04-10 02:32:02,827 Epoch 21 Step: 100100 Batch Loss: 2.139378 Tokens per Sec: 11046, Lr: 0.000300\n", "2020-04-10 02:32:20,973 Epoch 21 Step: 100200 Batch Loss: 1.826693 Tokens per Sec: 10968, Lr: 0.000300\n", "2020-04-10 02:32:39,195 Epoch 21 Step: 100300 Batch Loss: 1.991035 Tokens per Sec: 10757, Lr: 0.000300\n", "2020-04-10 02:32:57,599 Epoch 21 Step: 100400 Batch Loss: 1.827496 Tokens per Sec: 11025, Lr: 0.000300\n", "2020-04-10 02:33:15,857 Epoch 21 Step: 100500 Batch Loss: 1.770897 Tokens per Sec: 11092, Lr: 0.000300\n", "2020-04-10 02:33:34,029 Epoch 21 Step: 100600 Batch Loss: 1.712762 Tokens per Sec: 11162, Lr: 0.000300\n", "2020-04-10 02:33:52,368 Epoch 21 Step: 100700 Batch Loss: 1.831108 Tokens per Sec: 10835, Lr: 0.000300\n", "2020-04-10 02:34:01,605 Epoch 21: total training loss 8331.38\n", "2020-04-10 02:34:01,605 EPOCH 22\n", "2020-04-10 02:34:10,937 Epoch 22 Step: 100800 Batch Loss: 2.074594 Tokens per Sec: 10603, Lr: 0.000300\n", "2020-04-10 02:34:29,143 Epoch 22 Step: 100900 Batch Loss: 1.573670 Tokens per Sec: 10949, Lr: 0.000300\n", "2020-04-10 02:34:47,563 Epoch 22 Step: 101000 Batch Loss: 1.846040 Tokens per Sec: 10812, Lr: 0.000300\n", "2020-04-10 02:35:16,777 Hooray! New best validation result [ppl]!\n", "2020-04-10 02:35:16,777 Saving new checkpoint.\n", "2020-04-10 02:35:18,375 Example #0\n", "2020-04-10 02:35:18,375 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 02:35:18,375 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 02:35:18,375 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-10 02:35:18,376 Example #1\n", "2020-04-10 02:35:18,376 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 02:35:18,376 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 02:35:18,376 \tHypothesis: If We willed , We would have made it a blazing black , so why do you not give thanks ?\n", "2020-04-10 02:35:18,376 Example #2\n", "2020-04-10 02:35:18,377 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 02:35:18,377 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 02:35:18,377 \tHypothesis: And those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 02:35:18,377 Example #3\n", "2020-04-10 02:35:18,378 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 02:35:18,378 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 02:35:18,378 \tHypothesis: And He causes the ark to be safe , and when the leaders of his people passed away before him , they mock him .\n", "2020-04-10 02:35:18,378 Validation result (greedy) at epoch 22, step 101000: bleu: 23.99, loss: 48859.6250, ppl: 6.0946, duration: 30.8149s\n", "2020-04-10 02:35:36,827 Epoch 22 Step: 101100 Batch Loss: 1.868550 Tokens per Sec: 10970, Lr: 0.000300\n", "2020-04-10 02:35:55,158 Epoch 22 Step: 101200 Batch Loss: 2.067685 Tokens per Sec: 10906, Lr: 0.000300\n", "2020-04-10 02:36:13,213 Epoch 22 Step: 101300 Batch Loss: 1.842555 Tokens per Sec: 10810, Lr: 0.000300\n", "2020-04-10 02:36:31,467 Epoch 22 Step: 101400 Batch Loss: 1.851714 Tokens per Sec: 11070, Lr: 0.000300\n", "2020-04-10 02:36:49,582 Epoch 22 Step: 101500 Batch Loss: 2.095229 Tokens per Sec: 10891, Lr: 0.000300\n", "2020-04-10 02:37:07,988 Epoch 22 Step: 101600 Batch Loss: 1.827881 Tokens per Sec: 11274, Lr: 0.000300\n", "2020-04-10 02:37:26,173 Epoch 22 Step: 101700 Batch Loss: 2.174449 Tokens per Sec: 10895, Lr: 0.000300\n", "2020-04-10 02:37:44,297 Epoch 22 Step: 101800 Batch Loss: 2.040180 Tokens per Sec: 10934, Lr: 0.000300\n", "2020-04-10 02:38:02,780 Epoch 22 Step: 101900 Batch Loss: 1.982124 Tokens per Sec: 11336, Lr: 0.000300\n", "2020-04-10 02:38:20,743 Epoch 22 Step: 102000 Batch Loss: 1.865538 Tokens per Sec: 10499, Lr: 0.000300\n", "2020-04-10 02:38:51,135 Hooray! New best validation result [ppl]!\n", "2020-04-10 02:38:51,135 Saving new checkpoint.\n", "2020-04-10 02:38:52,337 Example #0\n", "2020-04-10 02:38:52,337 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 02:38:52,337 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 02:38:52,337 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 02:38:52,338 Example #1\n", "2020-04-10 02:38:52,338 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 02:38:52,338 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 02:38:52,338 \tHypothesis: If We willed , We could have made it rain , so why do you not give thanks ?\n", "2020-04-10 02:38:52,339 Example #2\n", "2020-04-10 02:38:52,339 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 02:38:52,339 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 02:38:52,339 \tHypothesis: And those who disbelieve in Our signs , they are the People of the Left Hand .\n", "2020-04-10 02:38:52,339 Example #3\n", "2020-04-10 02:38:52,340 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 02:38:52,340 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 02:38:52,340 \tHypothesis: And He causes the ship to be sound , and whenever the leaders of his people passed away before him , they ridicule him .\n", "2020-04-10 02:38:52,340 Validation result (greedy) at epoch 22, step 102000: bleu: 24.26, loss: 48840.1719, ppl: 6.0902, duration: 31.5965s\n", "2020-04-10 02:39:10,809 Epoch 22 Step: 102100 Batch Loss: 1.485612 Tokens per Sec: 10893, Lr: 0.000300\n", "2020-04-10 02:39:29,164 Epoch 22 Step: 102200 Batch Loss: 1.864814 Tokens per Sec: 11063, Lr: 0.000300\n", "2020-04-10 02:39:47,417 Epoch 22 Step: 102300 Batch Loss: 1.816631 Tokens per Sec: 10833, Lr: 0.000300\n", "2020-04-10 02:40:05,916 Epoch 22 Step: 102400 Batch Loss: 1.978672 Tokens per Sec: 11142, Lr: 0.000300\n", "2020-04-10 02:40:24,386 Epoch 22 Step: 102500 Batch Loss: 1.942218 Tokens per Sec: 11006, Lr: 0.000300\n", "2020-04-10 02:40:42,862 Epoch 22 Step: 102600 Batch Loss: 1.824738 Tokens per Sec: 11206, Lr: 0.000300\n", "2020-04-10 02:41:01,055 Epoch 22 Step: 102700 Batch Loss: 1.845015 Tokens per Sec: 10881, Lr: 0.000300\n", "2020-04-10 02:41:19,413 Epoch 22 Step: 102800 Batch Loss: 1.873487 Tokens per Sec: 11145, Lr: 0.000300\n", "2020-04-10 02:41:37,844 Epoch 22 Step: 102900 Batch Loss: 2.101802 Tokens per Sec: 11083, Lr: 0.000300\n", "2020-04-10 02:41:56,393 Epoch 22 Step: 103000 Batch Loss: 1.846867 Tokens per Sec: 11101, Lr: 0.000300\n", "2020-04-10 02:42:24,038 Example #0\n", "2020-04-10 02:42:24,039 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 02:42:24,039 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 02:42:24,039 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-10 02:42:24,039 Example #1\n", "2020-04-10 02:42:24,040 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 02:42:24,040 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 02:42:24,040 \tHypothesis: If We willed , We could have made it a rain , so why do you not give thanks ?\n", "2020-04-10 02:42:24,040 Example #2\n", "2020-04-10 02:42:24,041 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 02:42:24,041 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 02:42:24,041 \tHypothesis: And those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 02:42:24,041 Example #3\n", "2020-04-10 02:42:24,041 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 02:42:24,042 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 02:42:24,042 \tHypothesis: He sends the ship to the ship in security , and whenever some of his people passed away before him , they mock him .\n", "2020-04-10 02:42:24,042 Validation result (greedy) at epoch 22, step 103000: bleu: 23.82, loss: 48932.4727, ppl: 6.1111, duration: 27.6483s\n", "2020-04-10 02:42:42,600 Epoch 22 Step: 103100 Batch Loss: 1.764200 Tokens per Sec: 11105, Lr: 0.000300\n", "2020-04-10 02:43:01,086 Epoch 22 Step: 103200 Batch Loss: 1.640442 Tokens per Sec: 10803, Lr: 0.000300\n", "2020-04-10 02:43:19,598 Epoch 22 Step: 103300 Batch Loss: 1.797905 Tokens per Sec: 10864, Lr: 0.000300\n", "2020-04-10 02:43:37,995 Epoch 22 Step: 103400 Batch Loss: 1.913377 Tokens per Sec: 10957, Lr: 0.000300\n", "2020-04-10 02:43:55,982 Epoch 22 Step: 103500 Batch Loss: 1.977235 Tokens per Sec: 10500, Lr: 0.000300\n", "2020-04-10 02:44:14,635 Epoch 22 Step: 103600 Batch Loss: 1.965092 Tokens per Sec: 11103, Lr: 0.000300\n", "2020-04-10 02:44:33,026 Epoch 22 Step: 103700 Batch Loss: 1.992009 Tokens per Sec: 11100, Lr: 0.000300\n", "2020-04-10 02:44:51,254 Epoch 22 Step: 103800 Batch Loss: 1.977119 Tokens per Sec: 10831, Lr: 0.000300\n", "2020-04-10 02:45:09,807 Epoch 22 Step: 103900 Batch Loss: 1.736790 Tokens per Sec: 11067, Lr: 0.000300\n", "2020-04-10 02:45:28,305 Epoch 22 Step: 104000 Batch Loss: 1.956093 Tokens per Sec: 11098, Lr: 0.000300\n", "2020-04-10 02:45:57,936 Hooray! New best validation result [ppl]!\n", "2020-04-10 02:45:57,936 Saving new checkpoint.\n", "2020-04-10 02:45:59,180 Example #0\n", "2020-04-10 02:45:59,181 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 02:45:59,181 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 02:45:59,182 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 02:45:59,182 Example #1\n", "2020-04-10 02:45:59,182 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 02:45:59,182 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 02:45:59,182 \tHypothesis: Had We willed , We could have made it a rain , so why do you not give thanks ?\n", "2020-04-10 02:45:59,183 Example #2\n", "2020-04-10 02:45:59,183 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 02:45:59,183 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 02:45:59,183 \tHypothesis: And those who disbelieve in Our signs are the People of the Left Hand .\n", "2020-04-10 02:45:59,184 Example #3\n", "2020-04-10 02:45:59,184 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 02:45:59,184 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 02:45:59,184 \tHypothesis: And He causes the ship to be safe , and whenever there came to his people , they mock him .\n", "2020-04-10 02:45:59,185 Validation result (greedy) at epoch 22, step 104000: bleu: 24.15, loss: 48710.5469, ppl: 6.0611, duration: 30.8794s\n", "2020-04-10 02:46:17,604 Epoch 22 Step: 104100 Batch Loss: 1.877518 Tokens per Sec: 10914, Lr: 0.000300\n", "2020-04-10 02:46:36,126 Epoch 22 Step: 104200 Batch Loss: 1.778862 Tokens per Sec: 11099, Lr: 0.000300\n", "2020-04-10 02:46:54,484 Epoch 22 Step: 104300 Batch Loss: 1.858765 Tokens per Sec: 10956, Lr: 0.000300\n", "2020-04-10 02:47:12,719 Epoch 22 Step: 104400 Batch Loss: 1.777563 Tokens per Sec: 10796, Lr: 0.000300\n", "2020-04-10 02:47:31,073 Epoch 22 Step: 104500 Batch Loss: 2.008161 Tokens per Sec: 11158, Lr: 0.000300\n", "2020-04-10 02:47:49,227 Epoch 22 Step: 104600 Batch Loss: 1.958198 Tokens per Sec: 10690, Lr: 0.000300\n", "2020-04-10 02:48:07,653 Epoch 22 Step: 104700 Batch Loss: 1.958841 Tokens per Sec: 10947, Lr: 0.000300\n", "2020-04-10 02:48:25,903 Epoch 22 Step: 104800 Batch Loss: 1.995529 Tokens per Sec: 10663, Lr: 0.000300\n", "2020-04-10 02:48:44,227 Epoch 22 Step: 104900 Batch Loss: 1.963789 Tokens per Sec: 10835, Lr: 0.000300\n", "2020-04-10 02:49:02,811 Epoch 22 Step: 105000 Batch Loss: 1.944245 Tokens per Sec: 10977, Lr: 0.000300\n", "2020-04-10 02:49:32,096 Hooray! New best validation result [ppl]!\n", "2020-04-10 02:49:32,097 Saving new checkpoint.\n", "2020-04-10 02:49:33,368 Example #0\n", "2020-04-10 02:49:33,369 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 02:49:33,369 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 02:49:33,369 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 02:49:33,369 Example #1\n", "2020-04-10 02:49:33,370 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 02:49:33,370 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 02:49:33,370 \tHypothesis: If We had willed , We would have made it a rain , so why do you not give thanks ?\n", "2020-04-10 02:49:33,370 Example #2\n", "2020-04-10 02:49:33,371 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 02:49:33,371 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 02:49:33,371 \tHypothesis: Those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 02:49:33,371 Example #3\n", "2020-04-10 02:49:33,372 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 02:49:33,372 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 02:49:33,372 \tHypothesis: And He sends the ship into security , and whenever the leaders of his people passed away before him , they mock him .\n", "2020-04-10 02:49:33,372 Validation result (greedy) at epoch 22, step 105000: bleu: 24.10, loss: 48626.2383, ppl: 6.0422, duration: 30.5604s\n", "2020-04-10 02:49:52,023 Epoch 22 Step: 105100 Batch Loss: 2.144007 Tokens per Sec: 10770, Lr: 0.000300\n", "2020-04-10 02:50:02,893 Epoch 22: total training loss 8274.81\n", "2020-04-10 02:50:02,894 EPOCH 23\n", "2020-04-10 02:50:11,027 Epoch 23 Step: 105200 Batch Loss: 1.871672 Tokens per Sec: 10203, Lr: 0.000300\n", "2020-04-10 02:50:29,511 Epoch 23 Step: 105300 Batch Loss: 1.926469 Tokens per Sec: 10936, Lr: 0.000300\n", "2020-04-10 02:50:47,674 Epoch 23 Step: 105400 Batch Loss: 1.498001 Tokens per Sec: 10838, Lr: 0.000300\n", "2020-04-10 02:51:05,991 Epoch 23 Step: 105500 Batch Loss: 1.747415 Tokens per Sec: 10803, Lr: 0.000300\n", "2020-04-10 02:51:24,396 Epoch 23 Step: 105600 Batch Loss: 2.076372 Tokens per Sec: 10970, Lr: 0.000300\n", "2020-04-10 02:51:42,921 Epoch 23 Step: 105700 Batch Loss: 1.927302 Tokens per Sec: 11181, Lr: 0.000300\n", "2020-04-10 02:52:01,412 Epoch 23 Step: 105800 Batch Loss: 1.990465 Tokens per Sec: 11099, Lr: 0.000300\n", "2020-04-10 02:52:19,809 Epoch 23 Step: 105900 Batch Loss: 1.762620 Tokens per Sec: 10865, Lr: 0.000300\n", "2020-04-10 02:52:38,294 Epoch 23 Step: 106000 Batch Loss: 1.821693 Tokens per Sec: 11014, Lr: 0.000300\n", "2020-04-10 02:53:06,004 Hooray! New best validation result [ppl]!\n", "2020-04-10 02:53:06,004 Saving new checkpoint.\n", "2020-04-10 02:53:07,278 Example #0\n", "2020-04-10 02:53:07,279 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 02:53:07,279 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 02:53:07,279 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 02:53:07,279 Example #1\n", "2020-04-10 02:53:07,280 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 02:53:07,280 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 02:53:07,280 \tHypothesis: If We willed , We would have made it blazing , so why do you not give thanks ?\n", "2020-04-10 02:53:07,280 Example #2\n", "2020-04-10 02:53:07,280 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 02:53:07,281 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 02:53:07,281 \tHypothesis: And those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 02:53:07,281 Example #3\n", "2020-04-10 02:53:07,281 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 02:53:07,282 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 02:53:07,282 \tHypothesis: And He causes the ship to be safe , and whenever there came to his people , they mock him .\n", "2020-04-10 02:53:07,282 Validation result (greedy) at epoch 23, step 106000: bleu: 24.25, loss: 48522.8555, ppl: 6.0192, duration: 28.9875s\n", "2020-04-10 02:53:25,972 Epoch 23 Step: 106100 Batch Loss: 2.049596 Tokens per Sec: 10931, Lr: 0.000300\n", "2020-04-10 02:53:44,487 Epoch 23 Step: 106200 Batch Loss: 1.811247 Tokens per Sec: 11048, Lr: 0.000300\n", "2020-04-10 02:54:02,994 Epoch 23 Step: 106300 Batch Loss: 1.907853 Tokens per Sec: 11010, Lr: 0.000300\n", "2020-04-10 02:54:21,498 Epoch 23 Step: 106400 Batch Loss: 1.871131 Tokens per Sec: 10969, Lr: 0.000300\n", "2020-04-10 02:54:39,900 Epoch 23 Step: 106500 Batch Loss: 1.922667 Tokens per Sec: 10794, Lr: 0.000300\n", "2020-04-10 02:54:58,484 Epoch 23 Step: 106600 Batch Loss: 1.984406 Tokens per Sec: 10923, Lr: 0.000300\n", "2020-04-10 02:55:16,848 Epoch 23 Step: 106700 Batch Loss: 1.790109 Tokens per Sec: 10867, Lr: 0.000300\n", "2020-04-10 02:55:35,183 Epoch 23 Step: 106800 Batch Loss: 1.974879 Tokens per Sec: 10979, Lr: 0.000300\n", "2020-04-10 02:55:53,437 Epoch 23 Step: 106900 Batch Loss: 1.477202 Tokens per Sec: 10636, Lr: 0.000300\n", "2020-04-10 02:56:11,958 Epoch 23 Step: 107000 Batch Loss: 1.996345 Tokens per Sec: 10968, Lr: 0.000300\n", "2020-04-10 02:56:38,663 Example #0\n", "2020-04-10 02:56:38,663 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 02:56:38,663 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 02:56:38,664 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 02:56:38,664 Example #1\n", "2020-04-10 02:56:38,664 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 02:56:38,664 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 02:56:38,665 \tHypothesis: Had We willed , We could have made it a rain , so why do you not give thanks ?\n", "2020-04-10 02:56:38,665 Example #2\n", "2020-04-10 02:56:38,665 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 02:56:38,665 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 02:56:38,665 \tHypothesis: Those who disbelieve in Our signs are the people of the left hand .\n", "2020-04-10 02:56:38,666 Example #3\n", "2020-04-10 02:56:38,666 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 02:56:38,666 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 02:56:38,666 \tHypothesis: And He causes the ship to be safe , and whenever there came to his people , they mock him .\n", "2020-04-10 02:56:38,667 Validation result (greedy) at epoch 23, step 107000: bleu: 24.08, loss: 48525.9336, ppl: 6.0199, duration: 26.7084s\n", "2020-04-10 02:56:57,249 Epoch 23 Step: 107100 Batch Loss: 1.950733 Tokens per Sec: 11077, Lr: 0.000300\n", "2020-04-10 02:57:15,267 Epoch 23 Step: 107200 Batch Loss: 2.339120 Tokens per Sec: 10484, Lr: 0.000300\n", "2020-04-10 02:57:33,745 Epoch 23 Step: 107300 Batch Loss: 2.012876 Tokens per Sec: 11007, Lr: 0.000300\n", "2020-04-10 02:57:52,226 Epoch 23 Step: 107400 Batch Loss: 1.613550 Tokens per Sec: 10651, Lr: 0.000300\n", "2020-04-10 02:58:10,891 Epoch 23 Step: 107500 Batch Loss: 1.856424 Tokens per Sec: 11096, Lr: 0.000300\n", "2020-04-10 02:58:28,991 Epoch 23 Step: 107600 Batch Loss: 1.852674 Tokens per Sec: 10717, Lr: 0.000300\n", "2020-04-10 02:58:47,653 Epoch 23 Step: 107700 Batch Loss: 1.880745 Tokens per Sec: 11222, Lr: 0.000300\n", "2020-04-10 02:59:05,992 Epoch 23 Step: 107800 Batch Loss: 1.700163 Tokens per Sec: 10646, Lr: 0.000300\n", "2020-04-10 02:59:24,361 Epoch 23 Step: 107900 Batch Loss: 2.050034 Tokens per Sec: 11119, Lr: 0.000300\n", "2020-04-10 02:59:42,726 Epoch 23 Step: 108000 Batch Loss: 1.740442 Tokens per Sec: 11022, Lr: 0.000300\n", "2020-04-10 03:00:10,967 Hooray! New best validation result [ppl]!\n", "2020-04-10 03:00:10,968 Saving new checkpoint.\n", "2020-04-10 03:00:12,633 Example #0\n", "2020-04-10 03:00:12,634 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 03:00:12,634 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 03:00:12,634 \tHypothesis: See how We explain the signs , so that they may understand .\n", "2020-04-10 03:00:12,634 Example #1\n", "2020-04-10 03:00:12,635 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 03:00:12,635 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 03:00:12,635 \tHypothesis: Had We willed , We could have made it blazing , so why do you not give thanks ?\n", "2020-04-10 03:00:12,635 Example #2\n", "2020-04-10 03:00:12,636 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 03:00:12,636 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 03:00:12,636 \tHypothesis: Those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 03:00:12,636 Example #3\n", "2020-04-10 03:00:12,637 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 03:00:12,637 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 03:00:12,637 \tHypothesis: He sends the ship in security , and whenever some of his people passed away before him , they mock him .\n", "2020-04-10 03:00:12,637 Validation result (greedy) at epoch 23, step 108000: bleu: 24.09, loss: 48459.8047, ppl: 6.0051, duration: 29.9110s\n", "2020-04-10 03:00:31,306 Epoch 23 Step: 108100 Batch Loss: 1.976243 Tokens per Sec: 10899, Lr: 0.000300\n", "2020-04-10 03:00:49,989 Epoch 23 Step: 108200 Batch Loss: 1.871620 Tokens per Sec: 11104, Lr: 0.000300\n", "2020-04-10 03:01:08,309 Epoch 23 Step: 108300 Batch Loss: 2.023466 Tokens per Sec: 10957, Lr: 0.000300\n", "2020-04-10 03:01:26,690 Epoch 23 Step: 108400 Batch Loss: 2.001287 Tokens per Sec: 10955, Lr: 0.000300\n", "2020-04-10 03:01:45,378 Epoch 23 Step: 108500 Batch Loss: 1.667141 Tokens per Sec: 10959, Lr: 0.000300\n", "2020-04-10 03:02:03,802 Epoch 23 Step: 108600 Batch Loss: 1.893908 Tokens per Sec: 10751, Lr: 0.000300\n", "2020-04-10 03:02:22,272 Epoch 23 Step: 108700 Batch Loss: 1.617733 Tokens per Sec: 10769, Lr: 0.000300\n", "2020-04-10 03:02:40,689 Epoch 23 Step: 108800 Batch Loss: 1.729681 Tokens per Sec: 10719, Lr: 0.000300\n", "2020-04-10 03:02:59,126 Epoch 23 Step: 108900 Batch Loss: 1.843953 Tokens per Sec: 10878, Lr: 0.000300\n", "2020-04-10 03:03:17,528 Epoch 23 Step: 109000 Batch Loss: 2.020837 Tokens per Sec: 10828, Lr: 0.000300\n", "2020-04-10 03:03:47,614 Example #0\n", "2020-04-10 03:03:47,615 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 03:03:47,615 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 03:03:47,615 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 03:03:47,615 Example #1\n", "2020-04-10 03:03:47,616 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 03:03:47,616 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 03:03:47,616 \tHypothesis: If We willed , We could have made it rain , so why do you not give thanks ?\n", "2020-04-10 03:03:47,616 Example #2\n", "2020-04-10 03:03:47,616 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 03:03:47,617 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 03:03:47,617 \tHypothesis: And those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 03:03:47,617 Example #3\n", "2020-04-10 03:03:47,617 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 03:03:47,617 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 03:03:47,618 \tHypothesis: He sends the ships in security , and whenever the leaders of his people passed away before him , they mocked him .\n", "2020-04-10 03:03:47,618 Validation result (greedy) at epoch 23, step 109000: bleu: 24.33, loss: 48502.4297, ppl: 6.0146, duration: 30.0893s\n", "2020-04-10 03:04:06,276 Epoch 23 Step: 109100 Batch Loss: 1.838612 Tokens per Sec: 11007, Lr: 0.000300\n", "2020-04-10 03:04:24,828 Epoch 23 Step: 109200 Batch Loss: 1.931270 Tokens per Sec: 11158, Lr: 0.000300\n", "2020-04-10 03:04:43,013 Epoch 23 Step: 109300 Batch Loss: 1.957262 Tokens per Sec: 10570, Lr: 0.000300\n", "2020-04-10 03:05:01,475 Epoch 23 Step: 109400 Batch Loss: 1.714803 Tokens per Sec: 10899, Lr: 0.000300\n", "2020-04-10 03:05:19,935 Epoch 23 Step: 109500 Batch Loss: 1.960647 Tokens per Sec: 11001, Lr: 0.000300\n", "2020-04-10 03:05:31,698 Epoch 23: total training loss 8233.70\n", "2020-04-10 03:05:31,699 EPOCH 24\n", "2020-04-10 03:05:38,739 Epoch 24 Step: 109600 Batch Loss: 1.779984 Tokens per Sec: 10419, Lr: 0.000300\n", "2020-04-10 03:05:57,105 Epoch 24 Step: 109700 Batch Loss: 1.781981 Tokens per Sec: 10904, Lr: 0.000300\n", "2020-04-10 03:06:15,564 Epoch 24 Step: 109800 Batch Loss: 1.820411 Tokens per Sec: 10921, Lr: 0.000300\n", "2020-04-10 03:06:33,787 Epoch 24 Step: 109900 Batch Loss: 1.803962 Tokens per Sec: 10654, Lr: 0.000300\n", "2020-04-10 03:06:51,994 Epoch 24 Step: 110000 Batch Loss: 2.110221 Tokens per Sec: 10841, Lr: 0.000300\n", "2020-04-10 03:07:20,671 Example #0\n", "2020-04-10 03:07:20,672 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 03:07:20,672 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 03:07:20,672 \tHypothesis: See how We explain the signs that haply they may understand .\n", "2020-04-10 03:07:20,672 Example #1\n", "2020-04-10 03:07:20,673 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 03:07:20,673 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 03:07:20,673 \tHypothesis: If We willed , We would have made it rain , so why do you not give thanks ?\n", "2020-04-10 03:07:20,673 Example #2\n", "2020-04-10 03:07:20,674 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 03:07:20,674 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 03:07:20,674 \tHypothesis: And those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 03:07:20,674 Example #3\n", "2020-04-10 03:07:20,675 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 03:07:20,675 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 03:07:20,675 \tHypothesis: He causes the ark in security , and whenever there came to his people , they mocked him .\n", "2020-04-10 03:07:20,675 Validation result (greedy) at epoch 24, step 110000: bleu: 24.30, loss: 48468.0508, ppl: 6.0070, duration: 28.6804s\n", "2020-04-10 03:07:39,111 Epoch 24 Step: 110100 Batch Loss: 1.964252 Tokens per Sec: 11020, Lr: 0.000300\n", "2020-04-10 03:07:57,495 Epoch 24 Step: 110200 Batch Loss: 1.916517 Tokens per Sec: 10948, Lr: 0.000300\n", "2020-04-10 03:08:15,898 Epoch 24 Step: 110300 Batch Loss: 1.957351 Tokens per Sec: 11006, Lr: 0.000300\n", "2020-04-10 03:08:34,225 Epoch 24 Step: 110400 Batch Loss: 1.980502 Tokens per Sec: 10811, Lr: 0.000300\n", "2020-04-10 03:08:52,812 Epoch 24 Step: 110500 Batch Loss: 2.056383 Tokens per Sec: 11002, Lr: 0.000300\n", "2020-04-10 03:09:11,142 Epoch 24 Step: 110600 Batch Loss: 1.958437 Tokens per Sec: 10877, Lr: 0.000300\n", "2020-04-10 03:09:29,235 Epoch 24 Step: 110700 Batch Loss: 1.590593 Tokens per Sec: 10529, Lr: 0.000300\n", "2020-04-10 03:09:47,703 Epoch 24 Step: 110800 Batch Loss: 1.959534 Tokens per Sec: 10801, Lr: 0.000300\n", "2020-04-10 03:10:06,199 Epoch 24 Step: 110900 Batch Loss: 1.880054 Tokens per Sec: 11062, Lr: 0.000300\n", "2020-04-10 03:10:24,685 Epoch 24 Step: 111000 Batch Loss: 1.905140 Tokens per Sec: 11288, Lr: 0.000300\n", "2020-04-10 03:10:51,379 Hooray! New best validation result [ppl]!\n", "2020-04-10 03:10:51,379 Saving new checkpoint.\n", "2020-04-10 03:10:52,693 Example #0\n", "2020-04-10 03:10:52,694 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 03:10:52,694 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 03:10:52,695 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-10 03:10:52,695 Example #1\n", "2020-04-10 03:10:52,695 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 03:10:52,695 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 03:10:52,695 \tHypothesis: If We willed , We could have made it blazing , so why do you not give thanks ?\n", "2020-04-10 03:10:52,695 Example #2\n", "2020-04-10 03:10:52,696 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 03:10:52,696 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 03:10:52,696 \tHypothesis: And those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 03:10:52,696 Example #3\n", "2020-04-10 03:10:52,697 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 03:10:52,697 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 03:10:52,697 \tHypothesis: He causes the ship to be safe , and whenever there came to his people passed away before him , they mocked him .\n", "2020-04-10 03:10:52,698 Validation result (greedy) at epoch 24, step 111000: bleu: 24.29, loss: 48291.6445, ppl: 5.9679, duration: 28.0124s\n", "2020-04-10 03:11:11,292 Epoch 24 Step: 111100 Batch Loss: 1.824024 Tokens per Sec: 10692, Lr: 0.000300\n", "2020-04-10 03:11:29,795 Epoch 24 Step: 111200 Batch Loss: 2.497510 Tokens per Sec: 11002, Lr: 0.000300\n", "2020-04-10 03:11:48,035 Epoch 24 Step: 111300 Batch Loss: 1.931684 Tokens per Sec: 10768, Lr: 0.000300\n", "2020-04-10 03:12:06,262 Epoch 24 Step: 111400 Batch Loss: 1.946727 Tokens per Sec: 10917, Lr: 0.000300\n", "2020-04-10 03:12:24,423 Epoch 24 Step: 111500 Batch Loss: 1.872406 Tokens per Sec: 10881, Lr: 0.000300\n", "2020-04-10 03:12:42,780 Epoch 24 Step: 111600 Batch Loss: 2.026016 Tokens per Sec: 10850, Lr: 0.000300\n", "2020-04-10 03:13:01,290 Epoch 24 Step: 111700 Batch Loss: 1.989821 Tokens per Sec: 10986, Lr: 0.000300\n", "2020-04-10 03:13:19,326 Epoch 24 Step: 111800 Batch Loss: 1.720408 Tokens per Sec: 10691, Lr: 0.000300\n", "2020-04-10 03:13:37,754 Epoch 24 Step: 111900 Batch Loss: 1.941763 Tokens per Sec: 10848, Lr: 0.000300\n", "2020-04-10 03:13:56,001 Epoch 24 Step: 112000 Batch Loss: 1.962165 Tokens per Sec: 10846, Lr: 0.000300\n", "2020-04-10 03:14:24,331 Example #0\n", "2020-04-10 03:14:24,332 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 03:14:24,332 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 03:14:24,333 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 03:14:24,333 Example #1\n", "2020-04-10 03:14:24,333 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 03:14:24,333 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 03:14:24,333 \tHypothesis: Had We willed , We could have made it blazing , so why do you not give thanks ?\n", "2020-04-10 03:14:24,334 Example #2\n", "2020-04-10 03:14:24,334 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 03:14:24,334 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 03:14:24,334 \tHypothesis: And those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 03:14:24,334 Example #3\n", "2020-04-10 03:14:24,335 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 03:14:24,335 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 03:14:24,335 \tHypothesis: And He sends the ship in security , and whenever some of his people passed away before him , they mock at him .\n", "2020-04-10 03:14:24,335 Validation result (greedy) at epoch 24, step 112000: bleu: 24.35, loss: 48310.2227, ppl: 5.9720, duration: 28.3343s\n", "2020-04-10 03:14:42,972 Epoch 24 Step: 112100 Batch Loss: 1.841927 Tokens per Sec: 11166, Lr: 0.000300\n", "2020-04-10 03:15:01,171 Epoch 24 Step: 112200 Batch Loss: 1.873254 Tokens per Sec: 10661, Lr: 0.000300\n", "2020-04-10 03:15:19,676 Epoch 24 Step: 112300 Batch Loss: 1.691009 Tokens per Sec: 10981, Lr: 0.000300\n", "2020-04-10 03:15:38,171 Epoch 24 Step: 112400 Batch Loss: 1.932855 Tokens per Sec: 11013, Lr: 0.000300\n", "2020-04-10 03:15:56,342 Epoch 24 Step: 112500 Batch Loss: 1.989508 Tokens per Sec: 10776, Lr: 0.000300\n", "2020-04-10 03:16:14,772 Epoch 24 Step: 112600 Batch Loss: 1.860144 Tokens per Sec: 10833, Lr: 0.000300\n", "2020-04-10 03:16:33,173 Epoch 24 Step: 112700 Batch Loss: 1.821693 Tokens per Sec: 11047, Lr: 0.000300\n", "2020-04-10 03:16:51,743 Epoch 24 Step: 112800 Batch Loss: 2.010897 Tokens per Sec: 10946, Lr: 0.000300\n", "2020-04-10 03:17:10,130 Epoch 24 Step: 112900 Batch Loss: 1.883615 Tokens per Sec: 10721, Lr: 0.000300\n", "2020-04-10 03:17:28,467 Epoch 24 Step: 113000 Batch Loss: 1.904087 Tokens per Sec: 10893, Lr: 0.000300\n", "2020-04-10 03:17:54,553 Example #0\n", "2020-04-10 03:17:54,554 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 03:17:54,554 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 03:17:54,554 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-10 03:17:54,554 Example #1\n", "2020-04-10 03:17:54,555 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 03:17:54,555 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 03:17:54,555 \tHypothesis: If We willed , We could have made it rain , so why do you not give thanks ?\n", "2020-04-10 03:17:54,555 Example #2\n", "2020-04-10 03:17:54,556 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 03:17:54,556 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 03:17:54,556 \tHypothesis: And those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 03:17:54,556 Example #3\n", "2020-04-10 03:17:54,556 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 03:17:54,557 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 03:17:54,557 \tHypothesis: And He sends the ship in peace , and whenever the leaders of his people passed away before him , they mock him .\n", "2020-04-10 03:17:54,557 Validation result (greedy) at epoch 24, step 113000: bleu: 24.06, loss: 48346.8633, ppl: 5.9801, duration: 26.0892s\n", "2020-04-10 03:18:13,008 Epoch 24 Step: 113100 Batch Loss: 1.763944 Tokens per Sec: 10964, Lr: 0.000300\n", "2020-04-10 03:18:31,432 Epoch 24 Step: 113200 Batch Loss: 1.801926 Tokens per Sec: 11028, Lr: 0.000300\n", "2020-04-10 03:18:49,719 Epoch 24 Step: 113300 Batch Loss: 1.842804 Tokens per Sec: 10727, Lr: 0.000300\n", "2020-04-10 03:19:08,149 Epoch 24 Step: 113400 Batch Loss: 1.919477 Tokens per Sec: 10891, Lr: 0.000300\n", "2020-04-10 03:19:26,523 Epoch 24 Step: 113500 Batch Loss: 1.786272 Tokens per Sec: 10768, Lr: 0.000300\n", "2020-04-10 03:19:44,971 Epoch 24 Step: 113600 Batch Loss: 1.949725 Tokens per Sec: 11146, Lr: 0.000300\n", "2020-04-10 03:20:03,417 Epoch 24 Step: 113700 Batch Loss: 1.744419 Tokens per Sec: 11065, Lr: 0.000300\n", "2020-04-10 03:20:21,676 Epoch 24 Step: 113800 Batch Loss: 1.921000 Tokens per Sec: 10993, Lr: 0.000300\n", "2020-04-10 03:20:40,121 Epoch 24 Step: 113900 Batch Loss: 1.687077 Tokens per Sec: 11084, Lr: 0.000300\n", "2020-04-10 03:20:56,095 Epoch 24: total training loss 8234.60\n", "2020-04-10 03:20:56,095 EPOCH 25\n", "2020-04-10 03:20:58,895 Epoch 25 Step: 114000 Batch Loss: 1.658973 Tokens per Sec: 8955, Lr: 0.000300\n", "2020-04-10 03:21:27,096 Hooray! New best validation result [ppl]!\n", "2020-04-10 03:21:27,096 Saving new checkpoint.\n", "2020-04-10 03:21:28,303 Example #0\n", "2020-04-10 03:21:28,304 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 03:21:28,304 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 03:21:28,304 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 03:21:28,305 Example #1\n", "2020-04-10 03:21:28,305 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 03:21:28,305 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 03:21:28,305 \tHypothesis: If We willed , We would have made it black , so why do you not give thanks ?\n", "2020-04-10 03:21:28,305 Example #2\n", "2020-04-10 03:21:28,306 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 03:21:28,306 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 03:21:28,306 \tHypothesis: Those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 03:21:28,306 Example #3\n", "2020-04-10 03:21:28,307 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 03:21:28,307 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 03:21:28,307 \tHypothesis: And He sends the ship in security , and whenever the leaders of his people passed away before him , they mock him .\n", "2020-04-10 03:21:28,307 Validation result (greedy) at epoch 25, step 114000: bleu: 24.24, loss: 48094.4219, ppl: 5.9245, duration: 29.4123s\n", "2020-04-10 03:21:46,614 Epoch 25 Step: 114100 Batch Loss: 1.866067 Tokens per Sec: 10848, Lr: 0.000300\n", "2020-04-10 03:22:04,894 Epoch 25 Step: 114200 Batch Loss: 1.878180 Tokens per Sec: 10826, Lr: 0.000300\n", "2020-04-10 03:22:23,178 Epoch 25 Step: 114300 Batch Loss: 2.095686 Tokens per Sec: 11001, Lr: 0.000300\n", "2020-04-10 03:22:41,663 Epoch 25 Step: 114400 Batch Loss: 1.966977 Tokens per Sec: 11153, Lr: 0.000300\n", "2020-04-10 03:22:59,897 Epoch 25 Step: 114500 Batch Loss: 1.861874 Tokens per Sec: 10797, Lr: 0.000300\n", "2020-04-10 03:23:18,339 Epoch 25 Step: 114600 Batch Loss: 1.708859 Tokens per Sec: 11100, Lr: 0.000300\n", "2020-04-10 03:23:36,855 Epoch 25 Step: 114700 Batch Loss: 1.962713 Tokens per Sec: 10967, Lr: 0.000300\n", "2020-04-10 03:23:55,274 Epoch 25 Step: 114800 Batch Loss: 1.950596 Tokens per Sec: 11061, Lr: 0.000300\n", "2020-04-10 03:24:13,685 Epoch 25 Step: 114900 Batch Loss: 1.470534 Tokens per Sec: 10820, Lr: 0.000300\n", "2020-04-10 03:24:32,122 Epoch 25 Step: 115000 Batch Loss: 1.770441 Tokens per Sec: 11056, Lr: 0.000300\n", "2020-04-10 03:25:00,997 Example #0\n", "2020-04-10 03:25:00,997 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 03:25:00,997 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 03:25:00,998 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 03:25:00,998 Example #1\n", "2020-04-10 03:25:00,998 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 03:25:00,998 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 03:25:00,998 \tHypothesis: If We willed , We could have made it rain , so why do you not give thanks ?\n", "2020-04-10 03:25:00,999 Example #2\n", "2020-04-10 03:25:00,999 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 03:25:00,999 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 03:25:00,999 \tHypothesis: Those who disbelieve in Our signs are the companions of the Left .\n", "2020-04-10 03:25:00,999 Example #3\n", "2020-04-10 03:25:01,000 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 03:25:01,000 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 03:25:01,000 \tHypothesis: And He sends the ship in security , and whenever the leaders of his people passed away before him , they mock him .\n", "2020-04-10 03:25:01,000 Validation result (greedy) at epoch 25, step 115000: bleu: 24.31, loss: 48121.5273, ppl: 5.9305, duration: 28.8776s\n", "2020-04-10 03:25:19,405 Epoch 25 Step: 115100 Batch Loss: 2.168709 Tokens per Sec: 10946, Lr: 0.000300\n", "2020-04-10 03:25:37,808 Epoch 25 Step: 115200 Batch Loss: 1.594901 Tokens per Sec: 11157, Lr: 0.000300\n", "2020-04-10 03:25:56,306 Epoch 25 Step: 115300 Batch Loss: 1.841967 Tokens per Sec: 11153, Lr: 0.000300\n", "2020-04-10 03:26:14,687 Epoch 25 Step: 115400 Batch Loss: 1.736718 Tokens per Sec: 11039, Lr: 0.000300\n", "2020-04-10 03:26:33,151 Epoch 25 Step: 115500 Batch Loss: 1.914676 Tokens per Sec: 11110, Lr: 0.000300\n", "2020-04-10 03:26:51,414 Epoch 25 Step: 115600 Batch Loss: 1.694453 Tokens per Sec: 11049, Lr: 0.000300\n", "2020-04-10 03:27:09,625 Epoch 25 Step: 115700 Batch Loss: 1.863773 Tokens per Sec: 11040, Lr: 0.000300\n", "2020-04-10 03:27:27,967 Epoch 25 Step: 115800 Batch Loss: 1.894435 Tokens per Sec: 10975, Lr: 0.000300\n", "2020-04-10 03:27:46,399 Epoch 25 Step: 115900 Batch Loss: 1.648549 Tokens per Sec: 10955, Lr: 0.000300\n", "2020-04-10 03:28:04,792 Epoch 25 Step: 116000 Batch Loss: 1.785757 Tokens per Sec: 10975, Lr: 0.000300\n", "2020-04-10 03:28:35,300 Hooray! New best validation result [ppl]!\n", "2020-04-10 03:28:35,300 Saving new checkpoint.\n", "2020-04-10 03:28:36,893 Example #0\n", "2020-04-10 03:28:36,894 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 03:28:36,894 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 03:28:36,894 \tHypothesis: See how We explain the signs , so that they may understand .\n", "2020-04-10 03:28:36,895 Example #1\n", "2020-04-10 03:28:36,895 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 03:28:36,895 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 03:28:36,896 \tHypothesis: If We willed , We would have made it rain , so why do you not give thanks ?\n", "2020-04-10 03:28:36,896 Example #2\n", "2020-04-10 03:28:36,896 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 03:28:36,896 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 03:28:36,897 \tHypothesis: And those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 03:28:36,897 Example #3\n", "2020-04-10 03:28:36,897 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 03:28:36,897 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 03:28:36,898 \tHypothesis: He causes the ship to be safe , and whenever some of his people passed away before him , they ridicule him .\n", "2020-04-10 03:28:36,898 Validation result (greedy) at epoch 25, step 116000: bleu: 24.44, loss: 48085.6250, ppl: 5.9226, duration: 32.1050s\n", "2020-04-10 03:28:55,398 Epoch 25 Step: 116100 Batch Loss: 1.773732 Tokens per Sec: 10862, Lr: 0.000300\n", "2020-04-10 03:29:13,668 Epoch 25 Step: 116200 Batch Loss: 1.696605 Tokens per Sec: 10649, Lr: 0.000300\n", "2020-04-10 03:29:31,939 Epoch 25 Step: 116300 Batch Loss: 2.121222 Tokens per Sec: 10917, Lr: 0.000300\n", "2020-04-10 03:29:50,474 Epoch 25 Step: 116400 Batch Loss: 2.029610 Tokens per Sec: 10951, Lr: 0.000300\n", "2020-04-10 03:30:08,941 Epoch 25 Step: 116500 Batch Loss: 1.788030 Tokens per Sec: 10793, Lr: 0.000300\n", "2020-04-10 03:30:27,447 Epoch 25 Step: 116600 Batch Loss: 1.776874 Tokens per Sec: 11037, Lr: 0.000300\n", "2020-04-10 03:30:45,774 Epoch 25 Step: 116700 Batch Loss: 1.947770 Tokens per Sec: 10905, Lr: 0.000300\n", "2020-04-10 03:31:04,009 Epoch 25 Step: 116800 Batch Loss: 1.876894 Tokens per Sec: 10806, Lr: 0.000300\n", "2020-04-10 03:31:22,650 Epoch 25 Step: 116900 Batch Loss: 1.857099 Tokens per Sec: 11034, Lr: 0.000300\n", "2020-04-10 03:31:41,108 Epoch 25 Step: 117000 Batch Loss: 1.925053 Tokens per Sec: 11073, Lr: 0.000300\n", "2020-04-10 03:32:09,966 Hooray! New best validation result [ppl]!\n", "2020-04-10 03:32:09,967 Saving new checkpoint.\n", "2020-04-10 03:32:11,217 Example #0\n", "2020-04-10 03:32:11,218 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 03:32:11,218 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 03:32:11,218 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-10 03:32:11,218 Example #1\n", "2020-04-10 03:32:11,218 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 03:32:11,219 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 03:32:11,219 \tHypothesis: Had We willed , We could have made it rain , so why do you not give thanks ?\n", "2020-04-10 03:32:11,219 Example #2\n", "2020-04-10 03:32:11,219 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 03:32:11,219 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 03:32:11,220 \tHypothesis: And those who disbelieve in Our signs are the companions of the Left .\n", "2020-04-10 03:32:11,220 Example #3\n", "2020-04-10 03:32:11,220 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 03:32:11,220 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 03:32:11,220 \tHypothesis: He causes the ship to be safe , and whenever the leaders of his people passed away before him , they mock him .\n", "2020-04-10 03:32:11,220 Validation result (greedy) at epoch 25, step 117000: bleu: 24.31, loss: 48019.5000, ppl: 5.9081, duration: 30.1119s\n", "2020-04-10 03:32:29,408 Epoch 25 Step: 117100 Batch Loss: 1.456492 Tokens per Sec: 10508, Lr: 0.000300\n", "2020-04-10 03:32:47,852 Epoch 25 Step: 117200 Batch Loss: 1.925083 Tokens per Sec: 10946, Lr: 0.000300\n", "2020-04-10 03:33:06,154 Epoch 25 Step: 117300 Batch Loss: 1.920983 Tokens per Sec: 10826, Lr: 0.000300\n", "2020-04-10 03:33:24,420 Epoch 25 Step: 117400 Batch Loss: 1.655251 Tokens per Sec: 10737, Lr: 0.000300\n", "2020-04-10 03:33:42,939 Epoch 25 Step: 117500 Batch Loss: 1.861490 Tokens per Sec: 10991, Lr: 0.000300\n", "2020-04-10 03:34:01,456 Epoch 25 Step: 117600 Batch Loss: 1.835023 Tokens per Sec: 10842, Lr: 0.000300\n", "2020-04-10 03:34:19,636 Epoch 25 Step: 117700 Batch Loss: 1.965214 Tokens per Sec: 10999, Lr: 0.000300\n", "2020-04-10 03:34:37,869 Epoch 25 Step: 117800 Batch Loss: 2.058693 Tokens per Sec: 10475, Lr: 0.000300\n", "2020-04-10 03:34:56,370 Epoch 25 Step: 117900 Batch Loss: 1.696903 Tokens per Sec: 11021, Lr: 0.000300\n", "2020-04-10 03:35:14,966 Epoch 25 Step: 118000 Batch Loss: 1.869318 Tokens per Sec: 10811, Lr: 0.000300\n", "2020-04-10 03:35:45,245 Example #0\n", "2020-04-10 03:35:45,245 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 03:35:45,246 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 03:35:45,246 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 03:35:45,246 Example #1\n", "2020-04-10 03:35:45,246 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 03:35:45,246 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 03:35:45,246 \tHypothesis: If We had willed , We could have made it rain , so why do you not give thanks ?\n", "2020-04-10 03:35:45,247 Example #2\n", "2020-04-10 03:35:45,247 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 03:35:45,247 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 03:35:45,247 \tHypothesis: And those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 03:35:45,248 Example #3\n", "2020-04-10 03:35:45,249 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 03:35:45,249 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 03:35:45,249 \tHypothesis: He causes the ship to be safe , and whenever the leaders of his people passed away before him , they mock him .\n", "2020-04-10 03:35:45,249 Validation result (greedy) at epoch 25, step 118000: bleu: 24.40, loss: 48032.8828, ppl: 5.9111, duration: 30.2834s\n", "2020-04-10 03:36:03,264 Epoch 25 Step: 118100 Batch Loss: 1.950167 Tokens per Sec: 10919, Lr: 0.000300\n", "2020-04-10 03:36:21,665 Epoch 25 Step: 118200 Batch Loss: 1.958630 Tokens per Sec: 10987, Lr: 0.000300\n", "2020-04-10 03:36:40,082 Epoch 25 Step: 118300 Batch Loss: 1.810915 Tokens per Sec: 10761, Lr: 0.000300\n", "2020-04-10 03:36:58,373 Epoch 25 Step: 118400 Batch Loss: 1.687256 Tokens per Sec: 10870, Lr: 0.000300\n", "2020-04-10 03:36:58,948 Epoch 25: total training loss 8182.42\n", "2020-04-10 03:36:58,953 EPOCH 26\n", "2020-04-10 03:37:17,052 Epoch 26 Step: 118500 Batch Loss: 2.006658 Tokens per Sec: 10614, Lr: 0.000300\n", "2020-04-10 03:37:35,462 Epoch 26 Step: 118600 Batch Loss: 1.841850 Tokens per Sec: 10905, Lr: 0.000300\n", "2020-04-10 03:37:53,885 Epoch 26 Step: 118700 Batch Loss: 1.753000 Tokens per Sec: 11243, Lr: 0.000300\n", "2020-04-10 03:38:12,290 Epoch 26 Step: 118800 Batch Loss: 1.972871 Tokens per Sec: 10982, Lr: 0.000300\n", "2020-04-10 03:38:30,619 Epoch 26 Step: 118900 Batch Loss: 1.850541 Tokens per Sec: 10947, Lr: 0.000300\n", "2020-04-10 03:38:48,787 Epoch 26 Step: 119000 Batch Loss: 2.273704 Tokens per Sec: 10710, Lr: 0.000300\n", "2020-04-10 03:39:18,493 Hooray! New best validation result [ppl]!\n", "2020-04-10 03:39:18,493 Saving new checkpoint.\n", "2020-04-10 03:39:19,771 Example #0\n", "2020-04-10 03:39:19,771 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 03:39:19,772 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 03:39:19,772 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 03:39:19,772 Example #1\n", "2020-04-10 03:39:19,772 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 03:39:19,773 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 03:39:19,773 \tHypothesis: Had We willed , We would have made it rain , so why do you not give thanks ?\n", "2020-04-10 03:39:19,773 Example #2\n", "2020-04-10 03:39:19,773 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 03:39:19,774 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 03:39:19,774 \tHypothesis: Those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 03:39:19,774 Example #3\n", "2020-04-10 03:39:19,774 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 03:39:19,774 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 03:39:19,775 \tHypothesis: He sends the ships in security , and whenever some of his people passed away before him , they mock him .\n", "2020-04-10 03:39:19,775 Validation result (greedy) at epoch 26, step 119000: bleu: 24.13, loss: 47913.3906, ppl: 5.8850, duration: 30.9874s\n", "2020-04-10 03:39:38,252 Epoch 26 Step: 119100 Batch Loss: 1.902018 Tokens per Sec: 10820, Lr: 0.000300\n", "2020-04-10 03:39:56,492 Epoch 26 Step: 119200 Batch Loss: 1.968175 Tokens per Sec: 10979, Lr: 0.000300\n", "2020-04-10 03:40:14,793 Epoch 26 Step: 119300 Batch Loss: 1.581203 Tokens per Sec: 11004, Lr: 0.000300\n", "2020-04-10 03:40:33,205 Epoch 26 Step: 119400 Batch Loss: 1.769534 Tokens per Sec: 10955, Lr: 0.000300\n", "2020-04-10 03:40:51,335 Epoch 26 Step: 119500 Batch Loss: 1.868160 Tokens per Sec: 10733, Lr: 0.000300\n", "2020-04-10 03:41:09,846 Epoch 26 Step: 119600 Batch Loss: 1.970608 Tokens per Sec: 11028, Lr: 0.000300\n", "2020-04-10 03:41:28,169 Epoch 26 Step: 119700 Batch Loss: 1.843370 Tokens per Sec: 10903, Lr: 0.000300\n", "2020-04-10 03:41:46,799 Epoch 26 Step: 119800 Batch Loss: 1.783899 Tokens per Sec: 11211, Lr: 0.000300\n", "2020-04-10 03:42:05,273 Epoch 26 Step: 119900 Batch Loss: 1.950814 Tokens per Sec: 10848, Lr: 0.000300\n", "2020-04-10 03:42:23,550 Epoch 26 Step: 120000 Batch Loss: 1.866028 Tokens per Sec: 11013, Lr: 0.000300\n", "2020-04-10 03:42:51,855 Example #0\n", "2020-04-10 03:42:51,855 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 03:42:51,856 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 03:42:51,856 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 03:42:51,856 Example #1\n", "2020-04-10 03:42:51,856 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 03:42:51,856 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 03:42:51,856 \tHypothesis: If We willed , We would have made it blazing , so why do you not give thanks ?\n", "2020-04-10 03:42:51,857 Example #2\n", "2020-04-10 03:42:51,857 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 03:42:51,857 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 03:42:51,857 \tHypothesis: And those who disbelieve in Our signs are the companions of the Left Hand .\n", "2020-04-10 03:42:51,857 Example #3\n", "2020-04-10 03:42:51,858 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 03:42:51,858 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 03:42:51,858 \tHypothesis: He sends the ark in peace , and whenever some of his people passed away before him , they mocked at him .\n", "2020-04-10 03:42:51,858 Validation result (greedy) at epoch 26, step 120000: bleu: 24.16, loss: 48253.9102, ppl: 5.9596, duration: 28.3078s\n", "2020-04-10 03:43:10,267 Epoch 26 Step: 120100 Batch Loss: 1.810174 Tokens per Sec: 10953, Lr: 0.000300\n", "2020-04-10 03:43:28,753 Epoch 26 Step: 120200 Batch Loss: 1.801394 Tokens per Sec: 10897, Lr: 0.000300\n", "2020-04-10 03:43:47,093 Epoch 26 Step: 120300 Batch Loss: 1.913710 Tokens per Sec: 10806, Lr: 0.000300\n", "2020-04-10 03:44:05,198 Epoch 26 Step: 120400 Batch Loss: 1.822316 Tokens per Sec: 10502, Lr: 0.000300\n", "2020-04-10 03:44:23,546 Epoch 26 Step: 120500 Batch Loss: 1.882113 Tokens per Sec: 11052, Lr: 0.000300\n", "2020-04-10 03:44:42,040 Epoch 26 Step: 120600 Batch Loss: 1.915606 Tokens per Sec: 11214, Lr: 0.000300\n", "2020-04-10 03:45:00,613 Epoch 26 Step: 120700 Batch Loss: 2.177626 Tokens per Sec: 10980, Lr: 0.000300\n", "2020-04-10 03:45:18,641 Epoch 26 Step: 120800 Batch Loss: 1.660515 Tokens per Sec: 10770, Lr: 0.000300\n", "2020-04-10 03:45:37,145 Epoch 26 Step: 120900 Batch Loss: 1.395728 Tokens per Sec: 10878, Lr: 0.000300\n", "2020-04-10 03:45:55,542 Epoch 26 Step: 121000 Batch Loss: 1.852451 Tokens per Sec: 10910, Lr: 0.000300\n", "2020-04-10 03:46:24,412 Example #0\n", "2020-04-10 03:46:24,412 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 03:46:24,412 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 03:46:24,413 \tHypothesis: See how We explain Our signs so that they may understand .\n", "2020-04-10 03:46:24,413 Example #1\n", "2020-04-10 03:46:24,413 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 03:46:24,413 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 03:46:24,413 \tHypothesis: If We willed , We could have made it blazing , so why do you not give thanks ?\n", "2020-04-10 03:46:24,414 Example #2\n", "2020-04-10 03:46:24,414 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 03:46:24,414 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 03:46:24,414 \tHypothesis: And those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 03:46:24,414 Example #3\n", "2020-04-10 03:46:24,415 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 03:46:24,415 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 03:46:24,415 \tHypothesis: And He puts the ship in peace , and whenever the leaders of his people passed away before him , they mock him .\n", "2020-04-10 03:46:24,415 Validation result (greedy) at epoch 26, step 121000: bleu: 24.03, loss: 48035.3867, ppl: 5.9116, duration: 28.8724s\n", "2020-04-10 03:46:42,847 Epoch 26 Step: 121100 Batch Loss: 1.772023 Tokens per Sec: 11125, Lr: 0.000300\n", "2020-04-10 03:47:01,037 Epoch 26 Step: 121200 Batch Loss: 1.759276 Tokens per Sec: 10958, Lr: 0.000300\n", "2020-04-10 03:47:19,461 Epoch 26 Step: 121300 Batch Loss: 1.674344 Tokens per Sec: 11082, Lr: 0.000300\n", "2020-04-10 03:47:37,865 Epoch 26 Step: 121400 Batch Loss: 1.881211 Tokens per Sec: 11080, Lr: 0.000300\n", "2020-04-10 03:47:56,170 Epoch 26 Step: 121500 Batch Loss: 1.719445 Tokens per Sec: 10956, Lr: 0.000300\n", "2020-04-10 03:48:14,549 Epoch 26 Step: 121600 Batch Loss: 1.829804 Tokens per Sec: 10953, Lr: 0.000300\n", "2020-04-10 03:48:32,834 Epoch 26 Step: 121700 Batch Loss: 1.919584 Tokens per Sec: 10849, Lr: 0.000300\n", "2020-04-10 03:48:51,076 Epoch 26 Step: 121800 Batch Loss: 1.525952 Tokens per Sec: 10865, Lr: 0.000300\n", "2020-04-10 03:49:09,498 Epoch 26 Step: 121900 Batch Loss: 1.709751 Tokens per Sec: 10758, Lr: 0.000300\n", "2020-04-10 03:49:27,904 Epoch 26 Step: 122000 Batch Loss: 2.021675 Tokens per Sec: 11103, Lr: 0.000300\n", "2020-04-10 03:49:55,925 Hooray! New best validation result [ppl]!\n", "2020-04-10 03:49:55,926 Saving new checkpoint.\n", "2020-04-10 03:49:57,173 Example #0\n", "2020-04-10 03:49:57,173 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 03:49:57,174 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 03:49:57,174 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-10 03:49:57,174 Example #1\n", "2020-04-10 03:49:57,174 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 03:49:57,175 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 03:49:57,175 \tHypothesis: If We willed , We could have made it blazing , so why do you not give thanks ?\n", "2020-04-10 03:49:57,175 Example #2\n", "2020-04-10 03:49:57,175 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 03:49:57,175 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 03:49:57,175 \tHypothesis: And those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 03:49:57,175 Example #3\n", "2020-04-10 03:49:57,176 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 03:49:57,176 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 03:49:57,176 \tHypothesis: He sends the ships in peace , and whenever there came to him some of his people passed away before him , they mocked at him .\n", "2020-04-10 03:49:57,176 Validation result (greedy) at epoch 26, step 122000: bleu: 24.24, loss: 47881.0352, ppl: 5.8779, duration: 29.2714s\n", "2020-04-10 03:50:15,531 Epoch 26 Step: 122100 Batch Loss: 2.013350 Tokens per Sec: 10688, Lr: 0.000300\n", "2020-04-10 03:50:33,741 Epoch 26 Step: 122200 Batch Loss: 2.106317 Tokens per Sec: 11099, Lr: 0.000300\n", "2020-04-10 03:50:52,151 Epoch 26 Step: 122300 Batch Loss: 1.764923 Tokens per Sec: 11080, Lr: 0.000300\n", "2020-04-10 03:51:10,621 Epoch 26 Step: 122400 Batch Loss: 1.873826 Tokens per Sec: 11158, Lr: 0.000300\n", "2020-04-10 03:51:28,690 Epoch 26 Step: 122500 Batch Loss: 1.780244 Tokens per Sec: 10679, Lr: 0.000300\n", "2020-04-10 03:51:47,006 Epoch 26 Step: 122600 Batch Loss: 1.813673 Tokens per Sec: 10911, Lr: 0.000300\n", "2020-04-10 03:52:05,124 Epoch 26 Step: 122700 Batch Loss: 2.010910 Tokens per Sec: 10811, Lr: 0.000300\n", "2020-04-10 03:52:23,317 Epoch 26 Step: 122800 Batch Loss: 1.843598 Tokens per Sec: 10915, Lr: 0.000300\n", "2020-04-10 03:52:27,442 Epoch 26: total training loss 8158.77\n", "2020-04-10 03:52:27,442 EPOCH 27\n", "2020-04-10 03:52:42,073 Epoch 27 Step: 122900 Batch Loss: 1.831105 Tokens per Sec: 10434, Lr: 0.000300\n", "2020-04-10 03:53:00,407 Epoch 27 Step: 123000 Batch Loss: 1.854269 Tokens per Sec: 11073, Lr: 0.000300\n", "2020-04-10 03:53:29,127 Hooray! New best validation result [ppl]!\n", "2020-04-10 03:53:29,127 Saving new checkpoint.\n", "2020-04-10 03:53:30,373 Example #0\n", "2020-04-10 03:53:30,374 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 03:53:30,374 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 03:53:30,374 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 03:53:30,374 Example #1\n", "2020-04-10 03:53:30,375 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 03:53:30,375 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 03:53:30,375 \tHypothesis: If We willed , We could have made it rain , so why do you not give thanks ?\n", "2020-04-10 03:53:30,375 Example #2\n", "2020-04-10 03:53:30,376 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 03:53:30,376 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 03:53:30,376 \tHypothesis: Those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 03:53:30,376 Example #3\n", "2020-04-10 03:53:30,376 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 03:53:30,377 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 03:53:30,377 \tHypothesis: He sends the ark in peace , and whenever some of his people passed away before him , they mocked him .\n", "2020-04-10 03:53:30,377 Validation result (greedy) at epoch 27, step 123000: bleu: 24.14, loss: 47822.1875, ppl: 5.8652, duration: 29.9691s\n", "2020-04-10 03:53:48,840 Epoch 27 Step: 123100 Batch Loss: 1.785135 Tokens per Sec: 10964, Lr: 0.000300\n", "2020-04-10 03:54:07,317 Epoch 27 Step: 123200 Batch Loss: 1.895998 Tokens per Sec: 11126, Lr: 0.000300\n", "2020-04-10 03:54:25,799 Epoch 27 Step: 123300 Batch Loss: 1.982837 Tokens per Sec: 10888, Lr: 0.000300\n", "2020-04-10 03:54:44,342 Epoch 27 Step: 123400 Batch Loss: 1.553918 Tokens per Sec: 11003, Lr: 0.000300\n", "2020-04-10 03:55:02,598 Epoch 27 Step: 123500 Batch Loss: 1.847499 Tokens per Sec: 10867, Lr: 0.000300\n", "2020-04-10 03:55:20,956 Epoch 27 Step: 123600 Batch Loss: 1.941167 Tokens per Sec: 10985, Lr: 0.000300\n", "2020-04-10 03:55:39,494 Epoch 27 Step: 123700 Batch Loss: 1.914196 Tokens per Sec: 11033, Lr: 0.000300\n", "2020-04-10 03:55:57,729 Epoch 27 Step: 123800 Batch Loss: 2.028789 Tokens per Sec: 10956, Lr: 0.000300\n", "2020-04-10 03:56:15,954 Epoch 27 Step: 123900 Batch Loss: 1.865741 Tokens per Sec: 10739, Lr: 0.000300\n", "2020-04-10 03:56:34,267 Epoch 27 Step: 124000 Batch Loss: 1.856094 Tokens per Sec: 10868, Lr: 0.000300\n", "2020-04-10 03:57:03,690 Hooray! New best validation result [ppl]!\n", "2020-04-10 03:57:03,690 Saving new checkpoint.\n", "2020-04-10 03:57:04,991 Example #0\n", "2020-04-10 03:57:04,991 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 03:57:04,992 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 03:57:04,992 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-10 03:57:04,992 Example #1\n", "2020-04-10 03:57:04,992 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 03:57:04,992 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 03:57:04,992 \tHypothesis: If We willed , We would have made it blazing , so why do you not give thanks ?\n", "2020-04-10 03:57:04,993 Example #2\n", "2020-04-10 03:57:04,993 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 03:57:04,993 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 03:57:04,993 \tHypothesis: And those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 03:57:04,993 Example #3\n", "2020-04-10 03:57:04,994 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 03:57:04,994 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 03:57:04,994 \tHypothesis: He causes the ship to be safe , and whenever some of his people passed away before him , they mock him .\n", "2020-04-10 03:57:04,994 Validation result (greedy) at epoch 27, step 124000: bleu: 24.26, loss: 47750.2109, ppl: 5.8496, duration: 30.7266s\n", "2020-04-10 03:57:23,589 Epoch 27 Step: 124100 Batch Loss: 1.699356 Tokens per Sec: 11157, Lr: 0.000300\n", "2020-04-10 03:57:41,756 Epoch 27 Step: 124200 Batch Loss: 2.078734 Tokens per Sec: 11091, Lr: 0.000300\n", "2020-04-10 03:58:00,150 Epoch 27 Step: 124300 Batch Loss: 1.883550 Tokens per Sec: 11128, Lr: 0.000300\n", "2020-04-10 03:58:18,354 Epoch 27 Step: 124400 Batch Loss: 2.192124 Tokens per Sec: 10950, Lr: 0.000300\n", "2020-04-10 03:58:36,544 Epoch 27 Step: 124500 Batch Loss: 1.885268 Tokens per Sec: 11085, Lr: 0.000300\n", "2020-04-10 03:58:54,648 Epoch 27 Step: 124600 Batch Loss: 1.407257 Tokens per Sec: 10743, Lr: 0.000300\n", "2020-04-10 03:59:12,817 Epoch 27 Step: 124700 Batch Loss: 1.598969 Tokens per Sec: 11243, Lr: 0.000300\n", "2020-04-10 03:59:31,014 Epoch 27 Step: 124800 Batch Loss: 1.863965 Tokens per Sec: 10923, Lr: 0.000300\n", "2020-04-10 03:59:49,529 Epoch 27 Step: 124900 Batch Loss: 1.909608 Tokens per Sec: 11126, Lr: 0.000300\n", "2020-04-10 04:00:07,867 Epoch 27 Step: 125000 Batch Loss: 1.911790 Tokens per Sec: 11104, Lr: 0.000300\n", "2020-04-10 04:00:38,613 Hooray! New best validation result [ppl]!\n", "2020-04-10 04:00:38,614 Saving new checkpoint.\n", "2020-04-10 04:00:39,889 Example #0\n", "2020-04-10 04:00:39,890 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 04:00:39,890 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 04:00:39,890 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-10 04:00:39,890 Example #1\n", "2020-04-10 04:00:39,891 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 04:00:39,891 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 04:00:39,891 \tHypothesis: If We willed , We could have made it black , so why do you not give thanks ?\n", "2020-04-10 04:00:39,891 Example #2\n", "2020-04-10 04:00:39,891 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 04:00:39,891 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 04:00:39,892 \tHypothesis: Those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 04:00:39,892 Example #3\n", "2020-04-10 04:00:39,892 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 04:00:39,892 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 04:00:39,892 \tHypothesis: He causes the ship to be safe , and whenever the leaders of his people passed away before him , they mock him .\n", "2020-04-10 04:00:39,893 Validation result (greedy) at epoch 27, step 125000: bleu: 24.48, loss: 47717.1172, ppl: 5.8424, duration: 32.0250s\n", "2020-04-10 04:00:58,179 Epoch 27 Step: 125100 Batch Loss: 1.881755 Tokens per Sec: 10485, Lr: 0.000300\n", "2020-04-10 04:01:16,262 Epoch 27 Step: 125200 Batch Loss: 1.938323 Tokens per Sec: 10869, Lr: 0.000300\n", "2020-04-10 04:01:34,420 Epoch 27 Step: 125300 Batch Loss: 1.878189 Tokens per Sec: 11018, Lr: 0.000300\n", "2020-04-10 04:01:52,716 Epoch 27 Step: 125400 Batch Loss: 1.868928 Tokens per Sec: 10966, Lr: 0.000300\n", "2020-04-10 04:02:11,035 Epoch 27 Step: 125500 Batch Loss: 2.057715 Tokens per Sec: 10899, Lr: 0.000300\n", "2020-04-10 04:02:29,362 Epoch 27 Step: 125600 Batch Loss: 1.719055 Tokens per Sec: 11126, Lr: 0.000300\n", "2020-04-10 04:02:47,394 Epoch 27 Step: 125700 Batch Loss: 1.950153 Tokens per Sec: 10711, Lr: 0.000300\n", "2020-04-10 04:03:05,723 Epoch 27 Step: 125800 Batch Loss: 2.143820 Tokens per Sec: 11037, Lr: 0.000300\n", "2020-04-10 04:03:23,941 Epoch 27 Step: 125900 Batch Loss: 1.918655 Tokens per Sec: 10955, Lr: 0.000300\n", "2020-04-10 04:03:42,132 Epoch 27 Step: 126000 Batch Loss: 1.843726 Tokens per Sec: 10745, Lr: 0.000300\n", "2020-04-10 04:04:12,242 Example #0\n", "2020-04-10 04:04:12,242 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 04:04:12,242 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 04:04:12,242 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 04:04:12,243 Example #1\n", "2020-04-10 04:04:12,243 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 04:04:12,243 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 04:04:12,243 \tHypothesis: If We willed , We could have made it blazing , so why do you not give thanks ?\n", "2020-04-10 04:04:12,243 Example #2\n", "2020-04-10 04:04:12,244 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 04:04:12,244 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 04:04:12,244 \tHypothesis: And those who disbelieve in Our signs are the companions of the Left .\n", "2020-04-10 04:04:12,244 Example #3\n", "2020-04-10 04:04:12,244 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 04:04:12,244 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 04:04:12,245 \tHypothesis: He causes the ark to be safe , and whenever some of his people passed away before him , they mocked him .\n", "2020-04-10 04:04:12,245 Validation result (greedy) at epoch 27, step 126000: bleu: 24.45, loss: 47921.3711, ppl: 5.8867, duration: 30.1127s\n", "2020-04-10 04:04:30,381 Epoch 27 Step: 126100 Batch Loss: 1.626867 Tokens per Sec: 10989, Lr: 0.000300\n", "2020-04-10 04:04:48,725 Epoch 27 Step: 126200 Batch Loss: 1.794253 Tokens per Sec: 11198, Lr: 0.000300\n", "2020-04-10 04:05:07,008 Epoch 27 Step: 126300 Batch Loss: 1.917309 Tokens per Sec: 10928, Lr: 0.000300\n", "2020-04-10 04:05:25,308 Epoch 27 Step: 126400 Batch Loss: 1.979839 Tokens per Sec: 10864, Lr: 0.000300\n", "2020-04-10 04:05:43,456 Epoch 27 Step: 126500 Batch Loss: 1.455881 Tokens per Sec: 10749, Lr: 0.000300\n", "2020-04-10 04:06:01,850 Epoch 27 Step: 126600 Batch Loss: 1.913621 Tokens per Sec: 10996, Lr: 0.000300\n", "2020-04-10 04:06:20,221 Epoch 27 Step: 126700 Batch Loss: 2.121654 Tokens per Sec: 11013, Lr: 0.000300\n", "2020-04-10 04:06:38,620 Epoch 27 Step: 126800 Batch Loss: 1.935701 Tokens per Sec: 10881, Lr: 0.000300\n", "2020-04-10 04:06:56,677 Epoch 27 Step: 126900 Batch Loss: 1.467492 Tokens per Sec: 11013, Lr: 0.000300\n", "2020-04-10 04:07:15,058 Epoch 27 Step: 127000 Batch Loss: 1.852522 Tokens per Sec: 11033, Lr: 0.000300\n", "2020-04-10 04:07:43,362 Example #0\n", "2020-04-10 04:07:43,362 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 04:07:43,363 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 04:07:43,363 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 04:07:43,363 Example #1\n", "2020-04-10 04:07:43,363 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 04:07:43,363 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 04:07:43,363 \tHypothesis: If We willed , We could make it blazing , so why do you not give thanks ?\n", "2020-04-10 04:07:43,364 Example #2\n", "2020-04-10 04:07:43,364 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 04:07:43,364 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 04:07:43,364 \tHypothesis: Those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 04:07:43,364 Example #3\n", "2020-04-10 04:07:43,365 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 04:07:43,365 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 04:07:43,365 \tHypothesis: He sends the ark in peace , and whenever some of his people passed away before him , they mock him .\n", "2020-04-10 04:07:43,365 Validation result (greedy) at epoch 27, step 127000: bleu: 24.16, loss: 47805.6875, ppl: 5.8616, duration: 28.3068s\n", "2020-04-10 04:08:01,559 Epoch 27 Step: 127100 Batch Loss: 1.821951 Tokens per Sec: 10953, Lr: 0.000300\n", "2020-04-10 04:08:19,887 Epoch 27 Step: 127200 Batch Loss: 1.623057 Tokens per Sec: 10981, Lr: 0.000300\n", "2020-04-10 04:08:28,197 Epoch 27: total training loss 8140.89\n", "2020-04-10 04:08:28,198 EPOCH 28\n", "2020-04-10 04:08:38,552 Epoch 28 Step: 127300 Batch Loss: 2.145885 Tokens per Sec: 10518, Lr: 0.000300\n", "2020-04-10 04:08:56,982 Epoch 28 Step: 127400 Batch Loss: 2.014168 Tokens per Sec: 10925, Lr: 0.000300\n", "2020-04-10 04:09:15,074 Epoch 28 Step: 127500 Batch Loss: 1.797636 Tokens per Sec: 10914, Lr: 0.000300\n", "2020-04-10 04:09:33,632 Epoch 28 Step: 127600 Batch Loss: 1.913890 Tokens per Sec: 11178, Lr: 0.000300\n", "2020-04-10 04:09:51,643 Epoch 28 Step: 127700 Batch Loss: 1.616105 Tokens per Sec: 10851, Lr: 0.000300\n", "2020-04-10 04:10:09,805 Epoch 28 Step: 127800 Batch Loss: 1.582812 Tokens per Sec: 10795, Lr: 0.000300\n", "2020-04-10 04:10:28,045 Epoch 28 Step: 127900 Batch Loss: 1.980062 Tokens per Sec: 10780, Lr: 0.000300\n", "2020-04-10 04:10:46,414 Epoch 28 Step: 128000 Batch Loss: 1.835347 Tokens per Sec: 11082, Lr: 0.000300\n", "2020-04-10 04:11:14,385 Example #0\n", "2020-04-10 04:11:14,386 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 04:11:14,386 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 04:11:14,386 \tHypothesis: See how We explain the signs that haply they may understand .\n", "2020-04-10 04:11:14,387 Example #1\n", "2020-04-10 04:11:14,387 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 04:11:14,387 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 04:11:14,387 \tHypothesis: Had We willed , We could have made it black , so why do you not give thanks ?\n", "2020-04-10 04:11:14,388 Example #2\n", "2020-04-10 04:11:14,388 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 04:11:14,388 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 04:11:14,388 \tHypothesis: Those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 04:11:14,388 Example #3\n", "2020-04-10 04:11:14,389 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 04:11:14,389 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 04:11:14,389 \tHypothesis: He causes the ship to be safe , and whenever some of his people passed away before him , they mocked at him .\n", "2020-04-10 04:11:14,389 Validation result (greedy) at epoch 28, step 128000: bleu: 24.31, loss: 47892.7383, ppl: 5.8805, duration: 27.9751s\n", "2020-04-10 04:11:32,714 Epoch 28 Step: 128100 Batch Loss: 2.101567 Tokens per Sec: 10809, Lr: 0.000300\n", "2020-04-10 04:11:51,037 Epoch 28 Step: 128200 Batch Loss: 1.942591 Tokens per Sec: 10968, Lr: 0.000300\n", "2020-04-10 04:12:09,156 Epoch 28 Step: 128300 Batch Loss: 1.781545 Tokens per Sec: 10838, Lr: 0.000300\n", "2020-04-10 04:12:27,521 Epoch 28 Step: 128400 Batch Loss: 1.766939 Tokens per Sec: 11116, Lr: 0.000300\n", "2020-04-10 04:12:45,832 Epoch 28 Step: 128500 Batch Loss: 1.909710 Tokens per Sec: 11046, Lr: 0.000300\n", "2020-04-10 04:13:03,864 Epoch 28 Step: 128600 Batch Loss: 1.706258 Tokens per Sec: 10832, Lr: 0.000300\n", "2020-04-10 04:13:22,214 Epoch 28 Step: 128700 Batch Loss: 1.748190 Tokens per Sec: 11051, Lr: 0.000300\n", "2020-04-10 04:13:40,417 Epoch 28 Step: 128800 Batch Loss: 1.969501 Tokens per Sec: 10915, Lr: 0.000300\n", "2020-04-10 04:13:58,833 Epoch 28 Step: 128900 Batch Loss: 1.814145 Tokens per Sec: 11114, Lr: 0.000300\n", "2020-04-10 04:14:17,212 Epoch 28 Step: 129000 Batch Loss: 1.863478 Tokens per Sec: 11081, Lr: 0.000300\n", "2020-04-10 04:14:47,541 Example #0\n", "2020-04-10 04:14:47,541 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 04:14:47,541 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 04:14:47,542 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 04:14:47,542 Example #1\n", "2020-04-10 04:14:47,542 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 04:14:47,542 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 04:14:47,542 \tHypothesis: If We willed , We could make it black , so why do you not give thanks ?\n", "2020-04-10 04:14:47,542 Example #2\n", "2020-04-10 04:14:47,543 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 04:14:47,543 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 04:14:47,543 \tHypothesis: And those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 04:14:47,543 Example #3\n", "2020-04-10 04:14:47,543 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 04:14:47,544 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 04:14:47,544 \tHypothesis: And He sends the ark in peace , and whenever some of his people passed away before him , they mocked him .\n", "2020-04-10 04:14:47,544 Validation result (greedy) at epoch 28, step 129000: bleu: 24.36, loss: 47738.5039, ppl: 5.8470, duration: 30.3312s\n", "2020-04-10 04:15:05,714 Epoch 28 Step: 129100 Batch Loss: 1.815327 Tokens per Sec: 11027, Lr: 0.000300\n", "2020-04-10 04:15:24,197 Epoch 28 Step: 129200 Batch Loss: 1.552364 Tokens per Sec: 11236, Lr: 0.000300\n", "2020-04-10 04:15:42,472 Epoch 28 Step: 129300 Batch Loss: 1.505212 Tokens per Sec: 10983, Lr: 0.000300\n", "2020-04-10 04:16:00,788 Epoch 28 Step: 129400 Batch Loss: 1.939081 Tokens per Sec: 11161, Lr: 0.000300\n", "2020-04-10 04:16:18,971 Epoch 28 Step: 129500 Batch Loss: 1.850096 Tokens per Sec: 10970, Lr: 0.000300\n", "2020-04-10 04:16:37,251 Epoch 28 Step: 129600 Batch Loss: 1.757647 Tokens per Sec: 11021, Lr: 0.000300\n", "2020-04-10 04:16:55,427 Epoch 28 Step: 129700 Batch Loss: 1.919276 Tokens per Sec: 11133, Lr: 0.000300\n", "2020-04-10 04:17:13,684 Epoch 28 Step: 129800 Batch Loss: 1.646084 Tokens per Sec: 10707, Lr: 0.000300\n", "2020-04-10 04:17:32,081 Epoch 28 Step: 129900 Batch Loss: 1.897277 Tokens per Sec: 11119, Lr: 0.000300\n", "2020-04-10 04:17:50,320 Epoch 28 Step: 130000 Batch Loss: 1.857492 Tokens per Sec: 11273, Lr: 0.000300\n", "2020-04-10 04:18:18,512 Example #0\n", "2020-04-10 04:18:18,512 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 04:18:18,513 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 04:18:18,513 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 04:18:18,513 Example #1\n", "2020-04-10 04:18:18,513 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 04:18:18,513 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 04:18:18,513 \tHypothesis: If We willed , We could have made it blazing , so why do you not give thanks ?\n", "2020-04-10 04:18:18,514 Example #2\n", "2020-04-10 04:18:18,514 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 04:18:18,514 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 04:18:18,514 \tHypothesis: Those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 04:18:18,514 Example #3\n", "2020-04-10 04:18:18,515 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 04:18:18,515 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 04:18:18,515 \tHypothesis: He sends the ark in peace , and whenever some of his people passed away before him , they mock at him .\n", "2020-04-10 04:18:18,515 Validation result (greedy) at epoch 28, step 130000: bleu: 24.63, loss: 47821.0781, ppl: 5.8649, duration: 28.1945s\n", "2020-04-10 04:18:37,038 Epoch 28 Step: 130100 Batch Loss: 1.892731 Tokens per Sec: 11008, Lr: 0.000300\n", "2020-04-10 04:18:54,912 Epoch 28 Step: 130200 Batch Loss: 1.813517 Tokens per Sec: 10880, Lr: 0.000300\n", "2020-04-10 04:19:13,338 Epoch 28 Step: 130300 Batch Loss: 1.881487 Tokens per Sec: 11037, Lr: 0.000300\n", "2020-04-10 04:19:31,899 Epoch 28 Step: 130400 Batch Loss: 1.788428 Tokens per Sec: 11172, Lr: 0.000300\n", "2020-04-10 04:19:50,118 Epoch 28 Step: 130500 Batch Loss: 1.648577 Tokens per Sec: 10930, Lr: 0.000300\n", "2020-04-10 04:20:08,260 Epoch 28 Step: 130600 Batch Loss: 1.599141 Tokens per Sec: 10819, Lr: 0.000300\n", "2020-04-10 04:20:26,701 Epoch 28 Step: 130700 Batch Loss: 2.053214 Tokens per Sec: 10972, Lr: 0.000300\n", "2020-04-10 04:20:45,069 Epoch 28 Step: 130800 Batch Loss: 1.888891 Tokens per Sec: 11145, Lr: 0.000300\n", "2020-04-10 04:21:03,413 Epoch 28 Step: 130900 Batch Loss: 1.534948 Tokens per Sec: 10811, Lr: 0.000300\n", "2020-04-10 04:21:21,631 Epoch 28 Step: 131000 Batch Loss: 1.617226 Tokens per Sec: 10918, Lr: 0.000300\n", "2020-04-10 04:21:51,093 Example #0\n", "2020-04-10 04:21:51,093 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 04:21:51,093 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 04:21:51,094 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 04:21:51,094 Example #1\n", "2020-04-10 04:21:51,094 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 04:21:51,094 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 04:21:51,094 \tHypothesis: If We willed , We would have made it rain , so why do you not give thanks ?\n", "2020-04-10 04:21:51,094 Example #2\n", "2020-04-10 04:21:51,095 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 04:21:51,095 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 04:21:51,095 \tHypothesis: Those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 04:21:51,095 Example #3\n", "2020-04-10 04:21:51,095 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 04:21:51,096 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 04:21:51,096 \tHypothesis: He carries the ship in security , and whenever some of his people passed away before him , they mocked him .\n", "2020-04-10 04:21:51,096 Validation result (greedy) at epoch 28, step 131000: bleu: 24.53, loss: 47849.6602, ppl: 5.8711, duration: 29.4643s\n", "2020-04-10 04:22:09,335 Epoch 28 Step: 131100 Batch Loss: 1.959168 Tokens per Sec: 10842, Lr: 0.000210\n", "2020-04-10 04:22:27,771 Epoch 28 Step: 131200 Batch Loss: 1.885548 Tokens per Sec: 11035, Lr: 0.000210\n", "2020-04-10 04:22:45,937 Epoch 28 Step: 131300 Batch Loss: 1.840385 Tokens per Sec: 10833, Lr: 0.000210\n", "2020-04-10 04:23:04,167 Epoch 28 Step: 131400 Batch Loss: 1.742564 Tokens per Sec: 11150, Lr: 0.000210\n", "2020-04-10 04:23:22,308 Epoch 28 Step: 131500 Batch Loss: 1.475247 Tokens per Sec: 11008, Lr: 0.000210\n", "2020-04-10 04:23:40,649 Epoch 28 Step: 131600 Batch Loss: 1.916044 Tokens per Sec: 10900, Lr: 0.000210\n", "2020-04-10 04:23:51,913 Epoch 28: total training loss 8092.50\n", "2020-04-10 04:23:51,913 EPOCH 29\n", "2020-04-10 04:23:59,363 Epoch 29 Step: 131700 Batch Loss: 1.900550 Tokens per Sec: 10281, Lr: 0.000210\n", "2020-04-10 04:24:17,581 Epoch 29 Step: 131800 Batch Loss: 1.697192 Tokens per Sec: 10873, Lr: 0.000210\n", "2020-04-10 04:24:36,075 Epoch 29 Step: 131900 Batch Loss: 1.656174 Tokens per Sec: 11335, Lr: 0.000210\n", "2020-04-10 04:24:54,573 Epoch 29 Step: 132000 Batch Loss: 1.772596 Tokens per Sec: 11114, Lr: 0.000210\n", "2020-04-10 04:25:22,164 Hooray! New best validation result [ppl]!\n", "2020-04-10 04:25:22,165 Saving new checkpoint.\n", "2020-04-10 04:25:23,767 Example #0\n", "2020-04-10 04:25:23,767 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 04:25:23,767 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 04:25:23,768 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 04:25:23,768 Example #1\n", "2020-04-10 04:25:23,768 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 04:25:23,768 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 04:25:23,768 \tHypothesis: If We willed , We could have made it blazing , so why do you not give thanks ?\n", "2020-04-10 04:25:23,769 Example #2\n", "2020-04-10 04:25:23,769 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 04:25:23,769 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 04:25:23,769 \tHypothesis: Those who disbelieve in Our signs are the companions of the Left .\n", "2020-04-10 04:25:23,769 Example #3\n", "2020-04-10 04:25:23,770 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 04:25:23,770 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 04:25:23,770 \tHypothesis: He sends the ship in security , and whenever some of his people passed away before him , they mocked him .\n", "2020-04-10 04:25:23,770 Validation result (greedy) at epoch 29, step 132000: bleu: 24.51, loss: 47616.3242, ppl: 5.8207, duration: 29.1964s\n", "2020-04-10 04:25:42,079 Epoch 29 Step: 132100 Batch Loss: 1.943308 Tokens per Sec: 10715, Lr: 0.000210\n", "2020-04-10 04:26:00,326 Epoch 29 Step: 132200 Batch Loss: 1.935023 Tokens per Sec: 10884, Lr: 0.000210\n", "2020-04-10 04:26:18,859 Epoch 29 Step: 132300 Batch Loss: 1.693413 Tokens per Sec: 10848, Lr: 0.000210\n", "2020-04-10 04:26:37,438 Epoch 29 Step: 132400 Batch Loss: 1.864101 Tokens per Sec: 11093, Lr: 0.000210\n", "2020-04-10 04:26:55,919 Epoch 29 Step: 132500 Batch Loss: 1.800213 Tokens per Sec: 10754, Lr: 0.000210\n", "2020-04-10 04:27:14,358 Epoch 29 Step: 132600 Batch Loss: 1.806238 Tokens per Sec: 10988, Lr: 0.000210\n", "2020-04-10 04:27:32,898 Epoch 29 Step: 132700 Batch Loss: 1.679732 Tokens per Sec: 10967, Lr: 0.000210\n", "2020-04-10 04:27:50,972 Epoch 29 Step: 132800 Batch Loss: 1.844785 Tokens per Sec: 10774, Lr: 0.000210\n", "2020-04-10 04:28:09,449 Epoch 29 Step: 132900 Batch Loss: 1.867894 Tokens per Sec: 11014, Lr: 0.000210\n", "2020-04-10 04:28:27,678 Epoch 29 Step: 133000 Batch Loss: 1.478335 Tokens per Sec: 10787, Lr: 0.000210\n", "2020-04-10 04:28:55,854 Hooray! New best validation result [ppl]!\n", "2020-04-10 04:28:55,854 Saving new checkpoint.\n", "2020-04-10 04:28:57,173 Example #0\n", "2020-04-10 04:28:57,174 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 04:28:57,174 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 04:28:57,174 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 04:28:57,175 Example #1\n", "2020-04-10 04:28:57,175 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 04:28:57,175 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 04:28:57,176 \tHypothesis: If We willed , We would have made it blazing , so why do you not give thanks ?\n", "2020-04-10 04:28:57,176 Example #2\n", "2020-04-10 04:28:57,176 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 04:28:57,176 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 04:28:57,177 \tHypothesis: Those who disbelieve in Our signs are the people of the Left .\n", "2020-04-10 04:28:57,177 Example #3\n", "2020-04-10 04:28:57,177 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 04:28:57,177 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 04:28:57,177 \tHypothesis: And He puts the ships in security , and whenever there came to his people pass away before him , they mock him .\n", "2020-04-10 04:28:57,178 Validation result (greedy) at epoch 29, step 133000: bleu: 24.31, loss: 47466.9727, ppl: 5.7886, duration: 29.4987s\n", "2020-04-10 04:29:15,927 Epoch 29 Step: 133100 Batch Loss: 1.824124 Tokens per Sec: 10617, Lr: 0.000210\n", "2020-04-10 04:29:34,545 Epoch 29 Step: 133200 Batch Loss: 1.751584 Tokens per Sec: 10907, Lr: 0.000210\n", "2020-04-10 04:29:52,629 Epoch 29 Step: 133300 Batch Loss: 1.678895 Tokens per Sec: 10638, Lr: 0.000210\n", "2020-04-10 04:30:11,054 Epoch 29 Step: 133400 Batch Loss: 1.651203 Tokens per Sec: 10784, Lr: 0.000210\n", "2020-04-10 04:30:29,536 Epoch 29 Step: 133500 Batch Loss: 1.848131 Tokens per Sec: 10939, Lr: 0.000210\n", "2020-04-10 04:30:47,860 Epoch 29 Step: 133600 Batch Loss: 1.770941 Tokens per Sec: 10992, Lr: 0.000210\n", "2020-04-10 04:31:06,119 Epoch 29 Step: 133700 Batch Loss: 1.827007 Tokens per Sec: 10843, Lr: 0.000210\n", "2020-04-10 04:31:24,738 Epoch 29 Step: 133800 Batch Loss: 1.960996 Tokens per Sec: 11248, Lr: 0.000210\n", "2020-04-10 04:31:43,153 Epoch 29 Step: 133900 Batch Loss: 1.665458 Tokens per Sec: 10965, Lr: 0.000210\n", "2020-04-10 04:32:01,637 Epoch 29 Step: 134000 Batch Loss: 1.770740 Tokens per Sec: 10941, Lr: 0.000210\n", "2020-04-10 04:32:27,828 Hooray! New best validation result [ppl]!\n", "2020-04-10 04:32:27,828 Saving new checkpoint.\n", "2020-04-10 04:32:29,164 Example #0\n", "2020-04-10 04:32:29,165 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 04:32:29,165 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 04:32:29,165 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 04:32:29,165 Example #1\n", "2020-04-10 04:32:29,166 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 04:32:29,166 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 04:32:29,166 \tHypothesis: If We willed , We would have made it blazing , so why do you not give thanks ?\n", "2020-04-10 04:32:29,166 Example #2\n", "2020-04-10 04:32:29,166 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 04:32:29,166 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 04:32:29,167 \tHypothesis: And those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 04:32:29,167 Example #3\n", "2020-04-10 04:32:29,167 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 04:32:29,167 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 04:32:29,168 \tHypothesis: And He carries the ark in security , and whenever there came to him some of his people passed away before him , they mocked him .\n", "2020-04-10 04:32:29,168 Validation result (greedy) at epoch 29, step 134000: bleu: 24.79, loss: 47232.7969, ppl: 5.7387, duration: 27.5301s\n", "2020-04-10 04:32:47,865 Epoch 29 Step: 134100 Batch Loss: 1.649474 Tokens per Sec: 10851, Lr: 0.000210\n", "2020-04-10 04:33:06,592 Epoch 29 Step: 134200 Batch Loss: 1.796765 Tokens per Sec: 11033, Lr: 0.000210\n", "2020-04-10 04:33:25,083 Epoch 29 Step: 134300 Batch Loss: 1.654981 Tokens per Sec: 10774, Lr: 0.000210\n", "2020-04-10 04:33:43,884 Epoch 29 Step: 134400 Batch Loss: 2.062144 Tokens per Sec: 11147, Lr: 0.000210\n", "2020-04-10 04:34:02,298 Epoch 29 Step: 134500 Batch Loss: 1.758501 Tokens per Sec: 10792, Lr: 0.000210\n", "2020-04-10 04:34:20,805 Epoch 29 Step: 134600 Batch Loss: 1.901699 Tokens per Sec: 11018, Lr: 0.000210\n", "2020-04-10 04:34:39,106 Epoch 29 Step: 134700 Batch Loss: 1.935603 Tokens per Sec: 10824, Lr: 0.000210\n", "2020-04-10 04:34:57,503 Epoch 29 Step: 134800 Batch Loss: 1.825239 Tokens per Sec: 11123, Lr: 0.000210\n", "2020-04-10 04:35:15,437 Epoch 29 Step: 134900 Batch Loss: 1.466220 Tokens per Sec: 10641, Lr: 0.000210\n", "2020-04-10 04:35:33,789 Epoch 29 Step: 135000 Batch Loss: 1.946153 Tokens per Sec: 10757, Lr: 0.000210\n", "2020-04-10 04:36:00,563 Hooray! New best validation result [ppl]!\n", "2020-04-10 04:36:00,563 Saving new checkpoint.\n", "2020-04-10 04:36:02,182 Example #0\n", "2020-04-10 04:36:02,182 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 04:36:02,182 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 04:36:02,182 \tHypothesis: See how We explain the signs that they may understand .\n", "2020-04-10 04:36:02,183 Example #1\n", "2020-04-10 04:36:02,183 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 04:36:02,183 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 04:36:02,184 \tHypothesis: If We willed , We could have made it rain , so why do you not give thanks ?\n", "2020-04-10 04:36:02,184 Example #2\n", "2020-04-10 04:36:02,185 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 04:36:02,185 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 04:36:02,185 \tHypothesis: And those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 04:36:02,185 Example #3\n", "2020-04-10 04:36:02,186 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 04:36:02,186 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 04:36:02,186 \tHypothesis: And He carries the ark in security , and whenever some of his people passed away before him , they mock at him .\n", "2020-04-10 04:36:02,186 Validation result (greedy) at epoch 29, step 135000: bleu: 24.64, loss: 47172.3555, ppl: 5.7259, duration: 28.3973s\n", "2020-04-10 04:36:20,976 Epoch 29 Step: 135100 Batch Loss: 1.818374 Tokens per Sec: 11112, Lr: 0.000210\n", "2020-04-10 04:36:39,413 Epoch 29 Step: 135200 Batch Loss: 1.683642 Tokens per Sec: 10800, Lr: 0.000210\n", "2020-04-10 04:36:57,609 Epoch 29 Step: 135300 Batch Loss: 2.030717 Tokens per Sec: 10945, Lr: 0.000210\n", "2020-04-10 04:37:16,003 Epoch 29 Step: 135400 Batch Loss: 1.726725 Tokens per Sec: 10876, Lr: 0.000210\n", "2020-04-10 04:37:34,497 Epoch 29 Step: 135500 Batch Loss: 2.074475 Tokens per Sec: 11107, Lr: 0.000210\n", "2020-04-10 04:37:52,626 Epoch 29 Step: 135600 Batch Loss: 1.839821 Tokens per Sec: 10617, Lr: 0.000210\n", "2020-04-10 04:38:10,766 Epoch 29 Step: 135700 Batch Loss: 1.842852 Tokens per Sec: 10826, Lr: 0.000210\n", "2020-04-10 04:38:29,132 Epoch 29 Step: 135800 Batch Loss: 1.900749 Tokens per Sec: 10879, Lr: 0.000210\n", "2020-04-10 04:38:47,545 Epoch 29 Step: 135900 Batch Loss: 1.642637 Tokens per Sec: 10965, Lr: 0.000210\n", "2020-04-10 04:39:05,786 Epoch 29 Step: 136000 Batch Loss: 1.600109 Tokens per Sec: 10921, Lr: 0.000210\n", "2020-04-10 04:39:34,144 Example #0\n", "2020-04-10 04:39:34,145 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 04:39:34,146 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 04:39:34,146 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 04:39:34,146 Example #1\n", "2020-04-10 04:39:34,146 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 04:39:34,146 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 04:39:34,147 \tHypothesis: If We willed , We could have made it blazing , so why do you not give thanks ?\n", "2020-04-10 04:39:34,147 Example #2\n", "2020-04-10 04:39:34,147 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 04:39:34,147 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 04:39:34,147 \tHypothesis: And those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 04:39:34,147 Example #3\n", "2020-04-10 04:39:34,148 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 04:39:34,148 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 04:39:34,148 \tHypothesis: And He puts the ark in security , and whenever some of his people passed away before him , they mock at him .\n", "2020-04-10 04:39:34,148 Validation result (greedy) at epoch 29, step 136000: bleu: 24.68, loss: 47295.2344, ppl: 5.7519, duration: 28.3617s\n", "2020-04-10 04:39:48,046 Epoch 29: total training loss 7973.56\n", "2020-04-10 04:39:48,046 EPOCH 30\n", "2020-04-10 04:39:52,943 Epoch 30 Step: 136100 Batch Loss: 1.842678 Tokens per Sec: 9498, Lr: 0.000210\n", "2020-04-10 04:40:11,346 Epoch 30 Step: 136200 Batch Loss: 1.871600 Tokens per Sec: 11043, Lr: 0.000210\n", "2020-04-10 04:40:29,629 Epoch 30 Step: 136300 Batch Loss: 1.751481 Tokens per Sec: 10990, Lr: 0.000210\n", "2020-04-10 04:40:47,967 Epoch 30 Step: 136400 Batch Loss: 1.840236 Tokens per Sec: 11126, Lr: 0.000210\n", "2020-04-10 04:41:06,262 Epoch 30 Step: 136500 Batch Loss: 1.696874 Tokens per Sec: 10897, Lr: 0.000210\n", "2020-04-10 04:41:24,532 Epoch 30 Step: 136600 Batch Loss: 1.592871 Tokens per Sec: 10848, Lr: 0.000210\n", "2020-04-10 04:41:42,893 Epoch 30 Step: 136700 Batch Loss: 1.955695 Tokens per Sec: 10788, Lr: 0.000210\n", "2020-04-10 04:42:01,222 Epoch 30 Step: 136800 Batch Loss: 1.720689 Tokens per Sec: 11091, Lr: 0.000210\n", "2020-04-10 04:42:19,564 Epoch 30 Step: 136900 Batch Loss: 1.786544 Tokens per Sec: 10842, Lr: 0.000210\n", "2020-04-10 04:42:37,917 Epoch 30 Step: 137000 Batch Loss: 1.736498 Tokens per Sec: 10900, Lr: 0.000210\n", "2020-04-10 04:43:06,911 Example #0\n", "2020-04-10 04:43:06,911 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 04:43:06,912 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 04:43:06,912 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 04:43:06,912 Example #1\n", "2020-04-10 04:43:06,912 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 04:43:06,912 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 04:43:06,912 \tHypothesis: If We willed , We would have made it black , so why do you not give thanks ?\n", "2020-04-10 04:43:06,913 Example #2\n", "2020-04-10 04:43:06,913 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 04:43:06,913 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 04:43:06,913 \tHypothesis: And those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 04:43:06,913 Example #3\n", "2020-04-10 04:43:06,914 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 04:43:06,914 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 04:43:06,914 \tHypothesis: And He puts the ship in security , and whenever some of his people passed away before him , they mock him .\n", "2020-04-10 04:43:06,914 Validation result (greedy) at epoch 30, step 137000: bleu: 24.70, loss: 47199.4805, ppl: 5.7316, duration: 28.9966s\n", "2020-04-10 04:43:25,581 Epoch 30 Step: 137100 Batch Loss: 1.629669 Tokens per Sec: 11125, Lr: 0.000210\n", "2020-04-10 04:43:44,025 Epoch 30 Step: 137200 Batch Loss: 1.944280 Tokens per Sec: 10884, Lr: 0.000210\n", "2020-04-10 04:44:02,505 Epoch 30 Step: 137300 Batch Loss: 1.910910 Tokens per Sec: 10919, Lr: 0.000210\n", "2020-04-10 04:44:20,852 Epoch 30 Step: 137400 Batch Loss: 1.837475 Tokens per Sec: 10922, Lr: 0.000210\n", "2020-04-10 04:44:39,155 Epoch 30 Step: 137500 Batch Loss: 1.599319 Tokens per Sec: 10868, Lr: 0.000210\n", "2020-04-10 04:44:57,530 Epoch 30 Step: 137600 Batch Loss: 1.862972 Tokens per Sec: 10922, Lr: 0.000210\n", "2020-04-10 04:45:15,852 Epoch 30 Step: 137700 Batch Loss: 1.688215 Tokens per Sec: 11079, Lr: 0.000210\n", "2020-04-10 04:45:33,961 Epoch 30 Step: 137800 Batch Loss: 1.760016 Tokens per Sec: 10765, Lr: 0.000210\n", "2020-04-10 04:45:52,379 Epoch 30 Step: 137900 Batch Loss: 1.766687 Tokens per Sec: 10827, Lr: 0.000210\n", "2020-04-10 04:46:10,702 Epoch 30 Step: 138000 Batch Loss: 1.845708 Tokens per Sec: 10975, Lr: 0.000210\n", "2020-04-10 04:46:41,528 Hooray! New best validation result [ppl]!\n", "2020-04-10 04:46:41,528 Saving new checkpoint.\n", "2020-04-10 04:46:42,791 Example #0\n", "2020-04-10 04:46:42,792 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 04:46:42,792 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 04:46:42,792 \tHypothesis: See how We explain the signs , so that they may understand .\n", "2020-04-10 04:46:42,793 Example #1\n", "2020-04-10 04:46:42,793 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 04:46:42,793 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 04:46:42,793 \tHypothesis: If We willed , We could have made it black , so why do you not give thanks ?\n", "2020-04-10 04:46:42,794 Example #2\n", "2020-04-10 04:46:42,794 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 04:46:42,794 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 04:46:42,795 \tHypothesis: Those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 04:46:42,795 Example #3\n", "2020-04-10 04:46:42,795 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 04:46:42,795 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 04:46:42,795 \tHypothesis: He puts the ark in security , and whenever some of his people passed away before him , they mock at him .\n", "2020-04-10 04:46:42,796 Validation result (greedy) at epoch 30, step 138000: bleu: 24.55, loss: 46959.8359, ppl: 5.6810, duration: 32.0931s\n", "2020-04-10 04:47:01,529 Epoch 30 Step: 138100 Batch Loss: 1.845534 Tokens per Sec: 11037, Lr: 0.000210\n", "2020-04-10 04:47:19,894 Epoch 30 Step: 138200 Batch Loss: 1.661753 Tokens per Sec: 11028, Lr: 0.000210\n", "2020-04-10 04:47:38,238 Epoch 30 Step: 138300 Batch Loss: 1.669622 Tokens per Sec: 10814, Lr: 0.000210\n", "2020-04-10 04:47:56,611 Epoch 30 Step: 138400 Batch Loss: 1.871481 Tokens per Sec: 10663, Lr: 0.000210\n", "2020-04-10 04:48:15,062 Epoch 30 Step: 138500 Batch Loss: 1.851363 Tokens per Sec: 10958, Lr: 0.000210\n", "2020-04-10 04:48:33,265 Epoch 30 Step: 138600 Batch Loss: 1.668068 Tokens per Sec: 10808, Lr: 0.000210\n", "2020-04-10 04:48:51,685 Epoch 30 Step: 138700 Batch Loss: 1.784269 Tokens per Sec: 11061, Lr: 0.000210\n", "2020-04-10 04:49:09,870 Epoch 30 Step: 138800 Batch Loss: 1.740863 Tokens per Sec: 10724, Lr: 0.000210\n", "2020-04-10 04:49:28,321 Epoch 30 Step: 138900 Batch Loss: 1.469351 Tokens per Sec: 10916, Lr: 0.000210\n", "2020-04-10 04:49:46,601 Epoch 30 Step: 139000 Batch Loss: 1.709289 Tokens per Sec: 10771, Lr: 0.000210\n", "2020-04-10 04:50:15,026 Example #0\n", "2020-04-10 04:50:15,027 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 04:50:15,027 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 04:50:15,027 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 04:50:15,027 Example #1\n", "2020-04-10 04:50:15,028 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 04:50:15,028 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 04:50:15,028 \tHypothesis: If We willed , We could have made it black , so why do you not give thanks ?\n", "2020-04-10 04:50:15,028 Example #2\n", "2020-04-10 04:50:15,028 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 04:50:15,029 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 04:50:15,029 \tHypothesis: Those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 04:50:15,029 Example #3\n", "2020-04-10 04:50:15,029 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 04:50:15,029 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 04:50:15,030 \tHypothesis: He puts the ark in peace , and whenever some of his people passed away before him , they mocked at him .\n", "2020-04-10 04:50:15,030 Validation result (greedy) at epoch 30, step 139000: bleu: 24.36, loss: 47004.9727, ppl: 5.6905, duration: 28.4281s\n", "2020-04-10 04:50:33,464 Epoch 30 Step: 139100 Batch Loss: 1.781490 Tokens per Sec: 10860, Lr: 0.000210\n", "2020-04-10 04:50:51,616 Epoch 30 Step: 139200 Batch Loss: 1.855653 Tokens per Sec: 10621, Lr: 0.000210\n", "2020-04-10 04:51:09,910 Epoch 30 Step: 139300 Batch Loss: 1.891185 Tokens per Sec: 11028, Lr: 0.000210\n", "2020-04-10 04:51:28,253 Epoch 30 Step: 139400 Batch Loss: 1.802861 Tokens per Sec: 10885, Lr: 0.000210\n", "2020-04-10 04:51:46,701 Epoch 30 Step: 139500 Batch Loss: 2.048597 Tokens per Sec: 10991, Lr: 0.000210\n", "2020-04-10 04:52:05,201 Epoch 30 Step: 139600 Batch Loss: 1.767022 Tokens per Sec: 11068, Lr: 0.000210\n", "2020-04-10 04:52:23,426 Epoch 30 Step: 139700 Batch Loss: 1.814306 Tokens per Sec: 10811, Lr: 0.000210\n", "2020-04-10 04:52:41,866 Epoch 30 Step: 139800 Batch Loss: 1.895372 Tokens per Sec: 10771, Lr: 0.000210\n", "2020-04-10 04:53:00,325 Epoch 30 Step: 139900 Batch Loss: 1.889184 Tokens per Sec: 11036, Lr: 0.000210\n", "2020-04-10 04:53:18,999 Epoch 30 Step: 140000 Batch Loss: 1.749475 Tokens per Sec: 11198, Lr: 0.000210\n", "2020-04-10 04:53:48,255 Example #0\n", "2020-04-10 04:53:48,256 \tSource: \" Ka dũba yadda Muke sarrafa ãyõyi , tsammãninsu sunã fahimta ! \"\n", "2020-04-10 04:53:48,256 \tReference: Note how We explain the revelations , so that they may understand . ”\n", "2020-04-10 04:53:48,257 \tHypothesis: See how We explain the signs so that they may understand .\n", "2020-04-10 04:53:48,257 Example #1\n", "2020-04-10 04:53:48,257 \tSource: Dã Mun so , dã Mun mayar da shi ruwan zartsi . To don me bã ku gõdẽwa ?\n", "2020-04-10 04:53:48,258 \tReference: If We pleased , We would have made it salty ; why do you not then give thanks ?\n", "2020-04-10 04:53:48,258 \tHypothesis: If We willed , We could have made it blazing , so why do you not give thanks ?\n", "2020-04-10 04:53:48,258 Example #2\n", "2020-04-10 04:53:48,258 \tSource: Kuma waɗanda suka kãfirta da ãyõyinMu , sũ ne ma 'abũta shu 'umci\n", "2020-04-10 04:53:48,258 \tReference: and [ as for ] those who are bent on denying the truth of Our revelations , they are the people of the left hand ,\n", "2020-04-10 04:53:48,258 \tHypothesis: And those who disbelieve in Our signs are the people of the Left Hand .\n", "2020-04-10 04:53:48,259 Example #3\n", "2020-04-10 04:53:48,259 \tSource: Kuma Yanã sassaƙa jirgin cikin natsuwa , kuma a kõ yaushe waɗansu shugabanni daga mutãnensa suka shũɗe a gabansa , sai su yi izgili gare shi .\n", "2020-04-10 04:53:48,259 \tReference: And he was building the ship , and every time that chieftains of his people passed him , they made mock of him .\n", "2020-04-10 04:53:48,259 \tHypothesis: He puts the ark in peace , and whenever some of his people passed away before him , they mocked at him .\n", "2020-04-10 04:53:48,259 Validation result (greedy) at epoch 30, step 140000: bleu: 24.57, loss: 47084.7500, ppl: 5.7073, duration: 29.2595s\n", "2020-04-10 04:54:06,433 Epoch 30 Step: 140100 Batch Loss: 1.650761 Tokens per Sec: 10880, Lr: 0.000210\n", "2020-04-10 04:54:24,825 Epoch 30 Step: 140200 Batch Loss: 1.822501 Tokens per Sec: 11291, Lr: 0.000210\n", "2020-04-10 04:54:43,249 Epoch 30 Step: 140300 Batch Loss: 1.721434 Tokens per Sec: 10954, Lr: 0.000210\n", "2020-04-10 04:55:01,786 Epoch 30 Step: 140400 Batch Loss: 1.827833 Tokens per Sec: 11220, Lr: 0.000210\n", "2020-04-10 04:55:18,422 Epoch 30: total training loss 7940.81\n", "2020-04-10 04:55:18,422 Training ended after 30 epochs.\n", "2020-04-10 04:55:18,422 Best validation result (greedy) at step 138000: 5.68 ppl.\n", "2020-04-10 04:55:55,855 dev bleu: 24.95 [Beam search decoding with beam size = 5 and alpha = 1.0]\n", "2020-04-10 04:55:55,860 Translations saved to: /content/drive/My Drive/masakhane/model-temp/00138000.hyps.dev\n", "2020-04-10 04:59:17,225 test bleu: 25.27 [Beam search decoding with beam size = 5 and alpha = 1.0]\n", "2020-04-10 04:59:17,234 Translations saved to: /content/drive/My Drive/masakhane/model-temp/00138000.hyps.test\n" ], "name": "stdout" } ] }, { "cell_type": "code", "metadata": { "id": "a5YT00kykz6o", "colab_type": "code", "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}${tgt}_transformer/\"" ], "execution_count": 0, "outputs": [] }, { "cell_type": "code", "metadata": { "id": "L5MCFDe2dhx6", "colab_type": "code", "colab": {} }, "source": [ "\n", "\n", "# # Copy the created models from the notebook storage to google drive for persistant storage \n", "# !cp -r joeynmt/models/${src}${tgt}_transformer/* \"$gdrive_path/models/${src}${tgt}_transformer/\"\n" ], "execution_count": 0, "outputs": [] }, { "cell_type": "code", "metadata": { "id": "r-AWfR-CfZiS", "colab_type": "code", "outputId": "35f71b21-750e-4814-fb52-7b37a4d2b4a6", "colab": { "base_uri": "https://localhost:8080/", "height": 1000 } }, "source": [ "# Output our validation accuracy\n", "! cat \"$gdrive_path/models/${src}${tgt}_transformer/validations.txt\"" ], "execution_count": 70, "outputs": [ { "output_type": "stream", "text": [ "Steps: 9000\tLoss: 73550.10156\tPPL: 15.19175\tbleu: 11.57319\tLR: 0.00030000\t*\n", "Steps: 10000\tLoss: 71540.42188\tPPL: 14.10333\tbleu: 12.26218\tLR: 0.00030000\t*\n", "Steps: 11000\tLoss: 70733.12500\tPPL: 13.68838\tbleu: 12.87104\tLR: 0.00030000\t*\n", "Steps: 12000\tLoss: 69586.67969\tPPL: 13.12001\tbleu: 13.12258\tLR: 0.00030000\t*\n", "Steps: 13000\tLoss: 67706.30469\tPPL: 12.23841\tbleu: 14.33447\tLR: 0.00030000\t*\n", "Steps: 14000\tLoss: 66552.30469\tPPL: 11.72697\tbleu: 14.99448\tLR: 0.00030000\t*\n", "Steps: 15000\tLoss: 65527.41016\tPPL: 11.29069\tbleu: 15.48382\tLR: 0.00030000\t*\n", "Steps: 16000\tLoss: 64649.46094\tPPL: 10.92989\tbleu: 15.24217\tLR: 0.00030000\t*\n", "Steps: 17000\tLoss: 63707.59766\tPPL: 10.55564\tbleu: 15.90648\tLR: 0.00030000\t*\n", "Steps: 18000\tLoss: 62847.35156\tPPL: 10.22503\tbleu: 16.88917\tLR: 0.00030000\t*\n", "Steps: 19000\tLoss: 62223.76562\tPPL: 9.99186\tbleu: 16.97319\tLR: 0.00030000\t*\n", "Steps: 20000\tLoss: 61801.85156\tPPL: 9.83712\tbleu: 17.26285\tLR: 0.00030000\t*\n", "Steps: 21000\tLoss: 61139.49609\tPPL: 9.59902\tbleu: 17.67682\tLR: 0.00030000\t*\n", "Steps: 22000\tLoss: 60433.43750\tPPL: 9.35156\tbleu: 17.86495\tLR: 0.00030000\t*\n", "Steps: 23000\tLoss: 59807.67969\tPPL: 9.13758\tbleu: 18.40380\tLR: 0.00030000\t*\n", "Steps: 24000\tLoss: 59333.51172\tPPL: 8.97870\tbleu: 18.60992\tLR: 0.00030000\t*\n", "Steps: 25000\tLoss: 58789.05078\tPPL: 8.79967\tbleu: 18.62553\tLR: 0.00030000\t*\n", "Steps: 26000\tLoss: 58786.14844\tPPL: 8.79873\tbleu: 18.90870\tLR: 0.00030000\t*\n", "Steps: 27000\tLoss: 58117.37500\tPPL: 8.58372\tbleu: 19.12278\tLR: 0.00030000\t*\n", "Steps: 28000\tLoss: 57682.42969\tPPL: 8.44672\tbleu: 19.07431\tLR: 0.00030000\t*\n", "Steps: 29000\tLoss: 57607.88281\tPPL: 8.42346\tbleu: 19.27987\tLR: 0.00030000\t*\n", "Steps: 30000\tLoss: 57198.80859\tPPL: 8.29695\tbleu: 19.89630\tLR: 0.00030000\t*\n", "Steps: 31000\tLoss: 57024.42969\tPPL: 8.24360\tbleu: 19.61811\tLR: 0.00030000\t*\n", "Steps: 32000\tLoss: 56367.06250\tPPL: 8.04556\tbleu: 20.12995\tLR: 0.00030000\t*\n", "Steps: 33000\tLoss: 56370.41797\tPPL: 8.04656\tbleu: 19.51349\tLR: 0.00030000\t\n", "Steps: 34000\tLoss: 56096.91016\tPPL: 7.96556\tbleu: 20.56766\tLR: 0.00030000\t*\n", "Steps: 35000\tLoss: 55550.56250\tPPL: 7.80619\tbleu: 20.93779\tLR: 0.00030000\t*\n", "Steps: 36000\tLoss: 55508.80859\tPPL: 7.79414\tbleu: 20.96075\tLR: 0.00030000\t*\n", "Steps: 37000\tLoss: 55001.69141\tPPL: 7.64929\tbleu: 20.84088\tLR: 0.00030000\t*\n", "Steps: 38000\tLoss: 54860.39844\tPPL: 7.60942\tbleu: 20.68125\tLR: 0.00030000\t*\n", "Steps: 39000\tLoss: 54868.70312\tPPL: 7.61175\tbleu: 20.68774\tLR: 0.00030000\t\n", "Steps: 40000\tLoss: 54804.74609\tPPL: 7.59376\tbleu: 21.25522\tLR: 0.00030000\t*\n", "Steps: 41000\tLoss: 54244.32812\tPPL: 7.43796\tbleu: 21.21535\tLR: 0.00030000\t*\n", "Steps: 42000\tLoss: 54089.21094\tPPL: 7.39540\tbleu: 21.43270\tLR: 0.00030000\t*\n", "Steps: 43000\tLoss: 53769.90234\tPPL: 7.30856\tbleu: 21.51964\tLR: 0.00030000\t*\n", "Steps: 44000\tLoss: 53794.73047\tPPL: 7.31528\tbleu: 21.29646\tLR: 0.00030000\t\n", "Steps: 45000\tLoss: 53501.65234\tPPL: 7.23640\tbleu: 21.60419\tLR: 0.00030000\t*\n", "Steps: 46000\tLoss: 53400.79688\tPPL: 7.20945\tbleu: 21.43974\tLR: 0.00030000\t*\n", "Steps: 47000\tLoss: 53217.55078\tPPL: 7.16075\tbleu: 21.56809\tLR: 0.00030000\t*\n", "Steps: 48000\tLoss: 53295.61719\tPPL: 7.18146\tbleu: 21.66523\tLR: 0.00030000\t\n", "Steps: 49000\tLoss: 52955.17188\tPPL: 7.09158\tbleu: 22.08601\tLR: 0.00030000\t*\n", "Steps: 50000\tLoss: 52565.18750\tPPL: 6.99001\tbleu: 22.01732\tLR: 0.00030000\t*\n", "Steps: 51000\tLoss: 52727.73438\tPPL: 7.03217\tbleu: 22.08540\tLR: 0.00030000\t\n", "Steps: 52000\tLoss: 52320.33594\tPPL: 6.92699\tbleu: 22.08668\tLR: 0.00030000\t*\n", "Steps: 53000\tLoss: 52218.33594\tPPL: 6.90090\tbleu: 21.95706\tLR: 0.00030000\t*\n", "Steps: 54000\tLoss: 52261.88672\tPPL: 6.91202\tbleu: 21.81363\tLR: 0.00030000\t\n", "Steps: 55000\tLoss: 52147.53906\tPPL: 6.88285\tbleu: 22.27165\tLR: 0.00030000\t*\n", "Steps: 56000\tLoss: 52079.82031\tPPL: 6.86563\tbleu: 22.19836\tLR: 0.00030000\t*\n", "Steps: 57000\tLoss: 51867.77344\tPPL: 6.81198\tbleu: 22.21684\tLR: 0.00030000\t*\n", "Steps: 58000\tLoss: 51899.06250\tPPL: 6.81987\tbleu: 22.14502\tLR: 0.00030000\t\n", "Steps: 59000\tLoss: 51732.47266\tPPL: 6.77798\tbleu: 22.63271\tLR: 0.00030000\t*\n", "Steps: 60000\tLoss: 51570.83203\tPPL: 6.73757\tbleu: 22.11424\tLR: 0.00030000\t*\n", "Steps: 61000\tLoss: 51693.05859\tPPL: 6.76810\tbleu: 22.46565\tLR: 0.00030000\t\n", "Steps: 62000\tLoss: 51519.89453\tPPL: 6.72489\tbleu: 22.43105\tLR: 0.00030000\t*\n", "Steps: 63000\tLoss: 51195.34766\tPPL: 6.64463\tbleu: 22.70037\tLR: 0.00030000\t*\n", "Steps: 64000\tLoss: 51292.92969\tPPL: 6.66866\tbleu: 22.49982\tLR: 0.00030000\t\n", "Steps: 65000\tLoss: 51368.14453\tPPL: 6.68724\tbleu: 22.38450\tLR: 0.00030000\t\n", "Steps: 66000\tLoss: 51037.99609\tPPL: 6.60607\tbleu: 23.03994\tLR: 0.00030000\t*\n", "Steps: 67000\tLoss: 51285.60938\tPPL: 6.66685\tbleu: 22.18894\tLR: 0.00030000\t\n", "Steps: 68000\tLoss: 50936.96484\tPPL: 6.58142\tbleu: 22.96432\tLR: 0.00030000\t*\n", "Steps: 69000\tLoss: 50715.96875\tPPL: 6.52784\tbleu: 22.95148\tLR: 0.00030000\t*\n", "Steps: 70000\tLoss: 50636.30859\tPPL: 6.50863\tbleu: 22.91485\tLR: 0.00030000\t*\n", "Steps: 71000\tLoss: 50686.08203\tPPL: 6.52063\tbleu: 23.23181\tLR: 0.00030000\t\n", "Steps: 72000\tLoss: 50441.18750\tPPL: 6.46182\tbleu: 22.83605\tLR: 0.00030000\t*\n", "Steps: 73000\tLoss: 50537.30469\tPPL: 6.48484\tbleu: 23.01001\tLR: 0.00030000\t\n", "Steps: 74000\tLoss: 50347.60156\tPPL: 6.43949\tbleu: 23.24227\tLR: 0.00030000\t*\n", "Steps: 75000\tLoss: 50295.01172\tPPL: 6.42698\tbleu: 23.19973\tLR: 0.00030000\t*\n", "Steps: 76000\tLoss: 50363.28516\tPPL: 6.44323\tbleu: 23.28517\tLR: 0.00030000\t\n", "Steps: 77000\tLoss: 50225.05859\tPPL: 6.41037\tbleu: 22.80156\tLR: 0.00030000\t*\n", "Steps: 78000\tLoss: 50386.08984\tPPL: 6.44867\tbleu: 23.40086\tLR: 0.00030000\t\n", "Steps: 79000\tLoss: 50002.67969\tPPL: 6.35785\tbleu: 23.23236\tLR: 0.00030000\t*\n", "Steps: 80000\tLoss: 49990.48047\tPPL: 6.35498\tbleu: 23.38135\tLR: 0.00030000\t*\n", "Steps: 81000\tLoss: 49995.09766\tPPL: 6.35607\tbleu: 23.48620\tLR: 0.00030000\t\n", "Steps: 82000\tLoss: 49730.79297\tPPL: 6.29423\tbleu: 23.11307\tLR: 0.00030000\t*\n", "Steps: 83000\tLoss: 49861.58984\tPPL: 6.32475\tbleu: 23.36868\tLR: 0.00030000\t\n", "Steps: 84000\tLoss: 49811.26172\tPPL: 6.31299\tbleu: 23.25395\tLR: 0.00030000\t\n", "Steps: 85000\tLoss: 50014.10156\tPPL: 6.36054\tbleu: 23.40573\tLR: 0.00030000\t\n", "Steps: 86000\tLoss: 49752.88672\tPPL: 6.29937\tbleu: 23.39181\tLR: 0.00030000\t\n", "Steps: 87000\tLoss: 49512.03516\tPPL: 6.24350\tbleu: 23.49708\tLR: 0.00030000\t*\n", "Steps: 88000\tLoss: 49672.22266\tPPL: 6.28060\tbleu: 23.48113\tLR: 0.00030000\t\n", "Steps: 89000\tLoss: 49346.05078\tPPL: 6.20528\tbleu: 23.38552\tLR: 0.00030000\t*\n", "Steps: 90000\tLoss: 49344.75391\tPPL: 6.20498\tbleu: 23.58315\tLR: 0.00030000\t*\n", "Steps: 91000\tLoss: 49236.49219\tPPL: 6.18018\tbleu: 23.66106\tLR: 0.00030000\t*\n", "Steps: 92000\tLoss: 49448.01172\tPPL: 6.22873\tbleu: 23.52227\tLR: 0.00030000\t\n", "Steps: 93000\tLoss: 49437.21484\tPPL: 6.22624\tbleu: 23.37771\tLR: 0.00030000\t\n", "Steps: 94000\tLoss: 49196.59766\tPPL: 6.17107\tbleu: 23.89343\tLR: 0.00030000\t*\n", "Steps: 95000\tLoss: 49235.10547\tPPL: 6.17986\tbleu: 23.78882\tLR: 0.00030000\t\n", "Steps: 96000\tLoss: 49117.62109\tPPL: 6.15307\tbleu: 24.10978\tLR: 0.00030000\t*\n", "Steps: 97000\tLoss: 48887.02344\tPPL: 6.10080\tbleu: 24.04343\tLR: 0.00030000\t*\n", "Steps: 98000\tLoss: 49047.71094\tPPL: 6.13717\tbleu: 23.84139\tLR: 0.00030000\t\n", "Steps: 99000\tLoss: 48865.99609\tPPL: 6.09606\tbleu: 23.89779\tLR: 0.00030000\t*\n", "Steps: 100000\tLoss: 48984.08984\tPPL: 6.12275\tbleu: 23.64390\tLR: 0.00030000\t\n", "Steps: 101000\tLoss: 48859.62500\tPPL: 6.09462\tbleu: 23.98583\tLR: 0.00030000\t*\n", "Steps: 102000\tLoss: 48840.17188\tPPL: 6.09024\tbleu: 24.25505\tLR: 0.00030000\t*\n", "Steps: 103000\tLoss: 48932.47266\tPPL: 6.11107\tbleu: 23.81896\tLR: 0.00030000\t\n", "Steps: 104000\tLoss: 48710.54688\tPPL: 6.06110\tbleu: 24.14854\tLR: 0.00030000\t*\n", "Steps: 105000\tLoss: 48626.23828\tPPL: 6.04223\tbleu: 24.09812\tLR: 0.00030000\t*\n", "Steps: 106000\tLoss: 48522.85547\tPPL: 6.01917\tbleu: 24.25189\tLR: 0.00030000\t*\n", "Steps: 107000\tLoss: 48525.93359\tPPL: 6.01985\tbleu: 24.08381\tLR: 0.00030000\t\n", "Steps: 108000\tLoss: 48459.80469\tPPL: 6.00514\tbleu: 24.09373\tLR: 0.00030000\t*\n", "Steps: 109000\tLoss: 48502.42969\tPPL: 6.01462\tbleu: 24.32518\tLR: 0.00030000\t\n", "Steps: 110000\tLoss: 48468.05078\tPPL: 6.00698\tbleu: 24.29710\tLR: 0.00030000\t\n", "Steps: 111000\tLoss: 48291.64453\tPPL: 5.96791\tbleu: 24.28585\tLR: 0.00030000\t*\n", "Steps: 112000\tLoss: 48310.22266\tPPL: 5.97201\tbleu: 24.35188\tLR: 0.00030000\t\n", "Steps: 113000\tLoss: 48346.86328\tPPL: 5.98011\tbleu: 24.05949\tLR: 0.00030000\t\n", "Steps: 114000\tLoss: 48094.42188\tPPL: 5.92452\tbleu: 24.24045\tLR: 0.00030000\t*\n", "Steps: 115000\tLoss: 48121.52734\tPPL: 5.93047\tbleu: 24.31035\tLR: 0.00030000\t\n", "Steps: 116000\tLoss: 48085.62500\tPPL: 5.92260\tbleu: 24.43544\tLR: 0.00030000\t*\n", "Steps: 117000\tLoss: 48019.50000\tPPL: 5.90813\tbleu: 24.30652\tLR: 0.00030000\t*\n", "Steps: 118000\tLoss: 48032.88281\tPPL: 5.91105\tbleu: 24.39643\tLR: 0.00030000\t\n", "Steps: 119000\tLoss: 47913.39062\tPPL: 5.88498\tbleu: 24.13067\tLR: 0.00030000\t*\n", "Steps: 120000\tLoss: 48253.91016\tPPL: 5.95958\tbleu: 24.15701\tLR: 0.00030000\t\n", "Steps: 121000\tLoss: 48035.38672\tPPL: 5.91160\tbleu: 24.02503\tLR: 0.00030000\t\n", "Steps: 122000\tLoss: 47881.03516\tPPL: 5.87794\tbleu: 24.23874\tLR: 0.00030000\t*\n", "Steps: 123000\tLoss: 47822.18750\tPPL: 5.86516\tbleu: 24.13905\tLR: 0.00030000\t*\n", "Steps: 124000\tLoss: 47750.21094\tPPL: 5.84957\tbleu: 24.25684\tLR: 0.00030000\t*\n", "Steps: 125000\tLoss: 47717.11719\tPPL: 5.84241\tbleu: 24.47970\tLR: 0.00030000\t*\n", "Steps: 126000\tLoss: 47921.37109\tPPL: 5.88672\tbleu: 24.45313\tLR: 0.00030000\t\n", "Steps: 127000\tLoss: 47805.68750\tPPL: 5.86158\tbleu: 24.16186\tLR: 0.00030000\t\n", "Steps: 128000\tLoss: 47892.73828\tPPL: 5.88049\tbleu: 24.31078\tLR: 0.00030000\t\n", "Steps: 129000\tLoss: 47738.50391\tPPL: 5.84703\tbleu: 24.36292\tLR: 0.00030000\t\n", "Steps: 130000\tLoss: 47821.07812\tPPL: 5.86492\tbleu: 24.63265\tLR: 0.00030000\t\n", "Steps: 131000\tLoss: 47849.66016\tPPL: 5.87113\tbleu: 24.53310\tLR: 0.00021000\t\n", "Steps: 132000\tLoss: 47616.32422\tPPL: 5.82067\tbleu: 24.50851\tLR: 0.00021000\t*\n", "Steps: 133000\tLoss: 47466.97266\tPPL: 5.78860\tbleu: 24.30675\tLR: 0.00021000\t*\n", "Steps: 134000\tLoss: 47232.79688\tPPL: 5.73867\tbleu: 24.79230\tLR: 0.00021000\t*\n", "Steps: 135000\tLoss: 47172.35547\tPPL: 5.72585\tbleu: 24.64397\tLR: 0.00021000\t*\n", "Steps: 136000\tLoss: 47295.23438\tPPL: 5.75194\tbleu: 24.67831\tLR: 0.00021000\t\n", "Steps: 137000\tLoss: 47199.48047\tPPL: 5.73160\tbleu: 24.69814\tLR: 0.00021000\t\n", "Steps: 138000\tLoss: 46959.83594\tPPL: 5.68102\tbleu: 24.54855\tLR: 0.00021000\t*\n", "Steps: 139000\tLoss: 47004.97266\tPPL: 5.69051\tbleu: 24.36032\tLR: 0.00021000\t\n", "Steps: 140000\tLoss: 47084.75000\tPPL: 5.70733\tbleu: 24.57471\tLR: 0.00021000\t\n" ], "name": "stdout" } ] }, { "cell_type": "code", "metadata": { "id": "LUe7S9Gefd4F", "colab_type": "code", "outputId": "bb5c6412-20a6-46f4-c015-72ee6442bf64", "colab": { "base_uri": "https://localhost:8080/", "height": 68 } }, "source": [ "# Test our model\n", "! cd joeynmt; python3 -m joeynmt test \"$gdrive_path/models/${src}${tgt}_transformer/config.yaml\"" ], "execution_count": 71, "outputs": [ { "output_type": "stream", "text": [ "2020-04-10 04:59:29,486 Hello! This is Joey-NMT.\n", "2020-04-10 05:00:09,260 dev bleu: 24.95 [Beam search decoding with beam size = 5 and alpha = 1.0]\n", "2020-04-10 05:03:28,864 test bleu: 25.27 [Beam search decoding with beam size = 5 and alpha = 1.0]\n" ], "name": "stdout" } ] }, { "cell_type": "code", "metadata": { "id": "Z0ygGObugiYF", "colab_type": "code", "colab": {} }, "source": [ "# python3 plot_validation.py \"$gdrive_path/models/${src}${tgt}_transformer/\" --plot_values bleu PPL --output_path my_plot.pdf" ], "execution_count": 0, "outputs": [] }, { "cell_type": "code", "metadata": { "id": "elucrJ7NK3Iu", "colab_type": "code", "colab": {} }, "source": [ "" ], "execution_count": 0, "outputs": [] } ] }