agnews_cnn_TracIn_WE.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eWei9WKgj-Op"
      },
      "source": [
        "### Colab Setup"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "j400yhO7icjg",
        "outputId": "f81161fb-b867-4766-ffe9-1d49accb9308"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Sun Oct 16 04:22:02 2022       \n",
            "+-----------------------------------------------------------------------------+\n",
            "| NVIDIA-SMI 460.32.03    Driver Version: 460.32.03    CUDA Version: 11.2     |\n",
            "|-------------------------------+----------------------+----------------------+\n",
            "| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n",
            "| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n",
            "|                               |                      |               MIG M. |\n",
            "|===============================+======================+======================|\n",
            "|   0  Tesla T4            Off  | 00000000:00:04.0 Off |                    0 |\n",
            "| N/A   37C    P8     9W /  70W |      0MiB / 15109MiB |      0%      Default |\n",
            "|                               |                      |                  N/A |\n",
            "+-------------------------------+----------------------+----------------------+\n",
            "                                                                               \n",
            "+-----------------------------------------------------------------------------+\n",
            "| Processes:                                                                  |\n",
            "|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |\n",
            "|        ID   ID                                                   Usage      |\n",
            "|=============================================================================|\n",
            "|  No running processes found                                                 |\n",
            "+-----------------------------------------------------------------------------+\n",
            "Your runtime has 54.8 gigabytes of available RAM\n",
            "\n",
            "You are using a high-RAM runtime!\n"
          ]
        }
      ],
      "source": [
        "gpu_info = !nvidia-smi\n",
        "gpu_info = '\\n'.join(gpu_info)\n",
        "if gpu_info.find('failed') >= 0:\n",
        "  print('Not connected to a GPU')\n",
        "else:\n",
        "  print(gpu_info)\n",
        "\n",
        "from psutil import virtual_memory\n",
        "ram_gb = virtual_memory().total / 1e9\n",
        "print('Your runtime has {:.1f} gigabytes of available RAM\\n'.format(ram_gb))\n",
        "\n",
        "if ram_gb < 20:\n",
        "  print('Not using a high-RAM runtime')\n",
        "else:\n",
        "  print('You are using a high-RAM runtime!')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "k_ov6lMdr9uY",
        "outputId": "9d20204f-9560-4b68-dc35-13728c653027"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "3z2fR5jKqMNh",
        "outputId": "ee75f0ed-55e1-4009-8a2a-d05b03bd7270"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Requirement already satisfied: tensorflow==2.8 in /usr/local/lib/python3.7/dist-packages (2.8.0+zzzcolab20220506162203)\n",
            "Requirement already satisfied: wrapt>=1.11.0 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (1.14.1)\n",
            "Requirement already satisfied: keras-preprocessing>=1.1.1 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (1.1.2)\n",
            "Requirement already satisfied: gast>=0.2.1 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (0.4.0)\n",
            "Requirement already satisfied: flatbuffers>=1.12 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (1.12)\n",
            "Requirement already satisfied: tensorflow-io-gcs-filesystem>=0.23.1 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (0.27.0)\n",
            "Requirement already satisfied: astunparse>=1.6.0 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (1.6.3)\n",
            "Requirement already satisfied: protobuf>=3.9.2 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (3.17.3)\n",
            "Requirement already satisfied: keras<2.9,>=2.8.0rc0 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (2.8.0)\n",
            "Requirement already satisfied: tensorboard<2.9,>=2.8 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (2.8.0)\n",
            "Requirement already satisfied: typing-extensions>=3.6.6 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (4.1.1)\n",
            "Requirement already satisfied: libclang>=9.0.1 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (14.0.6)\n",
            "Requirement already satisfied: setuptools in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (57.4.0)\n",
            "Requirement already satisfied: tf-estimator-nightly==2.8.0.dev2021122109 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (2.8.0.dev2021122109)\n",
            "Requirement already satisfied: opt-einsum>=2.3.2 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (3.3.0)\n",
            "Requirement already satisfied: termcolor>=1.1.0 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (2.0.1)\n",
            "Requirement already satisfied: six>=1.12.0 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (1.15.0)\n",
            "Requirement already satisfied: numpy>=1.20 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (1.21.6)\n",
            "Requirement already satisfied: absl-py>=0.4.0 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (1.2.0)\n",
            "Requirement already satisfied: google-pasta>=0.1.1 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (0.2.0)\n",
            "Requirement already satisfied: h5py>=2.9.0 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (3.1.0)\n",
            "Requirement already satisfied: grpcio<2.0,>=1.24.3 in /usr/local/lib/python3.7/dist-packages (from tensorflow==2.8) (1.49.1)\n",
            "Requirement already satisfied: wheel<1.0,>=0.23.0 in /usr/local/lib/python3.7/dist-packages (from astunparse>=1.6.0->tensorflow==2.8) (0.37.1)\n",
            "Requirement already satisfied: cached-property in /usr/local/lib/python3.7/dist-packages (from h5py>=2.9.0->tensorflow==2.8) (1.5.2)\n",
            "Requirement already satisfied: google-auth-oauthlib<0.5,>=0.4.1 in /usr/local/lib/python3.7/dist-packages (from tensorboard<2.9,>=2.8->tensorflow==2.8) (0.4.6)\n",
            "Requirement already satisfied: google-auth<3,>=1.6.3 in /usr/local/lib/python3.7/dist-packages (from tensorboard<2.9,>=2.8->tensorflow==2.8) (1.35.0)\n",
            "Requirement already satisfied: werkzeug>=0.11.15 in /usr/local/lib/python3.7/dist-packages (from tensorboard<2.9,>=2.8->tensorflow==2.8) (1.0.1)\n",
            "Requirement already satisfied: markdown>=2.6.8 in /usr/local/lib/python3.7/dist-packages (from tensorboard<2.9,>=2.8->tensorflow==2.8) (3.4.1)\n",
            "Requirement already satisfied: tensorboard-plugin-wit>=1.6.0 in /usr/local/lib/python3.7/dist-packages (from tensorboard<2.9,>=2.8->tensorflow==2.8) (1.8.1)\n",
            "Requirement already satisfied: tensorboard-data-server<0.7.0,>=0.6.0 in /usr/local/lib/python3.7/dist-packages (from tensorboard<2.9,>=2.8->tensorflow==2.8) (0.6.1)\n",
            "Requirement already satisfied: requests<3,>=2.21.0 in /usr/local/lib/python3.7/dist-packages (from tensorboard<2.9,>=2.8->tensorflow==2.8) (2.23.0)\n",
            "Requirement already satisfied: rsa<5,>=3.1.4 in /usr/local/lib/python3.7/dist-packages (from google-auth<3,>=1.6.3->tensorboard<2.9,>=2.8->tensorflow==2.8) (4.9)\n",
            "Requirement already satisfied: cachetools<5.0,>=2.0.0 in /usr/local/lib/python3.7/dist-packages (from google-auth<3,>=1.6.3->tensorboard<2.9,>=2.8->tensorflow==2.8) (4.2.4)\n",
            "Requirement already satisfied: pyasn1-modules>=0.2.1 in /usr/local/lib/python3.7/dist-packages (from google-auth<3,>=1.6.3->tensorboard<2.9,>=2.8->tensorflow==2.8) (0.2.8)\n",
            "Requirement already satisfied: requests-oauthlib>=0.7.0 in /usr/local/lib/python3.7/dist-packages (from google-auth-oauthlib<0.5,>=0.4.1->tensorboard<2.9,>=2.8->tensorflow==2.8) (1.3.1)\n",
            "Requirement already satisfied: importlib-metadata>=4.4 in /usr/local/lib/python3.7/dist-packages (from markdown>=2.6.8->tensorboard<2.9,>=2.8->tensorflow==2.8) (5.0.0)\n",
            "Requirement already satisfied: zipp>=0.5 in /usr/local/lib/python3.7/dist-packages (from importlib-metadata>=4.4->markdown>=2.6.8->tensorboard<2.9,>=2.8->tensorflow==2.8) (3.9.0)\n",
            "Requirement already satisfied: pyasn1<0.5.0,>=0.4.6 in /usr/local/lib/python3.7/dist-packages (from pyasn1-modules>=0.2.1->google-auth<3,>=1.6.3->tensorboard<2.9,>=2.8->tensorflow==2.8) (0.4.8)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.7/dist-packages (from requests<3,>=2.21.0->tensorboard<2.9,>=2.8->tensorflow==2.8) (2022.9.24)\n",
            "Requirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.7/dist-packages (from requests<3,>=2.21.0->tensorboard<2.9,>=2.8->tensorflow==2.8) (3.0.4)\n",
            "Requirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.7/dist-packages (from requests<3,>=2.21.0->tensorboard<2.9,>=2.8->tensorflow==2.8) (2.10)\n",
            "Requirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.7/dist-packages (from requests<3,>=2.21.0->tensorboard<2.9,>=2.8->tensorflow==2.8) (1.24.3)\n",
            "Requirement already satisfied: oauthlib>=3.0.0 in /usr/local/lib/python3.7/dist-packages (from requests-oauthlib>=0.7.0->google-auth-oauthlib<0.5,>=0.4.1->tensorboard<2.9,>=2.8->tensorflow==2.8) (3.2.1)\n",
            "\u001b[K     |████████████████████████████████| 1.1 MB 29.3 MB/s \n",
            "\u001b[K     |████████████████████████████████| 43 kB 2.4 MB/s \n",
            "\u001b[K     |████████████████████████████████| 1.1 MB 59.6 MB/s \n",
            "\u001b[K     |████████████████████████████████| 352 kB 87.9 MB/s \n",
            "\u001b[K     |████████████████████████████████| 99 kB 11.2 MB/s \n",
            "\u001b[K     |████████████████████████████████| 238 kB 81.5 MB/s \n",
            "\u001b[K     |████████████████████████████████| 1.3 MB 69.9 MB/s \n",
            "\u001b[?25h  Building wheel for py-cpuinfo (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Building wheel for seqeval (setup.py) ... \u001b[?25l\u001b[?25hdone\n"
          ]
        }
      ],
      "source": [
        "import tensorflow_datasets as tfds\n",
        "!pip3 install tensorflow==2.8\n",
        "#!apt install --allow-change-held-packages libcudnn8=8.1.0.77-1+cuda11.2\n",
        "!pip3 install --quiet \"tensorflow-text==2.8.*\"\n",
        "import tensorflow_text as text\n",
        "import tensorflow_hub as hub\n",
        "import tensorflow as tf\n",
        "import numpy as np\n",
        "import time\n",
        "from tensorflow_text import HubModuleTokenizer\n",
        "tfhub_handle_preprocess = 'https://tfhub.dev/tensorflow/bert_en_uncased_preprocess/3'\n",
        "\n",
        "!pip install -q tf-models-official==2.4.0"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Vvob7PaY-hY3"
      },
      "source": [
        "### Build dataset\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "id": "zC83cSLT8hIH"
      },
      "outputs": [],
      "source": [
        "import tensorflow_datasets as tfds\n",
        "import tensorflow_hub as hub\n",
        "import tensorflow as tf\n",
        "import numpy as np\n",
        "import time\n",
        "from tensorflow_text import HubModuleTokenizer\n",
        "tfhub_handle_preprocess = 'https://tfhub.dev/tensorflow/bert_en_uncased_preprocess/3'\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 35,
      "metadata": {
        "id": "aOZnl9CzFseX"
      },
      "outputs": [],
      "source": [
        "bert_preprocess_model = hub.KerasLayer(tfhub_handle_preprocess)\n",
        "\n",
        "def prepare_bert_input(sentences, seq_len):\n",
        "  start = time.time()\n",
        "  segments = []\n",
        "  if len(sentences)>=64:\n",
        "    for i in range(int(len(sentences)-1/64)+1):\n",
        "      segment = bert_preprocess_model(sentences[64*i:min(len(sentences),64*(i+1))]) \n",
        "      segments.append(segment['input_word_ids'])\n",
        "    segment_concat = tf.concat(segments,axis=0)\n",
        "  else:\n",
        "    segment_concat = bert_preprocess_model(sentences)['input_word_ids']\n",
        "  print('time for tokenizing', time.time()-start)\n",
        "  return segment_concat\n",
        "\n",
        "def prepare_CNN_input(sentences, seq_len):\n",
        "  start = time.time()\n",
        "  segments = []\n",
        "  if len(sentences)>=64:\n",
        "    for i in range(int(len(sentences)-1/64)+1):\n",
        "      segment = bert_preprocess_model(sentences[64*i:min(len(sentences),64*(i+1))]) \n",
        "      segments.append(segment)\n",
        "    #print(segments)\n",
        "    segments_0 = tf.concat([segment['input_word_ids'] for segment in segments],axis=0)\n",
        "  else:\n",
        "    segments_0 = bert_preprocess_model(sentences)['input_word_ids'] \n",
        "  print('time for tokenizing', time.time()-start)\n",
        "  return segments_0\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 36,
      "metadata": {
        "id": "KB88rYKqkfQf"
      },
      "outputs": [],
      "source": [
        "from official.nlp.bert import tokenization\n",
        "\n",
        "bert_model = hub.KerasLayer('https://tfhub.dev/tensorflow/bert_en_uncased_L-12_H-768_A-12/3', trainable=True)\n",
        "vocab_file = bert_model.resolved_object.vocab_file.asset_path.numpy()\n",
        "do_lower_case = bert_model.resolved_object.do_lower_case.numpy()\n",
        "tokenizer = tokenization.FullTokenizer(vocab_file, do_lower_case)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 201,
          "referenced_widgets": [
            "026780de801140429fdaca2de8a0bbc9",
            "4260809d24724413af98285effa5b94d",
            "50811fea378e473c83b4c018a5672e58",
            "9dd5dba454e8499794b488ec2d344fa8",
            "17dd5c9d57b94d09934c8bc231e883f9",
            "b83c327336e84ccfb89ea24b02215176",
            "cb8e6b64031e4e55b6e5eee5ee5dd04b",
            "1823c3217eda40d9b76f81caa91741e6",
            "920a75e0cc5e42eca3f79d98f8fd516c",
            "5ccf97a4b03547339f9daadb53e97bec",
            "71ecf2dedad7489ca3d3e8905ec2e2ec",
            "f5937c5b546543d89c10816a22ef0644",
            "30d072fc93354dd8adec20e80c507262",
            "95d1c2da758e46bfb96edb92169ba3c3",
            "c8ca37a9a5654473b788e3780c45206f",
            "88a516d0d98a4919bf57588ecf9c5d0e",
            "c1109d62d719479db3c747142609d2d1",
            "cf171048374d4f9eae660d209d93eecc",
            "0d89d80422ab4b3a959ce12e39c05e41",
            "416f3deffae94a16bcb92df90b7d8bc3",
            "7be316d9b19942f3914d7167fca1e8ed",
            "387bc56902c8454fa9b274e868a50635",
            "8c976ef531d1472e942ac608ea957f7c",
            "a66f5f8ff5724d77a2e48ee08f63bc82",
            "ac313cf6f91c4c218f0931d3e35cab59",
            "6874e918b0244d279b66016ff7fe4a39",
            "78ea05bd650c48bd8068da925694f459",
            "ef78a0ac887245f8b3842dd236be1158",
            "eb1239b7b8d84226a4285352662612f7",
            "d7734b9698594e6dbfd67faa0872da76",
            "23bcdad689a44917b714bf40ca8efc88",
            "962e3828576b4c47b720e7b0b5b0a942",
            "0531c6048e20450f867d5dc4be1b2fca",
            "7f312183085942ddbee49ee68518b5a6",
            "4b301dd5436b4817a5902c666ea8720f",
            "9f4c3c35a8ab4a0b8a40acf41c668c80",
            "f42cac69a50946aa971d9720161303d5",
            "6ecda1493a8a4e8db04a7908b5f573e1",
            "86b949eee20a4f82b4eeb1862f9fe34c",
            "26088d3a2027402b8ad386041ec97c70",
            "d7742ebb54c74dd39847c004f03d5ba4",
            "072ac4984a974ab2b4a83c1e01df5bbd",
            "751c7c4a3b074ac783ed9778c3f15cf7",
            "89a773d97af04b4795476f6a8534631e",
            "22482f05ce424b0d969d1b5d67ca7562",
            "6904f800f1484c418ff43edfcb868dff",
            "e4902cf3945d4019b00c8d23e91b342a",
            "46c0dc7c65b5417ea0da28ebc159d3e7",
            "a8fcc0913d374d1993e16494fe5f90cd",
            "df9cea19a8494f6a98cb191a0cbe660f",
            "a79e5a29757e42f1b3acfc234790f0b1",
            "2c77f3549b99406e836b64b52079b1f8",
            "4deb2dd644944da2b2bcdc458b1d7818",
            "c9ed14f1afa4480d975c22ac6fae6ceb",
            "6027e75c591c446bbcc46cdc27b5d2ae",
            "2a503125f8354301942ea5ca8d3d99e8",
            "92d55759b0cb4d25910c194d2911f3c9",
            "3d8dafe9909347f0ae064a99dbc02171",
            "cbcca21e3a184e2e9d95e9268f18fce0",
            "413221ade0c84c91a7c76e6f283ad48f",
            "0d1f3589e8c3405784513b9d05b5bcdb",
            "ebb417f9454b4cd6b8102c199470950c",
            "a8090dab55054d478ec0c0262bd793a0",
            "b1fcba615c6e459eac5fa72efec04d59",
            "5c5b32bc53f84962820181c20d0b92c2",
            "6d9e12bc28454b92886704bc0883a704",
            "cc0365c9429142278f706a4f07bf34e8",
            "03bc494f9f9e45d69b1a9e38d36303c4",
            "d1c50e5ea8cf4d24a9b7adcefdb23b96",
            "3aeac441812f4252960f1420b17efa04",
            "2509cabd56364b7e9789bd5c260e06fb",
            "2b4b849656d9412aa3d379a80fdbfec7",
            "20b5342699514a6da13137e66db6f5f6",
            "89e41feb217749239c28248f466c361c",
            "2fa78e682214457599ad9c2cd8a8eac2",
            "d335c08c3aa74551a96e35831dc802c8",
            "ba1f3823b17e492db5a4ebebaf23e5a2",
            "07ab4bfe0f414e73a3bc141cc48faa38",
            "3667e623b9cb4fb981eb71a2ea0f2da7",
            "97c7aeef0e7c4dcfaca822ced5690ab9",
            "ec020c0988084ca0a7c7352dabb74702",
            "b160e728277142deb782141385016776",
            "98da7fb99cd844fbb24e18bf48d1a22f",
            "dd253db5b1cb4b94bb908acfd3277609",
            "11a289923a9845ac94c3cd9c95920edc",
            "c5959332e6c0442c96dd45c03b1237a6",
            "e194ab3ad9174e199a38f2075b59f2a9",
            "0ecfb382b7b342baae293fac6f26d867"
          ]
        },
        "id": "n43nWiz8FWiw",
        "outputId": "db7f9071-f995-4539-c8cd-981846fce310"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\u001b[1mDownloading and preparing dataset 11.24 MiB (download: 11.24 MiB, generated: 35.79 MiB, total: 47.03 MiB) to ~/tensorflow_datasets/ag_news_subset/1.0.0...\u001b[0m\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Dl Completed...: 0 url [00:00, ? url/s]"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "026780de801140429fdaca2de8a0bbc9"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Dl Size...: 0 MiB [00:00, ? MiB/s]"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "f5937c5b546543d89c10816a22ef0644"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Extraction completed...: 0 file [00:00, ? file/s]"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "8c976ef531d1472e942ac608ea957f7c"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Generating splits...:   0%|          | 0/2 [00:00<?, ? splits/s]"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "7f312183085942ddbee49ee68518b5a6"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Generating train examples...:   0%|          | 0/120000 [00:00<?, ? examples/s]"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "22482f05ce424b0d969d1b5d67ca7562"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Shuffling ~/tensorflow_datasets/ag_news_subset/1.0.0.incompleteO62OX8/ag_news_subset-train.tfrecord*...:   0%|…"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "2a503125f8354301942ea5ca8d3d99e8"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Generating test examples...:   0%|          | 0/7600 [00:00<?, ? examples/s]"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "cc0365c9429142278f706a4f07bf34e8"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Shuffling ~/tensorflow_datasets/ag_news_subset/1.0.0.incompleteO62OX8/ag_news_subset-test.tfrecord*...:   0%| …"
            ],
            "application/vnd.jupyter.widget-view+json": {
              "version_major": 2,
              "version_minor": 0,
              "model_id": "07ab4bfe0f414e73a3bc141cc48faa38"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\u001b[1mDataset ag_news_subset downloaded and prepared to ~/tensorflow_datasets/ag_news_subset/1.0.0. Subsequent calls will reuse this data.\u001b[0m\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'\\ny_train_orig = np.array(y_train_orig)\\ny_val_orig = np.array(y_val_orig)\\ny_test_orig = np.array(y_test_orig)\\n#y_test2_orig = np.array(y_test2_orig)\\n\\ny_train = np.sum(y_train_orig,axis=1,keepdims=True)\\ny_val = np.sum(y_val_orig,axis=1,keepdims=True)\\ny_test = np.sum(y_test_orig,axis=1,keepdims=True)\\n#y_test2 = np.sum(y_test2_orig,axis=1,keepdims=True)\\n\\n\\nprint(y_train[:10])\\n\\nprint(train_sentences_list[:10])\\n'"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            }
          },
          "metadata": {},
          "execution_count": 7
        }
      ],
      "source": [
        "MAX_SEQ_LEN = 128\n",
        "\n",
        "ds_train = tfds.load('ag_news_subset', split='train', shuffle_files=False)\n",
        "ds_test = tfds.load('ag_news_subset', split='test', shuffle_files=False)\n",
        "\n",
        "\n",
        "train_sentences_list = []\n",
        "y_train_orig = []\n",
        "val_sentences_list = []\n",
        "y_val_orig = []\n",
        "test_sentences_list = []\n",
        "y_test_orig = []\n",
        "test_sentences_list2 = []\n",
        "y_test2_orig = []\n",
        "\n",
        "for count,x in enumerate(ds_train):\n",
        "  if count<50000:\n",
        "    train_sentences_list.append(x['description'].numpy().decode(\"utf-8\") )\n",
        "    y_train_orig.append([x['label']])\n",
        "  elif count<55000:\n",
        "    val_sentences_list.append(x['description'].numpy().decode(\"utf-8\") )\n",
        "    y_val_orig.append([x['label']])\n",
        "\n",
        "for count,x in enumerate(ds_test):\n",
        "  if count<20000:\n",
        "    test_sentences_list.append(x['description'].numpy().decode(\"utf-8\") )\n",
        "    y_test_orig.append([x['label']])\n",
        "  else:\n",
        "    test_sentences_list2.append(x['description'].numpy().decode(\"utf-8\") )\n",
        "    y_test2_orig.append([x['label']])\n",
        "# 2) encode sentences following the BERT specifications\n",
        "base_dir = '/content/drive/MyDrive/tracin/agnews/'\n",
        "preprocessed = True\n",
        "\n",
        "if not preprocessed:\n",
        "  x_val = prepare_bert_input(val_sentences_list, MAX_SEQ_LEN)\n",
        "  x_test = prepare_bert_input(test_sentences_list, MAX_SEQ_LEN)\n",
        "  x_train = prepare_bert_input(train_sentences_list, MAX_SEQ_LEN)\n",
        "  y_train_orig = np.array(y_train_orig)\n",
        "  y_val_orig = np.array(y_val_orig)\n",
        "  y_test_orig = np.array(y_test_orig)\n",
        "\n",
        "  y_train = np.sum(y_train_orig,axis=1,keepdims=True)\n",
        "  y_val = np.sum(y_val_orig,axis=1,keepdims=True)\n",
        "  y_test = np.sum(y_test_orig,axis=1,keepdims=True)\n",
        "else:\n",
        "  x_train = np.load(base_dir+'x_train.npy')\n",
        "  x_test = np.load(base_dir+'x_test.npy')\n",
        "  x_val = np.load(base_dir+'x_val.npy')\n",
        "\n",
        "  y_train = np.load(base_dir+'y_train.npy')\n",
        "  y_test = np.load(base_dir+'y_test.npy')\n",
        "  y_val = np.load(base_dir+'y_val.npy')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Fxzcw3YJkCdJ"
      },
      "source": [
        "### Train CNN Model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "id": "m6guF74ATvM8"
      },
      "outputs": [],
      "source": [
        "def create_model(verbose=False):\n",
        "  filters = 50\n",
        "  kernel_size = 5\n",
        "  maxlen = 128\n",
        "  inputs = tf.keras.layers.Input(shape=(maxlen,))\n",
        "  emb1 = tf.keras.layers.Embedding(25000, 128, input_length=128)(inputs)\n",
        "  conv1 = tf.keras.layers.Conv1D(\n",
        "      filters, kernel_size, padding='valid', activation='relu')(\n",
        "          emb1)\n",
        "  conv2 = tf.keras.layers.Conv1D(\n",
        "      filters, kernel_size, padding='valid', activation='relu')(\n",
        "          conv1)\n",
        "  conv3 = tf.keras.layers.Conv1D(\n",
        "      filters, 1, padding='valid', activation='relu')(\n",
        "          conv2)\n",
        "  pool1 = tf.keras.layers.GlobalMaxPooling1D()(conv3)\n",
        "  pred = tf.keras.layers.Dense(4, activation='softmax')(pool1)\n",
        "  model = tf.keras.Model(inputs, pred)\n",
        "  if verbose:\n",
        "    model.summary()\n",
        "  return model\n",
        "  \"\"\"\n",
        "  filters = 50\n",
        "  kernel_size=5\n",
        "  maxlen = 128\n",
        "  inputs = tf.keras.layers.Input(shape=(maxlen,))\n",
        "  emb1 = tf.keras.layers.Embedding(30265,128,input_length=128)(inputs)\n",
        "  conv1 = tf.keras.layers.Conv1D(filters,kernel_size,padding='valid',activation='relu')(emb1)\n",
        "  conv2 = tf.keras.layers.Conv1D(filters,kernel_size,padding='valid',activation='relu')(conv1)\n",
        "  conv3 = tf.keras.layers.Conv1D(filters,1,padding='valid',activation='relu')(conv2)\n",
        "  pool1 = tf.keras.layers.GlobalMaxPooling1D()(conv3)\n",
        "  pred = tf.keras.layers.Dense(4,activation='softmax')(pool1)\n",
        "  model = tf.keras.Model(inputs,pred)\n",
        "  if verbose:\n",
        "    model.summary()\n",
        "  return model\n",
        "  \"\"\""
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "6yVL7m3NbE16",
        "outputId": "04575962-da1a-4699-ec16-23aeb1c55e15"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"model\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_1 (InputLayer)        [(None, 128)]             0         \n",
            "                                                                 \n",
            " embedding (Embedding)       (None, 128, 128)          3200000   \n",
            "                                                                 \n",
            " conv1d (Conv1D)             (None, 124, 50)           32050     \n",
            "                                                                 \n",
            " conv1d_1 (Conv1D)           (None, 120, 50)           12550     \n",
            "                                                                 \n",
            " conv1d_2 (Conv1D)           (None, 120, 50)           2550      \n",
            "                                                                 \n",
            " global_max_pooling1d (Globa  (None, 50)               0         \n",
            " lMaxPooling1D)                                                  \n",
            "                                                                 \n",
            " dense (Dense)               (None, 4)                 204       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 3,247,354\n",
            "Trainable params: 3,247,354\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Model: \"model_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_2 (InputLayer)        [(None, 128)]             0         \n",
            "                                                                 \n",
            " embedding_1 (Embedding)     (None, 128, 128)          3200000   \n",
            "                                                                 \n",
            " conv1d_3 (Conv1D)           (None, 124, 50)           32050     \n",
            "                                                                 \n",
            " conv1d_4 (Conv1D)           (None, 120, 50)           12550     \n",
            "                                                                 \n",
            " conv1d_5 (Conv1D)           (None, 120, 50)           2550      \n",
            "                                                                 \n",
            " global_max_pooling1d_1 (Glo  (None, 50)               0         \n",
            " balMaxPooling1D)                                                \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 4)                 204       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 3,247,354\n",
            "Trainable params: 3,247,354\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Model: \"model_2\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_3 (InputLayer)        [(None, 128)]             0         \n",
            "                                                                 \n",
            " embedding_2 (Embedding)     (None, 128, 128)          3200000   \n",
            "                                                                 \n",
            " conv1d_6 (Conv1D)           (None, 124, 50)           32050     \n",
            "                                                                 \n",
            " conv1d_7 (Conv1D)           (None, 120, 50)           12550     \n",
            "                                                                 \n",
            " conv1d_8 (Conv1D)           (None, 120, 50)           2550      \n",
            "                                                                 \n",
            " global_max_pooling1d_2 (Glo  (None, 50)               0         \n",
            " balMaxPooling1D)                                                \n",
            "                                                                 \n",
            " dense_2 (Dense)             (None, 4)                 204       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 3,247,354\n",
            "Trainable params: 3,247,354\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Model: \"model_3\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_4 (InputLayer)        [(None, 128)]             0         \n",
            "                                                                 \n",
            " embedding_3 (Embedding)     (None, 128, 128)          3200000   \n",
            "                                                                 \n",
            " conv1d_9 (Conv1D)           (None, 124, 50)           32050     \n",
            "                                                                 \n",
            " conv1d_10 (Conv1D)          (None, 120, 50)           12550     \n",
            "                                                                 \n",
            " conv1d_11 (Conv1D)          (None, 120, 50)           2550      \n",
            "                                                                 \n",
            " global_max_pooling1d_3 (Glo  (None, 50)               0         \n",
            " balMaxPooling1D)                                                \n",
            "                                                                 \n",
            " dense_3 (Dense)             (None, 4)                 204       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 3,247,354\n",
            "Trainable params: 3,247,354\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Model: \"model_4\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_5 (InputLayer)        [(None, 128)]             0         \n",
            "                                                                 \n",
            " embedding_4 (Embedding)     (None, 128, 128)          3200000   \n",
            "                                                                 \n",
            " conv1d_12 (Conv1D)          (None, 124, 50)           32050     \n",
            "                                                                 \n",
            " conv1d_13 (Conv1D)          (None, 120, 50)           12550     \n",
            "                                                                 \n",
            " conv1d_14 (Conv1D)          (None, 120, 50)           2550      \n",
            "                                                                 \n",
            " global_max_pooling1d_4 (Glo  (None, 50)               0         \n",
            " balMaxPooling1D)                                                \n",
            "                                                                 \n",
            " dense_4 (Dense)             (None, 4)                 204       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 3,247,354\n",
            "Trainable params: 3,247,354\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Model: \"model_5\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_6 (InputLayer)        [(None, 128)]             0         \n",
            "                                                                 \n",
            " embedding_5 (Embedding)     (None, 128, 128)          3200000   \n",
            "                                                                 \n",
            " conv1d_15 (Conv1D)          (None, 124, 50)           32050     \n",
            "                                                                 \n",
            " conv1d_16 (Conv1D)          (None, 120, 50)           12550     \n",
            "                                                                 \n",
            " conv1d_17 (Conv1D)          (None, 120, 50)           2550      \n",
            "                                                                 \n",
            " global_max_pooling1d_5 (Glo  (None, 50)               0         \n",
            " balMaxPooling1D)                                                \n",
            "                                                                 \n",
            " dense_5 (Dense)             (None, 4)                 204       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 3,247,354\n",
            "Trainable params: 3,247,354\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Model: \"model_6\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_7 (InputLayer)        [(None, 128)]             0         \n",
            "                                                                 \n",
            " embedding_6 (Embedding)     (None, 128, 128)          3200000   \n",
            "                                                                 \n",
            " conv1d_18 (Conv1D)          (None, 124, 50)           32050     \n",
            "                                                                 \n",
            " conv1d_19 (Conv1D)          (None, 120, 50)           12550     \n",
            "                                                                 \n",
            " conv1d_20 (Conv1D)          (None, 120, 50)           2550      \n",
            "                                                                 \n",
            " global_max_pooling1d_6 (Glo  (None, 50)               0         \n",
            " balMaxPooling1D)                                                \n",
            "                                                                 \n",
            " dense_6 (Dense)             (None, 4)                 204       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 3,247,354\n",
            "Trainable params: 3,247,354\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Model: \"model_7\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_8 (InputLayer)        [(None, 128)]             0         \n",
            "                                                                 \n",
            " embedding_7 (Embedding)     (None, 128, 128)          3200000   \n",
            "                                                                 \n",
            " conv1d_21 (Conv1D)          (None, 124, 50)           32050     \n",
            "                                                                 \n",
            " conv1d_22 (Conv1D)          (None, 120, 50)           12550     \n",
            "                                                                 \n",
            " conv1d_23 (Conv1D)          (None, 120, 50)           2550      \n",
            "                                                                 \n",
            " global_max_pooling1d_7 (Glo  (None, 50)               0         \n",
            " balMaxPooling1D)                                                \n",
            "                                                                 \n",
            " dense_7 (Dense)             (None, 4)                 204       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 3,247,354\n",
            "Trainable params: 3,247,354\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Model: \"model_8\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_9 (InputLayer)        [(None, 128)]             0         \n",
            "                                                                 \n",
            " embedding_8 (Embedding)     (None, 128, 128)          3200000   \n",
            "                                                                 \n",
            " conv1d_24 (Conv1D)          (None, 124, 50)           32050     \n",
            "                                                                 \n",
            " conv1d_25 (Conv1D)          (None, 120, 50)           12550     \n",
            "                                                                 \n",
            " conv1d_26 (Conv1D)          (None, 120, 50)           2550      \n",
            "                                                                 \n",
            " global_max_pooling1d_8 (Glo  (None, 50)               0         \n",
            " balMaxPooling1D)                                                \n",
            "                                                                 \n",
            " dense_8 (Dense)             (None, 4)                 204       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 3,247,354\n",
            "Trainable params: 3,247,354\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Model: \"model_9\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_10 (InputLayer)       [(None, 128)]             0         \n",
            "                                                                 \n",
            " embedding_9 (Embedding)     (None, 128, 128)          3200000   \n",
            "                                                                 \n",
            " conv1d_27 (Conv1D)          (None, 124, 50)           32050     \n",
            "                                                                 \n",
            " conv1d_28 (Conv1D)          (None, 120, 50)           12550     \n",
            "                                                                 \n",
            " conv1d_29 (Conv1D)          (None, 120, 50)           2550      \n",
            "                                                                 \n",
            " global_max_pooling1d_9 (Glo  (None, 50)               0         \n",
            " balMaxPooling1D)                                                \n",
            "                                                                 \n",
            " dense_9 (Dense)             (None, 4)                 204       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 3,247,354\n",
            "Trainable params: 3,247,354\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "batch_size = 128\n",
        "epochs = 10\n",
        "opt = tf.keras.optimizers.SGD(learning_rate=2e-2, momentum=0.9)\n",
        "for i in range(10):\n",
        "  trained = True\n",
        "  model = create_model(True)\n",
        "  loss = tf.keras.losses.SparseCategoricalCrossentropy()\n",
        "  model.compile(loss=loss, optimizer= opt, metrics=['accuracy'])\n",
        "  cp_callback = tf.keras.callbacks.ModelCheckpoint(base_dir + 'checkpoint_test_222_'+'{}'.format(i)+'_{epoch:02d}.hdf5', verbose=1, save_weights_only=False)\n",
        "  if not trained:\n",
        "    model.fit(x_train, y_train,\n",
        "              batch_size=batch_size,\n",
        "              epochs=10,\n",
        "              validation_data=(x_val, y_val),\n",
        "              callbacks=[cp_callback],\n",
        "              shuffle=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JYSTeia_viY7"
      },
      "source": [
        "###  Choose Test data list"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fv7fSO4fl2vL",
        "outputId": "4db50f71-9c53-462a-ab1b-7e1c8556e6e8"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"model_9\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_10 (InputLayer)       [(None, 128)]             0         \n",
            "                                                                 \n",
            " embedding_9 (Embedding)     (None, 128, 128)          3200000   \n",
            "                                                                 \n",
            " conv1d_27 (Conv1D)          (None, 124, 50)           32050     \n",
            "                                                                 \n",
            " conv1d_28 (Conv1D)          (None, 120, 50)           12550     \n",
            "                                                                 \n",
            " conv1d_29 (Conv1D)          (None, 120, 50)           2550      \n",
            "                                                                 \n",
            " global_max_pooling1d_9 (Glo  (None, 50)               0         \n",
            " balMaxPooling1D)                                                \n",
            "                                                                 \n",
            " dense_9 (Dense)             (None, 4)                 204       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 3,247,354\n",
            "Trainable params: 3,247,354\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "None\n"
          ]
        }
      ],
      "source": [
        "print(model.summary())"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_p3tmHGrv0I6",
        "outputId": "6b542960-b9ec-43bf-afaf-125bb3981085"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(7600, 4)\n",
            "(7600, 4)\n",
            "(7600, 4)\n",
            "(7600, 4)\n",
            "(7600, 4)\n",
            "(7600, 4)\n",
            "(7600, 4)\n",
            "(7600, 4)\n",
            "(7600, 4)\n",
            "(7600, 4)\n"
          ]
        }
      ],
      "source": [
        "prediction_tests = np.zeros((10,7600,4))\n",
        "model = create_model()\n",
        "for i in range(10):\n",
        "  model.load_weights(base_dir + 'checkpoint_test_222_{}_10.hdf5'.format(i))\n",
        "  prediction_list = model.predict(x_test)\n",
        "  print(prediction_list.shape)\n",
        "  prediction_tests[i] = prediction_list\n",
        "\n",
        "np.save(base_dir+'prediction_tests.npy', prediction_tests)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "qY-9OO-I6Ixa",
        "outputId": "1570cc81-44fe-4ce7-e511-d08ac83a3132"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(7600, 128)\n",
            "[6529 4672 3996 2967 5133 6871 1014 4700 2164 5935  273 4494 3304 2453\n",
            " 1824  152  613 4448 2232 4118]\n",
            "[[2]\n",
            " [2]\n",
            " [3]\n",
            " [1]\n",
            " [0]\n",
            " [3]\n",
            " [0]\n",
            " [1]\n",
            " [1]\n",
            " [0]\n",
            " [1]\n",
            " [3]\n",
            " [0]\n",
            " [2]\n",
            " [0]\n",
            " [0]\n",
            " [2]\n",
            " [0]\n",
            " [3]\n",
            " [1]]\n",
            "['Retail sales slid in August as people steered away from buying cars and shoppers kept a close eye on their spending after splurging in July.', 'The fallout from allegations of serious accounting problems at Fannie Mae has rattled investors and could even bump up mortgage rates down the road.', 'McData plans to introduce a new SAN router this week designed to connect the growing number of isolated SAN networks in corporations.&lt;p&gt;ADVERTISEMENT&lt;/p&gt;&lt;p&gt;&lt;img src=\"http://ad.doubleclick.net/ad/idg.us.ifw.general/sbcspotrssfeed;sz=1x1;ord=200301151450?\" width=\"1\" height=\"1\" border=\"0\"/&gt;&lt;a href=\"http://ad.doubleclick.net/clk;9228975;9651165;a?http://www.infoworld.com/spotlights/sbc/main.html?lpid0103035400730000idlp\"&gt;SBC Case Study: Crate   Barrel&lt;/a&gt;&lt;br/&gt;What sold them on improving their network? A system that could cut management costs from the get-go. Find out more.&lt;/p&gt;', 'Leon Washington ran for 104 yards and a touchdown and Florida State sacked Alabama-Birmingham #39;s Darrell Hackney eight times Saturday night to rebound from a disappointing loss to Miami with a 34-7 victory.', \"The European Union's decision to hold entry talks with Turkey receives a widespread welcome.\", \"Several Minnesota police departments are field testing a handheld device that scans a suspect's fingerprint and digitally checks it against Minnesota's criminal history and fingerprint database.\", 'A former Palestinian security minister who could be key to keeping order among rival factions in Gaza after an Israeli pullout held a fence-mending meeting Monday with President Yasser Arafat, officials said.', 'ESTONIA #39;s Markko Martin took the Rally of Catalonia lead today after newly-crowned world champion Sebastien Loeb, the overnight leader, was forced out of the race with a severe oil leak.', 'Javy Lopez drove in four runs, Daniel Cabrera became the first rookie to win 10 games this season, and the Baltimore Orioles held the Tampa Bay Devil Rays to two hits in an 8-0 victory.', 'SAN FRANCISCO - California\\'s constitution permits laws against gay marriage, the state\\'s attorney general declared Friday in a long-awaited legal opinion that sought to avoid offending either side of the debate.    While acknowledging that \"committed and loving relationships between two individuals deserve recognition under California law,\" Attorney General Bill Lockyer said it was up to the voters or the Legislature to decide questions about whether gay couples should be allowed to marry...', 'JL Lewis shot a 10-under 62 for his best start ever on the PGA Tour and a two-shot lead Thursday in the Funai Classic at Disney. Lewis putted for birdie on every hole and made 11 of them on the Magnolia course to match his career-low round.', 'In an attempt to become more useful, US researchers are developing new smartphone software which watches users calling and usage patterns and tries to learn how best to help.', 'AP - China has reached agreement with the Association of Southeast Asian Nations, or ASEAN, on completely removing tariffs on merchandise goods by 2010 as part of a proposed free trade agreement, the Chinese Ministry of Commerce says.', 'Crude oil fell to the lowest price in almost two months after Iran, OPEC #39;s second-biggest oil producer, said it would stop enriching uranium to ward off US calls for sanctions.', 'BRITAIN has warned Sudan that it still has a lot of work to do to satisfy the international community that it is tackling what the United Nations has described as the worlds worst humanitarian crisis.', 'Israeli jets force a flight to Tel Aviv to land in Cyprus after a bomb alert  German officials did not consider serious.', \"Abbey National shareholders will no longer need to fill in complex Spanish tax forms if BSCH's bid to buy the UK firm succeeds. \", 'Iran will pull out of the nuclear Non-Proliferation Treaty (NPT) and develop its atomic programme in secret if Western nations threaten or put pressure on Tehran, a senior Iran diplomat was quoted as saying on Wednesday.', 'SONY is going head-to-head with Nintendo in the battle for the handheld games console market. The company will price its long-awaited PlayStation Portable (PSP) at about 100 for its launch in Japan, when ', \" KANSAS CITY, Mo., (Sports Network) - Kansas City Chiefs  quarterback Trent Green is expected to start in Sunday's game  against the Oakland Raiders despite suffering from bruises to  his ribs and hip.\"]\n",
            "[[[2.37473822e-03 4.21195773e-06 9.80631590e-01 1.69895366e-02]\n",
            "  [6.34192897e-04 1.17770158e-08 9.99235749e-01 1.30111541e-04]\n",
            "  [2.63993861e-03 9.77412924e-07 4.24819905e-03 9.93110895e-01]\n",
            "  [5.66822477e-04 9.99427438e-01 5.06002516e-06 6.96880704e-07]\n",
            "  [9.99978662e-01 4.58430165e-08 1.73254612e-05 4.05118817e-06]\n",
            "  [2.53198951e-01 4.22035083e-02 5.63767366e-03 6.98959887e-01]\n",
            "  [9.99999404e-01 1.01626131e-10 2.24342159e-07 4.31237737e-07]\n",
            "  [2.82640420e-02 9.67386186e-01 4.28853929e-03 6.12134609e-05]\n",
            "  [5.99114428e-05 9.99936938e-01 3.53706810e-07 2.74946547e-06]\n",
            "  [9.96894956e-01 9.99155304e-07 1.78281174e-04 2.92580109e-03]\n",
            "  [9.24423148e-05 9.99902129e-01 3.24179950e-06 2.06915956e-06]\n",
            "  [1.00700834e-06 2.07567055e-07 1.89277416e-04 9.99809563e-01]\n",
            "  [9.95117426e-01 1.07894482e-06 4.02815966e-03 8.53368780e-04]\n",
            "  [4.31881584e-02 2.46395587e-07 9.56785560e-01 2.59521184e-05]\n",
            "  [9.99975443e-01 1.77795329e-07 1.79625840e-06 2.25998429e-05]\n",
            "  [9.99998689e-01 1.23146350e-07 5.91224762e-07 5.89492913e-07]\n",
            "  [8.18102970e-04 4.78806186e-08 9.98568654e-01 6.13171025e-04]\n",
            "  [9.99998927e-01 1.18967989e-07 6.69983606e-08 8.79620416e-07]\n",
            "  [5.97507751e-04 8.08013226e-07 4.08189535e-01 5.91212153e-01]\n",
            "  [3.72255755e-08 1.00000000e+00 4.22858137e-10 4.18033747e-10]]\n",
            "\n",
            " [[1.47927646e-03 5.38912209e-05 9.58537817e-01 3.99290398e-02]\n",
            "  [1.90597675e-05 3.59114516e-10 9.99902248e-01 7.86949458e-05]\n",
            "  [4.19909582e-02 8.68693917e-07 1.95519999e-03 9.56053019e-01]\n",
            "  [1.03097962e-04 9.99895692e-01 3.83502595e-07 8.01831902e-07]\n",
            "  [9.99887943e-01 1.94048584e-06 3.65006963e-05 7.35922295e-05]\n",
            "  [8.07721633e-03 9.82633412e-01 4.12524125e-04 8.87688715e-03]\n",
            "  [9.99997139e-01 4.52674378e-08 8.15803674e-08 2.79255619e-06]\n",
            "  [1.31043019e-02 9.84175742e-01 1.96466525e-03 7.55362620e-04]\n",
            "  [1.39310549e-04 9.99858141e-01 1.13075521e-06 1.38287066e-06]\n",
            "  [9.69736993e-01 1.13486385e-05 5.55572624e-04 2.96961572e-02]\n",
            "  [1.48921448e-04 9.99821961e-01 2.28863555e-05 6.34827893e-06]\n",
            "  [2.21169139e-05 1.97321959e-09 7.98890775e-04 9.99178946e-01]\n",
            "  [9.59139466e-01 1.88565537e-05 6.69676624e-03 3.41449305e-02]\n",
            "  [2.61369273e-02 9.48766257e-08 9.73108530e-01 7.54481298e-04]\n",
            "  [9.99395132e-01 4.09634322e-06 2.06017048e-05 5.80238935e-04]\n",
            "  [9.99966145e-01 1.19599588e-06 6.36531513e-06 2.63549064e-05]\n",
            "  [4.92406369e-04 1.72283615e-06 9.92522657e-01 6.98318565e-03]\n",
            "  [9.99920249e-01 6.62705716e-05 3.42046036e-07 1.31323941e-05]\n",
            "  [1.35351424e-04 2.33046592e-07 1.20955938e-03 9.98654842e-01]\n",
            "  [3.77633569e-06 9.99996066e-01 8.54449844e-09 9.11162061e-08]]\n",
            "\n",
            " [[1.38526200e-04 1.88488229e-07 9.42694843e-01 5.71664795e-02]\n",
            "  [4.79460759e-05 7.42421946e-09 9.97423649e-01 2.52850447e-03]\n",
            "  [5.47708152e-03 1.01736957e-08 6.91272458e-03 9.87610102e-01]\n",
            "  [4.58565861e-04 9.99505043e-01 3.59834485e-05 3.97419029e-07]\n",
            "  [9.99985814e-01 2.64631410e-07 1.13098113e-05 2.67853875e-06]\n",
            "  [4.35694426e-01 3.14199701e-02 2.62656230e-02 5.06619990e-01]\n",
            "  [9.99999285e-01 3.56046881e-09 3.12853075e-07 4.21803207e-07]\n",
            "  [4.70299907e-02 9.48763669e-01 3.66125442e-03 5.45100658e-04]\n",
            "  [2.10117782e-04 9.99763310e-01 2.48833421e-05 1.74175659e-06]\n",
            "  [7.76370287e-01 6.62891216e-06 1.26557366e-03 2.22357422e-01]\n",
            "  [5.39068023e-05 9.99897838e-01 4.22040539e-05 6.05692048e-06]\n",
            "  [2.67226733e-05 1.08081011e-09 2.16940558e-03 9.97803867e-01]\n",
            "  [9.74093974e-01 8.17515611e-05 1.85098387e-02 7.31447106e-03]\n",
            "  [2.06064321e-02 2.03231593e-06 9.79330242e-01 6.12817603e-05]\n",
            "  [9.98637021e-01 2.46645777e-06 3.98880686e-04 9.61620186e-04]\n",
            "  [9.99994874e-01 4.04660994e-07 4.20424522e-06 5.58080274e-07]\n",
            "  [1.93570543e-03 1.27826524e-06 9.95215058e-01 2.84788036e-03]\n",
            "  [9.99987483e-01 2.38619759e-06 5.65774656e-07 9.61460591e-06]\n",
            "  [1.14659859e-04 4.89276246e-08 7.05413446e-02 9.29343939e-01]\n",
            "  [9.59937825e-06 9.99990225e-01 8.19572250e-08 3.56765746e-08]]\n",
            "\n",
            " [[6.01569540e-04 8.07046672e-06 9.64535654e-01 3.48547548e-02]\n",
            "  [9.95153459e-06 1.25434569e-08 9.99864936e-01 1.25046296e-04]\n",
            "  [5.15724940e-04 4.93954303e-07 6.87329285e-03 9.92610395e-01]\n",
            "  [6.28415801e-05 9.99932051e-01 2.49239906e-06 2.66050301e-06]\n",
            "  [9.99934435e-01 3.72721934e-06 3.07992486e-05 3.11062977e-05]\n",
            "  [8.84454429e-01 1.60458609e-02 1.76353613e-04 9.93233398e-02]\n",
            "  [9.99999881e-01 3.51396273e-11 1.82936208e-08 5.88134448e-08]\n",
            "  [1.29868045e-01 8.69718969e-01 3.42668820e-04 7.03459227e-05]\n",
            "  [8.15027161e-05 9.99916911e-01 1.02147247e-07 1.40209363e-06]\n",
            "  [9.72311735e-01 4.81080860e-07 4.49406944e-04 2.72383317e-02]\n",
            "  [1.09856471e-03 9.98885691e-01 1.41087339e-05 1.62862398e-06]\n",
            "  [3.25918314e-04 4.17821830e-08 1.83351722e-03 9.97840524e-01]\n",
            "  [9.97124970e-01 3.41484497e-06 1.81579113e-03 1.05581409e-03]\n",
            "  [2.93663461e-02 3.33434514e-06 9.70626652e-01 3.65944311e-06]\n",
            "  [9.96041656e-01 3.34106553e-05 1.93706667e-03 1.98790408e-03]\n",
            "  [9.99998927e-01 4.29099565e-08 5.03310275e-07 5.34517540e-07]\n",
            "  [2.02154368e-02 1.52997075e-07 9.37768042e-01 4.20163870e-02]\n",
            "  [9.99996424e-01 1.41062026e-06 5.03485360e-08 2.04490266e-06]\n",
            "  [1.43395984e-04 2.79922745e-07 9.02037416e-03 9.90836024e-01]\n",
            "  [5.29304714e-07 9.99999523e-01 4.12607548e-09 4.75940309e-09]]\n",
            "\n",
            " [[1.31641813e-02 2.81963759e-04 9.64363873e-01 2.21898593e-02]\n",
            "  [3.83767883e-06 5.74698674e-07 9.99598682e-01 3.96982359e-04]\n",
            "  [1.21665956e-03 7.62719594e-07 1.38670742e-03 9.97395873e-01]\n",
            "  [9.97108640e-04 9.98958468e-01 1.15788571e-05 3.28228598e-05]\n",
            "  [9.99836922e-01 3.81813843e-05 4.67499922e-05 7.81691560e-05]\n",
            "  [3.30120698e-02 1.19707562e-01 8.10387253e-04 8.46469998e-01]\n",
            "  [9.99975324e-01 1.65174594e-08 1.55994570e-07 2.44749463e-05]\n",
            "  [1.77008528e-02 9.81723547e-01 4.05417522e-04 1.70190324e-04]\n",
            "  [1.83919880e-07 9.99999642e-01 3.56373109e-09 7.03110175e-08]\n",
            "  [9.94425297e-01 1.77870097e-05 1.12313055e-03 4.43372782e-03]\n",
            "  [1.12004846e-05 9.99842405e-01 1.38962831e-04 7.44273211e-06]\n",
            "  [1.98028865e-04 1.29230457e-07 5.99009404e-03 9.93811727e-01]\n",
            "  [9.66858327e-01 1.50322565e-04 3.05469590e-03 2.99367085e-02]\n",
            "  [6.55888095e-02 2.74964696e-05 9.34245884e-01 1.37716852e-04]\n",
            "  [9.96156633e-01 1.24750613e-06 1.36915032e-05 3.82838468e-03]\n",
            "  [9.92682993e-01 5.16532967e-03 1.35879964e-03 7.92975887e-04]\n",
            "  [1.29348170e-02 2.29942190e-04 9.82571721e-01 4.26351419e-03]\n",
            "  [9.99980569e-01 1.22557435e-06 1.12298842e-07 1.80703119e-05]\n",
            "  [1.51574408e-04 1.19885783e-06 2.80143589e-01 7.19703555e-01]\n",
            "  [4.86388387e-07 9.99999404e-01 2.46969787e-08 8.03012128e-08]]\n",
            "\n",
            " [[1.89140264e-04 4.73091313e-06 9.86104429e-01 1.37017667e-02]\n",
            "  [3.87453707e-04 4.10755348e-07 9.98988569e-01 6.23539207e-04]\n",
            "  [4.43777861e-03 4.23854945e-05 6.53827377e-03 9.88981605e-01]\n",
            "  [1.66845799e-04 9.99826491e-01 4.81340294e-06 1.80718257e-06]\n",
            "  [9.85322475e-01 1.03113183e-03 1.27236405e-02 9.22748353e-04]\n",
            "  [3.47623415e-02 8.72967839e-02 4.65658978e-02 8.31374943e-01]\n",
            "  [9.99980330e-01 7.02439849e-08 4.35663469e-06 1.51787071e-05]\n",
            "  [5.77178458e-03 9.94060814e-01 9.64998544e-05 7.07807339e-05]\n",
            "  [3.46257038e-05 9.99951243e-01 2.53996791e-06 1.16221745e-05]\n",
            "  [9.79308844e-01 7.99601839e-05 1.23118889e-03 1.93799250e-02]\n",
            "  [1.01568541e-04 9.99637485e-01 8.87856295e-05 1.72228058e-04]\n",
            "  [6.14997771e-06 1.07696820e-07 7.87336787e-04 9.99206364e-01]\n",
            "  [9.51977849e-01 4.10581488e-05 9.72157903e-03 3.82594876e-02]\n",
            "  [6.68068305e-02 1.08290114e-05 9.32537377e-01 6.44949090e-04]\n",
            "  [9.96192217e-01 1.14848034e-03 5.08522826e-05 2.60844734e-03]\n",
            "  [9.97352123e-01 1.67174754e-03 3.65277316e-04 6.10849704e-04]\n",
            "  [7.73896603e-03 2.93504445e-06 9.80254948e-01 1.20031368e-02]\n",
            "  [9.99464810e-01 3.40499042e-04 4.33168971e-05 1.51362503e-04]\n",
            "  [1.53008048e-04 1.63738332e-05 4.75613549e-02 9.52269316e-01]\n",
            "  [3.51040899e-06 9.99996424e-01 4.65599008e-08 1.08870417e-07]]\n",
            "\n",
            " [[4.56060320e-02 4.50534455e-04 9.50891614e-01 3.05191008e-03]\n",
            "  [2.11143088e-05 7.92216781e-07 9.99910951e-01 6.70958907e-05]\n",
            "  [2.60155741e-02 1.34161237e-04 2.67077773e-03 9.71179426e-01]\n",
            "  [7.70384431e-05 9.99916673e-01 6.05216746e-06 2.59997677e-07]\n",
            "  [9.99224186e-01 1.34490094e-07 2.03098989e-05 7.55352085e-04]\n",
            "  [5.10011911e-01 3.23900208e-02 3.13450210e-02 4.26253021e-01]\n",
            "  [9.99987960e-01 3.84447274e-09 1.92366301e-08 1.20025943e-05]\n",
            "  [5.28499018e-03 9.94582236e-01 1.31141685e-04 1.63177958e-06]\n",
            "  [4.89172526e-05 9.99948740e-01 2.18290370e-06 2.50505707e-07]\n",
            "  [5.05427837e-01 2.44760560e-03 2.25021574e-03 4.89874303e-01]\n",
            "  [3.93569382e-04 9.99478161e-01 1.27538922e-04 8.39359473e-07]\n",
            "  [1.41058015e-04 1.70648935e-08 1.84870895e-03 9.98010218e-01]\n",
            "  [9.54310834e-01 1.40845423e-05 3.98129411e-02 5.86205442e-03]\n",
            "  [7.29200989e-02 3.89114604e-04 9.26629782e-01 6.09580347e-05]\n",
            "  [9.91498172e-01 7.40246605e-07 6.73374627e-03 1.76733604e-03]\n",
            "  [9.98089492e-01 1.73214660e-03 1.26323474e-04 5.20490794e-05]\n",
            "  [1.88974792e-03 8.02060167e-05 9.93816435e-01 4.21359716e-03]\n",
            "  [9.99751151e-01 3.71783904e-06 3.80957863e-06 2.41387679e-04]\n",
            "  [1.47457875e-03 1.22631209e-05 2.25706756e-01 7.72806466e-01]\n",
            "  [2.97533279e-05 9.99970078e-01 7.24060101e-08 1.16761241e-08]]\n",
            "\n",
            " [[1.90412975e-04 8.49445423e-05 9.97659922e-01 2.06470978e-03]\n",
            "  [3.43348729e-05 3.33328728e-07 9.99636054e-01 3.29267466e-04]\n",
            "  [5.25371288e-04 4.07805807e-08 1.24994833e-02 9.86975074e-01]\n",
            "  [1.17029595e-05 9.99969721e-01 1.23878144e-05 6.25020311e-06]\n",
            "  [9.99946117e-01 3.58735122e-07 3.19648570e-05 2.15988020e-05]\n",
            "  [5.29019892e-01 3.50764334e-01 4.13605105e-03 1.16079718e-01]\n",
            "  [9.99957919e-01 3.43579654e-06 3.62331718e-07 3.82786893e-05]\n",
            "  [2.35698707e-02 9.75752890e-01 1.56922833e-04 5.20288071e-04]\n",
            "  [3.37603473e-04 9.99647737e-01 1.07216977e-06 1.36190774e-05]\n",
            "  [9.92171645e-01 7.82337395e-07 1.31781027e-03 6.50974037e-03]\n",
            "  [1.58274761e-05 9.99967813e-01 9.45953889e-06 6.83254802e-06]\n",
            "  [3.37269128e-04 1.49750401e-09 1.48878689e-03 9.98173952e-01]\n",
            "  [9.96066749e-01 1.78873123e-07 2.94014718e-03 9.92929214e-04]\n",
            "  [7.23778037e-03 2.52324880e-05 9.92717028e-01 1.99036094e-05]\n",
            "  [9.99969125e-01 1.29951604e-06 1.19836450e-05 1.75647638e-05]\n",
            "  [9.99988556e-01 9.45380089e-06 7.80986568e-07 1.20884295e-06]\n",
            "  [1.06206708e-05 4.58606280e-07 9.99734819e-01 2.54125480e-04]\n",
            "  [9.99931097e-01 1.61073913e-07 9.82894193e-08 6.86622152e-05]\n",
            "  [3.24054918e-06 5.58546231e-09 6.56801043e-03 9.93428767e-01]\n",
            "  [1.21984272e-06 9.99998331e-01 2.56709086e-07 1.98230424e-07]]\n",
            "\n",
            " [[1.29107002e-05 7.20129901e-05 9.90108013e-01 9.80697293e-03]\n",
            "  [6.60254409e-06 1.59010707e-08 9.99762833e-01 2.30466947e-04]\n",
            "  [1.15586594e-02 4.77904803e-04 1.17475986e-02 9.76215839e-01]\n",
            "  [1.25402830e-05 9.99980330e-01 3.99330384e-06 3.21383959e-06]\n",
            "  [9.99837279e-01 1.51186453e-06 1.57329632e-04 3.76124899e-06]\n",
            "  [7.20169902e-01 9.38636810e-02 9.20104212e-04 1.85046375e-01]\n",
            "  [9.99994159e-01 1.55244254e-06 1.75313528e-06 2.58988894e-06]\n",
            "  [1.30225584e-04 9.99868393e-01 1.18225432e-06 8.97850612e-08]\n",
            "  [1.35939374e-07 9.99999285e-01 6.10813089e-08 5.36868754e-07]\n",
            "  [9.53183770e-01 8.85303234e-05 6.88778528e-04 4.60389592e-02]\n",
            "  [1.31462075e-05 9.99986410e-01 2.55188525e-07 2.41387937e-07]\n",
            "  [5.60383851e-05 5.44155165e-09 4.43677825e-04 9.99500275e-01]\n",
            "  [9.94205058e-01 2.30078331e-06 4.79511917e-03 9.97627852e-04]\n",
            "  [2.38658674e-02 1.09922876e-04 9.75987315e-01 3.69235313e-05]\n",
            "  [9.99501348e-01 3.70618363e-05 1.28230720e-04 3.33328848e-04]\n",
            "  [9.99949098e-01 4.93250118e-05 9.74874638e-07 5.46376555e-07]\n",
            "  [4.60798838e-05 1.12130172e-08 9.99719679e-01 2.34205407e-04]\n",
            "  [9.99887347e-01 1.08689623e-04 1.67201131e-06 2.20613970e-06]\n",
            "  [1.77378552e-05 6.82067321e-06 7.98452869e-02 9.20130074e-01]\n",
            "  [9.00312997e-08 9.99999881e-01 1.28904070e-08 4.67369210e-09]]\n",
            "\n",
            " [[1.91883209e-05 1.47964863e-06 9.54776585e-01 4.52026986e-02]\n",
            "  [3.99434794e-06 1.54073838e-08 9.82282698e-01 1.77133065e-02]\n",
            "  [1.30109733e-03 2.73371370e-06 1.28644882e-02 9.85831618e-01]\n",
            "  [7.66443045e-05 9.99920130e-01 5.04508478e-07 2.74133458e-06]\n",
            "  [9.99764264e-01 1.38876465e-04 7.15037750e-05 2.54231491e-05]\n",
            "  [1.20877661e-02 9.81222570e-01 2.26962686e-04 6.46268995e-03]\n",
            "  [9.99996781e-01 1.71103159e-06 8.23273183e-07 7.13207612e-07]\n",
            "  [1.71265728e-03 9.98199046e-01 6.01947031e-05 2.80491477e-05]\n",
            "  [1.14818577e-04 9.99880433e-01 4.52773534e-08 4.67577138e-06]\n",
            "  [8.53575051e-01 5.15012907e-05 1.28181753e-04 1.46245226e-01]\n",
            "  [7.16891373e-05 9.99925256e-01 4.34110149e-07 2.63204083e-06]\n",
            "  [1.71668653e-05 3.36294590e-07 4.62227646e-04 9.99520302e-01]\n",
            "  [9.80977893e-01 2.91104207e-05 1.73676889e-02 1.62525859e-03]\n",
            "  [1.23168364e-01 1.32213074e-06 8.76804531e-01 2.57988340e-05]\n",
            "  [9.99940276e-01 3.00440561e-08 4.09408713e-05 1.88251943e-05]\n",
            "  [9.99827087e-01 6.17991054e-06 7.45450188e-06 1.59278512e-04]\n",
            "  [3.03454959e-04 1.82852951e-08 9.99346554e-01 3.49974056e-04]\n",
            "  [9.97925520e-01 1.56004995e-03 1.76169924e-04 3.38255748e-04]\n",
            "  [7.08665812e-06 1.98590237e-06 3.75817269e-02 9.62409198e-01]\n",
            "  [2.14770444e-06 9.99997377e-01 2.01629273e-08 4.22341003e-07]]]\n"
          ]
        }
      ],
      "source": [
        "np.random.seed(0)\n",
        "evaluate_test_list_highloss = np.random.permutation(7600)[:20]\n",
        "#evaluate_test_list_highloss = evaluate_test_list_highloss[-10:] + evaluate_test_list_highloss[10:20]\n",
        "\n",
        "print(x_test.shape)\n",
        "print(evaluate_test_list_highloss)\n",
        "x_eval = tf.convert_to_tensor(x_test[evaluate_test_list_highloss,:])\n",
        "y_eval = y_test[evaluate_test_list_highloss]\n",
        "eval_sentences_list = [test_sentences_list[i] for i in evaluate_test_list_highloss]\n",
        "\n",
        "prediction_eval = np.mean(prediction_tests[:, evaluate_test_list_highloss],0)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Hbm7Ek-MPSBo"
      },
      "source": [
        "## TracIn"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Mb6g_-BgL7fv"
      },
      "source": [
        "### Load models"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BQGUNQfZIIhv",
        "outputId": "66db6c10-7995-4421-bd0d-199f8202d47a"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"model_11\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_12 (InputLayer)       [(None, 128)]             0         \n",
            "                                                                 \n",
            " embedding_11 (Embedding)    (None, 128, 128)          3200000   \n",
            "                                                                 \n",
            " conv1d_33 (Conv1D)          (None, 124, 50)           32050     \n",
            "                                                                 \n",
            " conv1d_34 (Conv1D)          (None, 120, 50)           12550     \n",
            "                                                                 \n",
            " conv1d_35 (Conv1D)          (None, 120, 50)           2550      \n",
            "                                                                 \n",
            " global_max_pooling1d_11 (Gl  (None, 50)               0         \n",
            " obalMaxPooling1D)                                               \n",
            "                                                                 \n",
            " dense_11 (Dense)            (None, 4)                 204       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 3,247,354\n",
            "Trainable params: 3,247,354\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "None\n",
            "[3, 8]\n",
            "0\n",
            "1\n"
          ]
        }
      ],
      "source": [
        "model = create_model()\n",
        "print(model.summary())\n",
        "\n",
        "cp_list = [3,8]\n",
        "print(cp_list)\n",
        "models = []\n",
        "models_nosm = []\n",
        "models_penultimate = []\n",
        "\n",
        "for countm in range(1):\n",
        "  for count,i in enumerate(cp_list):\n",
        "    print(count)\n",
        "    model = create_model()\n",
        "    loss = tf.keras.losses.BinaryCrossentropy()\n",
        "    #best_weights_file = os.path.join('/content/gdrive/My Drive/models/', 'model.ckpt-{epoch:04d}')\n",
        "    model.load_weights(base_dir + 'checkpoint_test_222_{0:01d}_{1:02d}.hdf5'.format(countm,i+1))\n",
        "    model.compile(loss='sparse_categorical_crossentropy', optimizer='adam', metrics=[tf.keras.metrics.AUC(multi_label=False, curve=\"ROC\"),\n",
        "                                                        'accuracy'])\n",
        "    models.append(model)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fN3R4hYfR4bW",
        "outputId": "1f22b579-b999-410e-f2e2-4beeed10cc9c"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "9\n",
            "0 (25000, 128)\n",
            "1 (5, 128, 50)\n",
            "2 (50,)\n",
            "3 (5, 50, 50)\n",
            "4 (50,)\n",
            "5 (1, 50, 50)\n",
            "6 (50,)\n",
            "7 (50, 4)\n",
            "8 (4,)\n"
          ]
        }
      ],
      "source": [
        "print(len(model.trainable_variables))\n",
        "for count,layer in enumerate(model.trainable_variables):\n",
        "  print(count, layer.shape)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "E5aST9R7s908",
        "outputId": "84a21aca-61de-4036-e902-91d2243b94f0"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[  101  2572  2094 ...     0     0     0]\n",
            " [  101 26665  1011 ...     0     0     0]\n",
            " [  101  2343  5747 ...     0     0     0]\n",
            " ...\n",
            " [  101 18952  3619 ...     0     0     0]\n",
            " [  101  2149  2390 ...     0     0     0]\n",
            " [  101  4811  4484 ...     0     0     0]]\n"
          ]
        }
      ],
      "source": [
        "print(x_train)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tDr_e--lt5sf"
      },
      "source": [
        "### tracin Util"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 51,
      "metadata": {
        "id": "kQ4LgmIgeAXD"
      },
      "outputs": [],
      "source": [
        "tracin_evaluate = []\n",
        "tracin_train = []\n",
        "\n",
        "def gather_flat_grad(grads):\n",
        "    views = []\n",
        "    for p in grads:\n",
        "      view = tf.reshape(p,[-1])\n",
        "      views.append(view)\n",
        "    if len(views)>0:\n",
        "      return tf.concat(views, 0)\n",
        "    else:\n",
        "      return views\n",
        "\n",
        "@tf.function\n",
        "def run_withfirstgrad_pred1(X_train1, y_train1, layers='all'):\n",
        "  if layers == 'all' or 0 in layers:\n",
        "    we = True\n",
        "  else:\n",
        "    we = False\n",
        "  ind = []\n",
        "  val = []\n",
        "  grads = []\n",
        "  for i, ml in enumerate(models):\n",
        "    with tf.GradientTape(watch_accessed_variables=False,persistent=False) as tape1:\n",
        "      if layers == 'all':\n",
        "        w = ml.trainable_variables\n",
        "      else:\n",
        "        w = [ml.trainable_variables[layer] for layer in layers]\n",
        "      tape1.watch(w)\n",
        "      prediction = ml(X_train1)\n",
        "      loss = tf.keras.losses.SparseCategoricalCrossentropy(from_logits=False)(y_train1[0], prediction)\n",
        "      #print(prediction,y_train1[0])\n",
        "      #print(loss)\n",
        "    grad = tape1.gradient(loss,w)\n",
        "    if we:\n",
        "      grads.append(gather_flat_grad(grad[1:]))\n",
        "      ind.append(grad[0].indices)\n",
        "      val.append(grad[0].values)\n",
        "    else:\n",
        "      grads.append(gather_flat_grad(grad))\n",
        "  return tf.concat(grads, 0), ind, val\n",
        "\n",
        "def get_tracin_grad(x_full, y_full, layers='all', top=False):\n",
        "  batchsize= 1\n",
        "  grads = []\n",
        "  inds = []\n",
        "  vals = []\n",
        "  inds_10 = []\n",
        "  vals_10 = []\n",
        "  for i in range(x_full.shape[0]):\n",
        "    if i%1000 ==0:\n",
        "      print(i)  \n",
        "    x = x_full[i:i+1]\n",
        "    y = y_full[i:i+1]\n",
        "    grad, ind, val = run_withfirstgrad_pred1(x, y, layers)\n",
        "    grad = grad.numpy()\n",
        "    ind_reshape = np.array(ind)\n",
        "    val_reshape = np.array(val)\n",
        "    grads.append([grad])\n",
        "    if ((layers == 'all') or (0 in layers)):\n",
        "      model_size = len(models)\n",
        "      ind_repeat = np.array([[i*128]*10 for i in range(model_size)])\n",
        "      val_norm = np.linalg.norm(val_reshape,axis=2)\n",
        "      top_10_ind_loc = val_norm.argsort(axis=1)[:,-10:]\n",
        "      inds.append(ind_reshape)\n",
        "      vals.append(val_reshape)\n",
        "      top_10_ind_loc += ind_repeat\n",
        "      inds_10.append(ind_reshape.flatten()[top_10_ind_loc])\n",
        "      vals_10.append(val_reshape.reshape(-1,128)[top_10_ind_loc])\n",
        "\n",
        "  print('grad shape', np.concatenate(grads).shape)\n",
        "  if  top:\n",
        "    return {\n",
        "            'grads': np.concatenate(grads),\n",
        "            'inds': np.array(0),\n",
        "            'vals': np.array(0),\n",
        "            'inds_10': np.array(inds_10),\n",
        "            'vals_10': np.array(vals_10),\n",
        "            } \n",
        "  elif (layers=='all') or (0 in layers):\n",
        "    return {\n",
        "            'grads': np.concatenate(grads),\n",
        "            'inds': np.array(inds),\n",
        "            'vals': np.array(vals),\n",
        "            'inds_10': np.array(0),\n",
        "            'vals_10': np.array(0),\n",
        "            } \n",
        "  else:\n",
        "    return {\n",
        "            'grads': np.concatenate(grads),\n",
        "            'inds': np.array(0),\n",
        "            'vals': np.array(0),\n",
        "            'inds_10': np.array(0),\n",
        "            'vals_10': np.array(0),\n",
        "            } \n",
        "\n",
        "def sum_indexvalue(ind1s, val1s, ind2s, val2s):\n",
        "  sum=0.\n",
        "  max_list = [(-1,0), (-1,0), (-1,0), (-1,0), (-1,0)]\n",
        "  for m in range(ind1s.shape[0]):\n",
        "    ind1 = ind1s[m]\n",
        "    ind2 = ind2s[m]\n",
        "    if m==0:\n",
        "      indexset1 = set(ind1)\n",
        "      indexset2 = set(ind2)\n",
        "    else:\n",
        "      indexset1 = indexset1.union(set(ind1))\n",
        "      indexset2 = indexset2.union(set(ind2))\n",
        "  intersect = indexset1.intersection(indexset2)\n",
        "  for ind in intersect:\n",
        "    add = 0\n",
        "    for m in range(ind1s.shape[0]):\n",
        "      ind1 = ind1s[m]\n",
        "      ind2 = ind2s[m]\n",
        "      val1 = val1s[m]\n",
        "      val2 = val2s[m]\n",
        "      #if (ind !=0 and ind!= 101 and ind!=102):\n",
        "      #  sum+= np.sum(np.sum(tensor1[index1==ind],axis=0) * \\\n",
        "      add += np.sum(np.sum(val1[ind1==ind],axis=0) * \\\n",
        "          np.sum(val2[ind2==ind],axis=0))\n",
        "    max_list.append((np.abs(add),ind))\n",
        "    max_list.sort()\n",
        "    max_list.reverse()\n",
        "    if len(max_list)>5:\n",
        "      max_list.pop(-1)\n",
        "    sum+= add\n",
        "  \n",
        "\n",
        "  return sum, [item[1] for item in max_list]\n",
        "\n",
        "\n",
        "def get_tracin_sum(grad2, ind2, val2, grad1, ind1, val1, we):\n",
        "  if not we:\n",
        "    return np.sum(grad2*grad1), 0\n",
        "  elif grad2 and grad1:\n",
        "    val, max_ind = sum_indexvalue(ind1, val1, ind2, val2)\n",
        "    return np.sum(grad2*grad1)+val, max_ind\n",
        "  else:\n",
        "    val, max_ind = sum_indexvalue(ind1, val1, ind2, val2)\n",
        "    return val, max_ind\n",
        "\n",
        "def get_tracin_list(tracin_eval, tracin_train, layers='all', top = False):\n",
        "  if layers == 'all' or 0 in layers:\n",
        "    we = True\n",
        "  else:\n",
        "    we = False\n",
        "  sim_array = np.zeros((tracin_eval['grads'].shape[0], tracin_train['grads'].shape[0]))\n",
        "  ind_array = np.zeros((tracin_eval['grads'].shape[0], tracin_train['grads'].shape[0], 5))\n",
        "  for i in range(tracin_eval['grads'].shape[0]):\n",
        "    for j in range(tracin_train['grads'].shape[0]):\n",
        "      if we:\n",
        "        if top:\n",
        "          val, ind = get_tracin_sum(tracin_eval['grads'][i], tracin_eval['inds_10'][i], tracin_eval['vals_10'][i], tracin_train['grads'][j], tracin_train['inds_10'][j], tracin_train['vals_10'][j], we)\n",
        "        else:\n",
        "          val, ind = get_tracin_sum(tracin_eval['grads'][i], tracin_eval['inds'][i], tracin_eval['vals'][i], tracin_train['grads'][j], tracin_train['inds'][j], tracin_train['vals'][j], we)\n",
        "      else:\n",
        "          val, ind = get_tracin_sum(tracin_eval['grads'][i], 0,0,tracin_train['grads'][j],0,0, we)\n",
        "      sim_array[i,j] = val\n",
        "      ind_array[i,j,:] = ind\n",
        "\n",
        "  return sim_array, ind_array"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(tracin_evaluate['inds_10'].shape)\n",
        "print(tracin_evaluate['vals_10'].shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IP3LWOolNHT0",
        "outputId": "9b8f5f42-342f-4a00-97dc-7342035d5209"
      },
      "execution_count": 48,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(20, 2, 10)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ev9KQe2dC3S8"
      },
      "source": [
        "### TracIn firsttop\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 55,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "WqPDJx3yC1ws",
        "outputId": "dd8cb89e-0894-48de-8990-5db95854a2f3"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0\n",
            "grad shape (20, 0)\n",
            "0\n",
            "1000\n",
            "2000\n",
            "3000\n",
            "4000\n",
            "5000\n",
            "6000\n",
            "7000\n",
            "8000\n",
            "9000\n",
            "10000\n",
            "11000\n",
            "12000\n",
            "13000\n",
            "14000\n",
            "15000\n",
            "16000\n",
            "17000\n",
            "18000\n",
            "19000\n",
            "20000\n",
            "21000\n",
            "22000\n",
            "23000\n",
            "24000\n",
            "25000\n",
            "26000\n",
            "27000\n",
            "28000\n",
            "29000\n",
            "30000\n",
            "31000\n",
            "32000\n",
            "33000\n",
            "34000\n",
            "35000\n",
            "36000\n",
            "37000\n",
            "38000\n",
            "39000\n",
            "40000\n",
            "41000\n",
            "42000\n",
            "43000\n",
            "44000\n",
            "45000\n",
            "46000\n",
            "47000\n",
            "48000\n",
            "49000\n",
            "grad shape (50000, 0)\n",
            "preparation time on saving gradient: 131.29706478118896\n",
            "(20, 0)\n",
            "()\n",
            "()\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py:153: DeprecationWarning: The truth value of an empty array is ambiguous. Returning False, but in future this will result in an error. Use `array.size > 0` to check that an array is not empty.\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "time on calculating TracIn for 10 test * 50000 training: 60.36470437049866\n"
          ]
        }
      ],
      "source": [
        "tracin_evaluate = []\n",
        "tracin_train = []\n",
        "method = 'tracin_first_top10'\n",
        "tracin_evaluate = get_tracin_grad(x_eval, y_eval, [0], top=True)\n",
        "start = time.time()\n",
        "tracin_train = get_tracin_grad(x_train, y_train, [0], top=True)\n",
        "print('preparation time on saving gradient: {}'.format(time.time()-start))\n",
        "print(tracin_evaluate['grads'].shape)\n",
        "print(tracin_evaluate['inds'].shape)\n",
        "print(tracin_evaluate['vals'].shape)\n",
        "\n",
        "\n",
        "start = time.time()\n",
        "evaluate_train_score, evaluate_train_ind = get_tracin_list(tracin_evaluate, tracin_train, [0], top=True)\n",
        "print('time on calculating TracIn for 10 test * 50000 training: {}'.format(time.time()-start))\n",
        "np.save(base_dir + 'evaluate_train_score_testtest2_' + method + '.npy', evaluate_train_score)\n",
        "np.save(base_dir + 'evaluate_train_ind_testtest2_' + method + '.npy', evaluate_train_ind)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 58,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YS2J2r7SCE_y",
        "outputId": "3c53c0ec-74d8-4e44-d8d7-0b82074ff17b"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 2, prediction: [8.57603663e-06 9.99986172e-01 1.31222350e-06 3.96825226e-06]\n",
            "Retail sales slid in August as people steered away from buying cars and shoppers kept a close eye on their spending after splurging in July.\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [0],predicted_scores: [0], influence: -17.930963918566704\n",
            "The Passion of the Christ sells 2.4 million copies on DVD in its first few hours on sale in the US.\n",
            "top key words\n",
            "[['in'], ['on']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -16.542261362075806\n",
            "Ducks and geese coated in crude oil were carried to a national wildlife refuge yesterday by volunteers trying to save them from the largest oil spill on the Delaware River in nearly a decade.\n",
            "top key words\n",
            "[['in'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -14.66664457321167\n",
            "A businessman is sentenced to one year in jail in Japan for making a pirated film available on the internet.\n",
            "top key words\n",
            "[['in'], ['on']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -14.335768043994904\n",
            "US Dollar ended transactions in Cuba on Monday under an order by the authorities. From Monday dollar transactions in stores, hotels, restaurants and other business, which have \n",
            "top key words\n",
            "[['in'], ['on']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -12.195590436458588\n",
            "A relentless rise in the cost of employee health insurance has become a significant factor in the employment slump.\n",
            "top key words\n",
            "[['in'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -11.948430132120848\n",
            "A video posted Monday on a Web site showed the beheading of a man identified as American construction contractor Eugene Armstrong.\n",
            "top key words\n",
            "[['a'], ['as']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -11.7904521683231\n",
            "A ban on hunting critically endangered black rhinoceros has been lifted by the Convention on International Trade in Endangered Species, The Telegraph reported.\n",
            "top key words\n",
            "[['on'], ['in']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -11.625226054340601\n",
            "Applications for asylum in the UK rose by 9 in the last three months, according to official figures.\n",
            "top key words\n",
            "[['in'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -11.378149330615997\n",
            "Canada #39;s dollar surged past 83 US cents to a new 12-year high as a higher-than-forecast jobs report added to speculation the central bank needs to keep lifting its target interest rate.\n",
            "top key words\n",
            "[['as'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -11.205657303333282\n",
            "The trial starts of 10 suspects in a failed attempt to blow up a Christmas market in Strasbourg in 2000.\n",
            "top key words\n",
            "[['in'], ['a']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [2],predicted_scores: [2], influence: 18.096508026123047\n",
            "Two people were in custody Sunday following a shootout between two groups of people that left two men dead and a third man wounded, police said Sunday.\n",
            "top key words\n",
            "[['people'], ['in']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 16.660677686333656\n",
            "PHILADELPHIA -- A lawyer who was a central figure in an ongoing FBI investigation into city corruption died Thursday after a hospitalization of several days, his attorney said.\n",
            "top key words\n",
            "[['in'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 15.366823320277035\n",
            "LONDON - People in England and Wales can now discover at the click of a mouse whether they are living in areas liable to flooding, the Environment Agency says.\n",
            "top key words\n",
            "[['in'], ['people']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 13.518566608428955\n",
            "Bridgestone Firestone North American Tire launched a voluntary safety campaign Thursday to replace 27,000 tires, the same model as those involved in accidents that injured more than a dozen people and killed one.\n",
            "top key words\n",
            "[['people'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 13.143534252420068\n",
            "1. The toilets. What is that thing, you say? Well, its not a urinal, but rather what is known in English as a squat-toilet. It might seem hard to believe, but many people in Japan prefer using these floor holes to a standard Western toilet.\n",
            "top key words\n",
            "[['in'], ['people']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 13.05210460908711\n",
            "Reuters - Even the King of Beasts needs a hand\\staying alive in his own domain.\n",
            "top key words\n",
            "[['in'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 12.858000203967094\n",
            "When Gabrielle Wentworth was heading toward a new life in Seattle, she knew the job situation was tight, even for college graduates.\n",
            "top key words\n",
            "[['in'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 12.64328440465033\n",
            "Reuters - A volcano in central Japan spewed smoke\\into the sky as it erupted for a second straight day on\\Wednesday, but there were no immediate signs to suggest a major\\eruption was coming.\n",
            "top key words\n",
            "[['in'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 12.392805099487305\n",
            "Ray Brent Marsh, who is to enter the plea Friday, had faced up to 8,000 years in a case that shocked the nation two years ago when investigators found hundreds of rotting corpses stacked in storage sheds and scattered in woods outside his rural northwest \n",
            "top key words\n",
            "[['in'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 12.147090817277785\n",
            "Are cracks appearing in the Abbey-Santander romance? Certainly, the engaged couple do not seem as close as they were a few days ago.\n",
            "top key words\n",
            "[['as'], ['close']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 2, prediction: [9.99956608e-01 1.31112465e-05 2.81144585e-05 2.11895394e-06]\n",
            "The fallout from allegations of serious accounting problems at Fannie Mae has rattled investors and could even bump up mortgage rates down the road.\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [3],predicted_scores: [3], influence: -0.4158756136894226\n",
            "The European Commission has asked four more countries for information on how they buy computers. Rival AMD applauds the move.\n",
            "top key words\n",
            "[['has'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.38998787850141525\n",
            "MARK Latham attacked yesterday #39;s Coalition policy launch as a  quot;willy-nilly quot; spending spree aimed at buying votes, which would drive up interest rates.\n",
            "top key words\n",
            "[['rates'], ['up']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.388095922768116\n",
            "Tokyo, Nov. 15: The news that Princess Sayako, will wed at 35 after years of hesitation was welcomed today by many in Japan who fear that sagging marriage and birth rates has left the country headed for a demographic crisis.\n",
            "top key words\n",
            "[['rates'], ['has']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.3069351315498352\n",
            "Women are being infected with HIV at increasing rates in all regions of the world, and their numbers are now nearly equal to men, according to the United Nations and World Health Organization #39;s annual report on AIDS released Tuesday.\n",
            "top key words\n",
            "[['rates'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -0.2981923818588257\n",
            "The European Union has launched a 45 million euro (\\$60 million) plan to protect children from pornography and racist sites when they surf the Internet.\n",
            "top key words\n",
            "[['has'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -0.2932087332010269\n",
            "BRUSSELS -- The European Commission has confirmed that it wants to reach a decision on Oracle Corp.s hostile bid for PeopleSoft Inc. before the end of October, prompting speculation that it will approve the merger.\n",
            "top key words\n",
            "[['has'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.2913189586251974\n",
            "Australian Prime Minister John Howard is seeking a fourth term in power at today #39;s election, promising to extend 13 years of economic growth and keep mortgage rates low.\n",
            "top key words\n",
            "[['rates'], ['mortgage']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -0.29049743711948395\n",
            "A federal judge has ordered a permanent injunction against a Rhode Island man after he allegedly sent millions of unsolicited short text messages to Verizon Wireless customers.\n",
            "top key words\n",
            "[['has'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -0.28608478233218193\n",
            "TAMPA  Ever since Alonzo Mourning #39;s devastating revelation of kidney illness in September 2000, the Heat has been more than aware of life without a dominant big man.\n",
            "top key words\n",
            "[['has'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.27985143661499023\n",
            "AS Coalition strategists enjoyed a cold beer and the warm glow of a fourth term on Saturday night, the  quot;spin quot; over the election result had already begun -- it was about more than just interest rates.\n",
            "top key words\n",
            "[['rates'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [2],predicted_scores: [2], influence: 0.3468615710735321\n",
            "Honeywell announced late Wednesday it has filed a lawsuit against Apple, and 33 other electronics companies claiming claiming infringement of a liquid crystal display (LCD) patent.\n",
            "top key words\n",
            "[['has'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 0.3299623727798462\n",
            "A federal judge has ordered a man known as the  quot;Spam King quot; to disable so-called spyware programs that infiltrate people #39;s computers, track their Internet use and flood them with pop-up advertising.\n",
            "top key words\n",
            "[['has'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 0.3289223276078701\n",
            "Wireless carrier Sprint has signed contracts with Lucent, Nortel and Motorola for network upgrades, including the development of high-speed Internet services.\n",
            "top key words\n",
            "[['has'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 0.30208466202020645\n",
            "MUMBAI: It has been a long-running drama and it is Japanese electronics giant Sony Corp that has come out the winner at the end of it.\n",
            "top key words\n",
            "[['has'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 0.3017679490149021\n",
            "The European Union has asked Russia to explain the bloody end to the school siege, which resulted in a huge loss of life. In a statement in the name of the presidency of the 25-nation EU, Dutch Foreign Minister \n",
            "top key words\n",
            "[['has'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 0.28177787363529205\n",
            "Industrial conglomerate Honeywell International on Wednesday said it has filed a lawsuit against 34 electronics companies including Apple Computer and Eastman Kodak, claiming patent infringement of its liquid crystal display technology.\n",
            "top key words\n",
            "[['has'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 0.25467200204730034\n",
            "AP - It has been singled out as the second-leading cause of lung cancer in the United States, right behind smoking. But to Shiro Umeda, sprightly at 74, radon is the best thing since aspirin.\n",
            "top key words\n",
            "[['has'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 0.2492830455303192\n",
            "The success of a start-up magazine rests on many factors, not the least of which is a catchy name.\n",
            "top key words\n",
            "[['up'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 0.2460844786837697\n",
            "With the economy slowly turning up, upgrading hardware has been on businesses radar in the past 12 months as their number two priority.\n",
            "top key words\n",
            "[['up'], ['has']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 0.24311134219169617\n",
            "Microsoft Corp. has put the finishing touches on one of its biggest software overhauls ever after conceding that its Windows XP operating system -- which runs as many as 300 million computers worldwide -- is susceptible to security breaches.\n",
            "top key words\n",
            "[['has'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 3, prediction: [3.41259397e-06 3.46163489e-07 8.18924978e-03 9.91806924e-01]\n",
            "McData plans to introduce a new SAN router this week designed to connect the growing number of isolated SAN networks in corporations.&lt;p&gt;ADVERTISEMENT&lt;/p&gt;&lt;p&gt;&lt;img src=\"http://ad.doubleclick.net/ad/idg.us.ifw.general/sbcspotrssfeed;sz=1x1;ord=200301151450?\" width=\"1\" height=\"1\" border=\"0\"/&gt;&lt;a href=\"http://ad.doubleclick.net/clk;9228975;9651165;a?http://www.infoworld.com/spotlights/sbc/main.html?lpid0103035400730000idlp\"&gt;SBC Case Study: Crate   Barrel&lt;/a&gt;&lt;br/&gt;What sold them on improving their network? A system that could cut management costs from the get-go. Find out more.&lt;/p&gt;\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [0],predicted_scores: [0], influence: -24.04317808151245\n",
            "Scientists are hoping that \"playing\" a tiny guitar string will be able to help weigh molecules to discover more about them.\n",
            "top key words\n",
            "[['\"'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -23.447646230459213\n",
            "Actor William Shatner is shown in this undated publicity photo. (AP Photo/www.williamshatner.com). LONDON Oct 22, 2004 -  quot;Star Trek quot; star William Shatner and Red Hot Chili Peppers drummer Dave Navarro are \n",
            "top key words\n",
            "[['.'], ['/']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -21.3724308013916\n",
            "(08/30/04)-- A house fire in Oakland County #39;s Waterford Township has claimed the lives of a township police officer and his wife and child.\n",
            "top key words\n",
            "[['/'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -21.350353240966797\n",
            "TSA will collect domestic passenger records for June 2004; data will be used to test program aiming to hunt for terrorists. &lt;FONT face=\"verdana,MS Sans Serif,arial,helvetica\" size=\"-2\" color=\"#666666\"&gt;&lt;B&gt;-The Washington Post&lt;/b&gt;&lt;/font&gt;\n",
            "top key words\n",
            "[['\"'], ['=']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -21.299105525016785\n",
            "Grunting tennis pro. Evil brat. Marie Antoinette. What kind of roles are these for a young, blond Hollywood ingnue?\n",
            "top key words\n",
            "[['.'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -20.913507521152496\n",
            "If you had to ask what \"Qualia\" is, you probably weren't supposed to know about it.\n",
            "top key words\n",
            "[['\"'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -20.334396738559008\n",
            "The Hang Seng Index slipped 15.65 points, or 0.1 percent, to 13,886.16. On Friday, the index had slumped 107.01 points, or 0.8 percent.\n",
            "top key words\n",
            "[['.'], ['0']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -19.72606086730957\n",
            "Taipei plans to have a citywide connection by 2005. Also: MIT creates a voice operated plane that takes audio commands   hellip;.  Basketball players use video games to hone their skills   hellip;. and more.\n",
            "top key words\n",
            "[['.'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -17.275440603494644\n",
            "David Gardner launches our Foolanthropy campaign for 2004.\" /&gt;  &lt;meta name=\"keywords\" content=\"Foolanthropy Charity giving donate\" /&gt;  &lt;meta name=\"tickers\" content=\"\" /&gt;  &lt;meta name=\"date\" content=\"2004-10-15T15:33-04:00\" /&gt;  &lt;meta name=\"language\" content=\"en-us\" /&gt;  &lt;meta name=\"author\" content=\"David Gardner\" /&gt;  &lt;meta name=\"STORY_UID\" content=\"/USMF_Content/datasource/production2003/2004/foolanthropy/2004/foolanthropy041015.xml\" /&gt;  &lt;meta name=\"expiration\" content=\"Date  12 months\" /&gt;  &lt;meta name=\"articletype\" content=\"Foolanthropy\" /&gt;  &lt;meta name=\"contenttype\" content=\"a\" /&gt;   &lt;title&gt;Fool.com: \\$2,137,000 and Counting Foolanthropy&lt;/title&gt;    &lt;link rel=\"stylesheet\" type=\"text/css\" href=\"http://g.fool.com/includes/css/ie4.css\" title=\"ie4\" /&gt;    &lt;META Name=\"ArticleDate\" Content=\"2004/10/15\n",
            "top key words\n",
            "[['\"'], ['/']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -17.21932029724121\n",
            "The nation has always been independent from China. Not for a single day has it been ruled by China. Twenty-three million Taiwanese -- 7.7 times the number of \n",
            "top key words\n",
            "[['.'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [3],predicted_scores: [3], influence: 23.625885248184204\n",
            "During the nineties, our state saw an increase in population, due to \"The Californians\" (who we like to bitch about) migrating to our low tax/low smog state.  Though most of us natives like to bitch about their supposedly shallow, urban sprawl, rude soccer mom, cell phone, and shopping mall ways, they have done something useful and actually turned us backwards hillbilly, gun toting, Republican loving rednecks into a swing state. Now we have to deal with traffic on I-25 being snarled for hours whenever Dick Cheney decides to send his motorcade through rush hour.  The most important issue on the ballot in our state is Amendment 36, which would change our electoral voting from \"winner takes all\" to a proportional voting system.  \n",
            "top key words\n",
            "[['/'], ['\"']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 22.82476234436035\n",
            "JAY, Vt. -- From the top of the gondola building, 4,000 feet up on the summit of Jay Peak, the rolling hills of the Northeast Kingdom fade into the distance of hazy summer air. The forestland of this mountain gives way to a patchwork of fields, barns, and silos, with white steeples marking a village green. At the base, brown ...\n",
            "top key words\n",
            "[['.'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 22.416658878326416\n",
            "DreamWorks Animation Inc., riding on a wave of recent hit films including \"Shrek 2\" and \"Shark Tales\" was looking for friendly waters on Wall Street Thursday with its highly anticipated initial public offering.\n",
            "top key words\n",
            "[['\"'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 20.387184619903564\n",
            "STANFORD, CALIF. -- On Monday, the same day the California Senate passed a bill that would ban state agencies from contracting services out to companies that use overseas labor, opponents and proponents of offshore outsourcing clashed during a conference at Stanford University.&lt;p&gt;ADVERTISEMENT&lt;/p&gt;&lt;p&gt;&lt;img src=\"http://ad.doubleclick.net/ad/idg.us.ifw.general/ibmpseries;sz=1x1;ord=200301151450?\" width=\"1\" height=\"1\" border=\"0\"/&gt;&lt;a href=\"http://ad.doubleclick.net/clk;9824455;9690404;u?http://ad.doubleclick.net/clk;9473681;9688522;d?http://www.ibm.com/servers/eserver/pseries/campaigns/boardroom/index.html?ca=pSeries met=boardroom me=E P_Creative=P_InfoW_RSS\"&gt;Introducing IBM eServer p5 systems.&lt;/a&gt;&lt;br/&gt;Powered by IBMs most advanced 64-bit microprocessor (POWER5(tm)), p5 systems can run UNIX and Linux simultaneously. Learn more about eServer p5 systems.&lt;/p&gt;\n",
            "top key words\n",
            "[['general'], ['/']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 19.150104522705078\n",
            "AP - The webmasters got their wish. Florida finally decided to fireronzook.\n",
            "top key words\n",
            "[['.'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 17.28245973587036\n",
            "On Tuesday, millions of Americans will decide whether George W. Bush or John F. Kerry will be president of the United States for the next four years.\n",
            "top key words\n",
            "[['.'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 16.995624542236328\n",
            "While \"passive-aggressive\" has become an all-purpose label for almost any difficult character, it is a controversial concept in psychiatry.\n",
            "top key words\n",
            "[['\"'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 16.494765482842922\n",
            "Cicadas, a possible flooded future, and the discovery of a hobbit-like human were the most popular National Geographic News stories in 2004. Read the full list and \"vote\" by clicking on the stories.\n",
            "top key words\n",
            "[['\"'], ['.']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 16.13264301419258\n",
            "SAN FRANCISCO - As Sun Microsystems Inc. and Microsoft Corp. ready an early December unveiling of the first step in their highly publicized technology collaboration, industry observers are questioning whether the former foes will fail to address industry standards, particularly the work being done by the Sun-backed Liberty Alliance consortium.&lt;p&gt;ADVERTISEMENT&lt;/p&gt;&lt;p&gt;&lt;img src=\"http://ad.doubleclick.net/ad/idg.us.ifw.general/sbcspotrssfeed;sz=1x1;ord=200301151450?\" width=\"1\" height=\"1\" border=\"0\"/&gt;&lt;a href=\"http://ad.doubleclick.net/clk;11714255;9651165;g?http://infoworld.com/spotlights/sbc/main.html?lpid0101035400730403idlp\"&gt;SBC Datacomm White Paper&lt;/a&gt;&lt;br/&gt;Find out how Crate   Barrel expects to save \\$180,000 by moving to VoIP, compared to a traditional PBX.&lt;/p&gt;\n",
            "top key words\n",
            "[['/'], ['general']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 15.519453048706055\n",
            "ATHENS -- The US women #39;s trio of 100-meter sprinters, including five-time Olympian Gail Devers, all raced into the semifinals with strong, controlled performances yesterday in the debut of the track and ...\n",
            "top key words\n",
            "[['.'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 1, prediction: [1.98782422e-02 9.80081975e-01 1.70397652e-05 2.27062046e-05]\n",
            "Leon Washington ran for 104 yards and a touchdown and Florida State sacked Alabama-Birmingham #39;s Darrell Hackney eight times Saturday night to rebound from a disappointing loss to Miami with a 34-7 victory.\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [0],predicted_scores: [0], influence: -1.0051934598013759\n",
            "Sourav Ganguly files an appeal against a two-match ban imposed for time-wasting.\n",
            "top key words\n",
            "[['-'], ['for']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.7878651209175587\n",
            "South Africa A complete a seven-wicket victory against England in their final warm-up before the first Test.\n",
            "top key words\n",
            "[['a'], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.7819044440984726\n",
            "Europe go into the singles needing three-and-a-half points to win the Ryder Cup.\n",
            "top key words\n",
            "[['-'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.6870392663404346\n",
            "Reuters - It has only been four months\\since Stephanie Herseth became the first woman ever elected to\\the House of Representatives from South Dakota, but the\\prominence that win gave her may make the difference as she\\faces the same opponent again for a full two-year term.\n",
            "top key words\n",
            "[['for'], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.6337199211120605\n",
            "The first phase of the nationwide sit-at-home protest against the current increase in the prices of petroleum products will last four days -- Monday to Thursday.\n",
            "top key words\n",
            "[['-'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -0.616242378950119\n",
            "When Aroon Roy Padharia gathers today with family and friends for Thanksgiving dinner, the cuisine will be authentically Indian, with the same preparation methods and recipes that he loves from his native land.\n",
            "top key words\n",
            "[['and'], ['for']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.603452205657959\n",
            "AFP - England coach Sven-Goran Eriksson threw a three-pronged attack into Saturday's World Cup qualifier against Wales at Old Trafford, with Wayne Rooney in behind Michael Owen and Jermain Defoe.\n",
            "top key words\n",
            "[['-'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.5943562306347303\n",
            "The nation has always been independent from China. Not for a single day has it been ruled by China. Twenty-three million Taiwanese -- 7.7 times the number of \n",
            "top key words\n",
            "[['-'], ['7']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.5915703615173697\n",
            "New Zealand captain Stephen Fleming was offered 200,000 to join a match-fixing syndicate in 1999.\n",
            "top key words\n",
            "[['-'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -0.5794070598785765\n",
            "American tycoon Malcolm Glazer is reportedly poised to launch a takeover bid for Manchester United.\n",
            "top key words\n",
            "[['for'], ['a']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [1],predicted_scores: [1], influence: 1.0309289880096912\n",
            "Luck is a relative term for Mario Lemieux, who has endured back operations, hip operations, cancer, a premature retirement and a team bankruptcy.\n",
            "top key words\n",
            "[['and'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.6879368275403976\n",
            "THE guy exudes an aura of sure-fire confidence - the confidence that is based on his burning desire to excel. Shaikh Ahmed Mohammed Hasher Al Maktoum expects to be successful - and he invariably is. \n",
            "top key words\n",
            "[['and'], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.6523694833740592\n",
            "Muhammad Ali asked Congress yesterday to create a US Boxing Commission, saying oversight by the federal government is needed to protect boxers from exploitation and injury. Ali's testimony before a congressional panel was read by his wife, Lonnie Ali , because he suffers from Parkinson's disease.  quot;Reform measures are unlikely to succeed, quot; Ali said,  quot;unless a US Boxing Commission is ...\n",
            "top key words\n",
            "[['and'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.6162785657215863\n",
            "Caminiti died from acute intoxication due to the combined effects of cocaine and opiates, a spokeswoman for the New York City Medical Examiners Office told ESPN.\n",
            "top key words\n",
            "[['for'], ['and']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.5932039432227612\n",
            "The Gunners are delighted with their 100 million deal with the Dubai-based airline this week. But The Sun says Arsenal could have banked a total of 123m by separating the naming rights and shirt deal.\n",
            "top key words\n",
            "[['and'], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.5787112675607204\n",
            "On Tuesday, millions of Americans will decide whether George W. Bush or John F. Kerry will be president of the United States for the next four years.\n",
            "top key words\n",
            "[['for'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.5761093497276306\n",
            "Gavin Reynolds (Rochester, NY) 10/20/04 - Satellite radio was launched just two years go and already 2.5 million people subscribe.\n",
            "top key words\n",
            "[['and'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.575698774860939\n",
            "In the wee hours yesterday morning, promoter Don King stood at a podium inside Madison Square Garden and -- in the midst of a bombastic assault on his \n",
            "top key words\n",
            "[['and'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.565741408849135\n",
            "A group of Marine wives are running for the family of a Marine Corps officer who was killed in Iraq.\n",
            "top key words\n",
            "[['for'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.553159000352025\n",
            "USC: Reggie Bush is all that. It #39;s fitting that The President, as he #39;s nicknamed, chose the nation #39;s capital as the venue for his national coming-out party.\n",
            "top key words\n",
            "[['for'], ['-']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 0, prediction: [8.61599028e-01 5.96145128e-06 3.45460576e-04 1.38049632e-01]\n",
            "The European Union's decision to hold entry talks with Turkey receives a widespread welcome.\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [2],predicted_scores: [2], influence: -39.333360471297055\n",
            "The European Commission's former chief accountant, who said the EU budget was open to fraud, will fight her dismissal.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -35.236502254381776\n",
            "The European Union discussed proposals on Friday to help stabilize Iraq and organize elections there despite lingering differences among member states over the US-led war that ousted Saddam Hussein.\n",
            "top key words\n",
            "[['union'], ['european']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -33.97208881378174\n",
            "HOPKINTON -- If there's one room in the house that's his, Ron Sargent says it's the home office.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -31.375441074837\n",
            "Charles M. Vest, who will be stepping down next month as the president of MIT, personifies his school's famed ''engineering mentality. quot; Simply put, this is the inclination to find a problem and fix it.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -30.847001993097365\n",
            "Component problems meant Brillian's new big screens missed the NFL's kickoff party.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -30.55147389275953\n",
            "Coach joins the S P 500, and others stand to benefit from the leather in the weather.\n",
            "top key words\n",
            "[['s'], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -29.17387291789055\n",
            "AP - China's economic boom is still roaring despite efforts to cool sizzling growth, with the gross domestic product climbing 9.5 percent in the first three quarters of this year, the government reported Friday.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -26.774603674188256\n",
            "AFP - Britain's manufacturing industries experienced an unexpected drop in output in August, the first time since January 2002 that output has fallen for three months in a row.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -24.287745472043753\n",
            "Marvel's movies mean nothing to the bottom line without the right talent behind them.\n",
            "top key words\n",
            "[['s'], [\"'\"]]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -24.275877952575684\n",
            " BAGHDAD (Reuters) - Authorities kept a main oil pipeline in  southern Iraq shut on Sunday rather than risk it being  attacked, restricting the country's exports to half normal  levels, a South Oil Official said.\n",
            "top key words\n",
            "[['s'], [\"'\"]]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [0],predicted_scores: [0], influence: 41.41004591528326\n",
            "The European Union will begin its largest-ever peacekeeping operation on Thursday when it takes over from the Nato mission to Bosnia Hercegovina.\n",
            "top key words\n",
            "[['union'], ['european']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 33.38860278436914\n",
            "Small net firms warn they could be hit hard by BT's decision to raise prices for business broadband.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 32.097576815634966\n",
            "The European Union says it will to redouble its efforts to avert an epidemic of HIV/Aids across Europe.\n",
            "top key words\n",
            "[['union'], ['european']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 25.41851537534967\n",
            "Has a close relationship to the railroad industry dulled the Federal Railroad Administration's enforcement edge?\n",
            "top key words\n",
            "[[\"'\"], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 24.659145331243053\n",
            "WASHINGTON - Consumer prices rose just 0.1 percent in August, a sign that inflation poses no immediate risk to the economy.    The over-the-month-increase in the Consumer Price Index, the government's most closely watched inflation gauge, came after prices dipped by 0.1 percent in July, the Labor Department reported Thursday...\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 23.342768418602645\n",
            "In August, 1.2 million users visited the Patriots' website, www.patriots.com. Chances are, those who returned to the team's Internet home this week noticed only minor cosmetic changes, not the complete overhaul that has been done behind the scenes over the past seven months.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 23.107393488520756\n",
            "David Blunkett is set to back plans to work closely with other European Union states on asylum and immigration.\n",
            "top key words\n",
            "[['union'], ['european']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 22.593225037679076\n",
            "The European Union and China agreed to boost relations yesterday, but the EU made clear there can be no early lifting of its 15-year-old \n",
            "top key words\n",
            "[['union'], ['european']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 22.332589149475098\n",
            "Manchester United is to ask the Takeover Panel to clarify the company #39;s position regarding a potential takeover approach by US tycoon Malcolm Glazer.\n",
            "top key words\n",
            "[['to'], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 21.747611986473203\n",
            "HOUSTON (AP) -- Ivan's second foray into the United States came with little wind but plenty of the rain that became the 3-week-old system's calling card as it raked the Caribbean and eastern United States, while Floridians braced for another possible pounding as Hurricane Jeanne appeared to be gearing up for a weekend landing.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 3, prediction: [2.33804429e-04 5.62337561e-07 1.70764960e-02 9.82689142e-01]\n",
            "Several Minnesota police departments are field testing a handheld device that scans a suspect's fingerprint and digitally checks it against Minnesota's criminal history and fingerprint database.\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [1],predicted_scores: [1], influence: -261.3653144836426\n",
            "The rules seem to be the problem. The South Koreans are upset that the rules are too strict. The Americans are upset that the rules are being broken.\n",
            "top key words\n",
            "[['are'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -176.515531539917\n",
            "A group of Marine wives are running for the family of a Marine Corps officer who was killed in Iraq.\n",
            "top key words\n",
            "[['are'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -135.62760257720947\n",
            "Scientists are hoping that \"playing\" a tiny guitar string will be able to help weigh molecules to discover more about them.\n",
            "top key words\n",
            "[['are'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -129.3610801100731\n",
            "Howard's 0-2 start and the threat of rain from Hurricane Ivan have slowed ticket sales for Saturday's Prince George's Classic at FedEx Field, organizers said Tuesday.\n",
            "top key words\n",
            "[['field'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -125.91657161712646\n",
            "Online criminals are exploiting human hopes and fears to spread spam and viruses.\n",
            "top key words\n",
            "[['are'], ['and']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -114.93587493896484\n",
            "com. There are two weeks left in Octoberquest, and most of the attention today will be on a doubleheader in Florida that was never on the original schedule.\n",
            "top key words\n",
            "[['are'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -114.40446257591248\n",
            "The way cancer services are provided is a shambles and must be drastically reformed, a report by MPs says.\n",
            "top key words\n",
            "[['are'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -111.26926285028458\n",
            "Targeted attacks and a sluggish economy are pushing academics, Christians, and businessmen to move abroad.\n",
            "top key words\n",
            "[['are'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -108.15243339538574\n",
            "The new Liverpool are already looking remarkably similar to the previous Liverpool - right down to the scowl on Steven Gerrard #39;s face.\n",
            "top key words\n",
            "[['are'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -101.70275777578354\n",
            "Scientists are developing a device which could improve the lives of kidney dialysis patients.\n",
            "top key words\n",
            "[['a'], ['are']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [2],predicted_scores: [2], influence: 151.1771936416626\n",
            "The Lemon Bay Manta Rays were not going to let a hurricane get in the way of football. On Friday, they headed to the practice field for the first time in eight \n",
            "top key words\n",
            "[['field'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 146.89349460601807\n",
            "The Red Sox are the newest champions to grace the cover of a Wheaties box. But with what the team was paid for the deal, it couldn't even buy a box of Wheaties.\n",
            "top key words\n",
            "[['are'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 131.959228515625\n",
            "Reuters - These days Iranian women are not even\\allowed to watch men compete on the football field, but 2,000\\years ago they could have been carving the boys to pieces on\\the battlefield.\n",
            "top key words\n",
            "[['field'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 122.43327236175537\n",
            "CHICAGO - Like most of us, politicians don #39;t know much about isotopes. But because they know something about money and jobs, lawmakers from Illinois and Michigan are locked in a battle to convince the federal \n",
            "top key words\n",
            "[['are'], ['and']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 122.01340651512146\n",
            "Despite a record hurricane season, many Florida hot spots are relatively unscathed. And tourist bureaus, hotels, and others are pulling out all the stops to lure travelers back.\n",
            "top key words\n",
            "[['and'], ['are']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 108.94049453735352\n",
            "The evolution of a physique for long-distance running is what made humans look the way we do now -- whether winning a marathon, nursing a strained Achilles #39; tendon \n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 108.06662845611572\n",
            "AP - There is a good chance that the colorful carp you won at the county fair or picked up at the local pet shop came from Pool Fisheries, which sits in the middle of an Arkansas field in Lonoke County, the self-proclaimed baitfish capital of the world where more than 6 billion minnows and goldfish are bred each year.\n",
            "top key words\n",
            "[['field'], ['and']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 107.514200091362\n",
            "Bridgestone Firestone North American Tire launched a voluntary safety campaign Thursday to replace 27,000 tires, the same model as those involved in accidents that injured more than a dozen people and killed one.\n",
            "top key words\n",
            "[['a'], ['and']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 105.54267501831055\n",
            "As little girls, Yueqi  quot;Lucie quot; Guo and Xianlin Li dreamed of ending hunger, finding world peace and discovering a cure for cancer.\n",
            "top key words\n",
            "[['and'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 103.86315870285034\n",
            "This may be a calumny on Leesburg, Virginia, but somehow it doesn #39;t sound like the sort of place we would associate with a liberal judiciary.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 0, prediction: [3.64642688e-06 9.99996185e-01 6.63865691e-08 3.64009978e-08]\n",
            "A former Palestinian security minister who could be key to keeping order among rival factions in Gaza after an Israeli pullout held a fence-mending meeting Monday with President Yasser Arafat, officials said.\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [3],predicted_scores: [3], influence: -0.030418470763834193\n",
            "Comparing the state of security in 1994 versus 2004, has anything really changed over the course of ten years?\n",
            "top key words\n",
            "[['security'], ['in']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -0.02768912329338491\n",
            "AFP - President George W. Bush and Democratic rival John Kerry fought over who is the most wasteful spender in their televised election debate here.\n",
            "top key words\n",
            "[['rival'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -0.025006722222315148\n",
            "The Hague, Nov. 7: India did not expect any hiccups in its relationship with the US after the re-election of George W. Bush to the presidency, Prime Minister Manmohan Singh said here today.\n",
            "top key words\n",
            "[['in'], ['minister']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -0.024066922065685503\n",
            "Canada is interested in investing in Russia, the countrys prime minister, Paul Martin, said on Tuesday. Russia and Canada are two great northern countries rich in natural resources, Interfax news agency quoted him as saying.\n",
            "top key words\n",
            "[['in'], ['minister']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -0.02239430695772171\n",
            "Flute maker David C. Williams is known among the world's finest musicians for his artistry. But for 15 long years, he found it difficult to balance the expense of buying the materials he needs to craft his instruments against the unpredictability in orders for his handiwork.\n",
            "top key words\n",
            "[['among'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -0.021644405205734074\n",
            "WASHINGTON - Voting security advocates in the U.S. are bracing for a repeat of problems in the upcoming general election that could rival Florida during the 2000 presidential race.\n",
            "top key words\n",
            "[['security'], ['in']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -0.021144791622646153\n",
            "Despite a barrage of complaints, the Transportation Security Administration will continue to require physical pat-downs for some passengers at security checkpoints.\n",
            "top key words\n",
            "[['security'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -0.020608197897672653\n",
            "Ray Brent Marsh, who is to enter the plea Friday, had faced up to 8,000 years in a case that shocked the nation two years ago when investigators found hundreds of rotting corpses stacked in storage sheds and scattered in woods outside his rural northwest \n",
            "top key words\n",
            "[['in'], ['who']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -0.01962008490227163\n",
            "Toshiba has accused Hynix of infringing its DRAM and Flash patents, and has begun legal proceedings against its South Korean rival in the US and Japanese courts.\n",
            "top key words\n",
            "[['rival'], ['in']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -0.01828562506216258\n",
            "Contract with Department of Defense is part of homeland security boom.\n",
            "top key words\n",
            "[['security'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [1],predicted_scores: [1], influence: 0.02570471167564392\n",
            "Ducks and geese coated in crude oil were carried to a national wildlife refuge yesterday by volunteers trying to save them from the largest oil spill on the Delaware River in nearly a decade.\n",
            "top key words\n",
            "[['in'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.018701359629631042\n",
            "NEW DELHI : Top Indian and Pakistani officials approved proposals to break the impasse over Kashmir and strengthen ties on the eve of a meeting between the rival states #39; foreign ministers.\n",
            "top key words\n",
            "[['rival'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.016409412026405334\n",
            "In GAZA story headlined  quot;CNN producer seized by armed men in Gaza quot;, please read name in paragraph 3 as ... Ben Wedeman ... instead of ... Ben Wedemen ... and make conform throughout). By Nidal al-Mughrabi. GAZA \n",
            "top key words\n",
            "[['in'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.01596401259303093\n",
            "Athletes are marinated in preparation for years, and grilled in harsh evaluation or simply bask in recognition for decades. And during competition, each competitor can handle no more than the prescribed circumstances.\n",
            "top key words\n",
            "[['in'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.014629530131060164\n",
            "Instead of paying social security taxes and wondering whether the government will ever deliver promised retirement benefits, Chilean workers bankroll their own retirements and manage their nest eggs.\n",
            "top key words\n",
            "[['security'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.014020413160324097\n",
            "The colossal sign that is draped around one end of the American Airlines Arena is a rather imposing spectacle, to say the least. Theres little doubt that Dwyane Wade has achieved star status in south Florida.\n",
            "top key words\n",
            "[['in'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 0.01400986235239543\n",
            "Will Juniper Networks, Ciscos chief rival in the router market, make an entry into ethernet switching, and try to take yet another bite out of Ciscos business?\n",
            "top key words\n",
            "[['rival'], ['in']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.01381292287260294\n",
            "Applications for asylum in the UK rose by 9 in the last three months, according to official figures.\n",
            "top key words\n",
            "[['in'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.013264099368825555\n",
            "ISLAMABAD: The US government would not only maintain continuity in its policy towards Islamabad but also strive to enhance bilateral relations in economic, commercial, defence and security spheres, US Deputy Secretary of State Richard Armitage assured the \n",
            "top key words\n",
            "[['security'], ['in']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.012967976479558274\n",
            "The proposed legislation would break up the C.I.A. and divide its responsibilities among three new spy agencies.\n",
            "top key words\n",
            "[['among'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 1, prediction: [8.31022975e-04 9.99164462e-01 4.40380425e-08 4.46268496e-06]\n",
            "ESTONIA #39;s Markko Martin took the Rally of Catalonia lead today after newly-crowned world champion Sebastien Loeb, the overnight leader, was forced out of the race with a severe oil leak.\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [0],predicted_scores: [0], influence: -28.515435561537743\n",
            "Valentino Rossi is crowned world champion with victory at the Australian MotoGP.\n",
            "top key words\n",
            "[['champion'], ['world']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -19.324989888817072\n",
            "Reuters - A row over Indian TV rights, which threatens a test series next month with world champion Australia, is heading back to court after jilted broadcaster Zee Telefilms filed a petition in the Supreme Court on Wednesday.\n",
            "top key words\n",
            "[['champion'], ['world']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -17.590267956256866\n",
            "NORTH COUNTY ---- For many, it was like waking from a bad dream ---- and then discovering it was not a nightmare, but real, and that it was just beginning.\n",
            "top key words\n",
            "[['-'], [',']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -16.916219979524612\n",
            "Former super flyweight boxing champion Robert Quiroga has been stabbed to death in San Antonio, Texas.\n",
            "top key words\n",
            "[['champion'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -16.42599105834961\n",
            "The nation has always been independent from China. Not for a single day has it been ruled by China. Twenty-three million Taiwanese -- 7.7 times the number of \n",
            "top key words\n",
            "[['-'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -16.22093802690506\n",
            "Francoise Sagan, the rebellious French writer who won world fame as a teenager with Bonjour Tristesse, her 1954 bestseller about infidelity among the idle rich, has died of heart and lung failure at 69.\n",
            "top key words\n",
            "[['world'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -16.21757936477661\n",
            "So a few weeks ago, the two astronauts who live there tossed out some useless junk, like so many old hubcaps for the trash heap. And the throwing-away -- done \n",
            "top key words\n",
            "[['-'], [',']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -15.663779133930802\n",
            "Hicham El Guerrouj shocks world record holder Kenenisa Bekele to win the 5,000m.\n",
            "top key words\n",
            "[['world'], [',']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -14.752455711364746\n",
            "Sixth-seeded Guillermo Canas of Argentina beat wild card Julian Knowle of Austria 6-2, 6-4 Monday on the first day of the CA Trophy.\n",
            "top key words\n",
            "[['-'], [',']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -14.535448968410492\n",
            "&lt;p&gt;&lt;/p&gt;&lt;p&gt; By Barry Wood&lt;/p&gt;&lt;p&gt; VIENNA (Reuters) - Double Olympic champion Nicolas Massubattled for nearly three hours before overcoming Slovakqualifier Karol Kucera 6-4, 6-7, 6-4 Tuesday to claim his firstvictory in four attempts at the Vienna Trophy.&lt;/p&gt;\n",
            "top key words\n",
            "[['champion'], ['-']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [1],predicted_scores: [1], influence: 18.89185729622841\n",
            "ASPEN, COLO. - Figuring she had no chance of catching the leader, Finland #39;s Tanja Poutiainen charged down the hill without fear.\n",
            "top key words\n",
            "[['-'], ['leader']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 17.398149490356445\n",
            "ATHENS, Greece -- It was death by a thousand cuts. If you can describe taking silver instead of gold as  quot;death. quot; ...\n",
            "top key words\n",
            "[['-'], [',']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 16.610822275280952\n",
            "NAGOYA, Japan (AP-CP) - World champion Shizuka Arakawa of Japan overcame a case of jitters and a new pair of skates Friday to win the women #39;s short program at the NHK Trophy figure skating competition.\n",
            "top key words\n",
            "[['champion'], ['world']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 16.390033721923828\n",
            "SYRACUSE, NY - The world #39;s oldest man, according to Guinness World Records, has died less than two weeks before he would have turned 114.\n",
            "top key words\n",
            "[['world'], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 14.050442427396774\n",
            "COLLEGE STATION -- For the third year in a row, a group of current and former students continued the suspended Texas A amp;M University bonfire tradition off-campus.\n",
            "top key words\n",
            "[['-'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 13.2367480173707\n",
            "The coaches of high-performance athletes will benefit the most from the federal government #39;s announcement yesterday of how it will divide up \\$19.\n",
            "top key words\n",
            "[['-'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 12.65460979938507\n",
            "Digital technology has now replaced the old photo-finish cameras at Britain #39;s racecourses, but still there are times when even a computer cannot help.\n",
            "top key words\n",
            "[['-'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 12.17194452881813\n",
            "In cargo pants and a gray T-shirt, flanked by a couple of friends, Andy Roddick could have been a frat boy on the way to the movies.\n",
            "top key words\n",
            "[['-'], [',']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 11.644868647679687\n",
            "AP - Harvie Ward, a two-time U.S. Amateur champion and a top teaching pro whose clients included the late Payne Stewart, has died after a long illness.\n",
            "top key words\n",
            "[['champion'], [',']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 11.272650718688965\n",
            "SACRAMENTO, Calif. -- Although it was more competitive than most of their other trips to the West Coast in recent years, the result was the same.\n",
            "top key words\n",
            "[['-'], [',']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 1, prediction: [9.99986768e-01 1.00191005e-06 2.59937178e-06 9.64176979e-06]\n",
            "Javy Lopez drove in four runs, Daniel Cabrera became the first rookie to win 10 games this season, and the Baltimore Orioles held the Tampa Bay Devil Rays to two hits in an 8-0 victory.\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [0],predicted_scores: [0], influence: -0.08092516055330634\n",
            "Fan violence is worse -- and much more common -- at soccer \\games in Europe\n",
            "top key words\n",
            "[['games'], ['in']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.07526361003692728\n",
            "England are reduced to 95-4 in reply to South Africa's 441 in the third Test in Cape Town.\n",
            "top key words\n",
            "[['in'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.07433934323489666\n",
            "At least seven school children die when their bus crashed as they were travelling to the Paralympics Games in Athens.\n",
            "top key words\n",
            "[['games'], ['in']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -0.07043638222967274\n",
            "hen Donald J. Trump kicks off the second season of his hit reality television show  quot;The Apprentice quot; this Thursday evening, reality may be in short supply.\n",
            "top key words\n",
            "[['season'], ['in']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.060093265390605666\n",
            "King Abdullah II stripped his half brother and heir apparent of his title as crown prince in an abrupt shake-up Sunday aimed at redeeming the full power the king inherited from his late father.\n",
            "top key words\n",
            "[['in'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.05985444411635399\n",
            "Arrests at football games are down by 10 as more fans are subjected to banning orders, the Home Office says.\n",
            "top key words\n",
            "[['games'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.0580536148045212\n",
            "Reuters - Intense competition has depressed prices of everything from washing machines to refrigerators in India, dampening the spirits, and profits, of durable goods makers as they prepare for the busy festival season.\n",
            "top key words\n",
            "[['in'], ['season']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -0.057601820677518845\n",
            "A cable channel plans to resurrect each of the 1,230 regular-season games listed on the league's defunct 2004-2005 schedule by setting them in motion on a video game console.\n",
            "top key words\n",
            "[['games'], ['season']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.057314854580909014\n",
            "Credit card companies are changing the terms of their accounts at a historically high rate, costing Americans millions of dollars in fees that they did not expect.\n",
            "top key words\n",
            "[['in'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.05532846786081791\n",
            "A man is due in court accused of the murder of 14-year-old schoolgirl Danielle Beccan in a drive-by shooting. The teenager died after being shot yards from her home in the St Ann #39;s district of Nottingham as \n",
            "top key words\n",
            "[['in'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [1],predicted_scores: [1], influence: 0.08141490910202265\n",
            "In GAZA story headlined  quot;CNN producer seized by armed men in Gaza quot;, please read name in paragraph 3 as ... Ben Wedeman ... instead of ... Ben Wedemen ... and make conform throughout). By Nidal al-Mughrabi. GAZA \n",
            "top key words\n",
            "[['in'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.07731050252914429\n",
            "OLYMPIA  Amid the olive trees and just past the ancient ruins, on a hillside bathed in an unrelenting Mediterranean sun, 15,000 spectators from around the world converged on this little town to watch 12 men and 12 women compete in the shot put at the ...\n",
            "top key words\n",
            "[['in'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.07547869718200673\n",
            "Athletes are marinated in preparation for years, and grilled in harsh evaluation or simply bask in recognition for decades. And during competition, each competitor can handle no more than the prescribed circumstances.\n",
            "top key words\n",
            "[['in'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.06883451342582703\n",
            "Insurgents detonated three car bombs near a US military convoy in southern Baghdad on Thursday, killing at least 35 people and wounding around 120, many of them children, officials and doctors said.\n",
            "top key words\n",
            "[['in'], [',']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.06082708388566971\n",
            "The very pleasant Games most people experienced, superimposed on one of the world #39;s historic cities. The very expensive Games that Greeks will have to underwrite for decades.\n",
            "top key words\n",
            "[['games'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.05873467028141022\n",
            "Back in the days when the Washington Senators were losers, which was often, there was a saying:  quot;First in war, first in peace, and last in the American League.\n",
            "top key words\n",
            "[['in'], [',']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.057910255185561255\n",
            "Ducks and geese coated in crude oil were carried to a national wildlife refuge yesterday by volunteers trying to save them from the largest oil spill on the Delaware River in nearly a decade.\n",
            "top key words\n",
            "[['in'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.05596534942742437\n",
            "Instead of the fireworks and all night partying that usually accompanies such an event, it was a scaled back closing ceremony for the Athens Paralympics in tribute to seven teenagers killed in a bus crash on their way to the Games.\n",
            "top key words\n",
            "[['in'], ['games']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 0.05541239632293582\n",
            "Two people were in custody Sunday following a shootout between two groups of people that left two men dead and a third man wounded, police said Sunday.\n",
            "top key words\n",
            "[['in'], [',']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 0.05471119633875787\n",
            "A police officer in this Detroit suburb died with his wife and son in a Sunday morning house fire about the same time that two young children died in a blaze in Detroit, police and fire officials said.\n",
            "top key words\n",
            "[['in'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 0, prediction: [5.82023582e-04 1.10451708e-06 9.98848915e-01 5.67990413e-04]\n",
            "SAN FRANCISCO - California's constitution permits laws against gay marriage, the state's attorney general declared Friday in a long-awaited legal opinion that sought to avoid offending either side of the debate.    While acknowledging that \"committed and loving relationships between two individuals deserve recognition under California law,\" Attorney General Bill Lockyer said it was up to the voters or the Legislature to decide questions about whether gay couples should be allowed to marry...\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [3],predicted_scores: [3], influence: -43.62222743034363\n",
            "NORTH COUNTY ---- For many, it was like waking from a bad dream ---- and then discovering it was not a nightmare, but real, and that it was just beginning.\n",
            "top key words\n",
            "[['-'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -23.35471072047949\n",
            "So a few weeks ago, the two astronauts who live there tossed out some useless junk, like so many old hubcaps for the trash heap. And the throwing-away -- done \n",
            "top key words\n",
            "[['-'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -22.687966100871563\n",
            "AFP - Britain's manufacturing industries experienced an unexpected drop in output in August, the first time since January 2002 that output has fallen for three months in a row.\n",
            "top key words\n",
            "[[\"'\"], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -21.83906841278076\n",
            "Reuters - The world's whales, porpoises and\\dolphins have no standing to sue President Bush over the U.S.\n",
            "top key words\n",
            "[['-'], [\"'\"]]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -20.654954731464386\n",
            "AFP - Florida's orange production is expected to plunge 27 percent from a year ago in the wake of four hurricanes that lashed the state in August and September, the US Agriculture Department said.\n",
            "top key words\n",
            "[['-'], [\"'\"]]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -20.549907684326172\n",
            "WASHINGTON -- A flight from Paris bound for Dulles International Airport was diverted Saturday night after US Customs officials detained two passengers aboard, officials said.\n",
            "top key words\n",
            "[['-'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -20.315059572458267\n",
            "AP - China's economic boom is still roaring despite efforts to cool sizzling growth, with the gross domestic product climbing 9.5 percent in the first three quarters of this year, the government reported Friday.\n",
            "top key words\n",
            "[[\"'\"], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -19.447313707321882\n",
            "AFP - Japan's Mitsubishi Electric Corp. said that it plans to wind down production of large-size liquid crystal display (LCD) panels to focus on smaller panels because of price competition.\n",
            "top key words\n",
            "[['-'], [\"'\"]]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -19.16408747434616\n",
            "MOJAVE, California -- Nobody can claim that Burt Rutan, the innovative aerospace designer, doesn #39;t have his head in the cloudsand his eyes focused on the stars.\n",
            "top key words\n",
            "[['-'], ['california']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -19.066347524523735\n",
            "Reuters - Israel's Supreme Court has\\temporarily barred the removal of thousands of tons of rubble\\from Jerusalem's most sensitive holy site in response to a\\petition by archaeologists, officials said Tuesday.\n",
            "top key words\n",
            "[[\"'\"], ['-']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [0],predicted_scores: [0], influence: 25.664305984973907\n",
            "HOUSTON (AP) -- Ivan's second foray into the United States came with little wind but plenty of the rain that became the 3-week-old system's calling card as it raked the Caribbean and eastern United States, while Floridians braced for another possible pounding as Hurricane Jeanne appeared to be gearing up for a weekend landing.\n",
            "top key words\n",
            "[[\"'\"], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 17.68287965655327\n",
            "AFP - Online theft of financial information from individuals has become one of the fastest-growing non-violent crimes in the Asia-Pacific region, technology research house IDC says.\n",
            "top key words\n",
            "[['-'], ['in']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 17.375059366226196\n",
            "COLLEGE STATION -- For the third year in a row, a group of current and former students continued the suspended Texas A amp;M University bonfire tradition off-campus.\n",
            "top key words\n",
            "[['-'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 16.27298641204834\n",
            "In August, 1.2 million users visited the Patriots' website, www.patriots.com. Chances are, those who returned to the team's Internet home this week noticed only minor cosmetic changes, not the complete overhaul that has been done behind the scenes over the past seven months.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 15.177486538887024\n",
            "The coaches of high-performance athletes will benefit the most from the federal government #39;s announcement yesterday of how it will divide up \\$19.\n",
            "top key words\n",
            "[['-'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 14.838755130767822\n",
            "DUGWAY PROVING GROUND, Utah - Scientists with tweezers picked through the twisted wreckage of a space capsule that crash-landed on Earth, hoping that microscopic clues to the evolution of the solar system weren't completely lost in Utah's salt flats.    NASA engineers were stunned Wednesday when neither parachute deployed aboard the Genesis capsule and the craft plummeted to the ground at 193 mph, breaking open like a clamshell and exposing its collection of solar atoms to contamination...\n",
            "top key words\n",
            "[['in'], [\"'\"]]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 14.116061687469482\n",
            "Reuters - Japan's Sony Corp. said on Tuesday it\\would launch a new version of its popular PlayStation 2 game\\console in the United States and Europe on Nov. 1 at a price of\\ #36;149 and 149 euros ( #36;181) respectively.\n",
            "top key words\n",
            "[[\"'\"], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 14.038906395435333\n",
            "AFP - Japan's top mobile operator NTT DoCoMo believes Europe will embrace hi-tech telephones and expects a major boost in subscribers on the continent of its i-mode Internet service.\n",
            "top key words\n",
            "[['-'], [\"'\"]]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 13.984510660171509\n",
            "Microsoft's popular Internet Explorer has a serious rival in the Firefox browser.\n",
            "top key words\n",
            "[['in'], [\"'\"]]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 13.541757225990295\n",
            "FT.com - Europe's bourses are expected to track the climb on the Nasdaq overnight and open slightly firmer. Ericsson though is likley to suffer following a disappointing update and VW will be in focus after rejecting an union pay cuts offer as insufficient.\n",
            "top key words\n",
            "[['-'], [\"'\"]]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 1, prediction: [9.99999642e-01 6.33521680e-08 1.79364918e-07 1.91470235e-08]\n",
            "JL Lewis shot a 10-under 62 for his best start ever on the PGA Tour and a two-shot lead Thursday in the Funai Classic at Disney. Lewis putted for birdie on every hole and made 11 of them on the Magnolia course to match his career-low round.\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [0],predicted_scores: [0], influence: -0.6836832985281944\n",
            "Pop producer Phil Spector is indicted on a charge that he murdered an actress at his home 19 months ago.\n",
            "top key words\n",
            "[['his'], ['at']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.645688229938969\n",
            "Ernie Els holds a two-shot lead at the WGC Championship at Mount Juliet.\n",
            "top key words\n",
            "[['at'], ['shot']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -0.599546879529953\n",
            "US sports tycoon Malcolm Glazer pays 17m to increase his stake in Manchester Utd for the second time in four days.\n",
            "top key words\n",
            "[['to'], ['his']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.5986945647746325\n",
            "Sourav Ganguly files an appeal against a two-match ban imposed for time-wasting.\n",
            "top key words\n",
            "[['match'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.5079498738050461\n",
            "Reuters - When India play Pakistan, it is a classic hockey match, their historic rivalry sucking in the whole stadium.\n",
            "top key words\n",
            "[['match'], ['classic']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.5013938248157501\n",
            "After falling on the landing of his vault, Paul Hamm put on two of the most spectacular routines of his career to win Olympic gold, the first ever by a U.S. man.\n",
            "top key words\n",
            "[['.'], ['to']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -0.4706980735063553\n",
            "Successful businessman Here, But He Hasn't Forgotten His Native El Salvador.\n",
            "top key words\n",
            "[['his'], ['.']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.46307445578986517\n",
            "Gay bishop Gene Robinson has told the BBC he regrets the pain caused by his consecration but vows not to resign.\n",
            "top key words\n",
            "[['his'], ['to']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.4588313093408942\n",
            "Grunting tennis pro. Evil brat. Marie Antoinette. What kind of roles are these for a young, blond Hollywood ingnue?\n",
            "top key words\n",
            "[['.'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -0.4532451033592224\n",
            "MSU #39;s plan to move its medical school to west Michigan is impractical and too costly, according to a report that lays out an alternative arrangement.\n",
            "top key words\n",
            "[['to'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [1],predicted_scores: [1], influence: 0.7270639799535275\n",
            "JAY, Vt. -- From the top of the gondola building, 4,000 feet up on the summit of Jay Peak, the rolling hills of the Northeast Kingdom fade into the distance of hazy summer air. The forestland of this mountain gives way to a patchwork of fields, barns, and silos, with white steeples marking a village green. At the base, brown ...\n",
            "top key words\n",
            "[['.'], ['at']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.49053907834604615\n",
            "Listen to the voice of wisdom:  quot;It's early. quot; The speaker? Marcus Banks. He appears to have learned quite a bit in a very short period.\n",
            "top key words\n",
            "[['to'], ['.']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.47781363129615784\n",
            "Manchester United is to ask the Takeover Panel to clarify the company #39;s position regarding a potential takeover approach by US tycoon Malcolm Glazer.\n",
            "top key words\n",
            "[['to'], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.47247043903917074\n",
            "A Japanese trainer paid \\$8 million Tuesday for a colt by top sire Storm Cat at the Keeneland September Yearling Sale.\n",
            "top key words\n",
            "[['at'], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.4154619453474879\n",
            "Muhammad Ali asked Congress yesterday to create a US Boxing Commission, saying oversight by the federal government is needed to protect boxers from exploitation and injury. Ali's testimony before a congressional panel was read by his wife, Lonnie Ali , because he suffers from Parkinson's disease.  quot;Reform measures are unlikely to succeed, quot; Ali said,  quot;unless a US Boxing Commission is ...\n",
            "top key words\n",
            "[['.'], ['to']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.39548384770751\n",
            "LIVERPOOL seem set to become the target of a fresh takeover bid, from the clubs fourth suitor this year. A Los Angeles-based Liverpool-born Hollywood film executive has revealed he is ready to put together a new consortium to buy the club.\n",
            "top key words\n",
            "[['.'], ['to']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.3950549238361418\n",
            "ATHENS -- The US women #39;s trio of 100-meter sprinters, including five-time Olympian Gail Devers, all raced into the semifinals with strong, controlled performances yesterday in the debut of the track and ...\n",
            "top key words\n",
            "[['.'], ['and']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.38017674162983894\n",
            "As Papa Griffith used to say every Jan. 1, \"Well, it looks like we made it through another year.\"\n",
            "top key words\n",
            "[['.'], ['to']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.37795997504144907\n",
            "On Tuesday, millions of Americans will decide whether George W. Bush or John F. Kerry will be president of the United States for the next four years.\n",
            "top key words\n",
            "[['.'], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.35687604546546936\n",
            "Today #39;s The Day. Every blade of grass has been inspected and passed fit, every weather forecast scanned. Best Mate will, short of an overnight declaration of Armageddon, make his seasonal \n",
            "top key words\n",
            "[['.'], ['his']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 3, prediction: [6.96207208e-05 5.07197808e-04 1.29193490e-04 9.99293923e-01]\n",
            "In an attempt to become more useful, US researchers are developing new smartphone software which watches users calling and usage patterns and tries to learn how best to help.\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [2],predicted_scores: [2], influence: -4.679783344268799\n",
            "Rupert Murdoch #39;s brazen attempt to relocate to dodgy Delaware is in big trouble and the world #39;s most powerful man has blinked.\n",
            "top key words\n",
            "[['to'], ['attempt']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -3.5393576435744762\n",
            "Comcast is giving Microsoft something the software giant has struggled to get on its own: an easy way to expand from the desktop computer to the home-entertainment center.\n",
            "top key words\n",
            "[['software'], ['to']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -3.336185544729233\n",
            "MSU #39;s plan to move its medical school to west Michigan is impractical and too costly, according to a report that lays out an alternative arrangement.\n",
            "top key words\n",
            "[['to'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -3.2083777487277985\n",
            "Eurostar services to London will stop at St Pancras, instead of Waterloo, when the Channel Tunnel Rail Link is finished in 2007, it has been revealed.\n",
            "top key words\n",
            "[['to'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -3.0230801701545715\n",
            "It #39;s another level of security for America Online - but users will have to pay extra for it. AOL is offering a optional log-on service that will require more than just a password to get onto the service.\n",
            "top key words\n",
            "[['to'], ['users']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -3.0103649497032166\n",
            "Malcolm Glazer is preparing a new bid to buy Manchester United, according to reports.\n",
            "top key words\n",
            "[['to'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -2.8701428826898336\n",
            "The Russian Orthodox clergy are engaged in an intense debate over providing spiritual guide to internet users.\n",
            "top key words\n",
            "[['users'], ['an']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -2.8209137693047523\n",
            "The United States win the doubles rubber to cut Spain's Davis Cup final lead to 2-1.\n",
            "top key words\n",
            "[['to'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -2.778013154864311\n",
            "While taking a back seat to software and security efforts, hardware is still a big time concern according to our survey. Organizations are taking the opportunity to consolidate their environments as they replace \n",
            "top key words\n",
            "[['software'], ['to']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -2.6856493949890137\n",
            "America Online has launched an online shopping site it says will provide an easy and fast way to compare products from thousands of online merchants.\n",
            "top key words\n",
            "[['to'], ['an']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [3],predicted_scores: [3], influence: 3.316000685095787\n",
            "China hoped to become a highly competitive producer of crocodile meat, hides and shoes, but that dream has been difficult to realize.\n",
            "top key words\n",
            "[['to'], ['become']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 3.1228718757629395\n",
            "Manchester United is to ask the Takeover Panel to clarify the company #39;s position regarding a potential takeover approach by US tycoon Malcolm Glazer.\n",
            "top key words\n",
            "[['to'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 2.5409707562066615\n",
            "Freescale Semiconductor Inc. has won a long-term contract worth more than 200 million euros (\\$246 million) to supply car components to Siemens, the companies said on Monday.\n",
            "top key words\n",
            "[['to'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 2.2796695679426193\n",
            "An Internet company has agreed to forgive \\$17 million in customers' bills as part of an agreement with the government, the Federal Trade Commission announced Monday.\n",
            "top key words\n",
            "[['to'], ['an']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 2.2688398361206055\n",
            "Using video cameras to record films in movie theaters would become a federal crime punishable by up to six years in prison under a bill passed Tuesday by the House.\n",
            "top key words\n",
            "[['to'], ['become']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 2.212234251201153\n",
            "Investor's Business Daily - Oracle   Chief Executive Larry Ellison might well raise his offer of  #36;21 per share to buy business software rival PeopleSoft  , they say.\n",
            "top key words\n",
            "[['software'], ['to']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 2.051109552383423\n",
            "Slowdown will give big companies more time to hunt down compatibility problems, the company says.\n",
            "top key words\n",
            "[['more'], ['to']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 2.0076056122779846\n",
            "Revamp adds more oomph to a lineup that has sometimes proved trying for the chipmaker.\n",
            "top key words\n",
            "[['more'], ['to']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 1.9771280325949192\n",
            "Art collectors spend millions to buy masterpieces, but the masterpiece they buy today could prove to be virtually worthless tomorrow if they aren #39;t careful.\n",
            "top key words\n",
            "[['to'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 1.9433436933904886\n",
            "AP - Rick Ankiel's return to the major leagues might not be just a September cameo.\n",
            "top key words\n",
            "[['an'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 0, prediction: [2.71067140e-03 2.52148311e-05 9.93637621e-01 3.62649909e-03]\n",
            "AP - China has reached agreement with the Association of Southeast Asian Nations, or ASEAN, on completely removing tariffs on merchandise goods by 2010 as part of a proposed free trade agreement, the Chinese Ministry of Commerce says.\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [2],predicted_scores: [2], influence: -37.99079382419586\n",
            "SHANGHAI, China -- China has reached agreement with the Association of Southeast Asian Nations, or ASEAN, on completely removing tariffs on merchandise goods by 2010 as part of a proposed free trade agreement, the Chinese Ministry of Commerce says.\n",
            "top key words\n",
            "[[','], ['agreement']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -18.941321849822998\n",
            "New Zealand #39;s  quot;friends quot; in Washington will soon make another push to have the administration consider a free trade agreement (FTA) between the United States and NZ, Prime Minister Helen Clark said yesterday.\n",
            "top key words\n",
            "[['agreement'], ['trade']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -12.43070538341999\n",
            "New South Wales, Australia, is bracing for what authorities predict will be the worst locust plague in 25 years.\n",
            "top key words\n",
            "[[','], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -11.08427493646741\n",
            "Despite a record hurricane season, many Florida hot spots are relatively unscathed. And tourist bureaus, hotels, and others are pulling out all the stops to lure travelers back.\n",
            "top key words\n",
            "[[','], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -11.054184634238482\n",
            "Red Sox executives are considering creating a hall of fame or museum at Fenway Park to help keep the fever going, and cash flowing, all year long.\n",
            "top key words\n",
            "[[','], ['of']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -10.153970718383789\n",
            "For years, Continental Europeans have chastised the British for what they regard as a lack of European credentials. The British, other Europeans complain, stand outside the \n",
            "top key words\n",
            "[[','], ['of']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -9.981851816177368\n",
            "The European Court of Justice (ECJ) today issued a decision that paves the way for the Finnish Supreme Court to block Budejovicky Budvar #39;s use of an alleged trade name in Finland.\n",
            "top key words\n",
            "[['trade'], ['of']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -9.747129678726196\n",
            "AFP - Southeast Asian economic ministers began talks to discuss ways to speed up economic integration and free trade in an effort to move closer to establishing a regional economic community.\n",
            "top key words\n",
            "[['southeast'], ['asian']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -9.06082534790039\n",
            "The busiest hurricane season in years chased tourists from the South Carolina coast last summer. But in recent weeks, as hurricane season passes, tourists are again returning.\n",
            "top key words\n",
            "[[','], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -8.581124305725098\n",
            "This may be a calumny on Leesburg, Virginia, but somehow it doesn #39;t sound like the sort of place we would associate with a liberal judiciary.\n",
            "top key words\n",
            "[[','], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [0],predicted_scores: [0], influence: 15.486079305410385\n",
            "JAKARTA: Member countries of the Association of Southeast Asian Nations (ASEAN) announced on Saturday they recognize China #39;s full market economy status.\n",
            "top key words\n",
            "[['asian'], ['nations']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 13.519356802105904\n",
            "SHANGHAI : The Association of Southeast Asian Nations (ASEAN) has recognised China as a full market economy, Chinese state press reported following a meeting of ASEAN finance ministers in Indonesia.\n",
            "top key words\n",
            "[['asian'], ['nations']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 11.961124736815691\n",
            "The North American Free Trade Agreement (Nafta) is severely affecting rural Mexico, it is claimed.\n",
            "top key words\n",
            "[['agreement'], ['trade']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 11.333404192700982\n",
            "Not enough is being done to prevent the spread of HIV among Asian drug users, researchers say.\n",
            "top key words\n",
            "[['of'], ['asian']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 8.974592447280884\n",
            "Coldplay, Travis, The Darkness and Keane get behind a new version of charity single Do They Know It's Christmas?\n",
            "top key words\n",
            "[[','], ['of']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 8.692223966121674\n",
            "The luxury market, which, only a few years ago, was struggling, has come back with a bang for the holidays.\n",
            "top key words\n",
            "[[','], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 7.963413670659065\n",
            "CHICAGO  In July of 2002, community activist Derrick Mosley blasted State Sen. James Meeks, who had been named successor to the president of the Rainbow/PUSH coalition, for offering spiritual counseling to prominent people accused of sexual offenses.\n",
            "top key words\n",
            "[[','], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 7.854467198252678\n",
            "Defendants at the center of the BALCO steroid scandal asked a court to drop charges against them Friday, claiming misconduct by prosecutors in a case which has tarnished several top sporting names.\n",
            "top key words\n",
            "[[','], ['of']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 7.837362237274647\n",
            "JAY, Vt. -- From the top of the gondola building, 4,000 feet up on the summit of Jay Peak, the rolling hills of the Northeast Kingdom fade into the distance of hazy summer air. The forestland of this mountain gives way to a patchwork of fields, barns, and silos, with white steeples marking a village green. At the base, brown ...\n",
            "top key words\n",
            "[[','], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 7.6547306180000305\n",
            "Delta Air Lines is planning to cut 7,000 jobs, or about 10 of its workforce, as the US carrier fights to avoid bankruptcy.\n",
            "top key words\n",
            "[['of'], [',']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 2, prediction: [9.99999523e-01 3.81385128e-08 2.33277163e-07 1.88439842e-07]\n",
            "Crude oil fell to the lowest price in almost two months after Iran, OPEC #39;s second-biggest oil producer, said it would stop enriching uranium to ward off US calls for sanctions.\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [1],predicted_scores: [1], influence: -6.17302131652832\n",
            "Manchester United is to ask the Takeover Panel to clarify the company #39;s position regarding a potential takeover approach by US tycoon Malcolm Glazer.\n",
            "top key words\n",
            "[['to'], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -4.367173194885254\n",
            "A new swarm of locusts reached Eilat this morning, covering sidewalks, streets, beaches, gardens and parks. Some locusts entered peoples #39; homes.\n",
            "top key words\n",
            "[[','], ['#']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -4.346613883972168\n",
            "LIVERPOOL seem set to become the target of a fresh takeover bid, from the clubs fourth suitor this year. A Los Angeles-based Liverpool-born Hollywood film executive has revealed he is ready to put together a new consortium to buy the club.\n",
            "top key words\n",
            "[['to'], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -4.336056709289551\n",
            "Police said Wednesday a woman who lives at the home found the intruder Sept. 27 and called police. A police dog tried to track the intruder but the person got away.\n",
            "top key words\n",
            "[['the'], ['to']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -4.039589166641235\n",
            "The popularity of mobiles packed with features is set to explode as they get better and cheaper, say analysts.\n",
            "top key words\n",
            "[['to'], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -3.9418212175369263\n",
            "The Bahamas and Florida are readying themselves for the latest savage hurricane to hit the region.\n",
            "top key words\n",
            "[['to'], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -3.9156723618507385\n",
            "The M.T.A., in an effort to raise more than \\$1 billion, wants to sell or lease many of its 14,000 properties.\n",
            "top key words\n",
            "[['to'], [',']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -3.876283288002014\n",
            "A transport system used by the malaria parasite to infect blood cells could be a target for new treatments, scientists hope.\n",
            "top key words\n",
            "[['to'], [',']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -3.8302170634269714\n",
            "BOSTON -- They proudly call themselves  quot;idiots, quot; and the thing with idiots, you see, is they don #39;t often think rationally.\n",
            "top key words\n",
            "[['#'], ['39']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -3.829423666000366\n",
            "Mount Asama #39;s mid-sized eruption set off a loud blast and shook a nearby town. It also sprinkled small bits on ash on downtown Tokyo.\n",
            "top key words\n",
            "[['#'], ['39']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [2],predicted_scores: [2], influence: 5.426617503166199\n",
            "MSU #39;s plan to move its medical school to west Michigan is impractical and too costly, according to a report that lays out an alternative arrangement.\n",
            "top key words\n",
            "[['to'], ['39']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 5.275737524032593\n",
            "Malcolm Glazer is preparing a new bid to buy Manchester United, according to reports.\n",
            "top key words\n",
            "[['to'], [',']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 4.761557102203369\n",
            "American tycoon Malcolm Glazer is reportedly poised to launch a takeover bid for Manchester United.\n",
            "top key words\n",
            "[['to'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 4.646621197462082\n",
            "Coach joins the S P 500, and others stand to benefit from the leather in the weather.\n",
            "top key words\n",
            "[['to'], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 4.424274325370789\n",
            "A four-day general strike in Nigeria is expected to end shortly, with neither the unions nor government giving ground.\n",
            "top key words\n",
            "[['to'], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 4.41503170132637\n",
            "Eurostar services to London will stop at St Pancras, instead of Waterloo, when the Channel Tunnel Rail Link is finished in 2007, it has been revealed.\n",
            "top key words\n",
            "[['to'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 4.157512068748474\n",
            "By the numbers. By Shari Rudavsky. Jon Steinberg, 18, takes careful aim and shoots. Red spray fills the air, and the target falls dead, as Steinberg laughs lightly and says,  quot;There #39;s the blood.\n",
            "top key words\n",
            "[[','], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 4.0388181656599045\n",
            "After two weeks of focusing on Iraq, Democratic presidential challenger John Kerry turned his emphasis to the economy Saturday, delivering what he called a plan for  quot;middle-class families.\n",
            "top key words\n",
            "[['to'], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 3.9291313886642456\n",
            "The folks in charge of the United States president #39;s re-election campaign seem to have forgotten that the first two letters of WWW stand for  quot;world wide.\n",
            "top key words\n",
            "[['to'], ['two']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 3.745840311050415\n",
            "New South Wales, Australia, is bracing for what authorities predict will be the worst locust plague in 25 years.\n",
            "top key words\n",
            "[[','], ['the']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 0, prediction: [9.99962687e-01 7.88149901e-09 8.19499746e-06 2.91492051e-05]\n",
            "BRITAIN has warned Sudan that it still has a lot of work to do to satisfy the international community that it is tackling what the United Nations has described as the worlds worst humanitarian crisis.\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [3],predicted_scores: [3], influence: -5.247641075635329\n",
            "TEHRAN (IRNA) -- Iran #39;s top nuclear official went to parliament Monday to brief its National Security and Foreign Policy Committee on the process of work on the country #39;s peaceful nuclear program.\n",
            "top key words\n",
            "[['work'], ['of']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -4.557507961988449\n",
            "An international team of astronomers has observed the perfect cosmic storm: the most powerful collision of two galactic clusters ever recorded.\n",
            "top key words\n",
            "[['of'], ['has']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -3.9061639308929443\n",
            "Reuters - Egypt has killed swarms of a type of\\locust that has infested parts of Africa and Cyprus this year,\\the official Middle East News Agency (MENA) reported on Monday.\n",
            "top key words\n",
            "[['of'], ['has']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -3.779262065887451\n",
            "It was meant to bring a little levity to a potentially uncomfortable situation, a joke that John Bunting trotted out from time to time this fall.\n",
            "top key words\n",
            "[['that'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -3.418470412492752\n",
            "The rules seem to be the problem. The South Koreans are upset that the rules are too strict. The Americans are upset that the rules are being broken.\n",
            "top key words\n",
            "[['that'], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -3.369595766067505\n",
            "AP - Thomas Handwerker leans close to a young orchid, just over from China, to inspect its deep green leaves. He knows exactly where a long, flowery spindle will sprout. He's a career horticulturist, but he doesn't know a thing about growing phalaenopsis orchids. Handwerker is counting on farmers along the eastern shore region of Delaware, Maryland and Virginia to do the work of cultivating this brood of 30,000 orchids.\n",
            "top key words\n",
            "[['work'], ['of']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -3.1907452642917633\n",
            "Europe's Organization for Nuclear Research (Cern) in Geneva celebrates 50 years of its work.\n",
            "top key words\n",
            "[['work'], ['of']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -2.869636207818985\n",
            "AP - The number of days that beaches closed or posted warnings because of pollution rose sharply in 2003 due to more rainfall, increased monitoring and tougher standards, an environmental group said on Thursday.\n",
            "top key words\n",
            "[['that'], ['of']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -2.7026582434773445\n",
            "Two people were in custody Sunday following a shootout between two groups of people that left two men dead and a third man wounded, police said Sunday.\n",
            "top key words\n",
            "[['of'], ['that']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -2.6680784337222576\n",
            "Did you feel that extra jolt of electricity in the air when you arrived at work today? Yes, it's...\n",
            "top key words\n",
            "[['work'], ['of']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [0],predicted_scores: [0], influence: 4.81939460337162\n",
            "Paris partied to celebrate the 60th anniversary of liberation from Nazi occupation yesterday, but its mayor warned that, faced with a recent surge in racist and anti-semitic attacks, the city needed to recover the courage it had shown in the summer of 1944 \n",
            "top key words\n",
            "[['that'], ['it']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 4.091793339699507\n",
            "Health experts had warned that the nation's system for vaccine supply and distribution was growing increasingly fragile.\n",
            "top key words\n",
            "[['that'], ['warned']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 3.7324398308992386\n",
            "Scientists discover how the skin of the fingertips protects itself against a common bug that causes diarrhoea.\n",
            "top key words\n",
            "[['that'], ['of']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 3.351089373230934\n",
            "Credit card companies are changing the terms of their accounts at a historically high rate, costing Americans millions of dollars in fees that they did not expect.\n",
            "top key words\n",
            "[['that'], ['of']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 3.1020573377609253\n",
            " America risks losing its leadership of the global economy if government and industry fail to make changes that encourage innovation, a panel of leading executives and university presidents say in a study to be released today.\n",
            "top key words\n",
            "[['that'], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 3.0812777280807495\n",
            "Scientists are hoping that \"playing\" a tiny guitar string will be able to help weigh molecules to discover more about them.\n",
            "top key words\n",
            "[['that'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 3.0663284175097942\n",
            "Anthony Williams said today that he was deeply pessimistic about salvaging the deal to bring the Expos to the city.\n",
            "top key words\n",
            "[['that'], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 2.754070460796356\n",
            "\\The rate of childhood cancer has slowly increased over the last three decades, research has found.\n",
            "top key words\n",
            "[['of'], ['has']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 2.668061822419986\n",
            "One of five Cuban dissidents who have been released from prison this week has been told he should consider leaving the country. The Miami Herald reported Wednesday that writer Raul \n",
            "top key words\n",
            "[['that'], ['the']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 2.5276154577732086\n",
            "Walk up St. Giles, the great piazza that runs north of St. Mary Magdalen #39;s church here, with St. John #39;s College to your right and the Ashmolean Museum to your left, and \n",
            "top key words\n",
            "[['that'], ['of']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 0, prediction: [5.63149713e-03 9.94368255e-01 2.12432980e-07 6.99745417e-09]\n",
            "Israeli jets force a flight to Tel Aviv to land in Cyprus after a bomb alert  German officials did not consider serious.\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [1],predicted_scores: [1], influence: -0.5849386512381898\n",
            "In the wee hours yesterday morning, promoter Don King stood at a podium inside Madison Square Garden and -- in the midst of a bombastic assault on his \n",
            "top key words\n",
            "[['bomb'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -0.4461700969841331\n",
            "Despite a strong profit outlook from Ford Motor Co., shares of automotive stocks moved mostly lower Friday on concerns sales for the industry might not be as strong as previously expected.\n",
            "top key words\n",
            "[['not'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -0.2625826926450827\n",
            "Three aftershocks ultimately led to officials in India issuing what amounted to a false alarm.\n",
            "top key words\n",
            "[['officials'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -0.23557478189468384\n",
            "It was meant to bring a little levity to a potentially uncomfortable situation, a joke that John Bunting trotted out from time to time this fall.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -0.23382394647342153\n",
            "ANAHEIM, Calif. -- He's making it look easy, as if he's strolling down a supermarket aisle, checking off the items on his grocery list as he lobs them into his cart. And while some of these transactions were not yet official as of yesterday afternoon, no one doubts that Theo Epstein will have enough dough when he gets to the ...\n",
            "top key words\n",
            "[['not'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -0.22946033125845133\n",
            "The notion that young people should be seen, and not heard, has never caught on within the Twins organization. Not in Spring Training.\n",
            "top key words\n",
            "[['not'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -0.22412894666194916\n",
            "Police said Wednesday a woman who lives at the home found the intruder Sept. 27 and called police. A police dog tried to track the intruder but the person got away.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -0.2127815734129399\n",
            "You could be forgiven for thinking a military coup has just taken place on the south coast of England. Sven Goran Eriksson, Rupert Lowe and some other bloke.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -0.2053285837173462\n",
            "PHILADELPHIA -- A lawyer who was a central figure in an ongoing FBI investigation into city corruption died Thursday after a hospitalization of several days, his attorney said.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -0.1952843864564784\n",
            "A spate of recent deaths has stirred fears of a broader outbreak among people and raised the possibility of a human pandemic.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [0],predicted_scores: [0], influence: 0.7748540341854095\n",
            "A bus was hijacked today and shots were fired at police who surrounded it on the outskirts of Athens. Police did not know how many passengers were aboard the bus.\n",
            "top key words\n",
            "[['not'], ['did']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.6789533770061098\n",
            "AP - Ask not why so few inaugural speeches resonate long after they are given.\n",
            "top key words\n",
            "[['not'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.5067954962141812\n",
            "South Africa A complete a seven-wicket victory against England in their final warm-up before the first Test.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.4167471663977267\n",
            "Hours after a bomb ripped through his concert at Colombo, Shah Rukh Khan, who returned to Mumbai with his team mates, said the incident will not deter him from holding similar shows.\n",
            "top key words\n",
            "[['bomb'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.3856470505706966\n",
            "It was not exactly riveting: three candidates and the representatives of 12 others sat at a horseshoe-shaped table yesterday droning \n",
            "top key words\n",
            "[['not'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 0.35120001435279846\n",
            "How much does it cost to become a lion on the social scenes in New York, London and Toronto? Not as much as one might think.\n",
            "top key words\n",
            "[['not'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.3477348680535215\n",
            "In the many smaller cities that are reliant on hockey to enliven downtowns at night, the loss of a season is a significant blow.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.33695146889658645\n",
            "Oil prices fall as it emerges that supply from Iraq will not be affected by a bomb blast and Opec's  members are already producing more oil.\n",
            "top key words\n",
            "[['bomb'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.33522131596691906\n",
            "Sachin Tendulkar hits a career-best 248 not out as India gain a lead of 342 over Bangladesh.\n",
            "top key words\n",
            "[['not'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.3271635628771037\n",
            "Hundreds of thousands of priceless antique books were feared destroyed or badly damaged yesterday by a fire that swept through a 16th century German palace.\n",
            "top key words\n",
            "[['german'], ['a']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 2, prediction: [9.62134000e-05 9.99850750e-01 7.34704145e-07 5.23017916e-05]\n",
            "Abbey National shareholders will no longer need to fill in complex Spanish tax forms if BSCH's bid to buy the UK firm succeeds. \n",
            "==================================================\n",
            "Proponents: \n",
            "label: [0],predicted_scores: [0], influence: -3.3199883699417114\n",
            "Major League Baseball will return to the nation's capital for the first time in more than three decades next season.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -3.244499146938324\n",
            "The United States win the doubles rubber to cut Spain's Davis Cup final lead to 2-1.\n",
            "top key words\n",
            "[[\"'\"], ['to']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -2.602558732032776\n",
            "Reuters - Crude oil prices near  #36;50 a barrel\\spurred an attempt by John Kerry's campaign on Friday to\\re-inject gasoline price concerns into the election-year\\debate.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -2.5291383862495422\n",
            "\\Japan's professional baseball players are set to strike for the first time since the game was introduced 70 years ago.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -2.493003726005554\n",
            "Reuters - The ball burst on impact when midfielder Walter Baseggio scored Anderlecht's equaliser in their 2-1 home win\\over La Louviere in the Belgian first division.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -2.290620744228363\n",
            "Reuters - Japan's Nikkei average fell for a fifth\\straight day to hit a one-month closing low on Wednesday as key\\exporters and techs such as Kyocera Corp. extended losses on\\renewed caution about the country's economic outlook.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -2.253581553697586\n",
            "Oil giant Yukos has its tax bill for 2001 upped by Russia's tax ministry, a move that may push it closer to bankruptcy.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -2.1840237379074097\n",
            "Ian Thorpe sets the fastest time in the 200m freestyle as swimming's top names cruise into the semi-finals.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -2.14633709192276\n",
            "Reuters - Tokyo's Nikkei fell 1.66 percent by\\midday on Monday, extending losses into a third day as another\\surge in oil prices deepened worries about the global economic\\impact and knocked down exporters such as Toyota Motor Corp.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -2.0668560415506363\n",
            "Small net firms warn they could be hit hard by BT's decision to raise prices for business broadband.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [2],predicted_scores: [2], influence: 3.0884303152561188\n",
            "AP - China's economic boom is still roaring despite efforts to cool sizzling growth, with the gross domestic product climbing 9.5 percent in the first three quarters of this year, the government reported Friday.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 2.881294459104538\n",
            "The move came two months after Dan Rather acknowledged flaws in a report on President Bush's National Guard service.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 2.6914923787117004\n",
            "AP - Brazil's economy expanded 6.1 percent in the third quarter compared to same period a year ago, lifted by exports and rising domestic consumer demand, the government said Tuesday.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 2.509894222021103\n",
            "Harrah's, \"the leader in responsible gaming,\" gets fined for mailing marketing materials to known problem gamblers.\n",
            "top key words\n",
            "[[\"'\"], ['##h']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 2.3391661643981934\n",
            "AP - Japan's Yoko Shibui overcame a run-in with an official to win the Berlin Marathon on Sunday in a race-record 2 hours, 19 minutes, 41 seconds, the fourth-best time for a woman at the distance.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 1.97842538356781\n",
            "AFP - France's seasonally-adjusted unemployment rate rose to 9.9 percent of the workforce in August from 9.8 percent in July, data from the Labour Ministry showed several.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 1.9288132190704346\n",
            "There was steam on the windows. Outside, the town was in the grip of a freeze. It was time for a break, and this looked like a good place to thaw out. It was all red and chrome, a meatloaf-shaped building on a side street off Burlington's main drag. A neon sign said  quot;Oasis, quot; and that's exactly ...\n",
            "top key words\n",
            "[[\"'\"], ['to']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 1.8829843401908875\n",
            "Spencer Abraham joined the exodus from President Bush's  Cabinet yesterday, submitting his resignation as energy secretary after four years of running a department that faced a series of high-profile challenges.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 1.8413662314414978\n",
            "Something of a curate's egg of a lash out, this...\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 1.7381061688065529\n",
            "The European Commission's former chief accountant, who said the EU budget was open to fraud, will fight her dismissal.\n",
            "top key words\n",
            "[[\"'\"], ['s']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 0, prediction: [5.56042278e-03 9.49700620e-08 5.27872192e-03 9.89160836e-01]\n",
            "Iran will pull out of the nuclear Non-Proliferation Treaty (NPT) and develop its atomic programme in secret if Western nations threaten or put pressure on Tehran, a senior Iran diplomat was quoted as saying on Wednesday.\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [1],predicted_scores: [1], influence: -0.5356353678362211\n",
            "Race car driver Tommy Baldwin was killed in a crash at Thompson International Speedway during a Featherlite modified race. \n",
            "top key words\n",
            "[['was'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -0.49392699683085084\n",
            "Reuters - Iran admitted on Tuesday that poor\\enforcement of building regulations had added to the death toll\\from the country's regular devastating earthquakes.\n",
            "top key words\n",
            "[['iran'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -0.45560914278030396\n",
            "They were greeted at Beijing airport by a senior party leader, a brass band and a traditional drum and cymbal troop, as well as crowds of well-wishers.\n",
            "top key words\n",
            "[['senior'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -0.45526186656206846\n",
            "And you thought our nation #39;s president was a polarizing figure -- at least he had 59 million supporters chanting  quot;four more years quot; on election night.\n",
            "top key words\n",
            "[['was'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -0.4229351985268295\n",
            "WASHINGTON -- A flight from Paris bound for Dulles International Airport was diverted Saturday night after US Customs officials detained two passengers aboard, officials said.\n",
            "top key words\n",
            "[['was'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -0.4019758999347687\n",
            "PHILADELPHIA -- A lawyer who was a central figure in an ongoing FBI investigation into city corruption died Thursday after a hospitalization of several days, his attorney said.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: -0.39796851575374603\n",
            "Reuters - Foreign outrage at a Norwegian plan to let\\tourists go on seal-shooting trips is mostly \"emotional\\nonsense,\" a senior official said on Monday.\n",
            "top key words\n",
            "[['senior'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -0.39232784509658813\n",
            "It was meant to bring a little levity to a potentially uncomfortable situation, a joke that John Bunting trotted out from time to time this fall.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -0.3713500223129813\n",
            "In the wee hours yesterday morning, promoter Don King stood at a podium inside Madison Square Garden and -- in the midst of a bombastic assault on his \n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: -0.3636581003665924\n",
            "Police said Wednesday a woman who lives at the home found the intruder Sept. 27 and called police. A police dog tried to track the intruder but the person got away.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [0],predicted_scores: [0], influence: 0.794440892059356\n",
            "South Africa A complete a seven-wicket victory against England in their final warm-up before the first Test.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.5082531797961565\n",
            "In the many smaller cities that are reliant on hockey to enliven downtowns at night, the loss of a season is a significant blow.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.4513306678272784\n",
            "Data from Nasa's Mars rover Opportunity shows its unique landing site is a prime spot for a return mission to look for life, scientists say.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.4248623251914978\n",
            "Europe go into the singles needing three-and-a-half points to win the Ryder Cup.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.391290314924845\n",
            "Michael Schumacher wins a record 13th race in a season at the Japanese Grand Prix.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.38519438523144345\n",
            "EUROPE #39;S three main powers have given Iran a final chance to reassure the world it is not developing atomic weapons or it will face sanctions.\n",
            "top key words\n",
            "[['iran'], ['a']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.36712689232081175\n",
            "Ask university students about the practical way to get a job these days, and the answer comes back quickly: Study Chinese.  quot;A few years ago, the most popular major was English language and literature.\n",
            "top key words\n",
            "[['was'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.36060482263565063\n",
            "England boss Sven-Goran Eriksson says a decent start to the World Cup qualifying campaign is vital.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.3544212877750397\n",
            "In the two weeks since a student reported she had been raped by two football players, Montclair has been struggling to sift through the fallout and move on.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: 0.35121538676321507\n",
            "The gloomy day provided a fitting setting for the Longhorns, who lost because of a dismal offense and their inability to stop the freshman tailback Adrian Peterson.\n",
            "top key words\n",
            "[['a'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 3, prediction: [9.76232041e-05 6.74014245e-07 3.81041527e-01 6.18860185e-01]\n",
            "SONY is going head-to-head with Nintendo in the battle for the handheld games console market. The company will price its long-awaited PlayStation Portable (PSP) at about 100 for its launch in Japan, when \n",
            "==================================================\n",
            "Proponents: \n",
            "label: [2],predicted_scores: [2], influence: -101.44879341125488\n",
            "Manfred Gerdes, SCEE German Managing Director, may have made a blunder. He recently confirmed that Sony #39;s Playstation Portable (PSP) will launch in Japan with a price tag of 33,000 yen.\n",
            "top key words\n",
            "[['ps'], ['playstation']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -89.63464546203613\n",
            "Microsoft on Wednesday reached another milestone in its multiyear Dynamic Systems Initiative when it released to manufacturing the latest version of its Microsoft Operations Manager.\n",
            "top key words\n",
            "[['its'], ['in']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -89.11169052124023\n",
            "Canadian Press - LONDON (AP) - British defence contractor BAE Systems PLC said Monday it is buying DigitalNet Holdings Inc., a U.S. maker of secure computer networks and a leading information-technology supplier to the Pentagon.\n",
            "top key words\n",
            "[['-'], ['(']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -86.87302994728088\n",
            "MARYSVILLE, Ohio - Honda has decided to build a new line of sport utility vehicles from its Acura luxury division and make them in Ohio.\n",
            "top key words\n",
            "[['its'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -76.6795768737793\n",
            "Canadian Press - PARIS (AP) - Microsoft Corp. chairman Bill Gates signed a co-operation agreement with UNESCO on Wednesday to improve access to computers, the Internet and information technology training in developing countries.\n",
            "top key words\n",
            "[['-'], ['(']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -66.13622283935547\n",
            "Microsoft has unveiled its Windows XP Media Centre Edition 2005, its media-centric computing offering, which allows users to access the usual capabilities of a Windows XP-based PC but focussed on delivering media content such as music, digital photography \n",
            "top key words\n",
            "[['its'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -66.03860855102539\n",
            "With backwards compatibility, first-party games, and a cheaper price, Nintendo DS is the portable system to beat.\n",
            "top key words\n",
            "[['price'], ['portable']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -65.81705331802368\n",
            "Harvard, Oxford and others also will go online in the first deal of its kind. By Nick Bunkley / The Detroit News. By the end of the decade, anyone around the world will have instant access to 7 million volumes \n",
            "top key words\n",
            "[['its'], ['will']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -63.32090091705322\n",
            "Microsoft is expanding its indemnification program for Windows and other software products in order to shield users from legal claims and patent disputes.\n",
            "top key words\n",
            "[['its'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -63.317413330078125\n",
            "NEW YORK - IBM is claiming unofficial bragging rights as owner of the world #39;s fastest supercomputer. IBM says its still-unfinished  quot;BlueGene System quot; can sustain speeds of 36 teraflops.\n",
            "top key words\n",
            "[['-'], ['its']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [3],predicted_scores: [3], influence: 88.38316345214844\n",
            "During his career as a merger-and-acquisitions lawyer and independent-film executive, Scott Greenstein, has often bet big.\n",
            "top key words\n",
            "[['-'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 84.4381890296936\n",
            "AP - A panel from the association that accredits the University of Illinois chastised the school Wednesday for failing to resolve the long-standing controversy over its American Indian mascot, which supporters argue is an honored tradition and others say is racially offensive.\n",
            "top key words\n",
            "[['its'], ['long']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 72.2994978427887\n",
            "US internet giant Google launches its shopping price comparison website Froogle in the UK.\n",
            "top key words\n",
            "[['its'], ['price']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 71.76600223779678\n",
            "Google Inc. (GOOG) will have plenty to celebrate at its annual company summer picnic Friday - its debut as a public company gave it an immediate cash infusion of \\$1.16 billion, not to mention all the millionaires it made of employees and insiders.\n",
            "top key words\n",
            "[['('], ['its']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 70.514404296875\n",
            "AT T Corp. said it would restructure its business, cutting 20 percent of its jobs and writing down the value of its network by \\$11.4 billion, as it winds down its residential telephone business.\n",
            "top key words\n",
            "[['its'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 67.42203336954117\n",
            "FT.com - Good call. If Nextel can pull off a merger with Sprint, it would crown its recovery from the dark days of 2002 when its shares dipped below  #36;3 - compared with  #36;30 on Thursday. It would also fill important gaps for both mobile operators.\n",
            "top key words\n",
            "[['its'], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 65.89714622497559\n",
            "PROVO, Utah - Brigham Young University revamped its athletic department Wednesday, but said the move was made after a months-long internal review rather than as a reaction to the recent sexual abuse allegations that have rocked the steadfast Mormon \n",
            "top key words\n",
            "[['its'], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 65.27135562896729\n",
            "The PlayStation Portable, or PSP, goes on sale in Japan on December 12th. Prices for the United States and Europe are still undecided, although overseas sales are \n",
            "top key words\n",
            "[['portable'], ['playstation']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 65.25926780700684\n",
            "The Wall Street rumor mill is working overtime, spinning off speculation about how soon a decision will be announced in the US government lawsuit aimed at blocking Oracle #39;s (ORCL:Nasdaq - news - research) proposed takeover of PeopleSoft (PSFT:Nasdaq \n",
            "top key words\n",
            "[['ps'], ['(']]\n",
            "--------------------------------------------------\n",
            "label: [3],predicted_scores: [3], influence: 65.2118034362793\n",
            " NEW YORK (Reuters) - China, already a manufacturing  powerhouse, is flexing its financial muscles as companies based  there snatch up famous-name products from established companies  in the United States and Europe looking to divest low-margin  businesses.\n",
            "top key words\n",
            "[['-'], ['its']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n",
            "==================================================\n",
            "Evaluate Sentence: \n",
            "label: 1, prediction: [9.98808861e-01 1.69170233e-09 6.33581309e-04 5.57606108e-04]\n",
            " KANSAS CITY, Mo., (Sports Network) - Kansas City Chiefs  quarterback Trent Green is expected to start in Sunday's game  against the Oakland Raiders despite suffering from bruises to  his ribs and hip.\n",
            "==================================================\n",
            "Proponents: \n",
            "label: [0],predicted_scores: [0], influence: -0.0037560573690598176\n",
            "HEMPSTEAD, N.Y. - Quarterback Quincy Carter signed with the New York Jets on Tuesday, three weeks after his surprising release by Dallas...\n",
            "top key words\n",
            "[['quarterback'], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.003352728001118521\n",
            "Ian Thorpe sets the fastest time in the 200m freestyle as swimming's top names cruise into the semi-finals.\n",
            "top key words\n",
            "[[\"'\"], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.0031602206497609586\n",
            "David Beckham's spectacular strike seals a comfortable 2-0 victory for England at Old Trafford.\n",
            "top key words\n",
            "[[\"'\"], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.002935545590844413\n",
            "AP - The race in Colorado's newest congressional district has turned ugly in the homestretch, with swastikas scrawled on the signs of Republican Rep. Bob Beauprez and deer entrails left on the doorstep of a Democrat.\n",
            "top key words\n",
            "[['in'], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.002833144273608923\n",
            "The nation has always been independent from China. Not for a single day has it been ruled by China. Twenty-three million Taiwanese -- 7.7 times the number of \n",
            "top key words\n",
            "[['-'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.002698896130823414\n",
            "OAKLAND, Calif. - Kerry Collins, the amiable backup quarterback who has repeatedly promised to be ready when needed, replaced injured starter Rich Gannon and calmly directed the Oakland Raiders to an emotional 30-20 victory over the Tampa Bay Buccaneers on Sunday night...\n",
            "top key words\n",
            "[['quarterback'], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.002686052699573338\n",
            "England beat Zimbabwe by five wickets in the opening one-day international.\n",
            "top key words\n",
            "[['-'], ['in']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: -0.0026764540485828547\n",
            "MUNCIE, Ind. -- Ball State University officials said the college is bucking a nationwide trend toward declining enrollment in MBA degree programs.\n",
            "top key words\n",
            "[['-'], ['in']]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.002608678852993762\n",
            "Reuters - The ball burst on impact when midfielder Walter Baseggio scored Anderlecht's equaliser in their 2-1 home win\\over La Louviere in the Belgian first division.\n",
            "top key words\n",
            "[['-'], [\"'\"]]\n",
            "--------------------------------------------------\n",
            "label: [0],predicted_scores: [0], influence: -0.0025861709691525903\n",
            "New Zealand captain Stephen Fleming was offered 200,000 to join a match-fixing syndicate in 1999.\n",
            "top key words\n",
            "[['-'], ['in']]\n",
            "--------------------------------------------------\n",
            "==================================================\n",
            "Opponents: \n",
            "label: [1],predicted_scores: [1], influence: 0.0033666162034933222\n",
            "Paula Radcliffe is to have another medical check-up as a row brews over the marathon's scheduling.\n",
            "top key words\n",
            "[[\"'\"], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.00313947693348382\n",
            "Britain's Alison Sheppard coasts into the women's 50m freestyle semi-finals.\n",
            "top key words\n",
            "[['-'], [\"'\"]]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.0026812196797436627\n",
            "AP - Rick Ankiel's return to the major leagues might not be just a September cameo.\n",
            "top key words\n",
            "[[\"'\"], ['-']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.0024911020882427692\n",
            "Caminiti died from acute intoxication due to the combined effects of cocaine and opiates, a spokeswoman for the New York City Medical Examiners Office told ESPN.\n",
            "top key words\n",
            "[['city'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.0024826730973472877\n",
            "The reserves are the key reason for Baltimore's 2-1 record and its place atop the AFC North standings following Sunday's thumping of Cincinnati, 23-9.\n",
            "top key words\n",
            "[['-'], [\"'\"]]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.002323072971194051\n",
            "ATHENS, Greece -- It was death by a thousand cuts. If you can describe taking silver instead of gold as  quot;death. quot; ...\n",
            "top key words\n",
            "[['-'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.0023134455159379286\n",
            "With just four healthy wideouts heading into Sunday's home opener, the Ravens may give Deion Sanders a chance at wideout.\n",
            "top key words\n",
            "[[\"'\"], ['sunday']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.0021009143438277533\n",
            "The coaches of high-performance athletes will benefit the most from the federal government #39;s announcement yesterday of how it will divide up \\$19.\n",
            "top key words\n",
            "[['-'], ['[PAD]']]\n",
            "--------------------------------------------------\n",
            "label: [2],predicted_scores: [2], influence: 0.002094681141898036\n",
            "Two people were in custody Sunday following a shootout between two groups of people that left two men dead and a third man wounded, police said Sunday.\n",
            "top key words\n",
            "[['in'], ['sunday']]\n",
            "--------------------------------------------------\n",
            "label: [1],predicted_scores: [1], influence: 0.001984444640129368\n",
            "AP - Michigan's defensive backs have bullied and beaten Purdue's receivers in three straight wins over the Boilermakers.\n",
            "top key words\n",
            "[[\"'\"], ['-']]\n",
            "--------------------------------------------------\n",
            "\n",
            "\n",
            "\n"
          ]
        }
      ],
      "source": [
        "for i in range(20):\n",
        "  method = 'tracin_first_top10'\n",
        "  evaluate_train_score = np.load(base_dir + 'evaluate_train_score_testtest2_' + method + '.npy')\n",
        "  evaluate_train_ind = np.load(base_dir + 'evaluate_train_ind_testtest2_' + method + '.npy')\n",
        "\n",
        "\n",
        "  print(\"=\"*50) \n",
        "  print('Evaluate Sentence: ')\n",
        "  print('label: {}, prediction: {}'.format(\n",
        "     y_eval[i][0], prediction_tests[0][i]))\n",
        "  print(eval_sentences_list[i])\n",
        "  print(\"=\"*50)  \n",
        "  print('Proponents: ')\n",
        "  #print(x_eval[i])\n",
        "  rm_inds = np.argsort(evaluate_train_score[i])[:10]\n",
        "  for rm_ind in rm_inds:\n",
        "    print('label: {},predicted_scores: {}, influence: {}'.format(y_train[rm_ind], y_train[rm_ind], evaluate_train_score[i,rm_ind]))\n",
        "    print(train_sentences_list[rm_ind])\n",
        "    \n",
        "    print('top key words')\n",
        "    print([tokenizer.convert_ids_to_tokens([evaluate_train_ind[i, rm_ind,j]]) for j in range(2)])\n",
        "    print('-'*50)\n",
        "  print(\"=\"*50)  \n",
        "  print('Opponents: ')\n",
        "  #print(x_eval[i])\n",
        "  rm_inds = np.argsort(evaluate_train_score[i])[::-1][:10]\n",
        "  for rm_ind in rm_inds:\n",
        "    print('label: {},predicted_scores: {}, influence: {}'.format(y_train[rm_ind], y_train[rm_ind], evaluate_train_score[i,rm_ind]))\n",
        "    print(train_sentences_list[rm_ind])\n",
        "    \n",
        "    print('top key words')\n",
        "    print([tokenizer.convert_ids_to_tokens([evaluate_train_ind[i, rm_ind,j]]) for j in range(2)])\n",
        "    print('-'*50)\n",
        "  print('\\n\\n')"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### TracIn first"
      ],
      "metadata": {
        "id": "h2iE4wEcPNf5"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "tracin_evaluate = []\n",
        "tracin_train = []\n",
        "method = 'tracin_first'\n",
        "tracin_evaluate = get_tracin_grad(x_eval, y_eval, [0], top=False)\n",
        "start = time.time()\n",
        "tracin_train = get_tracin_grad(x_train, y_train, [0], top=False)\n",
        "print('preparation time on saving gradient: {}'.format(time.time()-start))\n",
        "print(tracin_evaluate['grads'].shape)\n",
        "print(tracin_evaluate['inds'].shape)\n",
        "print(tracin_evaluate['vals'].shape)\n",
        "\n",
        "\n",
        "start = time.time()\n",
        "evaluate_train_score, evaluate_train_ind = get_tracin_list(tracin_evaluate, tracin_train, [0], top=False)\n",
        "print('time on calculating TracIn for 10 test * 50000 training: {}'.format(time.time()-start))\n",
        "np.save(base_dir + 'evaluate_train_score_testtest2_' + method + '.npy', evaluate_train_score)\n",
        "np.save(base_dir + 'evaluate_train_ind_testtest2_' + method + '.npy', evaluate_train_ind)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XpEoTkW5PMrA",
        "outputId": "1c1f955f-9041-490d-8a8f-334591c536d0"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0\n",
            "grad shape (20, 0)\n",
            "0\n",
            "1000\n",
            "2000\n",
            "3000\n",
            "4000\n",
            "5000\n",
            "6000\n",
            "7000\n",
            "8000\n",
            "9000\n",
            "10000\n",
            "11000\n",
            "12000\n",
            "13000\n",
            "14000\n",
            "15000\n",
            "16000\n",
            "17000\n",
            "18000\n",
            "19000\n",
            "20000\n",
            "21000\n",
            "22000\n",
            "23000\n",
            "24000\n",
            "25000\n",
            "26000\n",
            "27000\n",
            "28000\n",
            "29000\n",
            "30000\n",
            "31000\n",
            "32000\n",
            "33000\n",
            "34000\n",
            "35000\n",
            "36000\n",
            "37000\n",
            "38000\n",
            "39000\n",
            "40000\n",
            "41000\n",
            "42000\n",
            "43000\n",
            "44000\n",
            "45000\n",
            "46000\n",
            "47000\n",
            "48000\n",
            "49000\n",
            "grad shape (50000, 0)\n",
            "preparation time on saving gradient: 129.33664202690125\n",
            "(20, 0)\n",
            "(20, 2, 128)\n",
            "(20, 2, 128, 128)\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py:153: DeprecationWarning: The truth value of an empty array is ambiguous. Returning False, but in future this will result in an error. Use `array.size > 0` to check that an array is not empty.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "for i in range(10):\n",
        "  method = 'tracin_first'\n",
        "  evaluate_train_score = np.load(base_dir + 'evaluate_train_score_testtest2_' + method + '.npy')\n",
        "  evaluate_train_ind = np.load(base_dir + 'evaluate_train_ind_testtest2_' + method + '.npy')\n",
        "\n",
        "\n",
        "  print(\"=\"*50) \n",
        "  print('Evaluate Sentence: ')\n",
        "  print('label: {}, prediction: {}'.format(\n",
        "     y_eval[i][0], prediction_tests[0][i]))\n",
        "  print(eval_sentences_list[i])\n",
        "  print(\"=\"*50)  \n",
        "  print('Proponents: ')\n",
        "  #print(x_eval[i])\n",
        "  rm_inds = np.argsort(evaluate_train_score[i])[:10]\n",
        "  for rm_ind in rm_inds:\n",
        "    print('label: {},predicted_scores: {}, influence: {}'.format(y_train[rm_ind], y_train[rm_ind], evaluate_train_score[i,rm_ind]))\n",
        "    print(train_sentences_list[rm_ind])\n",
        "    \n",
        "    print('top key words')\n",
        "    print([tokenizer.convert_ids_to_tokens([evaluate_train_ind[i, rm_ind,j]]) for j in range(2)])\n",
        "    print('-'*50)\n",
        "  print(\"=\"*50)  \n",
        "  print('Opponents: ')\n",
        "  #print(x_eval[i])\n",
        "  rm_inds = np.argsort(evaluate_train_score[i])[::-1][:10]\n",
        "  for rm_ind in rm_inds:\n",
        "    print('label: {},predicted_scores: {}, influence: {}'.format(y_train[rm_ind], y_train[rm_ind], evaluate_train_score[i,rm_ind]))\n",
        "    print(train_sentences_list[rm_ind])\n",
        "    \n",
        "    print('top key words')\n",
        "    print([tokenizer.convert_ids_to_tokens([evaluate_train_ind[i, rm_ind,j]]) for j in range(2)])\n",
        "    print('-'*50)\n",
        "  print('\\n\\n')"
      ],
      "metadata": {
        "id": "8fuPtdhSPYwy"
      },
      "execution_count": null,
      "outputs": []
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "background_execution": "on",
      "collapsed_sections": [
        "eWei9WKgj-Op",
        "Vvob7PaY-hY3",
        "Fxzcw3YJkCdJ",
        "Mb6g_-BgL7fv",
        "Ev9KQe2dC3S8",
        "h2iE4wEcPNf5"
      ],
      "machine_shape": "hm",
      "provenance": []
    },
    "gpuClass": "standard",
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    },
    "widgets": {
      "application/vnd.jupyter.widget-state+json": {
        "026780de801140429fdaca2de8a0bbc9": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_4260809d24724413af98285effa5b94d",
              "IPY_MODEL_50811fea378e473c83b4c018a5672e58",
              "IPY_MODEL_9dd5dba454e8499794b488ec2d344fa8"
            ],
            "layout": "IPY_MODEL_17dd5c9d57b94d09934c8bc231e883f9"
          }
        },
        "4260809d24724413af98285effa5b94d": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_b83c327336e84ccfb89ea24b02215176",
            "placeholder": "​",
            "style": "IPY_MODEL_cb8e6b64031e4e55b6e5eee5ee5dd04b",
            "value": "Dl Completed...: 100%"
          }
        },
        "50811fea378e473c83b4c018a5672e58": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "success",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_1823c3217eda40d9b76f81caa91741e6",
            "max": 1,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_920a75e0cc5e42eca3f79d98f8fd516c",
            "value": 1
          }
        },
        "9dd5dba454e8499794b488ec2d344fa8": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_5ccf97a4b03547339f9daadb53e97bec",
            "placeholder": "​",
            "style": "IPY_MODEL_71ecf2dedad7489ca3d3e8905ec2e2ec",
            "value": " 1/1 [00:10&lt;00:00, 10.56s/ url]"
          }
        },
        "17dd5c9d57b94d09934c8bc231e883f9": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "b83c327336e84ccfb89ea24b02215176": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "cb8e6b64031e4e55b6e5eee5ee5dd04b": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "1823c3217eda40d9b76f81caa91741e6": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": "20px"
          }
        },
        "920a75e0cc5e42eca3f79d98f8fd516c": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "5ccf97a4b03547339f9daadb53e97bec": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "71ecf2dedad7489ca3d3e8905ec2e2ec": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "f5937c5b546543d89c10816a22ef0644": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_30d072fc93354dd8adec20e80c507262",
              "IPY_MODEL_95d1c2da758e46bfb96edb92169ba3c3",
              "IPY_MODEL_c8ca37a9a5654473b788e3780c45206f"
            ],
            "layout": "IPY_MODEL_88a516d0d98a4919bf57588ecf9c5d0e"
          }
        },
        "30d072fc93354dd8adec20e80c507262": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_c1109d62d719479db3c747142609d2d1",
            "placeholder": "​",
            "style": "IPY_MODEL_cf171048374d4f9eae660d209d93eecc",
            "value": "Dl Size...: 100%"
          }
        },
        "95d1c2da758e46bfb96edb92169ba3c3": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "success",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_0d89d80422ab4b3a959ce12e39c05e41",
            "max": 1,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_416f3deffae94a16bcb92df90b7d8bc3",
            "value": 1
          }
        },
        "c8ca37a9a5654473b788e3780c45206f": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_7be316d9b19942f3914d7167fca1e8ed",
            "placeholder": "​",
            "style": "IPY_MODEL_387bc56902c8454fa9b274e868a50635",
            "value": " 11/11 [00:10&lt;00:00,  2.02s/ MiB]"
          }
        },
        "88a516d0d98a4919bf57588ecf9c5d0e": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "c1109d62d719479db3c747142609d2d1": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "cf171048374d4f9eae660d209d93eecc": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "0d89d80422ab4b3a959ce12e39c05e41": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": "20px"
          }
        },
        "416f3deffae94a16bcb92df90b7d8bc3": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "7be316d9b19942f3914d7167fca1e8ed": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "387bc56902c8454fa9b274e868a50635": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "8c976ef531d1472e942ac608ea957f7c": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_a66f5f8ff5724d77a2e48ee08f63bc82",
              "IPY_MODEL_ac313cf6f91c4c218f0931d3e35cab59",
              "IPY_MODEL_6874e918b0244d279b66016ff7fe4a39"
            ],
            "layout": "IPY_MODEL_78ea05bd650c48bd8068da925694f459"
          }
        },
        "a66f5f8ff5724d77a2e48ee08f63bc82": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_ef78a0ac887245f8b3842dd236be1158",
            "placeholder": "​",
            "style": "IPY_MODEL_eb1239b7b8d84226a4285352662612f7",
            "value": "Extraction completed...: 100%"
          }
        },
        "ac313cf6f91c4c218f0931d3e35cab59": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "success",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_d7734b9698594e6dbfd67faa0872da76",
            "max": 1,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_23bcdad689a44917b714bf40ca8efc88",
            "value": 1
          }
        },
        "6874e918b0244d279b66016ff7fe4a39": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_962e3828576b4c47b720e7b0b5b0a942",
            "placeholder": "​",
            "style": "IPY_MODEL_0531c6048e20450f867d5dc4be1b2fca",
            "value": " 1/1 [00:10&lt;00:00, 10.82s/ file]"
          }
        },
        "78ea05bd650c48bd8068da925694f459": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "ef78a0ac887245f8b3842dd236be1158": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "eb1239b7b8d84226a4285352662612f7": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "d7734b9698594e6dbfd67faa0872da76": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": "20px"
          }
        },
        "23bcdad689a44917b714bf40ca8efc88": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "962e3828576b4c47b720e7b0b5b0a942": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "0531c6048e20450f867d5dc4be1b2fca": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "7f312183085942ddbee49ee68518b5a6": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_4b301dd5436b4817a5902c666ea8720f",
              "IPY_MODEL_9f4c3c35a8ab4a0b8a40acf41c668c80",
              "IPY_MODEL_f42cac69a50946aa971d9720161303d5"
            ],
            "layout": "IPY_MODEL_6ecda1493a8a4e8db04a7908b5f573e1"
          }
        },
        "4b301dd5436b4817a5902c666ea8720f": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_86b949eee20a4f82b4eeb1862f9fe34c",
            "placeholder": "​",
            "style": "IPY_MODEL_26088d3a2027402b8ad386041ec97c70",
            "value": "Generating splits...: 100%"
          }
        },
        "9f4c3c35a8ab4a0b8a40acf41c668c80": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_d7742ebb54c74dd39847c004f03d5ba4",
            "max": 2,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_072ac4984a974ab2b4a83c1e01df5bbd",
            "value": 2
          }
        },
        "f42cac69a50946aa971d9720161303d5": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_751c7c4a3b074ac783ed9778c3f15cf7",
            "placeholder": "​",
            "style": "IPY_MODEL_89a773d97af04b4795476f6a8534631e",
            "value": " 2/2 [00:13&lt;00:00,  5.63s/ splits]"
          }
        },
        "6ecda1493a8a4e8db04a7908b5f573e1": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": "hidden",
            "width": null
          }
        },
        "86b949eee20a4f82b4eeb1862f9fe34c": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "26088d3a2027402b8ad386041ec97c70": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "d7742ebb54c74dd39847c004f03d5ba4": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "072ac4984a974ab2b4a83c1e01df5bbd": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "751c7c4a3b074ac783ed9778c3f15cf7": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "89a773d97af04b4795476f6a8534631e": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "22482f05ce424b0d969d1b5d67ca7562": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_6904f800f1484c418ff43edfcb868dff",
              "IPY_MODEL_e4902cf3945d4019b00c8d23e91b342a",
              "IPY_MODEL_46c0dc7c65b5417ea0da28ebc159d3e7"
            ],
            "layout": "IPY_MODEL_a8fcc0913d374d1993e16494fe5f90cd"
          }
        },
        "6904f800f1484c418ff43edfcb868dff": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_df9cea19a8494f6a98cb191a0cbe660f",
            "placeholder": "​",
            "style": "IPY_MODEL_a79e5a29757e42f1b3acfc234790f0b1",
            "value": "Generating train examples...: 100%"
          }
        },
        "e4902cf3945d4019b00c8d23e91b342a": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_2c77f3549b99406e836b64b52079b1f8",
            "max": 120000,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_4deb2dd644944da2b2bcdc458b1d7818",
            "value": 120000
          }
        },
        "46c0dc7c65b5417ea0da28ebc159d3e7": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_c9ed14f1afa4480d975c22ac6fae6ceb",
            "placeholder": "​",
            "style": "IPY_MODEL_6027e75c591c446bbcc46cdc27b5d2ae",
            "value": " 119578/120000 [00:12&lt;00:00, 10266.24 examples/s]"
          }
        },
        "a8fcc0913d374d1993e16494fe5f90cd": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": "hidden",
            "width": null
          }
        },
        "df9cea19a8494f6a98cb191a0cbe660f": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "a79e5a29757e42f1b3acfc234790f0b1": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "2c77f3549b99406e836b64b52079b1f8": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "4deb2dd644944da2b2bcdc458b1d7818": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "c9ed14f1afa4480d975c22ac6fae6ceb": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "6027e75c591c446bbcc46cdc27b5d2ae": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "2a503125f8354301942ea5ca8d3d99e8": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_92d55759b0cb4d25910c194d2911f3c9",
              "IPY_MODEL_3d8dafe9909347f0ae064a99dbc02171",
              "IPY_MODEL_cbcca21e3a184e2e9d95e9268f18fce0"
            ],
            "layout": "IPY_MODEL_413221ade0c84c91a7c76e6f283ad48f"
          }
        },
        "92d55759b0cb4d25910c194d2911f3c9": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_0d1f3589e8c3405784513b9d05b5bcdb",
            "placeholder": "​",
            "style": "IPY_MODEL_ebb417f9454b4cd6b8102c199470950c",
            "value": "Shuffling ~/tensorflow_datasets/ag_news_subset/1.0.0.incompleteO62OX8/ag_news_subset-train.tfrecord*...:  74%"
          }
        },
        "3d8dafe9909347f0ae064a99dbc02171": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_a8090dab55054d478ec0c0262bd793a0",
            "max": 120000,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_b1fcba615c6e459eac5fa72efec04d59",
            "value": 120000
          }
        },
        "cbcca21e3a184e2e9d95e9268f18fce0": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_5c5b32bc53f84962820181c20d0b92c2",
            "placeholder": "​",
            "style": "IPY_MODEL_6d9e12bc28454b92886704bc0883a704",
            "value": " 88785/120000 [00:00&lt;00:00, 339727.14 examples/s]"
          }
        },
        "413221ade0c84c91a7c76e6f283ad48f": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": "hidden",
            "width": null
          }
        },
        "0d1f3589e8c3405784513b9d05b5bcdb": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "ebb417f9454b4cd6b8102c199470950c": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "a8090dab55054d478ec0c0262bd793a0": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "b1fcba615c6e459eac5fa72efec04d59": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "5c5b32bc53f84962820181c20d0b92c2": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "6d9e12bc28454b92886704bc0883a704": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "cc0365c9429142278f706a4f07bf34e8": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_03bc494f9f9e45d69b1a9e38d36303c4",
              "IPY_MODEL_d1c50e5ea8cf4d24a9b7adcefdb23b96",
              "IPY_MODEL_3aeac441812f4252960f1420b17efa04"
            ],
            "layout": "IPY_MODEL_2509cabd56364b7e9789bd5c260e06fb"
          }
        },
        "03bc494f9f9e45d69b1a9e38d36303c4": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_2b4b849656d9412aa3d379a80fdbfec7",
            "placeholder": "​",
            "style": "IPY_MODEL_20b5342699514a6da13137e66db6f5f6",
            "value": "Generating test examples...:  88%"
          }
        },
        "d1c50e5ea8cf4d24a9b7adcefdb23b96": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_89e41feb217749239c28248f466c361c",
            "max": 7600,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_2fa78e682214457599ad9c2cd8a8eac2",
            "value": 7600
          }
        },
        "3aeac441812f4252960f1420b17efa04": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_d335c08c3aa74551a96e35831dc802c8",
            "placeholder": "​",
            "style": "IPY_MODEL_ba1f3823b17e492db5a4ebebaf23e5a2",
            "value": " 6695/7600 [00:00&lt;00:00, 9909.24 examples/s]"
          }
        },
        "2509cabd56364b7e9789bd5c260e06fb": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": "hidden",
            "width": null
          }
        },
        "2b4b849656d9412aa3d379a80fdbfec7": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "20b5342699514a6da13137e66db6f5f6": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "89e41feb217749239c28248f466c361c": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "2fa78e682214457599ad9c2cd8a8eac2": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "d335c08c3aa74551a96e35831dc802c8": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "ba1f3823b17e492db5a4ebebaf23e5a2": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "07ab4bfe0f414e73a3bc141cc48faa38": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HBoxModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HBoxModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HBoxView",
            "box_style": "",
            "children": [
              "IPY_MODEL_3667e623b9cb4fb981eb71a2ea0f2da7",
              "IPY_MODEL_97c7aeef0e7c4dcfaca822ced5690ab9",
              "IPY_MODEL_ec020c0988084ca0a7c7352dabb74702"
            ],
            "layout": "IPY_MODEL_b160e728277142deb782141385016776"
          }
        },
        "3667e623b9cb4fb981eb71a2ea0f2da7": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_98da7fb99cd844fbb24e18bf48d1a22f",
            "placeholder": "​",
            "style": "IPY_MODEL_dd253db5b1cb4b94bb908acfd3277609",
            "value": "Shuffling ~/tensorflow_datasets/ag_news_subset/1.0.0.incompleteO62OX8/ag_news_subset-test.tfrecord*...:   0%"
          }
        },
        "97c7aeef0e7c4dcfaca822ced5690ab9": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "FloatProgressModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "FloatProgressModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "ProgressView",
            "bar_style": "",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_11a289923a9845ac94c3cd9c95920edc",
            "max": 7600,
            "min": 0,
            "orientation": "horizontal",
            "style": "IPY_MODEL_c5959332e6c0442c96dd45c03b1237a6",
            "value": 7600
          }
        },
        "ec020c0988084ca0a7c7352dabb74702": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "HTMLModel",
          "model_module_version": "1.5.0",
          "state": {
            "_dom_classes": [],
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "HTMLModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/controls",
            "_view_module_version": "1.5.0",
            "_view_name": "HTMLView",
            "description": "",
            "description_tooltip": null,
            "layout": "IPY_MODEL_e194ab3ad9174e199a38f2075b59f2a9",
            "placeholder": "​",
            "style": "IPY_MODEL_0ecfb382b7b342baae293fac6f26d867",
            "value": " 0/7600 [00:00&lt;?, ? examples/s]"
          }
        },
        "b160e728277142deb782141385016776": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": "hidden",
            "width": null
          }
        },
        "98da7fb99cd844fbb24e18bf48d1a22f": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "dd253db5b1cb4b94bb908acfd3277609": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        },
        "11a289923a9845ac94c3cd9c95920edc": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "c5959332e6c0442c96dd45c03b1237a6": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "ProgressStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "ProgressStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "bar_color": null,
            "description_width": ""
          }
        },
        "e194ab3ad9174e199a38f2075b59f2a9": {
          "model_module": "@jupyter-widgets/base",
          "model_name": "LayoutModel",
          "model_module_version": "1.2.0",
          "state": {
            "_model_module": "@jupyter-widgets/base",
            "_model_module_version": "1.2.0",
            "_model_name": "LayoutModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "LayoutView",
            "align_content": null,
            "align_items": null,
            "align_self": null,
            "border": null,
            "bottom": null,
            "display": null,
            "flex": null,
            "flex_flow": null,
            "grid_area": null,
            "grid_auto_columns": null,
            "grid_auto_flow": null,
            "grid_auto_rows": null,
            "grid_column": null,
            "grid_gap": null,
            "grid_row": null,
            "grid_template_areas": null,
            "grid_template_columns": null,
            "grid_template_rows": null,
            "height": null,
            "justify_content": null,
            "justify_items": null,
            "left": null,
            "margin": null,
            "max_height": null,
            "max_width": null,
            "min_height": null,
            "min_width": null,
            "object_fit": null,
            "object_position": null,
            "order": null,
            "overflow": null,
            "overflow_x": null,
            "overflow_y": null,
            "padding": null,
            "right": null,
            "top": null,
            "visibility": null,
            "width": null
          }
        },
        "0ecfb382b7b342baae293fac6f26d867": {
          "model_module": "@jupyter-widgets/controls",
          "model_name": "DescriptionStyleModel",
          "model_module_version": "1.5.0",
          "state": {
            "_model_module": "@jupyter-widgets/controls",
            "_model_module_version": "1.5.0",
            "_model_name": "DescriptionStyleModel",
            "_view_count": null,
            "_view_module": "@jupyter-widgets/base",
            "_view_module_version": "1.2.0",
            "_view_name": "StyleView",
            "description_width": ""
          }
        }
      }
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}