process_seq_50000_word2vec_share.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "process_seq_50000_word2vec_share",
      "provenance": [],
      "collapsed_sections": [],
      "machine_shape": "hm"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "TPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "qziHAa8Fq5KS",
        "colab_type": "code",
        "outputId": "c2c64303-9773-4b49-8a9f-6525e5ff9c99",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "import tensorflow as tf\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "print(tf.__version__)\n",
        "if tf.__version__ != '1.13.1':\n",
        "  !pip install tensorflow==1.13.1"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "1.13.1\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "87woRq0CrGk7",
        "colab_type": "code",
        "outputId": "c43482fd-3f62-4958-db77-9e920bf89c01",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 121
        }
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/gdrive')\n",
        "\n",
        "data_folder_path = '/content/gdrive/My Drive/model'\n",
        "\n",
        "train_seq_path = 'train_word2vec_w5_whole_data.pkl'\n",
        "test_seq_path = 'test_word2vec_w5_whole_data.pkl'\n",
        "\n",
        "embedding_matrix_path = 'word_seg_vectors_arr_whole_data.pkl' #已替换\n",
        "\n",
        "train_labels_path = 'train_label.pkl'\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Go to this URL in a browser: https://accounts.google.com/o/oauth2/auth?client_id=947318989803-6bn6qk8qdgf4n4g3pfee6491hc0brc4i.apps.googleusercontent.com&redirect_uri=urn%3Aietf%3Awg%3Aoauth%3A2.0%3Aoob&scope=email%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdocs.test%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive.photos.readonly%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fpeopleapi.readonly&response_type=code\n",
            "\n",
            "Enter your authorization code:\n",
            "··········\n",
            "Mounted at /content/gdrive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pFouwiUXrfPR",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import os\n",
        "import pickle\n",
        "\n",
        "train_seq = pickle.load(open(os.path.join(data_folder_path, train_seq_path), 'rb' ) )\n",
        "test_seq = pickle.load(open(os.path.join(data_folder_path, test_seq_path), 'rb' ))\n",
        "\n",
        "train_labels = pickle.load(open(os.path.join(data_folder_path, train_labels_path), 'rb'))\n",
        "embedding_matrix = pickle.load(open(os.path.join(data_folder_path, embedding_matrix_path), 'rb'))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "claPFElFuKxA",
        "colab_type": "code",
        "outputId": "0b1e27c5-1a16-400e-98fd-7c493cf5f087",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "np.max(test_seq)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "301"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 5
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5t2z4Dwc6r9s",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "train_labels = train_labels.astype(np.float32)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-kuA4Pbk71IH",
        "colab_type": "code",
        "outputId": "1b51c3d4-7c78-4acf-b077-4a8bfae6fa6e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 134
        }
      },
      "source": [
        "train_labels"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[0., 0., 0., ..., 1., 0., 0.],\n",
              "       [0., 0., 1., ..., 0., 0., 0.],\n",
              "       [1., 0., 0., ..., 0., 0., 0.],\n",
              "       ...,\n",
              "       [1., 0., 0., ..., 0., 0., 0.],\n",
              "       [0., 1., 0., ..., 0., 0., 0.],\n",
              "       [0., 0., 1., ..., 0., 0., 0.]], dtype=float32)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zU7oWq8o3h9Q",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for i in range(len(train_labels)):\n",
        "    for j in range(len(train_labels[i])):\n",
        "        if train_labels[i][j] == 0:\n",
        "            train_labels[i][j] = 0.005 / 7 \n",
        "      \n",
        "        else:\n",
        "            train_labels[i][j] = 0.995"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FUgR5h4G3iCy",
        "colab_type": "code",
        "outputId": "8b6ff46d-6e53-45d4-ce62-c9000d735c4b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 235
        }
      },
      "source": [
        "train_labels"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[7.142857e-04, 7.142857e-04, 7.142857e-04, ..., 9.950000e-01,\n",
              "        7.142857e-04, 7.142857e-04],\n",
              "       [7.142857e-04, 7.142857e-04, 9.950000e-01, ..., 7.142857e-04,\n",
              "        7.142857e-04, 7.142857e-04],\n",
              "       [9.950000e-01, 7.142857e-04, 7.142857e-04, ..., 7.142857e-04,\n",
              "        7.142857e-04, 7.142857e-04],\n",
              "       ...,\n",
              "       [9.950000e-01, 7.142857e-04, 7.142857e-04, ..., 7.142857e-04,\n",
              "        7.142857e-04, 7.142857e-04],\n",
              "       [7.142857e-04, 9.950000e-01, 7.142857e-04, ..., 7.142857e-04,\n",
              "        7.142857e-04, 7.142857e-04],\n",
              "       [7.142857e-04, 7.142857e-04, 9.950000e-01, ..., 7.142857e-04,\n",
              "        7.142857e-04, 7.142857e-04]], dtype=float32)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "s5egNDAM3h6U",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import tensorflow as tf\n",
        "import numpy as np\n",
        "\n",
        "from tensorflow.keras.layers import Dense, Input, Embedding, Dropout, Activation\n",
        "from tensorflow.keras.layers import Conv1D, GlobalMaxPooling1D, Flatten, GlobalAveragePooling1D\n",
        "from tensorflow.keras.layers import SpatialDropout1D, BatchNormalization\n",
        "from tensorflow.keras.layers import concatenate\n",
        "from tensorflow.keras.models import Model\n",
        "from tensorflow.keras.optimizers import Adam\n",
        "from tensorflow.keras.callbacks import EarlyStopping, ModelCheckpoint\n",
        "\n",
        "from sklearn.preprocessing import LabelBinarizer, LabelEncoder\n",
        "from sklearn.model_selection import StratifiedKFold\n",
        "from sklearn.metrics import confusion_matrix\n",
        "import matplotlib.pyplot as plt"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fQgfLziD3h1d",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def TextCNN_add_pool_wordvec_not_trainable(max_len, max_cnt, embed_size,\n",
        "            num_filters, kernel_size,\n",
        "            conv_action,\n",
        "            mask_zero, embedding_matrix):\n",
        "    _input = Input(shape=(max_len,), dtype='int32')\n",
        "    _embed = Embedding(max_cnt, embed_size, input_length=max_len, mask_zero=mask_zero, weights=[embedding_matrix], trainable=False)(_input)\n",
        "    _embed = SpatialDropout1D(0.2)(_embed)\n",
        "    warppers = []\n",
        "#     print('warppers is', warppers)\n",
        "    \n",
        "    for _kernel_size in kernel_size:\n",
        "        conv1d = Conv1D(filters=num_filters, kernel_size=_kernel_size, activation=conv_action)(_embed)\n",
        "        warppers.append(GlobalMaxPooling1D()(conv1d))\n",
        "        warppers.append(GlobalAveragePooling1D()(conv1d))\n",
        "#         warppers.append(GlobalMaxPooling1D()(conv1d))\n",
        "                        \n",
        "    fc = concatenate(warppers)\n",
        "    fc = Dropout(0.5)(fc)\n",
        "    #fc = BatchNormalization()(fc)\n",
        "    fc = Dense(256, activation='relu')(fc)\n",
        "    fc = Dropout(0.25)(fc)\n",
        "    #fc = BatchNormalization()(fc) \n",
        "    preds = Dense(8, activation = 'softmax')(fc)\n",
        "    \n",
        "    model = Model(inputs=_input, outputs=preds)\n",
        "    #model.summary()\n",
        "    return model"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LU6SH4wiKDul",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class_weights=[1, 0.8, 0.5, 0.81,  1.22, 1, 1.22, 0.81]\n",
        "from sklearn.metrics import accuracy_score"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gw7QnThaxaQv",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# from sklearn.utils import class_weight\n",
        "# from sklearn.metrics import accuracy_score\n",
        "\n",
        "# class_weights = class_weight.compute_class_weight('balanced',\n",
        "#                                                  np.unique(np.argmax(train_labels, axis = 1)),\n",
        "#                                                  np.argmax(train_labels, axis = 1))\n",
        "# print(class_weights)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QC6ugg0VWeO6",
        "colab_type": "code",
        "outputId": "898e1b10-d357-43ef-ff39-ac14b0e8a55c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "class_weights"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[1, 0.8, 0.5, 0.81, 1.22, 1, 1.22, 0.81]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gQNNUp7V3hyj",
        "colab_type": "code",
        "outputId": "4b298383-7989-4b54-c3ef-75ef04cdf7e2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "seeds = [0]\n",
        "total_test = np.zeros(shape = (len(test_seq), 8))\n",
        "for seed in seeds:\n",
        "    skf = StratifiedKFold(n_splits=5, shuffle=True, random_state=seed)\n",
        "    max_len = 50000\n",
        "    max_cnt = 302\n",
        "    embed_size = 100\n",
        "    num_filters = 24\n",
        "    kernel_size_short = [2, 4, 6, 8, 10, 12, 14, 16, 18, 20]\n",
        "    conv_action = 'relu'\n",
        "    mask_zero = False\n",
        "    TRAIN = True\n",
        "\n",
        "\n",
        "    tf.logging.set_verbosity(0)\n",
        "\n",
        "    meta_train = np.zeros(shape = (len(train_seq), 8))\n",
        "    meta_test = np.zeros(shape = (len(test_seq), 8))\n",
        "\n",
        "    multi_train_labels = [np.argmax(i) for i in train_labels]\n",
        "    split_data = skf.split(train_seq, multi_train_labels)\n",
        "\n",
        "    for i,(tr_ind,te_ind) in enumerate(split_data):\n",
        "        print('FOLD: ', i + 1)\n",
        "        print(len(te_ind),len(tr_ind))\n",
        "        model = TextCNN_add_pool_wordvec_not_trainable(max_len,max_cnt,embed_size,num_filters,kernel_size_short,conv_action,mask_zero, embedding_matrix)\n",
        "        model_name = 'benchmark_cnn_fold_' + str(i)\n",
        "        X_train, X_train_label = train_seq[tr_ind], train_labels[tr_ind]\n",
        "        X_val, X_val_label = train_seq[te_ind], train_labels[te_ind]\n",
        "        model_save_path = '%s_%s.ckpt'%(model_name, embed_size)\n",
        "        early_stopping = EarlyStopping(monitor='val_loss', patience=10)\n",
        "        model_checkpoint = ModelCheckpoint(model_save_path, save_best_only=True, save_weights_only=True)\n",
        "\n",
        "        strategy = tf.contrib.tpu.TPUDistributionStrategy(\n",
        "            tf.contrib.cluster_resolver.TPUClusterResolver(tpu='grpc://' + os.environ['COLAB_TPU_ADDR']))\n",
        "        model = tf.contrib.tpu.keras_to_tpu_model(model, strategy)\n",
        "        model.compile(loss='categorical_crossentropy',\n",
        "                    optimizer='adam',\n",
        "                    metrics=['accuracy'])\n",
        "\n",
        "        if TRAIN:\n",
        "            history = model.fit(X_train, X_train_label,\n",
        "                    validation_data=(X_val, X_val_label),\n",
        "                    epochs=100, batch_size=64,\n",
        "                    shuffle=True,\n",
        "                    callbacks=[early_stopping],\n",
        "                    class_weight=class_weights\n",
        "                    )\n",
        "            \n",
        "            # summarize history for accuracy\n",
        "            plt.plot(history.history['acc'])\n",
        "            plt.plot(history.history['val_acc'])\n",
        "            plt.title('model accuracy')\n",
        "            plt.ylabel('accuracy')\n",
        "            plt.xlabel('epoch')\n",
        "            plt.legend(['train', 'test'], loc='upper left')\n",
        "            plt.show()\n",
        "            # summarize history for loss\n",
        "            plt.plot(history.history['loss'])\n",
        "            plt.plot(history.history['val_loss'])\n",
        "            plt.title('model loss')\n",
        "            plt.ylabel('loss')\n",
        "            plt.xlabel('epoch')\n",
        "            plt.legend(['train', 'test'], loc='upper left')\n",
        "            plt.show()\n",
        "        \n",
        "        #model.load_weights(model_save_path)\n",
        "        #model.save_weights('/tmp/%s.h5' % model_name, overwrite=True)\n",
        "        cpu_model = model.sync_to_cpu()\n",
        "        pred_val = cpu_model.predict(X_val, batch_size=64, verbose=1)\n",
        "        pred_test = cpu_model.predict(test_seq, batch_size=64, verbose=1)\n",
        "        pred_train = cpu_model.predict(X_train, batch_size=64, verbose=1)\n",
        "        \n",
        "        print('train confuse matrix is')\n",
        "        print(confusion_matrix(X_train_label.argmax(1), pred_train.argmax(1)))\n",
        "        print('val confuse matrix is')\n",
        "        print(confusion_matrix(X_val_label.argmax(1), pred_val.argmax(1)))\n",
        "        print('train accuracy is', accuracy_score(X_train_label.argmax(1), pred_train.argmax(1)))\n",
        "        print('valid accuracy is', accuracy_score(X_val_label.argmax(1), pred_val.argmax(1)))\n",
        "        #print(confusion_matrix(binary_labels.argmax(1), pred_train.argmax(1)))\n",
        "            \n",
        "        meta_train[te_ind] = pred_val\n",
        "        meta_test += pred_test\n",
        "        tf.keras.backend.clear_session()\n",
        "    meta_test /= 5.0\n",
        "    total_test += meta_test\n",
        "\n",
        "total_test /= len(seeds)"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "FOLD:  1\n",
            "2780 11107\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/resource_variable_ops.py:435: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Colocations handled automatically by placer.\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/layers/core.py:143: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_1 (InputLayer)            (None, 50000)        0                                            \n",
            "__________________________________________________________________________________________________\n",
            "embedding (Embedding)           (None, 50000, 100)   30200       input_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "spatial_dropout1d (SpatialDropo (None, 50000, 100)   0           embedding[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "conv1d (Conv1D)                 (None, 49999, 24)    4824        spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_1 (Conv1D)               (None, 49997, 24)    9624        spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_2 (Conv1D)               (None, 49995, 24)    14424       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_3 (Conv1D)               (None, 49993, 24)    19224       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_4 (Conv1D)               (None, 49991, 24)    24024       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_5 (Conv1D)               (None, 49989, 24)    28824       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_6 (Conv1D)               (None, 49987, 24)    33624       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_7 (Conv1D)               (None, 49985, 24)    38424       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_8 (Conv1D)               (None, 49983, 24)    43224       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_9 (Conv1D)               (None, 49981, 24)    48024       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d (GlobalMax (None, 24)           0           conv1d[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d (Globa (None, 24)           0           conv1d[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_1 (GlobalM (None, 24)           0           conv1d_1[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_1 (Glo (None, 24)           0           conv1d_1[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_2 (GlobalM (None, 24)           0           conv1d_2[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_2 (Glo (None, 24)           0           conv1d_2[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_3 (GlobalM (None, 24)           0           conv1d_3[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_3 (Glo (None, 24)           0           conv1d_3[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_4 (GlobalM (None, 24)           0           conv1d_4[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_4 (Glo (None, 24)           0           conv1d_4[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_5 (GlobalM (None, 24)           0           conv1d_5[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_5 (Glo (None, 24)           0           conv1d_5[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_6 (GlobalM (None, 24)           0           conv1d_6[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_6 (Glo (None, 24)           0           conv1d_6[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_7 (GlobalM (None, 24)           0           conv1d_7[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_7 (Glo (None, 24)           0           conv1d_7[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_8 (GlobalM (None, 24)           0           conv1d_8[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_8 (Glo (None, 24)           0           conv1d_8[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_9 (GlobalM (None, 24)           0           conv1d_9[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_9 (Glo (None, 24)           0           conv1d_9[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "concatenate (Concatenate)       (None, 480)          0           global_max_pooling1d[0][0]       \n",
            "                                                                 global_average_pooling1d[0][0]   \n",
            "                                                                 global_max_pooling1d_1[0][0]     \n",
            "                                                                 global_average_pooling1d_1[0][0] \n",
            "                                                                 global_max_pooling1d_2[0][0]     \n",
            "                                                                 global_average_pooling1d_2[0][0] \n",
            "                                                                 global_max_pooling1d_3[0][0]     \n",
            "                                                                 global_average_pooling1d_3[0][0] \n",
            "                                                                 global_max_pooling1d_4[0][0]     \n",
            "                                                                 global_average_pooling1d_4[0][0] \n",
            "                                                                 global_max_pooling1d_5[0][0]     \n",
            "                                                                 global_average_pooling1d_5[0][0] \n",
            "                                                                 global_max_pooling1d_6[0][0]     \n",
            "                                                                 global_average_pooling1d_6[0][0] \n",
            "                                                                 global_max_pooling1d_7[0][0]     \n",
            "                                                                 global_average_pooling1d_7[0][0] \n",
            "                                                                 global_max_pooling1d_8[0][0]     \n",
            "                                                                 global_average_pooling1d_8[0][0] \n",
            "                                                                 global_max_pooling1d_9[0][0]     \n",
            "                                                                 global_average_pooling1d_9[0][0] \n",
            "__________________________________________________________________________________________________\n",
            "dropout (Dropout)               (None, 480)          0           concatenate[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "dense (Dense)                   (None, 256)          123136      dropout[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "dropout_1 (Dropout)             (None, 256)          0           dense[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "dense_1 (Dense)                 (None, 8)            2056        dropout_1[0][0]                  \n",
            "==================================================================================================\n",
            "Total params: 419,632\n",
            "Trainable params: 389,432\n",
            "Non-trainable params: 30,200\n",
            "__________________________________________________________________________________________________\n",
            "\n",
            "WARNING: The TensorFlow contrib module will not be included in TensorFlow 2.0.\n",
            "For more information, please see:\n",
            "  * https://github.com/tensorflow/community/blob/master/rfcs/20180907-contrib-sunset.md\n",
            "  * https://github.com/tensorflow/addons\n",
            "If you depend on functionality not listed there, please file an issue.\n",
            "\n",
            "WARNING:tensorflow:tpu_model (from tensorflow.contrib.tpu.python.tpu.keras_support) is experimental and may change or be removed at any time, and without warning.\n",
            "Train on 11107 samples, validate on 2780 samples\n",
            "Epoch 1/100\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/contrib/tpu/python/tpu/keras_support.py:302: to_float (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use tf.cast instead.\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use tf.cast instead.\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_grad.py:102: div (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Deprecated in favor of operator or tf.math.divide.\n",
            "WARNING:tensorflow:Cannot update non-variable config: epsilon\n",
            "WARNING:tensorflow:Cannot update non-variable config: amsgrad\n",
            "11107/11107 [==============================] - 129s 12ms/sample - loss: 1.1767 - acc: 0.5908 - val_loss: 0.6567 - val_acc: 0.8112\n",
            "Epoch 2/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.6855 - acc: 0.7893 - val_loss: 0.5044 - val_acc: 0.8487\n",
            "Epoch 3/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.5718 - acc: 0.8293 - val_loss: 0.4591 - val_acc: 0.8584\n",
            "Epoch 4/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.5126 - acc: 0.8497 - val_loss: 0.4375 - val_acc: 0.8703\n",
            "Epoch 5/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.4736 - acc: 0.8603 - val_loss: 0.4203 - val_acc: 0.8739\n",
            "Epoch 6/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.4470 - acc: 0.8656 - val_loss: 0.4077 - val_acc: 0.8793\n",
            "Epoch 7/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.4289 - acc: 0.8754 - val_loss: 0.3855 - val_acc: 0.8876\n",
            "Epoch 8/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.4073 - acc: 0.8798 - val_loss: 0.3726 - val_acc: 0.8898\n",
            "Epoch 9/100\n",
            "11107/11107 [==============================] - 37s 3ms/sample - loss: 0.3930 - acc: 0.8845 - val_loss: 0.3711 - val_acc: 0.8937\n",
            "Epoch 10/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.3811 - acc: 0.8886 - val_loss: 0.3657 - val_acc: 0.8883\n",
            "Epoch 11/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.3669 - acc: 0.8941 - val_loss: 0.3509 - val_acc: 0.8988\n",
            "Epoch 12/100\n",
            "11107/11107 [==============================] - 37s 3ms/sample - loss: 0.3597 - acc: 0.8932 - val_loss: 0.3536 - val_acc: 0.8959\n",
            "Epoch 13/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.3502 - acc: 0.8989 - val_loss: 0.3536 - val_acc: 0.8981\n",
            "Epoch 14/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.3358 - acc: 0.9003 - val_loss: 0.3491 - val_acc: 0.8934\n",
            "Epoch 15/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.3340 - acc: 0.9006 - val_loss: 0.3575 - val_acc: 0.8923\n",
            "Epoch 16/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.3334 - acc: 0.9047 - val_loss: 0.3513 - val_acc: 0.8930\n",
            "Epoch 17/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.3225 - acc: 0.9050 - val_loss: 0.3570 - val_acc: 0.8984\n",
            "Epoch 18/100\n",
            "11107/11107 [==============================] - 37s 3ms/sample - loss: 0.3145 - acc: 0.9069 - val_loss: 0.3556 - val_acc: 0.8984\n",
            "Epoch 19/100\n",
            "11107/11107 [==============================] - 37s 3ms/sample - loss: 0.3086 - acc: 0.9102 - val_loss: 0.3404 - val_acc: 0.8973\n",
            "Epoch 20/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.3080 - acc: 0.9098 - val_loss: 0.3397 - val_acc: 0.8981\n",
            "Epoch 21/100\n",
            "11107/11107 [==============================] - 37s 3ms/sample - loss: 0.3008 - acc: 0.9093 - val_loss: 0.3479 - val_acc: 0.9024\n",
            "Epoch 22/100\n",
            "11107/11107 [==============================] - 37s 3ms/sample - loss: 0.2963 - acc: 0.9117 - val_loss: 0.3425 - val_acc: 0.8973\n",
            "Epoch 23/100\n",
            "11107/11107 [==============================] - 37s 3ms/sample - loss: 0.2910 - acc: 0.9136 - val_loss: 0.3411 - val_acc: 0.9031\n",
            "Epoch 24/100\n",
            "11107/11107 [==============================] - 37s 3ms/sample - loss: 0.2899 - acc: 0.9137 - val_loss: 0.3421 - val_acc: 0.9017\n",
            "Epoch 25/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.2862 - acc: 0.9146 - val_loss: 0.3393 - val_acc: 0.8984\n",
            "Epoch 26/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.2846 - acc: 0.9163 - val_loss: 0.3360 - val_acc: 0.9038\n",
            "Epoch 27/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.2742 - acc: 0.9174 - val_loss: 0.3363 - val_acc: 0.9027\n",
            "Epoch 28/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.2662 - acc: 0.9197 - val_loss: 0.3318 - val_acc: 0.9027\n",
            "Epoch 29/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.2698 - acc: 0.9190 - val_loss: 0.3350 - val_acc: 0.8973\n",
            "Epoch 30/100\n",
            "11107/11107 [==============================] - 37s 3ms/sample - loss: 0.2635 - acc: 0.9207 - val_loss: 0.3371 - val_acc: 0.9020\n",
            "Epoch 31/100\n",
            "11107/11107 [==============================] - 37s 3ms/sample - loss: 0.2614 - acc: 0.9213 - val_loss: 0.3324 - val_acc: 0.9009\n",
            "Epoch 32/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.2567 - acc: 0.9262 - val_loss: 0.3489 - val_acc: 0.8973\n",
            "Epoch 33/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.2535 - acc: 0.9281 - val_loss: 0.3426 - val_acc: 0.9009\n",
            "Epoch 34/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.2529 - acc: 0.9236 - val_loss: 0.3484 - val_acc: 0.9017\n",
            "Epoch 35/100\n",
            "11107/11107 [==============================] - 37s 3ms/sample - loss: 0.2485 - acc: 0.9279 - val_loss: 0.3307 - val_acc: 0.9027\n",
            "Epoch 36/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.2483 - acc: 0.9269 - val_loss: 0.3270 - val_acc: 0.9035\n",
            "Epoch 37/100\n",
            "11107/11107 [==============================] - 37s 3ms/sample - loss: 0.2431 - acc: 0.9291 - val_loss: 0.3394 - val_acc: 0.9045\n",
            "Epoch 38/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.2451 - acc: 0.9305 - val_loss: 0.3398 - val_acc: 0.9006\n",
            "Epoch 39/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.2419 - acc: 0.9299 - val_loss: 0.3387 - val_acc: 0.9009\n",
            "Epoch 40/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.2364 - acc: 0.9295 - val_loss: 0.3429 - val_acc: 0.9042\n",
            "Epoch 41/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.2341 - acc: 0.9338 - val_loss: 0.3458 - val_acc: 0.9024\n",
            "Epoch 42/100\n",
            "11107/11107 [==============================] - 38s 3ms/sample - loss: 0.2374 - acc: 0.9308 - val_loss: 0.3382 - val_acc: 0.9013\n",
            "Epoch 43/100\n",
            "11107/11107 [==============================] - 37s 3ms/sample - loss: 0.2321 - acc: 0.9333 - val_loss: 0.3547 - val_acc: 0.8984\n",
            "Epoch 44/100\n",
            "11107/11107 [==============================] - 37s 3ms/sample - loss: 0.2257 - acc: 0.9374 - val_loss: 0.3354 - val_acc: 0.9031\n",
            "Epoch 45/100\n",
            "11107/11107 [==============================] - 37s 3ms/sample - loss: 0.2253 - acc: 0.9329 - val_loss: 0.3422 - val_acc: 0.9042\n",
            "Epoch 46/100\n",
            "11107/11107 [==============================] - 37s 3ms/sample - loss: 0.2232 - acc: 0.9371 - val_loss: 0.3456 - val_acc: 0.9024\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\niEjIKQhEREJOQSAiEnIKAhGRkPv/AY7JDqP9oVzmAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Cannot update non-variable config: epsilon\n",
            "WARNING:tensorflow:Cannot update non-variable config: amsgrad\n",
            "2780/2780 [==============================] - 58s 21ms/sample\n",
            "12955/12955 [==============================] - 272s 21ms/sample\n",
            "11107/11107 [==============================] - 228s 21ms/sample\n",
            "train confuse matrix is\n",
            "[[3976    0    1    0    0    4    0    1]\n",
            " [   0  397    0    0    0    4    0    0]\n",
            " [   2    0  947    0    0    5    0    2]\n",
            " [   3    8    1  577   17   13    5   32]\n",
            " [   0    0    0    8   65    3    1    3]\n",
            " [   2    0    1   43    1 3380    1    3]\n",
            " [   1    0    1    2    0    3  381   24]\n",
            " [   5   20   12  103   21   23   23  982]]\n",
            "val confuse matrix is\n",
            "[[990   0   1   0   0   2   0   3]\n",
            " [  2  96   1   0   0   1   0   1]\n",
            " [ 11   0 213   0   0   2   2  12]\n",
            " [  0   1   2 125   4  10   1  21]\n",
            " [  1   0   0   7   5   2   0   5]\n",
            " [  4   0   3  12   0 834   3   2]\n",
            " [  5   0   3   3   0  11  54  27]\n",
            " [ 13   3  13  44   8   5  22 190]]\n",
            "train accuracy is 0.9638066084451247\n",
            "valid accuracy is 0.9017985611510791\n",
            "FOLD:  2\n",
            "2779 11108\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_1 (InputLayer)            (None, 50000)        0                                            \n",
            "__________________________________________________________________________________________________\n",
            "embedding (Embedding)           (None, 50000, 100)   30200       input_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "spatial_dropout1d (SpatialDropo (None, 50000, 100)   0           embedding[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "conv1d (Conv1D)                 (None, 49999, 24)    4824        spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_1 (Conv1D)               (None, 49997, 24)    9624        spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_2 (Conv1D)               (None, 49995, 24)    14424       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_3 (Conv1D)               (None, 49993, 24)    19224       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_4 (Conv1D)               (None, 49991, 24)    24024       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_5 (Conv1D)               (None, 49989, 24)    28824       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_6 (Conv1D)               (None, 49987, 24)    33624       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_7 (Conv1D)               (None, 49985, 24)    38424       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_8 (Conv1D)               (None, 49983, 24)    43224       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_9 (Conv1D)               (None, 49981, 24)    48024       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d (GlobalMax (None, 24)           0           conv1d[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d (Globa (None, 24)           0           conv1d[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_1 (GlobalM (None, 24)           0           conv1d_1[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_1 (Glo (None, 24)           0           conv1d_1[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_2 (GlobalM (None, 24)           0           conv1d_2[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_2 (Glo (None, 24)           0           conv1d_2[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_3 (GlobalM (None, 24)           0           conv1d_3[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_3 (Glo (None, 24)           0           conv1d_3[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_4 (GlobalM (None, 24)           0           conv1d_4[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_4 (Glo (None, 24)           0           conv1d_4[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_5 (GlobalM (None, 24)           0           conv1d_5[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_5 (Glo (None, 24)           0           conv1d_5[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_6 (GlobalM (None, 24)           0           conv1d_6[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_6 (Glo (None, 24)           0           conv1d_6[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_7 (GlobalM (None, 24)           0           conv1d_7[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_7 (Glo (None, 24)           0           conv1d_7[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_8 (GlobalM (None, 24)           0           conv1d_8[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_8 (Glo (None, 24)           0           conv1d_8[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_9 (GlobalM (None, 24)           0           conv1d_9[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_9 (Glo (None, 24)           0           conv1d_9[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "concatenate (Concatenate)       (None, 480)          0           global_max_pooling1d[0][0]       \n",
            "                                                                 global_average_pooling1d[0][0]   \n",
            "                                                                 global_max_pooling1d_1[0][0]     \n",
            "                                                                 global_average_pooling1d_1[0][0] \n",
            "                                                                 global_max_pooling1d_2[0][0]     \n",
            "                                                                 global_average_pooling1d_2[0][0] \n",
            "                                                                 global_max_pooling1d_3[0][0]     \n",
            "                                                                 global_average_pooling1d_3[0][0] \n",
            "                                                                 global_max_pooling1d_4[0][0]     \n",
            "                                                                 global_average_pooling1d_4[0][0] \n",
            "                                                                 global_max_pooling1d_5[0][0]     \n",
            "                                                                 global_average_pooling1d_5[0][0] \n",
            "                                                                 global_max_pooling1d_6[0][0]     \n",
            "                                                                 global_average_pooling1d_6[0][0] \n",
            "                                                                 global_max_pooling1d_7[0][0]     \n",
            "                                                                 global_average_pooling1d_7[0][0] \n",
            "                                                                 global_max_pooling1d_8[0][0]     \n",
            "                                                                 global_average_pooling1d_8[0][0] \n",
            "                                                                 global_max_pooling1d_9[0][0]     \n",
            "                                                                 global_average_pooling1d_9[0][0] \n",
            "__________________________________________________________________________________________________\n",
            "dropout (Dropout)               (None, 480)          0           concatenate[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "dense (Dense)                   (None, 256)          123136      dropout[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "dropout_1 (Dropout)             (None, 256)          0           dense[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "dense_1 (Dense)                 (None, 8)            2056        dropout_1[0][0]                  \n",
            "==================================================================================================\n",
            "Total params: 419,632\n",
            "Trainable params: 389,432\n",
            "Non-trainable params: 30,200\n",
            "__________________________________________________________________________________________________\n",
            "WARNING:tensorflow:tpu_model (from tensorflow.contrib.tpu.python.tpu.keras_support) is experimental and may change or be removed at any time, and without warning.\n",
            "Train on 11108 samples, validate on 2779 samples\n",
            "Epoch 1/100\n",
            "WARNING:tensorflow:Cannot update non-variable config: epsilon\n",
            "WARNING:tensorflow:Cannot update non-variable config: amsgrad\n",
            "11108/11108 [==============================] - 133s 12ms/sample - loss: 1.1626 - acc: 0.6034 - val_loss: 0.6344 - val_acc: 0.8120\n",
            "Epoch 2/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.6919 - acc: 0.7905 - val_loss: 0.5064 - val_acc: 0.8530\n",
            "Epoch 3/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.5702 - acc: 0.8284 - val_loss: 0.4386 - val_acc: 0.8779\n",
            "Epoch 4/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.5129 - acc: 0.8485 - val_loss: 0.4112 - val_acc: 0.8840\n",
            "Epoch 5/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.4781 - acc: 0.8578 - val_loss: 0.3734 - val_acc: 0.8890\n",
            "Epoch 6/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.4516 - acc: 0.8655 - val_loss: 0.3716 - val_acc: 0.8916\n",
            "Epoch 7/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.4307 - acc: 0.8721 - val_loss: 0.3690 - val_acc: 0.8912\n",
            "Epoch 8/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.4051 - acc: 0.8803 - val_loss: 0.3535 - val_acc: 0.8977\n",
            "Epoch 9/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.3908 - acc: 0.8836 - val_loss: 0.3476 - val_acc: 0.8981\n",
            "Epoch 10/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.3829 - acc: 0.8860 - val_loss: 0.3445 - val_acc: 0.8995\n",
            "Epoch 11/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.3643 - acc: 0.8926 - val_loss: 0.3428 - val_acc: 0.9031\n",
            "Epoch 12/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.3561 - acc: 0.8945 - val_loss: 0.3350 - val_acc: 0.9013\n",
            "Epoch 13/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.3533 - acc: 0.8957 - val_loss: 0.3293 - val_acc: 0.9006\n",
            "Epoch 14/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.3426 - acc: 0.8988 - val_loss: 0.3286 - val_acc: 0.9020\n",
            "Epoch 15/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.3350 - acc: 0.9002 - val_loss: 0.3318 - val_acc: 0.9045\n",
            "Epoch 16/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.3313 - acc: 0.9026 - val_loss: 0.3298 - val_acc: 0.9027\n",
            "Epoch 17/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.3281 - acc: 0.9044 - val_loss: 0.3269 - val_acc: 0.9053\n",
            "Epoch 18/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.3216 - acc: 0.9036 - val_loss: 0.3325 - val_acc: 0.9053\n",
            "Epoch 19/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.3044 - acc: 0.9101 - val_loss: 0.3260 - val_acc: 0.9063\n",
            "Epoch 20/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.3075 - acc: 0.9057 - val_loss: 0.3352 - val_acc: 0.9078\n",
            "Epoch 21/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.3020 - acc: 0.9090 - val_loss: 0.3315 - val_acc: 0.9045\n",
            "Epoch 22/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.2982 - acc: 0.9099 - val_loss: 0.3475 - val_acc: 0.9053\n",
            "Epoch 23/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.2913 - acc: 0.9127 - val_loss: 0.3367 - val_acc: 0.9074\n",
            "Epoch 24/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.2869 - acc: 0.9125 - val_loss: 0.3349 - val_acc: 0.9078\n",
            "Epoch 25/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.2839 - acc: 0.9170 - val_loss: 0.3362 - val_acc: 0.9085\n",
            "Epoch 26/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.2796 - acc: 0.9180 - val_loss: 0.3321 - val_acc: 0.9096\n",
            "Epoch 27/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.2739 - acc: 0.9192 - val_loss: 0.3357 - val_acc: 0.9074\n",
            "Epoch 28/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.2686 - acc: 0.9207 - val_loss: 0.3287 - val_acc: 0.9089\n",
            "Epoch 29/100\n",
            "11108/11108 [==============================] - 37s 3ms/sample - loss: 0.2737 - acc: 0.9187 - val_loss: 0.3389 - val_acc: 0.9078\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nt4ru5rWMdHQAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Cannot update non-variable config: epsilon\n",
            "WARNING:tensorflow:Cannot update non-variable config: amsgrad\n",
            "2779/2779 [==============================] - 56s 20ms/sample\n",
            "12955/12955 [==============================] - 266s 21ms/sample\n",
            "11108/11108 [==============================] - 222s 20ms/sample\n",
            "train confuse matrix is\n",
            "[[3972    0    2    0    0    6    0    2]\n",
            " [   1  397    0    0    0    3    0    0]\n",
            " [   5    0  940    0    0    9    0    3]\n",
            " [   1    8    4  523    0   18    3   99]\n",
            " [   2    0    1   52    3    4    2   16]\n",
            " [   4    0    1   48    0 3367    1   10]\n",
            " [   2    0    3    2    0   11  299   95]\n",
            " [   6   18   19  113    0   29   19  985]]\n",
            "val confuse matrix is\n",
            "[[985   0   5   0   0   1   1   4]\n",
            " [  2  97   0   0   0   2   0   0]\n",
            " [  5   0 228   0   0   2   0   4]\n",
            " [  2   1   2 110   0   5   2  42]\n",
            " [  0   0   0  12   0   0   0   8]\n",
            " [  3   0   4   9   0 834   0   8]\n",
            " [  5   0   2   1   0   3  59  33]\n",
            " [ 20   6  15  29   0   4  14 210]]\n",
            "train accuracy is 0.9440043212099388\n",
            "valid accuracy is 0.9078805325656711\n",
            "FOLD:  3\n",
            "2777 11110\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_1 (InputLayer)            (None, 50000)        0                                            \n",
            "__________________________________________________________________________________________________\n",
            "embedding (Embedding)           (None, 50000, 100)   30200       input_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "spatial_dropout1d (SpatialDropo (None, 50000, 100)   0           embedding[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "conv1d (Conv1D)                 (None, 49999, 24)    4824        spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_1 (Conv1D)               (None, 49997, 24)    9624        spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_2 (Conv1D)               (None, 49995, 24)    14424       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_3 (Conv1D)               (None, 49993, 24)    19224       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_4 (Conv1D)               (None, 49991, 24)    24024       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_5 (Conv1D)               (None, 49989, 24)    28824       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_6 (Conv1D)               (None, 49987, 24)    33624       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_7 (Conv1D)               (None, 49985, 24)    38424       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_8 (Conv1D)               (None, 49983, 24)    43224       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_9 (Conv1D)               (None, 49981, 24)    48024       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d (GlobalMax (None, 24)           0           conv1d[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d (Globa (None, 24)           0           conv1d[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_1 (GlobalM (None, 24)           0           conv1d_1[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_1 (Glo (None, 24)           0           conv1d_1[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_2 (GlobalM (None, 24)           0           conv1d_2[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_2 (Glo (None, 24)           0           conv1d_2[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_3 (GlobalM (None, 24)           0           conv1d_3[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_3 (Glo (None, 24)           0           conv1d_3[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_4 (GlobalM (None, 24)           0           conv1d_4[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_4 (Glo (None, 24)           0           conv1d_4[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_5 (GlobalM (None, 24)           0           conv1d_5[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_5 (Glo (None, 24)           0           conv1d_5[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_6 (GlobalM (None, 24)           0           conv1d_6[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_6 (Glo (None, 24)           0           conv1d_6[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_7 (GlobalM (None, 24)           0           conv1d_7[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_7 (Glo (None, 24)           0           conv1d_7[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_8 (GlobalM (None, 24)           0           conv1d_8[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_8 (Glo (None, 24)           0           conv1d_8[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_9 (GlobalM (None, 24)           0           conv1d_9[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_9 (Glo (None, 24)           0           conv1d_9[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "concatenate (Concatenate)       (None, 480)          0           global_max_pooling1d[0][0]       \n",
            "                                                                 global_average_pooling1d[0][0]   \n",
            "                                                                 global_max_pooling1d_1[0][0]     \n",
            "                                                                 global_average_pooling1d_1[0][0] \n",
            "                                                                 global_max_pooling1d_2[0][0]     \n",
            "                                                                 global_average_pooling1d_2[0][0] \n",
            "                                                                 global_max_pooling1d_3[0][0]     \n",
            "                                                                 global_average_pooling1d_3[0][0] \n",
            "                                                                 global_max_pooling1d_4[0][0]     \n",
            "                                                                 global_average_pooling1d_4[0][0] \n",
            "                                                                 global_max_pooling1d_5[0][0]     \n",
            "                                                                 global_average_pooling1d_5[0][0] \n",
            "                                                                 global_max_pooling1d_6[0][0]     \n",
            "                                                                 global_average_pooling1d_6[0][0] \n",
            "                                                                 global_max_pooling1d_7[0][0]     \n",
            "                                                                 global_average_pooling1d_7[0][0] \n",
            "                                                                 global_max_pooling1d_8[0][0]     \n",
            "                                                                 global_average_pooling1d_8[0][0] \n",
            "                                                                 global_max_pooling1d_9[0][0]     \n",
            "                                                                 global_average_pooling1d_9[0][0] \n",
            "__________________________________________________________________________________________________\n",
            "dropout (Dropout)               (None, 480)          0           concatenate[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "dense (Dense)                   (None, 256)          123136      dropout[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "dropout_1 (Dropout)             (None, 256)          0           dense[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "dense_1 (Dense)                 (None, 8)            2056        dropout_1[0][0]                  \n",
            "==================================================================================================\n",
            "Total params: 419,632\n",
            "Trainable params: 389,432\n",
            "Non-trainable params: 30,200\n",
            "__________________________________________________________________________________________________\n",
            "WARNING:tensorflow:tpu_model (from tensorflow.contrib.tpu.python.tpu.keras_support) is experimental and may change or be removed at any time, and without warning.\n",
            "Train on 11110 samples, validate on 2777 samples\n",
            "Epoch 1/100\n",
            "WARNING:tensorflow:Cannot update non-variable config: epsilon\n",
            "WARNING:tensorflow:Cannot update non-variable config: amsgrad\n",
            "11110/11110 [==============================] - 140s 13ms/sample - loss: 1.1923 - acc: 0.5847 - val_loss: 0.6871 - val_acc: 0.7875\n",
            "Epoch 2/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.7183 - acc: 0.7784 - val_loss: 0.5489 - val_acc: 0.8332\n",
            "Epoch 3/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.5898 - acc: 0.8228 - val_loss: 0.4816 - val_acc: 0.8577\n",
            "Epoch 4/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.5231 - acc: 0.8409 - val_loss: 0.4556 - val_acc: 0.8628\n",
            "Epoch 5/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.4800 - acc: 0.8555 - val_loss: 0.4393 - val_acc: 0.8746\n",
            "Epoch 6/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.4454 - acc: 0.8652 - val_loss: 0.4311 - val_acc: 0.8754\n",
            "Epoch 7/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.4269 - acc: 0.8743 - val_loss: 0.4046 - val_acc: 0.8775\n",
            "Epoch 8/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.4117 - acc: 0.8801 - val_loss: 0.3840 - val_acc: 0.8858\n",
            "Epoch 9/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.3945 - acc: 0.8809 - val_loss: 0.3777 - val_acc: 0.8872\n",
            "Epoch 10/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.3825 - acc: 0.8862 - val_loss: 0.3700 - val_acc: 0.8880\n",
            "Epoch 11/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.3716 - acc: 0.8890 - val_loss: 0.3670 - val_acc: 0.8865\n",
            "Epoch 12/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.3570 - acc: 0.8959 - val_loss: 0.3621 - val_acc: 0.8880\n",
            "Epoch 13/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.3502 - acc: 0.8959 - val_loss: 0.3598 - val_acc: 0.8916\n",
            "Epoch 14/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.3423 - acc: 0.9004 - val_loss: 0.3616 - val_acc: 0.8912\n",
            "Epoch 15/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.3369 - acc: 0.8985 - val_loss: 0.3564 - val_acc: 0.8898\n",
            "Epoch 16/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.3227 - acc: 0.9033 - val_loss: 0.3581 - val_acc: 0.8945\n",
            "Epoch 17/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.3235 - acc: 0.9030 - val_loss: 0.3557 - val_acc: 0.8919\n",
            "Epoch 18/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.3174 - acc: 0.9069 - val_loss: 0.3511 - val_acc: 0.8984\n",
            "Epoch 19/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.3123 - acc: 0.9074 - val_loss: 0.3474 - val_acc: 0.8984\n",
            "Epoch 20/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.3091 - acc: 0.9113 - val_loss: 0.3468 - val_acc: 0.8981\n",
            "Epoch 21/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2993 - acc: 0.9118 - val_loss: 0.3558 - val_acc: 0.8898\n",
            "Epoch 22/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2928 - acc: 0.9124 - val_loss: 0.3448 - val_acc: 0.8995\n",
            "Epoch 23/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2911 - acc: 0.9151 - val_loss: 0.3529 - val_acc: 0.8898\n",
            "Epoch 24/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2838 - acc: 0.9135 - val_loss: 0.3448 - val_acc: 0.9002\n",
            "Epoch 25/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2775 - acc: 0.9194 - val_loss: 0.3470 - val_acc: 0.8970\n",
            "Epoch 26/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2751 - acc: 0.9166 - val_loss: 0.3512 - val_acc: 0.8995\n",
            "Epoch 27/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2687 - acc: 0.9190 - val_loss: 0.3421 - val_acc: 0.9002\n",
            "Epoch 28/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2689 - acc: 0.9192 - val_loss: 0.3468 - val_acc: 0.8984\n",
            "Epoch 29/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2583 - acc: 0.9233 - val_loss: 0.3457 - val_acc: 0.8977\n",
            "Epoch 30/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2595 - acc: 0.9228 - val_loss: 0.3457 - val_acc: 0.8991\n",
            "Epoch 31/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2531 - acc: 0.9223 - val_loss: 0.3476 - val_acc: 0.8999\n",
            "Epoch 32/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2532 - acc: 0.9230 - val_loss: 0.3501 - val_acc: 0.9002\n",
            "Epoch 33/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2527 - acc: 0.9265 - val_loss: 0.3468 - val_acc: 0.9002\n",
            "Epoch 34/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2456 - acc: 0.9310 - val_loss: 0.3478 - val_acc: 0.8999\n",
            "Epoch 35/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2463 - acc: 0.9282 - val_loss: 0.3412 - val_acc: 0.9027\n",
            "Epoch 36/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2392 - acc: 0.9313 - val_loss: 0.3490 - val_acc: 0.8999\n",
            "Epoch 37/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2418 - acc: 0.9301 - val_loss: 0.3404 - val_acc: 0.9013\n",
            "Epoch 38/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2371 - acc: 0.9313 - val_loss: 0.3427 - val_acc: 0.9024\n",
            "Epoch 39/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2376 - acc: 0.9309 - val_loss: 0.3465 - val_acc: 0.9009\n",
            "Epoch 40/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2360 - acc: 0.9325 - val_loss: 0.3435 - val_acc: 0.9017\n",
            "Epoch 41/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2329 - acc: 0.9338 - val_loss: 0.3484 - val_acc: 0.9009\n",
            "Epoch 42/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2335 - acc: 0.9325 - val_loss: 0.3464 - val_acc: 0.9002\n",
            "Epoch 43/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2260 - acc: 0.9382 - val_loss: 0.3532 - val_acc: 0.9006\n",
            "Epoch 44/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2260 - acc: 0.9369 - val_loss: 0.3537 - val_acc: 0.9031\n",
            "Epoch 45/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2158 - acc: 0.9386 - val_loss: 0.3413 - val_acc: 0.9024\n",
            "Epoch 46/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2228 - acc: 0.9381 - val_loss: 0.3399 - val_acc: 0.9053\n",
            "Epoch 47/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2194 - acc: 0.9387 - val_loss: 0.3415 - val_acc: 0.9056\n",
            "Epoch 48/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2208 - acc: 0.9361 - val_loss: 0.3545 - val_acc: 0.8999\n",
            "Epoch 49/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2133 - acc: 0.9386 - val_loss: 0.3463 - val_acc: 0.9074\n",
            "Epoch 50/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2097 - acc: 0.9398 - val_loss: 0.3524 - val_acc: 0.9031\n",
            "Epoch 51/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2119 - acc: 0.9393 - val_loss: 0.3632 - val_acc: 0.9009\n",
            "Epoch 52/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2102 - acc: 0.9392 - val_loss: 0.3565 - val_acc: 0.9045\n",
            "Epoch 53/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2080 - acc: 0.9418 - val_loss: 0.3461 - val_acc: 0.9067\n",
            "Epoch 54/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2052 - acc: 0.9416 - val_loss: 0.3533 - val_acc: 0.9042\n",
            "Epoch 55/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2080 - acc: 0.9453 - val_loss: 0.3526 - val_acc: 0.9038\n",
            "Epoch 56/100\n",
            "11110/11110 [==============================] - 37s 3ms/sample - loss: 0.2062 - acc: 0.9426 - val_loss: 0.3558 - val_acc: 0.9045\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0\ndHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nO3deXxc1X338c9vpJFGu2RJXiUvGNvY\nGGODccwSYgIEGxKgDymEYBLSNCRt0iZPEhpoljZp+zR5pUnTtJCEJGQPS0hYEkxZEhsIm22MARsb\nb9iW5E2WJWtfZuY8f5wrW7YlIRuNxtL9vl+vec1y78ycK83c75xz7j3HnHOIiEh4RdJdABERSS8F\ngYhIyCkIRERCTkEgIhJyCgIRkZBTEIiIhJyCQGSAzOynZvavA1x3u5ld8nZfR2QoKAhEREJOQSAi\nEnIKAhlRgiaZW8zsVTNrMbMfm9kYM3vUzJrM7EkzK+mx/pVmtt7MGsxshZnN7LFsnpmtCZ53LxA7\n6r3ea2Zrg+c+Z2ZzTrDMHzOzLWZ2wMweNrPxweNmZv9pZvvMrNHMXjOz2cGyy83s9aBsNWb2+RP6\ng4mgIJCR6RrgUmA68D7gUeAfgXL8Z/7vAcxsOnA38Jlg2TLg92aWZWZZwIPAL4BRwG+C1yV47jzg\nLuDjQCnwA+BhM8s+noKa2buBfweuBcYBO4B7gsXvAS4MtqMoWKcuWPZj4OPOuQJgNvCn43lfkZ4U\nBDIS/bdzbq9zrgZ4BnjROfeyc64deACYF6x3HfCIc+4J51wX8B9ADnAesBCIAt9xznU55+4HVvV4\nj5uBHzjnXnTOJZxzPwM6gucdjxuAu5xza5xzHcBtwLlmNhnoAgqA0wBzzm1wzu0OntcFzDKzQudc\nvXNuzXG+r8ghCgIZifb2uN3Wy/384PZ4/C9wAJxzSaAKmBAsq3FHjsq4o8ftScDngmahBjNrACqD\n5x2Po8vQjP/VP8E59yfgf4DbgX1mdqeZFQarXgNcDuwws6fM7NzjfF+RQxQEEma78Dt0wLfJ43fm\nNcBuYELwWLeJPW5XAf/mnCvuccl1zt39NsuQh29qqgFwzn3XOXc2MAvfRHRL8Pgq59xVwGh8E9Z9\nx/m+IocoCCTM7gOuMLOLzSwKfA7fvPMc8DwQB/7ezKJm9n+ABT2e+0PgE2b2jqBTN8/MrjCzguMs\nw93AR8xsbtC/8P/wTVnbzeyc4PWjQAvQDiSDPowbzKwoaNJqBJJv4+8gIacgkNByzr0BLAX+G9iP\n71h+n3Ou0znXCfwf4CbgAL4/4Xc9nrsa+Bi+6aYe2BKse7xleBL4MvBbfC1kKvCBYHEhPnDq8c1H\ndcA3g2U3AtvNrBH4BL6vQeSEmCamEREJN9UIRERCTkEgIhJyCgIRkZBTEIiIhFxmugtwvMrKytzk\nyZPTXQwRkWHlpZde2u+cK+9t2bALgsmTJ7N69ep0F0NEZFgxsx19LVPTkIhIyCkIRERCTkEgIhJy\nw66PoDddXV1UV1fT3t6e7qKkXCwWo6Kigmg0mu6iiMgIkbIgMLO7gPcC+5xzs3tZfgPwBcCAJuBv\nnHOvnMh7VVdXU1BQwOTJkzlysMiRxTlHXV0d1dXVTJkyJd3FEZERIpVNQz8FFvez/E3gXc65M4B/\nAe480Tdqb2+ntLR0RIcAgJlRWloaipqPiAydlNUInHNPB7Ms9bX8uR53XwAq3s77jfQQ6BaW7RSR\noXOydBZ/FD+vbK/M7GYzW21mq2tra0/oDdq7Euw52E48oWHbRUR6SnsQmNlF+CD4Ql/rOOfudM7N\nd87NLy/v9cS4t9TRlWBfUztdKQiChoYG7rjjjuN+3uWXX05DQ8Ogl0dE5HikNQjMbA7wI+Aq51xd\nKt8rEvFNKokUTL/QVxDE4/F+n7ds2TKKi4sHv0AiIschbYePmtlE/IxPNzrnNqX6/TKCIEgmBz8J\nbr31VrZu3crcuXOJRqPEYjFKSkrYuHEjmzZt4uqrr6aqqor29nY+/elPc/PNNwOHh8tobm5myZIl\nXHDBBTz33HNMmDCBhx56iJycnEEvq4jI0VJ5+OjdwCKgzMyqgX8CogDOue8DX8FP0n1H0AEad87N\nf7vv+9Xfr+f1XY3HPO6co7UzQXY0g8zI8XW4zhpfyD+97/Q+l3/9619n3bp1rF27lhUrVnDFFVew\nbt26Q4d43nXXXYwaNYq2tjbOOeccrrnmGkpLS494jc2bN3P33Xfzwx/+kGuvvZbf/va3LF269LjK\nKSJyIlJ51ND1b7H8r4G/TtX7HyM42sZPzZnaI28WLFhwxHH+3/3ud3nggQcAqKqqYvPmzccEwZQp\nU5g7dy4AZ599Ntu3b09pGUVEuo2IM4t76uuXezLpWLfrIGMLY4wujKW0DHl5eYdur1ixgieffJLn\nn3+e3NxcFi1a1Ot5ANnZ2YduZ2Rk0NbWltIyioh0S/tRQ0MlEjHMjIQb/D6CgoICmpqael128OBB\nSkpKyM3NZePGjbzwwguD/v4iIm/HiKsR9CfDjEQKOotLS0s5//zzmT17Njk5OYwZM+bQssWLF/P9\n73+fmTNnMmPGDBYuXDjo7y8i8naYS8Ev5FSaP3++O3pimg0bNjBz5sy3fO4bexrJiWYysTQ3VcUb\nEgPdXhGRbmb2Ul8H5ISmaQggkqKmIRGR4SxUQZARSU3TkIjIcBa6IEiqRiAicoRQBUEkRZ3FIiLD\nWaiCICNiKRliQkRkOAtVEHR3Fg+3I6VERFIpVEFwaOC5QQ6CEx2GGuA73/kOra2tg1oeEZHjEbIg\n8NeDPSWBgkBEhrNQnVkcse45CQa3RtBzGOpLL72U0aNHc99999HR0cFf/MVf8NWvfpWWlhauvfZa\nqqurSSQSfPnLX2bv3r3s2rWLiy66iLKyMpYvXz6o5RIRGYiRFwSP3gp7Xut1UX4yySldSbKyMg6N\nRjogY8+AJV/vc3HPYagff/xx7r//flauXIlzjiuvvJKnn36a2tpaxo8fzyOPPAL4MYiKior49re/\nzfLlyykrKzuuzRQRGSyhahqyI4aiTo3HH3+cxx9/nHnz5nHWWWexceNGNm/ezBlnnMETTzzBF77w\nBZ555hmKiopSVgYRkeMx8moE/fxy7+pKsG1vExNH5VKcm5WSt3fOcdttt/Hxj3/8mGVr1qxh2bJl\nfOlLX+Liiy/mK1/5SkrKICJyPEJVI+g+amiwTyrrOQz1ZZddxl133UVzczMANTU17Nu3j127dpGb\nm8vSpUu55ZZbWLNmzTHPFRFJh5FXI+hHRoo6i3sOQ71kyRI++MEPcu655wKQn5/PL3/5S7Zs2cIt\nt9xCJBIhGo3yve99D4Cbb76ZxYsXM378eHUWi0hahGoYaucc62oaKSvIYlzR8J0YXsNQi8jx0jDU\nATMjI4KGmRAR6SFUQQB+ysqEckBE5JAREwQDbeLKsOE98Nxwa8oTkZPfiAiCWCxGXV3dgHaSkWE8\nOY1zjrq6OmKxWLqLIiIjyIg4aqiiooLq6mpqa2vfct265g7iSUdn3fDcmcZiMSoqKtJdDBEZQUZE\nEESjUaZMmTKgdT9771pefLOeZ299d4pLJSIyPIyIpqHjURDLpKm9K93FEBE5aYQwCKI0d8TV6Soi\nEghhEGSSdNDSmUh3UURETgohDIIogJqHREQCIQwC3z/e1B5Pc0lERE4OIQ4C1QhERCCFQWBmd5nZ\nPjNb18dyM7PvmtkWM3vVzM5KVVl66m4aalSNQEQESG2N4KfA4n6WLwGmBZebge+lsCyHFKppSETk\nCCkLAufc08CBfla5Cvi5814Ais1sXKrK002dxSIiR0pnH8EEoKrH/ergsWOY2c1mttrMVg9kGIn+\nqLNYRORIw6Kz2Dl3p3NuvnNufnl5+dt6rdysDCKmGoGISLd0BkENUNnjfkXwWEqZGfnZmaoRiIgE\n0hkEDwMfCo4eWggcdM7tHoo3LohFaVYQiIgAKRx91MzuBhYBZWZWDfwTEAVwzn0fWAZcDmwBWoGP\npKosRyuIZerwURGRQMqCwDl3/Vssd8AnU/X+/SmMRdVHICISGBadxYPND0WtGoGICIQ5CDpUIxAR\ngdAGQVQ1AhGRQEiDwDcNaXIaEZHQBkGURNLR1qXJaUREQhoEGmZCRKRbyINAHcYiIqEMgkLNSSAi\nckgog0BNQyIih4UyCPLVNCQickgog+Dw5DSqEYiIhDQIfI1AI5CKiIQ0CPKzMjFNTiMiAoQ0CCIR\nIz9LQ1GLiEBIgwA0AqmISLcQB4HmJBARgVAHgWoEIiIQ9iDQnAQiImEOAs1JICICoQ4CNQ2JiECo\ng8B3FmtyGhEJuxAHQSZdCUdHPJnuooiIpFVog6AwGGaiUYeQikjIhTYI8jUUtYgIEOIgKMj2I5Bq\n4DkRCbvwBoFqBCIiQKiDoHtOAvURiEi4hTgIVCMQEYEQB8HhCexVIxCRcAttEOioIRERL7RBkBEx\n8rIyFAQiEnopDQIzW2xmb5jZFjO7tZflE81suZm9bGavmtnlqSzP0TQngYhICoPAzDKA24ElwCzg\nejObddRqXwLuc87NAz4A3JGq8vRGA8+JiKS2RrAA2OKc2+ac6wTuAa46ah0HFAa3i4BdKSzPMTQn\ngYhIaoNgAlDV43518FhP/wwsNbNqYBnwd729kJndbGarzWx1bW3toBVQcxKIiKS/s/h64KfOuQrg\ncuAXZnZMmZxzdzrn5jvn5peXlw/am6tpSEQktUFQA1T2uF8RPNbTR4H7AJxzzwMxoCyFZTqCDwI1\nDYlIuKUyCFYB08xsipll4TuDHz5qnZ3AxQBmNhMfBIPX9vMWCmJRGlUjEJGQS1kQOOfiwKeAx4AN\n+KOD1pvZ18zsymC1zwEfM7NXgLuBm9wQThlWkJ1JZzxJRzwxVG8pInLSyUzlizvnluE7gXs+9pUe\nt18Hzk9lGfrTPd5Qc3uc7PyMdBVDRCSt0t1ZnFaHRyBV85CIhFfIg0DjDYmIhDwINCeBiEjIg6B7\nAnvVCEQkvEIdBIWqEYiIhDsI1EcgIhLyINDkNCIiIQ+CaEaEnGiGmoZEJNQGFARm9mkzKzTvx2a2\nxszek+rCDQUNPCciYTfQGsFfOecagfcAJcCNwNdTVqohpDkJRCTsBhoEFlxfDvzCObe+x2PD2pjC\nGNX1bekuhohI2gw0CF4ys8fxQfCYmRUAydQVa+jMqShmw+5G2rs08JyIhNNAg+CjwK3AOc65ViAK\nfCRlpRpCcyuL6Uo41u9qTHdRRETSYqBBcC7whnOuwcyW4iedP5i6Yg2deROLAVhb1ZDmkoiIpMdA\ng+B7QKuZnYmfQ2Ar8POUlWoIjSmMMa4opiAQkdAaaBDEgwljrgL+xzl3O1CQumKlQNMeeOUeiHcc\ns2huZTFrq+rTUCgRkfQbaBA0mdlt+MNGHwkmmI+mrlgpsOM5eODjsO/1YxbNrSym6kAbdc3HhoSI\nyEg30CC4DujAn0+wBz8R/TdTVqpUGD/XX+9ae8yiuZXqJxCR8BpQEAQ7/18BRWb2XqDdOTe8+ghK\npkCsCHYfGwRnVBSRETEFgYiE0kCHmLgWWAn8JXAt8KKZvT+VBRt0ZjBuLux6+ZhFuVmZTB9ToCAQ\nkVAa6OT1X8SfQ7APwMzKgSeB+1NVsJQYPxeev8N3GGdmH7FobmUxf3h1F8mkIxIZESdNi4gMyED7\nCCLdIRCoO47nnjzGzYVkV68dxvMqi2lqj7Ntf0saCiYikj4D3Zn/r5k9ZmY3mdlNwCPAstQVK0X6\n6zDWiWUiElID7Sy+BbgTmBNc7nTOfSGVBUuJfjqMp5bnk5+dqfMJRCR0BtpHgHPut8BvU1iW1Oun\nwzgjYsypKFKNQERCp98agZk1mVljL5cmMxueo7SNnwt7X+/zDOONu5s0EqmIhEq/QeCcK3DOFfZy\nKXDOFQ5VIQdVfx3GE0uIJx3rakbEeHoiIgMy/I78ebt0hrGIyBHCFwT9dBiXF2QzoTiHlxUEIhIi\n4QuCQx3GxwYB+MNI1+5UEIhIeIQvCCDoMF7fa4fxvMpiahraqG3SSKQiEg4pDQIzW2xmb5jZFjO7\ntY91rjWz181svZn9OpXlOaSfDmP1E4hI2KQsCMwsA7gdWALMAq43s1lHrTMNuA043zl3OvCZVJXn\nCP10GM+eUERmxHRimYiERiprBAuALc65bc65TuAe/AxnPX0MuN05Vw9w1HhGqdNPh3EsmsFp4zQS\nqYiERyqDYAJQ1eN+dfBYT9OB6Wb2rJm9YGaLe3shM7vZzFab2era2tq3X7K36jCuLObVqoMkk+7t\nv5eIyEku3Z3FmcA0YBFwPfBDMys+eiXn3J3OufnOufnl5eWD8879dBjPrSyhqSPOpn1Ng/NeIiIn\nsVQGQQ1Q2eN+RfBYT9XAw865Lufcm8AmfDCkXj8dxu+cVkZGxHhgzdHFFREZeVIZBKuAaWY2xcyy\ngA8ADx+1zoP42gBmVoZvKtqWwjId1k+H8ZjCGJedPoZ7V1dp3CERGfFSFgTOuTjwKeAxYANwn3Nu\nvZl9zcyuDFZ7DKgzs9eB5cAtzrm6VJXpCP10GAMsXTiJhtYu/vDq7iEpjohIugx4GOoT4ZxbxlET\n2DjnvtLjtgM+G1yG1lt0GJ97Simnjs7nFy/s4P1nVwxx4UREhk66O4vTq58OYzPjxoWTeKWqgVer\ndSipiIxc4Q6CfjqMAf7irAnkZmXwi+d3DHHBRESGTriDoJ8OY4DCWJSr503g4Vd20dDaOYQFExEZ\nOuEOgu4O416mrux248JJdMST3P9S9RAWTERk6IQ7CMxg8jvhjWUQ7/0X/8xxhZwzuYRfvrBDZxqL\nyIgU7iAAOPsmaKmFjb/vc5WlCyexva6VZ7bsH7pyiYgMEQXB1IuheBKsuqvPVRbPHktZfpY6jUVk\nRFIQRCIw/69gx59h38ZeV8nOzOC6cyr508a9VNe3DnEBRURSS0EAMG8pZGTB6h/3ucoH3zEJgF+/\nuHOoSiUiMiQUBAB5ZTDranjlHuho7nWVCcU5XHb6WH763HbVCkRkRFEQdDvno9DRCOvu73OVL14x\nE4AvPbgOPzqGiMjwpyDoVvkOGH06rPox9LGTryjJ5ZbLZrDijVoeWrtriAsoIpIaCoJuZr5WsOdV\nqHmpz9U+dO5k5lYW89Xfr6eu+dgxikREhhsFQU9zroWsfF8r6ENGxPjGNXNoao/zr49sGMLCiYik\nhoKgp+wCmHMdrPsttB7oc7UZYwv420VTeeDlGla8sW8ICygiMvgUBEc756OQ6IC1v+p3tU+++1Sm\nlufxxQfW0dIRH6LCiYgMPgXB0cacDpULYfVdkEz2uVp2ZgZfv2YONQ1tfOvxTUNYQBGRwaUg6M05\nfw0HtsHyf+3zCCKAcyaP4saFk/jJc2+y8s2+m5JERE5mCoLezL4GzvoQPPMtePQL/dYM/mHxDCaX\n5vG3v1rDnoPtQ1hIEZHBoSDoTSQC7/sunPspWPkDeOhvIdF7P0BBLMqdN55NW2ecT/zyJTriiSEu\nrIjI26Mg6IsZvOdf4aIvwSt3w28+3OvcxgDTxhTwrWvPZG1VA195cL3OOhaRYUVB0B8zeNctsPgb\nsPEP8OvroLOl11UXzx7HJy+ayr2rq/j1Sg1MJyLDh4JgIBZ+Aq66A958Cn50Cex4rtfVPnvpDBbN\nKOefH17P6u3qPBaR4UFBMFDzboDr74WOJvjJErj/o9B45HhDGRHjv66bx/jiHP7mV2vY26jOYxE5\n+SkIjsf098AnV8KF/wAbfg//PR+e+fYRfQdFuVHuvHE+LR1xPvTjlWzc05jGAouIvDUFwfHKyoV3\nfxE++SKcsgj++FW4YyE8/R+wZx04x4yxBXx/6dnsb+7gff/9Z/7nT5uJJ/o+BFVEJJ1suB3hMn/+\nfLd69ep0F+OwLU/C8v93eMTSokqYfhlMX0Jd+QK+smwLj7y6mzkVRfzHX57J9DEF6S2viISSmb3k\nnJvf6zIFwSBp3A2bH4dN/wtbl0O8DUadAh96iEd2RvnyQ+tobo/z6Uum8fELTyEzQ5UxERk6CoKh\n1tUGm5+Ahz8FWQXw4YfZn13Blx9cx6Pr9jC1PI9bl8zkkpmjMbN0l1ZEQqC/INDP0lSI5sCsK+HD\nf/A1g58soax1G3fccBY//NB8HPCxn6/mujtf4JWqhnSXVkRCTkGQSuPmwE3LAIOfXI7teZVLZ43h\nsc9cyL9cdTpb9zVz1e3P8vd3v0zVgdZ0l1ZEQiqlQWBmi83sDTPbYma39rPeNWbmzKzXasuwNvo0\n+MgyyMqDn74PqlYRzYhw47mTWXHLIj510ak8/voeLvn2U9y+fAudcR1dJCJDK2VBYGYZwO3AEmAW\ncL2ZzeplvQLg08CLqSpL2pVO9WGQOwp+cTW88SjgB6z7/GUzWP75RVw8czTffOwNLv/uM7y4rS7N\nBRaRMElljWABsMU5t8051wncA1zVy3r/AnwDGNmn4RZPhI886o8kuvsD8NgXId4JwLiiHO644Wx+\nctM5tHcluO7OF7jlN69woKUzzYUWkTBIZRBMAKp63K8OHjvEzM4CKp1zj/T3QmZ2s5mtNrPVtbW1\ng1/SoVI4Dj76BCy4GZ7/H/jJYqjffmjxRaeN5on/+y4+8S4/H/LF31rB91ZspaFVgSAiqZO2zmIz\niwDfBj73Vus65+50zs13zs0vLy9PfeFSKRqDy78J1/4c9m+B718Irz90aHFOVga3LjmNP/z9Bcwa\nX8g3/ncjC//9j9z2u9fYtLcpjQUXkZEqZecRmNm5wD875y4L7t8G4Jz79+B+EbAVaA6eMhY4AFzp\nnOvzRIFhcR7BQNVvh998BHatgdnvh2mXQuU7oGSyHwIb2LC7kZ8+u50H19bQEU9y/qml3HTeFC6a\nUa6T0kRkwNJyQpmZZQKbgIuBGmAV8EHn3Po+1l8BfL6/EIARFgTg+wn+9C/w0k+hIxigLn8MVC7w\noTDlXTBmNgfa4ty9cie/eH4HexrbGVsY49r5FVx7TiUVJblp3QQROfml7cxiM7sc+A6QAdzlnPs3\nM/sasNo59/BR664gjEHQLZmAfRug6kV/2fkCNOzwy/JGw9R3w6kX0zX5Xfxxp+OeVTt5apPvL7lw\nWjnXL6jkkpljVEsQkV5piInhqnGXH7do6x/9dVsw2U3ZDCiupDl7NK825rN8VyZvtBXQWTyVpZdd\nwOVnjCcS0dAVInKYgmAkSCZg91rY8ifY9TI01vigaNl3xGqNLofq6GSKJ5/JuOnzsfIZvqkptwxy\nSiByVI3BOehqhbYGSHZB0cRj1xGRIyUTsPVP0LQbKhdC2bRD/Xonq/6CIHOoCyMnKJIBE872l57i\nnf7DeLCaZO0b7F//Ip07XyVv88PYlnuOXNcy/EltuWXgEtBWfzgAukXzYOxsGDvHD5Ex9gw/cF77\nQWhvCK4P+oH1ckdBXpl/vbxyfzszO/V/C5F0qd8BL/8S1v7K/xjrllsGk86FSefDpPP892ewgyGZ\n9N/VFHzHVCMYgTrjSe5ZuYN7//giRW07GJ/ZzJmlcWYWdDAp1kqpNRLJiEJOsa8lxIr9bQz2vQ67\nX4U9r0HnCRyumj8Gyk+D0TN7XM/w73M8nIM9r/pDa19/2M8CN/d6OOtDUFTR+3PinbDzed+ENuVd\nPqhORLwDMrJO+l94I0YyCbUb/FzgrXWQjEOiy18n434Qx7k3+F/dg/J+ieBz9RCMOxOmL/af077+\n38374M2nfQBsW+EfO/Vi/1ksnwlVL/iy73gWGnb65SVTfJnP/AAUV/Zehj2v+c9rV5v/vGVkQUbU\nXye74GANHKyGg1X+unEXnP9pPzHWCVDTUEi1dsb508Z9vLjtAC++Wcemvf5I3Vg0wjumlLJoRjmL\nZoxmSlnesU9OJqH+Tb8zjnf6oIgVHb5Ec3yNomU/tNT6S3Otf86+DVD7BnS1HH69nBJ/WGzJlOB6\nMuSP9h/8SPTwFyHeDpsf81/S+u2+FjPlnWAR309iBqdeCmffBNPe43f6mx+HTY/55d3hZRGoWOCn\nF512GYw5/fAXPdEFrQf8Tqdplz+fo24z7N8MdVv8L73iSTDrKph1NUw4q/edRDLhO/Tjnf7vkZXn\nrzNz+m5eS8QP167a6v2l9YBv4mve63c63ZdIhp/oqLjSXxdVQOF4v3PsavN/q662oHZWChMX9h1+\nyQTUrPETKXW1+p3fxIX+PY5HMunLmZUHscK+1zm4E/a+DrUb/f8ip8SXLacEckZBIgjt7X/2O9Hu\n/i8ALPhcZPrPRleL3+bpS+C8T/lf3ScS0l3t8Mrd8Ox/+c9pbqn/DIBvEp1+mf+75JZA1SqoXgXV\nKw/v3IsqYd5Sv4PvbecOfoe9bQW8eq8PDwymXOifM3a239Y3n4I3n/Gfgf5YBArG+f9792Xqu/3M\niCdAQSAA1DV3sGr7AV7YdoCnN9eyrdbvqCeV5rJoejkXTi/nzMpiyvIHoeqZTPpfMrUb/aV+++FL\nw07/xe5LJNN/2GddBTOugLxS/3j9dljzC//LrHmPr8l0f5kKxvlgmPYeHzCbn/CBsvuVYPl4X6Vu\nPQAdB499z+wiKDsVSqdBySS/09y23JezqNKXpXIBHNgG+zb6mtP+TX5n3JuM7N53Vn2tD5AZ82XP\nH+OPFEt2QUOV/zt2Nvf9vJ7KT4OJ5wbNE2f4/qTNT/j27PYGv3OJZPodcW4pzFgCp73X/70zY36d\nljpo3e9Dvmm33+YDb/rr+u2QCObozsr3wVQwDgonQEam/xGwb8PAy1s8ESZdAJMvgMnn995H1VwL\nq34Iq37kd9zj5sJ5fwenXgLZBX2HmXPQ2eLDdv3v4PnbfYiNnwcXfBZOu8IHbveEUttW+JDsVjgB\nKuZDxTn+R0XF/OMLzvod8Mo9vhmp+whA8Nt4yoW+1jr5Ah+OiU7/IyHR6S8WgYKxPhAHiYJAerWz\nrpUVm/ax4o1antu6n/YuP/Lp2MIYsycUcvr4ImZPKOIdp4yiMDZ4H0gScWis9l/Qnh/+RBfg/A63\nv6akRJevAWx42DcXTLvM7/R62/E27oYtT8C2p/zy3NLgMspf54+B0lN9H8fRz2+r9wMErn/Q70i7\n+1IKxvumhO7mr6zcw7/MO+CP2lYAAA+4SURBVFuCX+ttx5bFueCXdNAk1/OSXw7Zhb1vg3N+B91Q\n5Xdkkcyg5hGDaK4/W/1gDex8DnY87w8/7j4nBfw2nnqJb8445SK/c9nyJGx8xP8dOxp9cLnkkf1F\n3aK5foysksnB9STobPVNFU27/HXjLh9y5TNhzKzg73N60OQSCWo/Bw7XgFzS/5+LJ/b9fz5aV5v/\nRf/87b7m1rN8WfmQne9rY51N0N4IHU2+L6zbKYt8AEy5sPe/c1e7r6F0tcCE+VA04dh1TkQy6f83\nDTt9SI+aMjive5wUBPKW2rsSrK1qYF3NQdbvamRdzUG21jaTdFAYy+TmC0/hpvOnkJ8d0uML2g/6\nJqTSqUF/ykksmYC9630b9NgzYMzsvpuq4p2w/RkfdBnRHh3/pf52wVgfJCdTf0ky6Q+prn3D1zw6\nmoLrZh9GWXm++TK70DdfZRf6WsD4uekueVopCOSEtHbGebX6ID96ZhtPbtjHqLws/uZdU7nx3EnE\nosfZtiwiaaUgkLft5Z31fPuJTTyzeT+jC7L5yPlTKMqJ0hlP0JlI0tGVpDORpCw/m3kTi5k5rpCo\nznIWOWkoCGTQvLitjm89vomV2w/0u152ZoQ5FUXMm1jC3MpiTh9fSGVJrs54FkkTBYEMKuccexrb\nMYyszAjZmRGyMiNkRozdB9t5eWcDa3bW8/LOetbVNNKZ8J3Q+dmZzBhbwMxxBcwcV8ip5fmML85h\nbFFMtQeRFFMQSNp0xBNs3N3Eht2NwcXfbuo4fPioGYwpiDG+OEZFSS6LZpTzntPHhrdjWiQFNMSE\npE12ZgZnVhZzZuXhI22cc1TXt7G9roVdDW3UNLSzq6GNXQ1tvLCtjodf2UUs+hqXzhrLVWeO58Lp\n5WRlqsYgkioKAhlyZkblqFwqRx07j0Iy6XhpZz0PvlzDI6/t5vev7KI4N8olM8ewYMooFkwexaTS\nXOxkOpxRZJhT05CctDrjSZ7ZXMuDa3fx9KZaDrb5k53KC7I5Z3IJZ08axbiiGIWxKIU5mRTEohTG\nMinKiWpeBpGjqGlIhqWszAgXzxzDxTPHkEw6ttQ2s/LNA6zafoDV2+tZ9tqeXp+Xm5XBgimjOH9q\nGeefWsZpYwt0tJJIP1QjkGFrX2M7dS2dNLZ10dQep7HdX2+tbebZLfvZGoylNCovi/OmlvKOKaM4\na1IJp40tJEPBICGjGoGMSKMLY4wujPW5fPfBNp7bUsezW/bz7Nb9/OHV3QDkZWUwb2IJZ08qYfaE\nIkYXZFOan0VZfrbOmJZQUo1AQqH7SKWXdtQfumzc00jyqI9/fnYmpflZTB9TwPxJJcyf7MMiO1MB\nIcObagQSej2PVLp6nh9Vsqm9i621LdQ1d7C/uYP9zZ3UNXeyr6mddTUHeeL1vYDvq5gzoYiZ43yT\nknOOpIOkczjgjAlFvHfOOAoGc4RWkSGkGoFIH2qbOnhph++YXr2jnq21zRgQiRgRMyIGiaSjvrWL\nnGgGl58xjuvOqeScySU6vFVOOjqzWCRFnHOsrWrgvtVV/P6V3TR3xDmlLI8lZ4wlYkZLR4LWzjgt\nnQnaOuOMLoz5JqdJo6gclaPAkCGjIBAZAq2dcZa9tof7VlWxcvsBIgZ5WZnkZGWQl51JLJpB9YHW\nQ8NrdJ8PccaEYrIyI3R/F7u/kmaQGTEyMyKHrmPRCHMri6koOfZkPJH+KAhEhlhnPEk0w475xZ9I\nOjbtbWL1jnpWB+dD1DT0MpvZW5g4KpfzppZybnAZXeCPnnLOEU86uhJJkg6N1ySHKAhETmKN7V24\nJGC+FmD4zu2kc8QTjngy6a8Tjsb2LlZtP8BzW+t4YVsdTe2+dpGfnUlnIklXIknPr/SUsjzOnVrK\neVNLWXhK6THzUTvnaO6IU9/SxehCHT47kikIREagRNKxftdBnttax97GdrIyI2RlRIhm+GHBE0nH\nmh31rHzzwKHmqNPGFlA5Kpe65g5qmzuobeo4NFd1xGBSaR7Tx+QzfUwB08cUMK4oRjzpQ6grCJp4\n0pETzaAwJ5PCWNQP7ZGTSU40Q30eJzEdPioyAmVEjDkVxcyp6H8O5XgiybpdjTy3dT/PbaljR10L\n5QXZnD2xhPKCbMoLsinOyaKmoY1Ne5vYtLeJJzfsI3H0SRZvISszwtjCGOOK/GVsUQ7jimJkZtih\nIOlM+NpNdmaEd5xSyhkTinSW90lANQIROUZHPMG22hb2NXUQjRjRYOKhaEaEzAyjrTNBY3v8iOE9\nDrR0sudgO3sOtrPrYBt7G9vpSvS/fynKiXLe1FIumFbGO08tp6IkBwiayFS7GFSqEYjIccnOzGDm\nuEJmjjvx10gmHXUtnSSdOxQgWcERUA1tXTy7ZT9/3ryfP2/Zz6Preh9AEPyQIJfNHss1Z1Ww8JTS\nXmsQzjm27W9h4+4mSvOzqCjJYWxhTKPQDpBqBCKSVs45tta28NzW/dS3+KHGHQ7nwAF7Drbx6Lo9\nNLXHGVcU46q5E7jmrAlEMyI8v813mj+/tY59TR1HvG5GxBhbGGNCcQ6xrAxaO/z5HK2dcVo64nQl\nHGdWFvPOU8t45/QyZowpGNG1EHUWi8iw1t6V4MkNe/ndmhqe2lR7RP9FWX62P4z2lFLmVBRR39pJ\nTX0b1fVt1DS0UVPfRkc8QV52JrlZmeRl+/M6nHOsfPPAoVFqywuyueDUMs6aVMLEUblUluQwoSTn\nmHGmnHO0dCaob+mkK5EkP+Y7zU/2I67SFgRmthj4LyAD+JFz7utHLf8s8NdAHKgF/so5t6O/11QQ\niIRbbVMHj67bjQHnTi1lann+2/olv6uhjT9v2c8zm/fz7Jb9HGjpPLTMjEMd4K2dCepbO6lv6aIz\nkTzmdbIyIhTEMinMiVKcG2VUbhYleVmMysuiJDeL0vwsxhTGGBtcCnMyD5W7I55gX2MHexvb2dvY\nQXNHFwWxKEU5hy+FOVEKsjNPeG6NtASBmWUAm4BLgWpgFXC9c+71HutcBLzonGs1s78BFjnnruvv\ndRUEIpIqyaRjb1M7VQfaqDrQSlV9KzsPtLLnYDt52ZmU5Eb9zj3YyWdlRGhq76KxPU5Te/zQ7YbW\nTg60+EtdSyed8WODIxaNUF6QTUtH4ojw6c/H3jmFL14x64S2LV2dxQuALc65bUEh7gGuAg4FgXNu\neY/1XwCWprA8IiL9ikSMcUU5jCvKYcGUUYPyms452roS1DZ1sK+pgz0H24Nf/u3UNnWQl53J2MIY\nYwpjjCmKMaYwm/zsTJra4xxs6zp0aWzrYvaEokEp09FSGQQTgKoe96uBd/Sz/keBR3tbYGY3AzcD\nTJw4cbDKJyKScmZGblYmk0ozmVSal+7i9OqkOLbKzJYC84Fv9rbcOXenc26+c25+eXn50BZORGSE\nS2WNoAao7HG/InjsCGZ2CfBF4F3OuY6jl4uISGqlskawCphmZlPMLAv4APBwzxXMbB7wA+BK59y+\nFJZFRET6kLIgcM7FgU8BjwEbgPucc+vN7GtmdmWw2jeBfOA3ZrbWzB7u4+VERCRFUjrEhHNuGbDs\nqMe+0uP2Jal8fxEReWsnRWexiIikj4JARCTkFAQiIiE37AadM7NaoN/xiPpRBuwfxOKcbEby9mnb\nhq+RvH3DadsmOed6PRFr2AXB22Fmq/saa2MkGMnbp20bvkby9o2UbVPTkIhIyCkIRERCLmxBcGe6\nC5BiI3n7tG3D10jevhGxbaHqIxARkWOFrUYgIiJHURCIiIRcaILAzBab2RtmtsXMbk13ed4uM7vL\nzPaZ2boej40ysyfMbHNwXZLOMp4oM6s0s+Vm9rqZrTezTwePD/vtM7OYma00s1eCbftq8PgUM3sx\n+HzeG4zYOyyZWYaZvWxmfwjuj6Rt225mrwWDZK4OHhv2n8tQBEEwf/LtwBJgFnC9mZ3YxJ8nj58C\ni4967Fbgj865acAfg/vDURz4nHNuFrAQ+GTw/xoJ29cBvNs5dyYwF1hsZguBbwD/6Zw7FajHz9g3\nXH0aP+Jwt5G0bQAXOefm9jh/YNh/LkMRBPSYP9k51wl0z588bDnnngYOHPXwVcDPgts/A64e0kIN\nEufcbufcmuB2E36nMoERsH3Oaw7uRoOLA94N3B88Piy3DcDMKoArgB8F940Rsm39GPafy7AEQW/z\nJ09IU1lSaYxzbndwew8wJp2FGQxmNhmYB7zICNm+oOlkLbAPeALYCjQEc3jA8P58fgf4ByAZ3C9l\n5Gwb+NB+3MxeCuZShxHwuUzpfASSPs45Z2bD+thgM8sHfgt8xjnX6H9cesN5+5xzCWCumRUDDwCn\npblIg8LM3gvsc869ZGaL0l2eFLnAOVdjZqOBJ8xsY8+Fw/VzGZYawYDmTx4B9prZOIDgethO/2lm\nUXwI/Mo597vg4RGzfQDOuQZgOXAuUGxm3T/Mhuvn83zgSjPbjm9+fTfwX4yMbQPAOVcTXO/Dh/gC\nRsDnMixB8JbzJ48QDwMfDm5/GHgojWU5YUG78o+BDc65b/dYNOy3z8zKg5oAZpYDXIrvA1kOvD9Y\nbVhum3PuNudchXNuMv479ifn3A2MgG0DMLM8Myvovg28B1jHSPhchuXMYjO7HN9+mQHc5Zz7tzQX\n6W0xs7uBRfhhcPcC/wQ8CNwHTMQP1X2tc+7oDuWTnpldADwDvMbhtuZ/xPcTDOvtM7M5+A7FDPwP\nsfucc18zs1Pwv6JHAS8DS51zHekr6dsTNA193jn33pGybcF2PBDczQR+7Zz7NzMrZbh/LsMSBCIi\n0ruwNA2JiEgfFAQiIiGnIBARCTkFgYhIyCkIRERCTkEgMoTMbFH3qJwiJwsFgYhIyCkIRHphZkuD\neQPWmtkPgoHims3sP4N5BP5oZuXBunPN7AUze9XMHugej97MTjWzJ4O5B9aY2dTg5fPN7H4z22hm\nv7KegyiJpIGCQOQoZjYTuA443zk3F0gANwB5wGrn3OnAU/izuQF+DnzBOTcHfzZ09+O/Am4P5h44\nD+geoXIe8Bn83Bin4MfoEUkbjT4qcqyLgbOBVcGP9Rz8QGJJ4N5gnV8CvzOzIqDYOfdU8PjPgN8E\nY9JMcM49AOCcawcIXm+lc646uL8WmAz8OfWbJdI7BYHIsQz4mXPutiMeNPvyUeud6PgsPcfZSaDv\noaSZmoZEjvVH4P3BmPPdc9JOwn9fukfR/CDwZ+fcQaDezN4ZPH4j8FQws1q1mV0dvEa2meUO6VaI\nDJB+iYgcxTn3upl9CT8TVQToAj4JtAALgmX78P0I4Ice/n6wo98GfCR4/EbgB2b2teA1/nIIN0Nk\nwDT6qMgAmVmzcy4/3eUQGWxqGhIRCTnVCEREQk41AhGRkFMQiIiEnIJARCTkFAQiIiGnIBARCbn/\nDytSdejck55aAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Cannot update non-variable config: epsilon\n",
            "WARNING:tensorflow:Cannot update non-variable config: amsgrad\n",
            "2777/2777 [==============================] - 53s 19ms/sample\n",
            "12955/12955 [==============================] - 246s 19ms/sample\n",
            "11110/11110 [==============================] - 207s 19ms/sample\n",
            "train confuse matrix is\n",
            "[[3978    0    1    0    0    1    0    2]\n",
            " [   0  399    0    0    0    3    0    0]\n",
            " [   0    0  951    0    0    3    0    3]\n",
            " [   1    7    2  597    7   16    2   24]\n",
            " [   0    0    0   13   59    4    1    3]\n",
            " [   1    0    2   40    0 3384    2    2]\n",
            " [   0    0    0    3    0    2  399    8]\n",
            " [   2   19   10   97   16   18   38  990]]\n",
            "val confuse matrix is\n",
            "[[991   0   1   0   0   3   0   1]\n",
            " [  3  95   0   0   0   2   0   0]\n",
            " [  9   0 213   0   0   6   0  11]\n",
            " [  2   2   1 130   5   4   1  19]\n",
            " [  3   0   0   5   8   1   1   2]\n",
            " [  4   0   1  14   1 831   3   4]\n",
            " [  8   0   1   3   0   4  69  18]\n",
            " [ 12   5  11  54   6   9  26 174]]\n",
            "train accuracy is 0.9682268226822682\n",
            "valid accuracy is 0.9042131796903133\n",
            "FOLD:  4\n",
            "2776 11111\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_1 (InputLayer)            (None, 50000)        0                                            \n",
            "__________________________________________________________________________________________________\n",
            "embedding (Embedding)           (None, 50000, 100)   30200       input_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "spatial_dropout1d (SpatialDropo (None, 50000, 100)   0           embedding[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "conv1d (Conv1D)                 (None, 49999, 24)    4824        spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_1 (Conv1D)               (None, 49997, 24)    9624        spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_2 (Conv1D)               (None, 49995, 24)    14424       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_3 (Conv1D)               (None, 49993, 24)    19224       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_4 (Conv1D)               (None, 49991, 24)    24024       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_5 (Conv1D)               (None, 49989, 24)    28824       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_6 (Conv1D)               (None, 49987, 24)    33624       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_7 (Conv1D)               (None, 49985, 24)    38424       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_8 (Conv1D)               (None, 49983, 24)    43224       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "conv1d_9 (Conv1D)               (None, 49981, 24)    48024       spatial_dropout1d[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d (GlobalMax (None, 24)           0           conv1d[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d (Globa (None, 24)           0           conv1d[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_1 (GlobalM (None, 24)           0           conv1d_1[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_1 (Glo (None, 24)           0           conv1d_1[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_2 (GlobalM (None, 24)           0           conv1d_2[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_2 (Glo (None, 24)           0           conv1d_2[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_3 (GlobalM (None, 24)           0           conv1d_3[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_3 (Glo (None, 24)           0           conv1d_3[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_4 (GlobalM (None, 24)           0           conv1d_4[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_4 (Glo (None, 24)           0           conv1d_4[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_5 (GlobalM (None, 24)           0           conv1d_5[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_5 (Glo (None, 24)           0           conv1d_5[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_6 (GlobalM (None, 24)           0           conv1d_6[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_6 (Glo (None, 24)           0           conv1d_6[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_7 (GlobalM (None, 24)           0           conv1d_7[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_7 (Glo (None, 24)           0           conv1d_7[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_8 (GlobalM (None, 24)           0           conv1d_8[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_8 (Glo (None, 24)           0           conv1d_8[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling1d_9 (GlobalM (None, 24)           0           conv1d_9[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling1d_9 (Glo (None, 24)           0           conv1d_9[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "concatenate (Concatenate)       (None, 480)          0           global_max_pooling1d[0][0]       \n",
            "                                                                 global_average_pooling1d[0][0]   \n",
            "                                                                 global_max_pooling1d_1[0][0]     \n",
            "                                                                 global_average_pooling1d_1[0][0] \n",
            "                                                                 global_max_pooling1d_2[0][0]     \n",
            "                                                                 global_average_pooling1d_2[0][0] \n",
            "                                                                 global_max_pooling1d_3[0][0]     \n",
            "                                                                 global_average_pooling1d_3[0][0] \n",
            "                                                                 global_max_pooling1d_4[0][0]     \n",
            "                                                                 global_average_pooling1d_4[0][0] \n",
            "                                                                 global_max_pooling1d_5[0][0]     \n",
            "                                                                 global_average_pooling1d_5[0][0] \n",
            "                                                                 global_max_pooling1d_6[0][0]     \n",
            "                                                                 global_average_pooling1d_6[0][0] \n",
            "                                                                 global_max_pooling1d_7[0][0]     \n",
            "                                                                 global_average_pooling1d_7[0][0] \n",
            "                                                                 global_max_pooling1d_8[0][0]     \n",
            "                                                                 global_average_pooling1d_8[0][0] \n",
            "                                                                 global_max_pooling1d_9[0][0]     \n",
            "                                                                 global_average_pooling1d_9[0][0] \n",
            "__________________________________________________________________________________________________\n",
            "dropout (Dropout)               (None, 480)          0           concatenate[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "dense (Dense)                   (None, 256)          123136      dropout[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "dropout_1 (Dropout)             (None, 256)          0           dense[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "dense_1 (Dense)                 (None, 8)            2056        dropout_1[0][0]                  \n",
            "==================================================================================================\n",
            "Total params: 419,632\n",
            "Trainable params: 389,432\n",
            "Non-trainable params: 30,200\n",
            "__________________________________________________________________________________________________\n",
            "WARNING:tensorflow:tpu_model (from tensorflow.contrib.tpu.python.tpu.keras_support) is experimental and may change or be removed at any time, and without warning.\n",
            "Train on 11111 samples, validate on 2776 samples\n",
            "Epoch 1/100\n",
            "WARNING:tensorflow:Cannot update non-variable config: epsilon\n",
            "WARNING:tensorflow:Cannot update non-variable config: amsgrad\n",
            "11111/11111 [==============================] - 139s 13ms/sample - loss: 1.1547 - acc: 0.6107 - val_loss: 0.6499 - val_acc: 0.8051\n",
            "Epoch 2/100\n",
            "11111/11111 [==============================] - 37s 3ms/sample - loss: 0.6708 - acc: 0.8028 - val_loss: 0.5164 - val_acc: 0.8408\n",
            "Epoch 3/100\n",
            "11111/11111 [==============================] - 37s 3ms/sample - loss: 0.5554 - acc: 0.8373 - val_loss: 0.5015 - val_acc: 0.8527\n",
            "Epoch 4/100\n",
            "11111/11111 [==============================] - 37s 3ms/sample - loss: 0.5077 - acc: 0.8484 - val_loss: 0.4501 - val_acc: 0.8674\n",
            "Epoch 5/100\n",
            "11111/11111 [==============================] - 37s 3ms/sample - loss: 0.4696 - acc: 0.8619 - val_loss: 0.4256 - val_acc: 0.8696\n",
            "Epoch 6/100\n",
            "11111/11111 [==============================] - 37s 3ms/sample - loss: 0.4409 - acc: 0.8700 - val_loss: 0.4062 - val_acc: 0.8818\n",
            "Epoch 7/100\n",
            "11111/11111 [==============================] - 37s 3ms/sample - loss: 0.4256 - acc: 0.8764 - val_loss: 0.3826 - val_acc: 0.8898\n",
            "Epoch 8/100\n",
            " 8576/11111 [======================>.......] - ETA: 7s - loss: 0.4067 - acc: 0.8809"
          ],
          "name": "stdout"
        },
        {
          "output_type": "error",
          "ename": "KeyboardInterrupt",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-14-bec20ee6dd94>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     45\u001b[0m                     \u001b[0mshuffle\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     46\u001b[0m                     \u001b[0mcallbacks\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mearly_stopping\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 47\u001b[0;31m                     \u001b[0mclass_weight\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mclass_weights\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     48\u001b[0m                     )\n\u001b[1;32m     49\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/tensorflow/contrib/tpu/python/tpu/keras_support.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, **kwargs)\u001b[0m\n\u001b[1;32m   1530\u001b[0m                                   \u001b[0mvalidation_split\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvalidation_data\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mshuffle\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1531\u001b[0m                                   \u001b[0mclass_weight\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msample_weight\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minitial_epoch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1532\u001b[0;31m                                   steps_per_epoch, validation_steps, **kwargs)\n\u001b[0m\u001b[1;32m   1533\u001b[0m       \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1534\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_numpy_to_infeed_manager_list\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/tensorflow/contrib/tpu/python/tpu/keras_support.py\u001b[0m in \u001b[0;36m_pipeline_fit\u001b[0;34m(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, **kwargs)\u001b[0m\n\u001b[1;32m   1631\u001b[0m         \u001b[0minitial_epoch\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0minitial_epoch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1632\u001b[0m         \u001b[0msteps_per_epoch\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msteps_per_epoch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1633\u001b[0;31m         validation_steps=validation_steps)\n\u001b[0m\u001b[1;32m   1634\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1635\u001b[0m   def _pipeline_fit_loop(self,\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/tensorflow/contrib/tpu/python/tpu/keras_support.py\u001b[0m in \u001b[0;36m_pipeline_fit_loop\u001b[0;34m(self, inputs, targets, sample_weights, batch_size, epochs, verbose, callbacks, val_inputs, val_targets, val_sample_weights, shuffle, initial_epoch, steps_per_epoch, validation_steps)\u001b[0m\n\u001b[1;32m   1730\u001b[0m             \u001b[0mval_sample_weights\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mval_sample_weights\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1731\u001b[0m             \u001b[0mvalidation_steps\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mvalidation_steps\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1732\u001b[0;31m             epoch_logs=epoch_logs)\n\u001b[0m\u001b[1;32m   1733\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1734\u001b[0m       \u001b[0mcallbacks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mon_epoch_end\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mepoch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mepoch_logs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/tensorflow/contrib/tpu/python/tpu/keras_support.py\u001b[0m in \u001b[0;36m_pipeline_fit_loop_sample_wise\u001b[0;34m(self, ins, callbacks, index_array, shuffle, batch_size, num_training_samples, indices_for_conversion_to_dense, do_validation, val_inputs, val_targets, val_sample_weights, validation_steps, epoch_logs)\u001b[0m\n\u001b[1;32m   1786\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1787\u001b[0m       outs = f.pipeline_run(\n\u001b[0;32m-> 1788\u001b[0;31m           cur_step_inputs=ins_last_batch, next_step_inputs=ins_batch)\n\u001b[0m\u001b[1;32m   1789\u001b[0m       \u001b[0mins_last_batch\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mins_batch\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1790\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/tensorflow/contrib/tpu/python/tpu/keras_support.py\u001b[0m in \u001b[0;36mpipeline_run\u001b[0;34m(***failed resolving arguments***)\u001b[0m\n\u001b[1;32m   1344\u001b[0m           \u001b[0mnext_tpu_model_ops\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0minfeed_op\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcur_tpu_model_ops\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mexecute_op\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1345\u001b[0m           \u001b[0mcur_tpu_model_ops\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moutfeed_op\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1346\u001b[0;31m       ], infeed_dict)\n\u001b[0m\u001b[1;32m   1347\u001b[0m       \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_process_outputs\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0moutfeed_outputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1348\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m    927\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    928\u001b[0m       result = self._run(None, fetches, feed_dict, options_ptr,\n\u001b[0;32m--> 929\u001b[0;31m                          run_metadata_ptr)\n\u001b[0m\u001b[1;32m    930\u001b[0m       \u001b[0;32mif\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    931\u001b[0m         \u001b[0mproto_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtf_session\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTF_GetBuffer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrun_metadata_ptr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_run\u001b[0;34m(self, handle, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1150\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mfinal_fetches\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mfinal_targets\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mhandle\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mfeed_dict_tensor\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1151\u001b[0m       results = self._do_run(handle, final_targets, final_fetches,\n\u001b[0;32m-> 1152\u001b[0;31m                              feed_dict_tensor, options, run_metadata)\n\u001b[0m\u001b[1;32m   1153\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1154\u001b[0m       \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_do_run\u001b[0;34m(self, handle, target_list, fetch_list, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1326\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mhandle\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1327\u001b[0m       return self._do_call(_run_fn, feeds, fetches, targets, options,\n\u001b[0;32m-> 1328\u001b[0;31m                            run_metadata)\n\u001b[0m\u001b[1;32m   1329\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1330\u001b[0m       \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_prun_fn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeeds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetches\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_do_call\u001b[0;34m(self, fn, *args)\u001b[0m\n\u001b[1;32m   1332\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1333\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1334\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1335\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0merrors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mOpError\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1336\u001b[0m       \u001b[0mmessage\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompat\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mas_text\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmessage\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_run_fn\u001b[0;34m(feed_dict, fetch_list, target_list, options, run_metadata)\u001b[0m\n\u001b[1;32m   1317\u001b[0m       \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_extend_graph\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1318\u001b[0m       return self._call_tf_sessionrun(\n\u001b[0;32m-> 1319\u001b[0;31m           options, feed_dict, fetch_list, target_list, run_metadata)\n\u001b[0m\u001b[1;32m   1320\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1321\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_prun_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_call_tf_sessionrun\u001b[0;34m(self, options, feed_dict, fetch_list, target_list, run_metadata)\u001b[0m\n\u001b[1;32m   1405\u001b[0m     return tf_session.TF_SessionRun_wrapper(\n\u001b[1;32m   1406\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_session\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moptions\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtarget_list\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1407\u001b[0;31m         run_metadata)\n\u001b[0m\u001b[1;32m   1408\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1409\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m_call_tf_sessionprun\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MEzh38V23hvi",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import pandas as pd\n",
        "\n",
        "test_df = pd.DataFrame(meta_test)\n",
        "test_df.columns = ['prob0', 'prob1', 'prob2', 'prob3', 'prob4', 'prob5', 'prob6', 'prob7']\n",
        "test_df.index += 1 \n",
        "test_df.index.name = 'file_id'\n",
        "\n",
        "test_df.to_csv(os.path.join(data_folder_path, 'secruity_submit_keep_50000_word2vec_adam_new_weights_seed3.csv'))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "piqPwfHl3hrc",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from google.colab import files\n",
        "files.download(os.path.join(data_folder_path, 'secruity_submit_keep_50000_word2vec_adam_new_weights_seed3.csv'))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jJTHrUxptbWj",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}