BiLSTM_74__WRI_text.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": " BiLSTM_74%_WRI_text.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/JoanneBurke/NLP-Projects/blob/master/BiLSTM_74__WRI_text.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hYwdEwMb-y6n",
        "colab_type": "code",
        "outputId": "0ee9e1b1-d28d-4c59-8728-87f4b16730b2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yX_07-Q5__Jz",
        "colab_type": "code",
        "outputId": "9b59b492-31b0-4ab6-c026-70a92946738c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "import pandas as pd\n",
        "import nltk\n",
        "nltk.download('stopwords')\n",
        "from nltk.corpus import stopwords\n",
        "\n",
        "import gensim\n",
        "from gensim.utils import simple_preprocess\n",
        "from tqdm import tqdm as tqdm\n",
        "\n",
        "from collections import Counter\n",
        "import numpy as np"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[nltk_data] Downloading package stopwords to /root/nltk_data...\n",
            "[nltk_data]   Package stopwords is already up-to-date!\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_XazWdlbODrt",
        "colab_type": "code",
        "outputId": "2f0d2933-5d15-4c2f-bf6b-568ae1ff5ef6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "pwd"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'/content'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 30
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fK51t6RAwIaA",
        "colab_type": "code",
        "outputId": "f3bf30bc-9c7c-4a3d-cf30-b8e6dbd26d0f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 170
        }
      },
      "source": [
        "import os\n",
        "\n",
        "% cd drive/My Drive/Colab Notebooks/Omdena_WRI_JB/\n",
        "\n",
        "% pwd\n",
        "% ls"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[Errno 2] No such file or directory: 'drive/My Drive/Colab Notebooks/Omdena_WRI_JB/'\n",
            "/content/drive/My Drive/Colab Notebooks/Omdena_WRI_JB\n",
            "'Copy of Attempt Bert.ipynb'          NER_ents.csv\n",
            " \u001b[0m\u001b[01;34menvironmental-conflict-tracker\u001b[0m/      PolicyMatch.csv\n",
            "'Gabriella -  LSTM_WRI.ipynb'         PolicyMatch.gsheet\n",
            "' Gabriella -  LSTM_WRI_text.ipynb'   Unlabelled1023.csv\n",
            " gold_standard.csv                   'wri_102919 (1).csv'\n",
            " manually_annotated.csv               wri.csv\n",
            "'NER and Matcher attempts.ipynb'     'WRI titles NN - 1st pass.ipynb'\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RyuoI7dD_h08",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "total= pd.read_csv('/content/drive/My Drive/Colab Notebooks/Omdena_WRI_JB/wri_102919 (1).csv', encoding='utf-8')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SshloRUOAkc1",
        "colab_type": "code",
        "outputId": "4e53ecc9-fcf8-4429-8815-89a60e2344f2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        }
      },
      "source": [
        "total['labels'].value_counts()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "negative    1344\n",
              "positive     459\n",
              "Name: labels, dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 51
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GFCdFn-ATElt",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "total['class_binary']=total['labels'].map({'negative':0,'positive':1})"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KvS_rVwHYkAS",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "total['text']= total['text'].str.replace(r'\\d+','')\n",
        "total['text ']= total['text'].str.replace(r'\\s+',' ')\n",
        "total['text']= total['text'].str.lower().str.strip()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "y3wxk5G9fROq",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "total= total[~total.text.isna()]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NMuORBFyH-_j",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "X= total.text.values\n",
        "y= total['class_binary'].values"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "o-TLtO-MazPK",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def sent_to_words(X):\n",
        "    '''Tokenizar las normas y eliminar puntuacion restante (acentos)'''\n",
        "    for linea in X:\n",
        "        yield(gensim.utils.simple_preprocess(str(linea), deacc=True))  # deacc=True removes punctuations\n",
        "\n",
        "X_token = list(sent_to_words(X))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "K9NgmXwKH8xg",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "stop_words= stopwords.words('english')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xwSja1EDTRFX",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def no_stopwords(X):\n",
        "    return [[word for word in line if word not in stop_words] for line in tqdm(X)]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5ZUYEXO7cGbv",
        "colab_type": "code",
        "outputId": "bd91c645-c568-4bb8-fde2-3c85dc8fca90",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "X_no_stopwords= no_stopwords(X_token)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "100%|██████████| 1803/1803 [00:01<00:00, 1463.27it/s]\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SVitL_szeMSH",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "vocab= Counter()\n",
        "for line in X_no_stopwords:\n",
        "    for word in line:\n",
        "        vocab[word]+=1\n",
        "\n",
        "vocab= sorted(vocab, key= vocab.get, reverse=True)\n",
        "word_to_id= {word:i for i,word in enumerate(vocab,1)}"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "P6njOGL-eg50",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "encoded_x= [[word_to_id[word] for word in line] for line in X_no_stopwords]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "H1Hu5HSRe28v",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "largo=[]\n",
        "for i,line in enumerate(X_no_stopwords):\n",
        "  largo.append(len(line))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kirsy3ehfAfC",
        "colab_type": "code",
        "outputId": "41f5221b-0392-4ac1-96eb-de35b7cc27a3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        }
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "plt.hist(largo)\n",
        "plt.title('Length of sentences')\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYoAAAEICAYAAABBBrPDAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0FI8XiovgYg9",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "MAX_LENGTH = 300\n",
        "EPOCHS = 15\n",
        "S_DROPOUT = .05\n",
        "DROPOUT = .05\n",
        "R_DROPOUT = .05\n",
        "\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9nKcnK4xewxj",
        "colab_type": "code",
        "outputId": "dbe819fb-bbe7-4bdf-bfb7-6b1b9d20635d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "\"\"\"def cut_pad(text, length_seq):\n",
        "    seq=[]\n",
        "    for line in text:\n",
        "        if len(line)>=length_seq:\n",
        "            seq.append(line[:length_seq])\n",
        "        else:\n",
        "            padding=[0]*(length_seq-len(line))\n",
        "            seq.append(padding+line)\n",
        "    return seq\n",
        "  \n",
        "padded_X = cut_pad(encoded_x,12)\n",
        "padded_X= np.array(padded_X)\"\"\"\n",
        "\n",
        "from keras.preprocessing.sequence import pad_sequences\n",
        "\n",
        "padded_X= pad_sequences(encoded_x,maxlen=MAX_LENGTH)\n",
        "padded_X.shape"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(1803, 300)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 65
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FdONuJdSH53i",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "X_train, X_test, y_train, y_test = train_test_split(padded_X, y,\n",
        "                                                    stratify=y, \n",
        "                                                    test_size=0.2)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Kcv33icAcObn",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from keras.datasets import imdb\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Dense, LSTM, SpatialDropout1D, Bidirectional\n",
        "from keras.layers.embeddings import Embedding\n",
        "from keras.preprocessing import sequence"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UxK09WR2kjws",
        "colab_type": "code",
        "outputId": "2177c7d5-62f3-40a6-a87c-9d8f288dd9fc",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 833
        }
      },
      "source": [
        "np.random.seed(7)\n",
        "\n",
        "embedding_vector_length = 32\n",
        "\n",
        "model = Sequential()\n",
        "model.add(Embedding(len(word_to_id)+1, embedding_vector_length, input_length=MAX_LENGTH))\n",
        "model.add(SpatialDropout1D(S_DROPOUT))\n",
        "model.add(Bidirectional(LSTM(60,dropout = DROPOUT, recurrent_dropout = R_DROPOUT)))\n",
        "model.add(Dense(1, activation='sigmoid'))\n",
        "model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
        "print(model.summary())\n",
        "model.fit(X_train, y_train, epochs=EPOCHS, batch_size=128)\n",
        "# Final evaluation of the model\n",
        "scores = model.evaluate(X_test, y_test, verbose=0)\n",
        "print(\"Accuracy: %.2f%%\" % (scores[1]*100))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_3\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "embedding_3 (Embedding)      (None, 300, 32)           1120192   \n",
            "_________________________________________________________________\n",
            "spatial_dropout1d_3 (Spatial (None, 300, 32)           0         \n",
            "_________________________________________________________________\n",
            "bidirectional_3 (Bidirection (None, 120)               44640     \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 1)                 121       \n",
            "=================================================================\n",
            "Total params: 1,164,953\n",
            "Trainable params: 1,164,953\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "None\n",
            "Epoch 1/15\n",
            "1442/1442 [==============================] - 15s 10ms/step - loss: 0.6595 - acc: 0.7393\n",
            "Epoch 2/15\n",
            "1442/1442 [==============================] - 12s 8ms/step - loss: 0.5641 - acc: 0.7455\n",
            "Epoch 3/15\n",
            "1442/1442 [==============================] - 12s 8ms/step - loss: 0.5349 - acc: 0.7455\n",
            "Epoch 4/15\n",
            "1442/1442 [==============================] - 12s 8ms/step - loss: 0.4875 - acc: 0.7455\n",
            "Epoch 5/15\n",
            "1442/1442 [==============================] - 12s 8ms/step - loss: 0.3518 - acc: 0.8162\n",
            "Epoch 6/15\n",
            "1442/1442 [==============================] - 12s 8ms/step - loss: 0.2384 - acc: 0.9542\n",
            "Epoch 7/15\n",
            "1442/1442 [==============================] - 12s 8ms/step - loss: 0.1426 - acc: 0.9730\n",
            "Epoch 8/15\n",
            "1442/1442 [==============================] - 12s 8ms/step - loss: 0.0840 - acc: 0.9834\n",
            "Epoch 9/15\n",
            "1442/1442 [==============================] - 12s 8ms/step - loss: 0.0549 - acc: 0.9903\n",
            "Epoch 10/15\n",
            "1442/1442 [==============================] - 12s 8ms/step - loss: 0.0469 - acc: 0.9958\n",
            "Epoch 11/15\n",
            "1442/1442 [==============================] - 12s 8ms/step - loss: 0.0343 - acc: 0.9903\n",
            "Epoch 12/15\n",
            "1442/1442 [==============================] - 12s 8ms/step - loss: 0.0241 - acc: 0.9979\n",
            "Epoch 13/15\n",
            "1442/1442 [==============================] - 12s 8ms/step - loss: 0.0234 - acc: 0.9945\n",
            "Epoch 14/15\n",
            "1442/1442 [==============================] - 12s 8ms/step - loss: 0.0182 - acc: 0.9972\n",
            "Epoch 15/15\n",
            "1442/1442 [==============================] - 12s 8ms/step - loss: 0.0154 - acc: 0.9972\n",
            "Accuracy: 87.81%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "K6MqNgp-l6fZ",
        "colab_type": "code",
        "outputId": "04d87fb3-ee69-43eb-d41c-a31408cf9a01",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 153
        }
      },
      "source": [
        "prediction = np.array([1 if pred>=0.5 else 0 for pred in model.predict(X_test)])\n",
        "\n",
        "true_positive= np.sum(prediction*y_test)\n",
        "false_positive= np.sum((y_test[np.where(prediction==1)]*prediction[np.where(prediction==1)])==0)\n",
        "false_negative=np.sum((y_test[np.where(prediction==0)]+prediction[np.where(prediction==0)])==1)\n",
        "true_negative = len(prediction) - true_positive - false_negative - false_positive\n",
        "\n",
        "precision= true_positive/(true_positive+false_positive)\n",
        "recall= true_positive/(true_positive+false_negative)\n",
        "\n",
        "f1= 2* precision* recall/(precision+recall)\n",
        "\n",
        "print('Precision: ', precision)\n",
        "print('Recall: ', recall)\n",
        "print('F1: ', f1)\n",
        "print('-'*65)\n",
        "print('True Positives: ', true_positive)\n",
        "print('True Negatives: ', true_negative)\n",
        "print('False Positives: ', false_positive)\n",
        "print('False Negatives: ', false_negative)\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Precision:  0.8076923076923077\n",
            "Recall:  0.6847826086956522\n",
            "F1:  0.7411764705882354\n",
            "-----------------------------------------------------------------\n",
            "True Positives:  63\n",
            "True Negatives:  254\n",
            "False Positives:  15\n",
            "False Negatives:  29\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "M9Q0b6j4zN1e",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from tensorflow.keras.models import save_model\n",
        "model.save_weights('text_bilstm_weights.hdf5')\n",
        "model.save('text_bilstm_model.hdf5')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "suEHAZXQ7KA0",
        "colab_type": "code",
        "outputId": "8ba92e01-6c25-4f63-b975-aed0aac9fc20",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "dir(model)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['__call__',\n",
              " '__class__',\n",
              " '__delattr__',\n",
              " '__dict__',\n",
              " '__dir__',\n",
              " '__doc__',\n",
              " '__eq__',\n",
              " '__format__',\n",
              " '__ge__',\n",
              " '__getattribute__',\n",
              " '__getstate__',\n",
              " '__gt__',\n",
              " '__hash__',\n",
              " '__init__',\n",
              " '__init_subclass__',\n",
              " '__le__',\n",
              " '__lt__',\n",
              " '__module__',\n",
              " '__ne__',\n",
              " '__new__',\n",
              " '__reduce__',\n",
              " '__reduce_ex__',\n",
              " '__repr__',\n",
              " '__setattr__',\n",
              " '__setstate__',\n",
              " '__sizeof__',\n",
              " '__str__',\n",
              " '__subclasshook__',\n",
              " '__weakref__',\n",
              " '_add_inbound_node',\n",
              " '_base_init',\n",
              " '_build_input_shape',\n",
              " '_built',\n",
              " '_check_trainable_weights_consistency',\n",
              " '_collected_trainable_weights',\n",
              " '_compute_previous_mask',\n",
              " '_expects_training_arg',\n",
              " '_feed_input_names',\n",
              " '_feed_input_shapes',\n",
              " '_feed_inputs',\n",
              " '_feed_loss_fns',\n",
              " '_feed_output_names',\n",
              " '_feed_output_shapes',\n",
              " '_feed_outputs',\n",
              " '_feed_sample_weight_modes',\n",
              " '_feed_sample_weights',\n",
              " '_feed_targets',\n",
              " '_function_kwargs',\n",
              " '_get_callback_model',\n",
              " '_get_node_attribute_at_index',\n",
              " '_inbound_nodes',\n",
              " '_init_graph_network',\n",
              " '_init_subclassed_network',\n",
              " '_initial_weights',\n",
              " '_input_coordinates',\n",
              " '_input_layers',\n",
              " '_is_compiled',\n",
              " '_is_graph_network',\n",
              " '_layers',\n",
              " '_layers_by_depth',\n",
              " '_losses',\n",
              " '_make_predict_function',\n",
              " '_make_test_function',\n",
              " '_make_train_function',\n",
              " '_network_nodes',\n",
              " '_node_key',\n",
              " '_nodes_by_depth',\n",
              " '_non_trainable_weights',\n",
              " '_outbound_nodes',\n",
              " '_output_coordinates',\n",
              " '_output_layers',\n",
              " '_output_mask_cache',\n",
              " '_output_shape_cache',\n",
              " '_output_tensor_cache',\n",
              " '_per_input_losses',\n",
              " '_per_input_updates',\n",
              " '_set_inputs',\n",
              " '_standardize_user_data',\n",
              " '_trainable_weights',\n",
              " '_updated_config',\n",
              " '_updates',\n",
              " '_uses_dynamic_learning_phase',\n",
              " '_uses_inputs_arg',\n",
              " '_validate_or_infer_batch_size',\n",
              " 'add',\n",
              " 'add_loss',\n",
              " 'add_update',\n",
              " 'add_weight',\n",
              " 'assert_input_compatibility',\n",
              " 'build',\n",
              " 'built',\n",
              " 'call',\n",
              " 'compile',\n",
              " 'compute_mask',\n",
              " 'compute_output_shape',\n",
              " 'count_params',\n",
              " 'dtype',\n",
              " 'evaluate',\n",
              " 'evaluate_generator',\n",
              " 'fit',\n",
              " 'fit_generator',\n",
              " 'from_config',\n",
              " 'get_config',\n",
              " 'get_input_at',\n",
              " 'get_input_mask_at',\n",
              " 'get_input_shape_at',\n",
              " 'get_layer',\n",
              " 'get_losses_for',\n",
              " 'get_output_at',\n",
              " 'get_output_mask_at',\n",
              " 'get_output_shape_at',\n",
              " 'get_updates_for',\n",
              " 'get_weights',\n",
              " 'history',\n",
              " 'input',\n",
              " 'input_mask',\n",
              " 'input_names',\n",
              " 'input_shape',\n",
              " 'input_spec',\n",
              " 'inputs',\n",
              " 'layers',\n",
              " 'load_weights',\n",
              " 'loss',\n",
              " 'loss_functions',\n",
              " 'loss_weights',\n",
              " 'losses',\n",
              " 'metrics',\n",
              " 'metrics_names',\n",
              " 'metrics_tensors',\n",
              " 'metrics_updates',\n",
              " 'model',\n",
              " 'name',\n",
              " 'non_trainable_weights',\n",
              " 'optimizer',\n",
              " 'output',\n",
              " 'output_mask',\n",
              " 'output_names',\n",
              " 'output_shape',\n",
              " 'outputs',\n",
              " 'pop',\n",
              " 'predict',\n",
              " 'predict_classes',\n",
              " 'predict_function',\n",
              " 'predict_generator',\n",
              " 'predict_on_batch',\n",
              " 'predict_proba',\n",
              " 'reset_states',\n",
              " 'run_internal_graph',\n",
              " 'sample_weight_mode',\n",
              " 'sample_weight_modes',\n",
              " 'sample_weights',\n",
              " 'save',\n",
              " 'save_weights',\n",
              " 'set_weights',\n",
              " 'state_updates',\n",
              " 'stateful',\n",
              " 'stateful_metric_functions',\n",
              " 'stateful_metric_names',\n",
              " 'stop_training',\n",
              " 'summary',\n",
              " 'supports_masking',\n",
              " 'targets',\n",
              " 'test_function',\n",
              " 'test_on_batch',\n",
              " 'to_json',\n",
              " 'to_yaml',\n",
              " 'total_loss',\n",
              " 'train_function',\n",
              " 'train_on_batch',\n",
              " 'trainable',\n",
              " 'trainable_weights',\n",
              " 'updates',\n",
              " 'uses_learning_phase',\n",
              " 'weighted_metrics',\n",
              " 'weights']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 75
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kwLyS1537349",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}