Memory_Networks.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Memory Networks.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/pranaya-mathur/Deep-Learning/blob/master/Memory_Networks.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PHOMor2DXcoM",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "2cab6ab2-a3d8-4170-d56d-925624f14d7a"
      },
      "source": [
        "import numpy as np\n",
        "import keras.backend as K\n",
        "import matplotlib.pyplot as plt\n",
        "import re\n",
        "import tarfile\n",
        "\n",
        "from keras.models import Model\n",
        "from keras.layers import Dense, Embedding, Input, Lambda, Reshape, add, dot, Activation\n",
        "from keras.preprocessing.sequence import pad_sequences\n",
        "from keras.optimizers import Adam, RMSprop\n",
        "from keras.utils.data_utils import get_file"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Lgv_qY5QX2lk",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "5a354dba-9289-46dc-b73d-58533cbc0fa6"
      },
      "source": [
        "path = get_file(\n",
        "  'babi-tasks-v1-2.tar.gz', \n",
        "  origin='https://s3.amazonaws.com/text-datasets/babi_tasks_1-20_v1-2.tar.gz')\n",
        "tar = tarfile.open(path)"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://s3.amazonaws.com/text-datasets/babi_tasks_1-20_v1-2.tar.gz\n",
            "11747328/11745123 [==============================] - 0s 0us/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "efi4veU9X7Ns",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "challenges = {\n",
        "  # QA1 with 10,000 samples\n",
        "  'single_supporting_fact_10k': 'tasks_1-20_v1-2/en-10k/qa1_single-supporting-fact_{}.txt',\n",
        "  # QA2 with 10,000 samples\n",
        "  'two_supporting_facts_10k': 'tasks_1-20_v1-2/en-10k/qa2_two-supporting-facts_{}.txt',\n",
        "}"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JxhgSAxZazBX",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 2531
        },
        "outputId": "df6822fc-ddaf-4ac5-8bd1-bc185a5c59b0"
      },
      "source": [
        "def tokenize(sent):\n",
        "  '''Return the tokens of a sentence including punctuation.\n",
        "\n",
        "  >>> tokenize('Bob dropped the apple. Where is the apple?')\n",
        "  ['Bob', 'dropped', 'the', 'apple', '.', 'Where', 'is', 'the', 'apple', '?']\n",
        "  '''\n",
        "  return [x.strip() for x in re.split('(\\W+)?', sent) if x.strip()]\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "def get_stories(f):\n",
        "  # data will return a list of triples\n",
        "  # each triple contains:\n",
        "  #   1. a story\n",
        "  #   2. a question about the story\n",
        "  #   3. the answer to the question\n",
        "  data = []\n",
        "\n",
        "  # use this list to keep track of the story so far\n",
        "  story = []\n",
        "\n",
        "  # print a random story, helpful to see the data\n",
        "  printed = False\n",
        "  for line in f:\n",
        "    line = line.decode('utf-8').strip()\n",
        "\n",
        "    # split the line number from the rest of the line\n",
        "    nid, line = line.split(' ', 1)\n",
        "\n",
        "    # see if we should begin a new story\n",
        "    if int(nid) == 1:\n",
        "      story = []\n",
        "\n",
        "    # this line contains a question and answer if it has a tab\n",
        "    #       question<TAB>answer\n",
        "    # it also tells us which line in the story is relevant to the answer\n",
        "    # Note: we actually ignore this fact, since the model will learn\n",
        "    #       which lines are important\n",
        "    # Note: the max line number is not the number of lines of the story\n",
        "    #       since lines with questions do not contain any story\n",
        "    # one story may contain MULTIPLE questions\n",
        "    if '\\t' in line:\n",
        "      q, a, supporting = line.split('\\t')\n",
        "      q = tokenize(q)\n",
        "\n",
        "      # numbering each line is very useful\n",
        "      # it's the equivalent of adding a unique token to the front\n",
        "      # of each sentence\n",
        "      story_so_far = [[str(i)] + s for i, s in enumerate(story) if s]\n",
        "\n",
        "      # uncomment if you want to see what a story looks like\n",
        "      # if not printed and np.random.rand() < 0.5:\n",
        "      #     print(\"story_so_far:\", story_so_far)\n",
        "      #     printed = True\n",
        "      data.append((story_so_far, q, a))\n",
        "      story.append('')\n",
        "    else:\n",
        "      # just add the line to the current story\n",
        "      story.append(tokenize(line))\n",
        "  return data\n",
        "\n",
        "\n",
        "# recursively flatten a list\n",
        "def should_flatten(el): \n",
        "  return not isinstance(el, (str, bytes))\n",
        "\n",
        "def flatten(l):\n",
        "  for el in l:\n",
        "    if should_flatten(el):\n",
        "      yield from flatten(el)\n",
        "    else:\n",
        "      yield el\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "# convert stories from words into lists of word indexes (integers)\n",
        "# pad each sequence so that they are the same length\n",
        "# we will need to re-pad the stories later so that each story\n",
        "# is the same length\n",
        "def vectorize_stories(data, word2idx, story_maxlen, query_maxlen):\n",
        "  inputs, queries, answers = [], [], []\n",
        "  for story, query, answer in data:\n",
        "    inputs.append([[word2idx[w] for w in s] for s in story])\n",
        "    queries.append([word2idx[w] for w in query])\n",
        "    answers.append([word2idx[answer]])\n",
        "  return (\n",
        "    [pad_sequences(x, maxlen=story_maxlen) for x in inputs],\n",
        "    pad_sequences(queries, maxlen=query_maxlen),\n",
        "    np.array(answers)\n",
        "  )\n",
        "\n",
        "\n",
        "\n",
        "# this is like 'pad_sequences' but for entire stories\n",
        "# we are padding each story with zeros so every story\n",
        "# has the same number of sentences\n",
        "# append an array of zeros of size:\n",
        "# (max_sentences - num sentences in story, max words in sentence)\n",
        "def stack_inputs(inputs, story_maxsents, story_maxlen):\n",
        "  for i, story in enumerate(inputs):\n",
        "    inputs[i] = np.concatenate(\n",
        "      [\n",
        "        story, \n",
        "        np.zeros((story_maxsents - story.shape[0], story_maxlen), 'int')\n",
        "      ]\n",
        "    )\n",
        "  return np.stack(inputs)\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "# make a function to get the data since\n",
        "# we want to load both the single supporting fact data\n",
        "# and the two supporting fact data later\n",
        "def get_data(challenge_type):\n",
        "  # input should either be 'single_supporting_fact_10k' or 'two_supporting_facts_10k'\n",
        "  challenge = challenges[challenge_type]\n",
        "\n",
        "\n",
        "  # returns a list of triples of:\n",
        "  # (story, question, answer)\n",
        "  # story is a list of sentences\n",
        "  # question is a sentence\n",
        "  # answer is a word\n",
        "  train_stories = get_stories(tar.extractfile(challenge.format('train')))\n",
        "  test_stories = get_stories(tar.extractfile(challenge.format('test')))\n",
        "\n",
        "\n",
        "  # group all the stories together\n",
        "  stories = train_stories + test_stories\n",
        "\n",
        "  # so we can get the max length of each story, of each sentence, and of each question\n",
        "  story_maxlen = max((len(s) for x, _, _ in stories for s in x))\n",
        "  story_maxsents = max((len(x) for x, _, _ in stories))\n",
        "  query_maxlen = max(len(x) for _, x, _ in stories)\n",
        "\n",
        "  # Create vocabulary of corpus and find size, including a padding element.\n",
        "  vocab = sorted(set(flatten(stories)))\n",
        "  vocab.insert(0, '<PAD>')\n",
        "  vocab_size = len(vocab)\n",
        "\n",
        "  # Create an index mapping for the vocabulary.\n",
        "  word2idx = {c:i for i, c in enumerate(vocab)}\n",
        "\n",
        "  # convert stories from strings to lists of integers\n",
        "  inputs_train, queries_train, answers_train = vectorize_stories(\n",
        "    train_stories, \n",
        "    word2idx,\n",
        "    story_maxlen,\n",
        "    query_maxlen\n",
        "  )\n",
        "  inputs_test, queries_test, answers_test = vectorize_stories(\n",
        "    test_stories, \n",
        "    word2idx,\n",
        "    story_maxlen,\n",
        "    query_maxlen\n",
        "  )\n",
        "\n",
        "  # convert inputs into 3-D numpy arrays\n",
        "  inputs_train = stack_inputs(inputs_train, story_maxsents, story_maxlen)\n",
        "  inputs_test = stack_inputs(inputs_test, story_maxsents, story_maxlen)\n",
        "  print(\"inputs_train.shape, inputs_test.shape\", inputs_train.shape, inputs_test.shape)\n",
        "\n",
        "\n",
        "  # return model inputs for keras\n",
        "  return train_stories, test_stories, \\\n",
        "    inputs_train, queries_train, answers_train, \\\n",
        "    inputs_test, queries_test, answers_test, \\\n",
        "    story_maxsents, story_maxlen, query_maxlen, \\\n",
        "    vocab, vocab_size\n",
        "\n",
        "\n",
        "# get the single supporting fact data\n",
        "train_stories, test_stories, \\\n",
        "  inputs_train, queries_train, answers_train, \\\n",
        "  inputs_test, queries_test, answers_test, \\\n",
        "  story_maxsents, story_maxlen, query_maxlen, \\\n",
        "  vocab, vocab_size = get_data('single_supporting_fact_10k')\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "##### create the model #####\n",
        "embedding_dim = 15\n",
        "\n",
        "\n",
        "# turn the story into a sequence of embedding vectors\n",
        "# one for each story line\n",
        "# treating each story line like a \"bag of words\"\n",
        "input_story_ = Input((story_maxsents, story_maxlen))\n",
        "embedded_story = Embedding(vocab_size, embedding_dim)(input_story_)\n",
        "embedded_story = Lambda(lambda x: K.sum(x, axis=2))(embedded_story)\n",
        "print(\"input_story_.shape, embedded_story.shape:\", input_story_.shape, embedded_story.shape)\n",
        "\n",
        "\n",
        "# turn the question into an embedding\n",
        "# also a bag of words\n",
        "input_question_ = Input((query_maxlen,))\n",
        "embedded_question = Embedding(vocab_size, embedding_dim)(input_question_)\n",
        "embedded_question = Lambda(lambda x: K.sum(x, axis=1))(embedded_question)\n",
        "\n",
        "# add a \"sequence length\" of 1 so that it can\n",
        "# be dotted with the story later\n",
        "embedded_question = Reshape((1, embedding_dim))(embedded_question)\n",
        "print(\"inp_q.shape, emb_q.shape:\", input_question_.shape, embedded_question.shape)\n",
        "\n",
        "\n",
        "# calculate the weights for each story line\n",
        "# embedded_story.shape        = (N, num sentences, embedding_dim)\n",
        "# embedded_question.shape     = (N, 1, embedding_dim)\n",
        "x = dot([embedded_story, embedded_question], 2)\n",
        "x = Reshape((story_maxsents,))(x) # flatten the vector\n",
        "x = Activation('softmax')(x)\n",
        "story_weights = Reshape((story_maxsents, 1))(x) # unflatten it again to be dotted later\n",
        "print(\"story_weights.shape:\", story_weights.shape)\n",
        "\n",
        "\n",
        "\n",
        "x = dot([story_weights, embedded_story], 1)\n",
        "x = Reshape((embedding_dim,))(x) # flatten it again\n",
        "ans = Dense(vocab_size, activation='softmax')(x)\n",
        "\n",
        "# make the model\n",
        "model = Model([input_story_, input_question_], ans)\n",
        "\n",
        "# compile the model\n",
        "model.compile(\n",
        "  optimizer=RMSprop(lr=1e-2),\n",
        "  loss='sparse_categorical_crossentropy',\n",
        "  metrics=['accuracy']\n",
        ")\n",
        "\n",
        "# train the model\n",
        "r = model.fit(\n",
        "  [inputs_train, queries_train],\n",
        "  answers_train,\n",
        "  epochs=4,\n",
        "  batch_size=32,\n",
        "  validation_data=([inputs_test, queries_test], answers_test)\n",
        ")\n",
        "\n",
        "\n",
        "# Check how we weight each input sentence given a story and question\n",
        "debug_model = Model([input_story_, input_question_], story_weights)\n",
        "\n",
        "# choose a random story\n",
        "story_idx = np.random.choice(len(train_stories))\n",
        "\n",
        "# get weights from debug model\n",
        "i = inputs_train[story_idx:story_idx+1]\n",
        "q = queries_train[story_idx:story_idx+1]\n",
        "w = debug_model.predict([i, q]).flatten()\n",
        "\n",
        "story, question, ans = train_stories[story_idx]\n",
        "print(\"story:\\n\")\n",
        "for i, line in enumerate(story):\n",
        "  print(\"{:1.5f}\".format(w[i]), \"\\t\", \" \".join(line))\n",
        "\n",
        "print(\"question:\", \" \".join(question))\n",
        "print(\"answer:\", ans)\n",
        "\n",
        "\n",
        "# pause so we can see the output\n",
        "input(\"Hit enter to continue\\n\\n\")\n",
        "\n",
        "\n",
        "\n",
        "##### two supporting facts #####\n",
        "\n",
        "\n",
        "# get the two supporting fact data\n",
        "train_stories, test_stories, \\\n",
        "  inputs_train, queries_train, answers_train, \\\n",
        "  inputs_test, queries_test, answers_test, \\\n",
        "  story_maxsents, story_maxlen, query_maxlen, \\\n",
        "  vocab, vocab_size = get_data('two_supporting_facts_10k')\n",
        "\n",
        "\n",
        "\n",
        "##### create the model #####\n",
        "embedding_dim = 30\n",
        "\n",
        "\n",
        "# make a function for this so we can use it again\n",
        "def embed_and_sum(x, axis=2):\n",
        "  x = Embedding(vocab_size, embedding_dim)(x)\n",
        "  x = Lambda(lambda x: K.sum(x, axis))(x)\n",
        "  return x\n",
        "\n",
        "# define the inputs\n",
        "input_story_ = Input((story_maxsents, story_maxlen))\n",
        "input_question_ = Input((query_maxlen,))\n",
        "\n",
        "\n",
        "# embed the inputs\n",
        "embedded_story = embed_and_sum(input_story_)\n",
        "embedded_question = embed_and_sum(input_question_, 1)\n",
        "\n",
        "\n",
        "# final dense will be used in each hop\n",
        "dense_layer = Dense(embedding_dim, activation='elu')\n",
        "\n",
        "\n",
        "# define one hop\n",
        "# the \"query\" can be the question, or the answer from the previous hop\n",
        "def hop(query, story):\n",
        "  # query.shape = (embedding_dim,)\n",
        "  # story.shape = (num sentences, embedding_dim)\n",
        "  x = Reshape((1, embedding_dim))(query) # make it (1, embedding_dim)\n",
        "  x = dot([story, x], 2)\n",
        "  x = Reshape((story_maxsents,))(x) # flatten it for softmax\n",
        "  x = Activation('softmax')(x)\n",
        "  story_weights = Reshape((story_maxsents, 1))(x) # unflatten for dotting\n",
        "\n",
        "  # makes a new embedding\n",
        "  story_embedding2 = embed_and_sum(input_story_)\n",
        "  x = dot([story_weights, story_embedding2], 1)\n",
        "  x = Reshape((embedding_dim,))(x)\n",
        "  x = dense_layer(x)\n",
        "  return x, story_embedding2, story_weights\n",
        "\n",
        "\n",
        "# do the hops\n",
        "ans1, embedded_story, story_weights1 = hop(embedded_question, embedded_story)\n",
        "ans2, _,              story_weights2 = hop(ans1,              embedded_story)\n",
        "\n",
        "# get the final answer\n",
        "ans = Dense(vocab_size, activation='softmax')(ans2)\n",
        "\n",
        "\n",
        "# build the model\n",
        "model2 = Model([input_story_, input_question_], ans)\n",
        "\n",
        "# compile the model\n",
        "model2.compile(\n",
        "  optimizer=RMSprop(lr=5e-3),\n",
        "  loss='sparse_categorical_crossentropy',\n",
        "  metrics=['accuracy']\n",
        ")\n",
        "\n",
        "# fit the model\n",
        "r = model2.fit(\n",
        "  [inputs_train, queries_train],\n",
        "  answers_train,\n",
        "  epochs=30,\n",
        "  batch_size=32,\n",
        "  validation_data=([inputs_test, queries_test], answers_test)\n",
        ")\n",
        "\n",
        "\n",
        "### print story line weights again ###\n",
        "debug_model2 = Model(\n",
        "  [input_story_, input_question_],\n",
        "  [story_weights1, story_weights2]\n",
        ")\n",
        "\n",
        "# choose a random story\n",
        "story_idx = np.random.choice(len(train_stories))\n",
        "\n",
        "# get weights from debug model\n",
        "i = inputs_train[story_idx:story_idx+1]\n",
        "q = queries_train[story_idx:story_idx+1]\n",
        "w1, w2 = debug_model2.predict([i, q])\n",
        "w1 = w1.flatten()\n",
        "w2 = w2.flatten()\n",
        "\n",
        "story, question, ans = train_stories[story_idx]\n",
        "print(\"story:\\n\")\n",
        "for j, line in enumerate(story):\n",
        "  print(\"{:1.5f}\".format(w1[j]), \"\\t\", \"{:1.5f}\".format(w2[j]), \"\\t\", \" \".join(line))\n",
        "\n",
        "print(\"question:\", \" \".join(question))\n",
        "print(\"answer:\", ans)\n",
        "print(\"prediction:\", vocab[ np.argmax(model2.predict([i, q])[0]) ])\n",
        "\n",
        "\n",
        "# plot some data\n",
        "plt.plot(r.history['loss'], label='loss')\n",
        "plt.plot(r.history['val_loss'], label='val_loss')\n",
        "plt.legend()\n",
        "plt.show()\n",
        "\n",
        "# accuracies\n",
        "plt.plot(r.history['acc'], label='acc')\n",
        "plt.plot(r.history['val_acc'], label='val_acc')\n",
        "plt.legend()\n",
        "plt.show()\n"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/lib/python3.6/re.py:212: FutureWarning: split() requires a non-empty pattern match.\n",
            "  return _compile(pattern, flags).split(string, maxsplit)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "inputs_train.shape, inputs_test.shape (10000, 10, 8) (1000, 10, 8)\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/op_def_library.py:263: 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",
            "input_story_.shape, embedded_story.shape: (?, 10, 8) (?, 10, 15)\n",
            "inp_q.shape, emb_q.shape: (?, 4) (?, 1, 15)\n",
            "story_weights.shape: (?, 10, 1)\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",
            "Train on 10000 samples, validate on 1000 samples\n",
            "Epoch 1/4\n",
            "10000/10000 [==============================] - 3s 348us/step - loss: 0.7711 - acc: 0.7393 - val_loss: 0.0277 - val_acc: 0.9940\n",
            "Epoch 2/4\n",
            "10000/10000 [==============================] - 1s 90us/step - loss: 0.0084 - acc: 0.9971 - val_loss: 0.0021 - val_acc: 1.0000\n",
            "Epoch 3/4\n",
            "10000/10000 [==============================] - 1s 88us/step - loss: 0.0036 - acc: 0.9991 - val_loss: 0.0285 - val_acc: 0.9900\n",
            "Epoch 4/4\n",
            "10000/10000 [==============================] - 1s 88us/step - loss: 0.0019 - acc: 0.9994 - val_loss: 3.7163e-06 - val_acc: 1.0000\n",
            "story:\n",
            "\n",
            "0.00000 \t 0 Mary went to the hallway .\n",
            "0.00000 \t 1 John went back to the bathroom .\n",
            "0.00000 \t 3 Daniel journeyed to the kitchen .\n",
            "0.00000 \t 4 Sandra travelled to the bedroom .\n",
            "0.00000 \t 6 Daniel moved to the bathroom .\n",
            "0.00000 \t 7 Sandra travelled to the hallway .\n",
            "0.00000 \t 9 Daniel journeyed to the garden .\n",
            "0.00000 \t 10 Daniel travelled to the bedroom .\n",
            "0.07755 \t 12 Mary went back to the bedroom .\n",
            "0.92245 \t 13 Mary went back to the kitchen .\n",
            "question: Where is Mary ?\n",
            "answer: kitchen\n",
            "Hit enter to continue\n",
            "\n",
            "\n",
            "inputs_train.shape, inputs_test.shape (10000, 88, 8) (1000, 88, 8)\n",
            "Train on 10000 samples, validate on 1000 samples\n",
            "Epoch 1/30\n",
            "10000/10000 [==============================] - 2s 162us/step - loss: 1.7467 - acc: 0.2595 - val_loss: 1.4791 - val_acc: 0.3960\n",
            "Epoch 2/30\n",
            "10000/10000 [==============================] - 1s 131us/step - loss: 1.0403 - acc: 0.5989 - val_loss: 0.7758 - val_acc: 0.7040\n",
            "Epoch 3/30\n",
            "10000/10000 [==============================] - 1s 129us/step - loss: 0.6750 - acc: 0.7486 - val_loss: 0.6231 - val_acc: 0.7570\n",
            "Epoch 4/30\n",
            "10000/10000 [==============================] - 1s 129us/step - loss: 0.5581 - acc: 0.7926 - val_loss: 0.6667 - val_acc: 0.7410\n",
            "Epoch 5/30\n",
            "10000/10000 [==============================] - 1s 129us/step - loss: 0.4767 - acc: 0.8288 - val_loss: 0.5181 - val_acc: 0.8090\n",
            "Epoch 6/30\n",
            "10000/10000 [==============================] - 1s 130us/step - loss: 0.4083 - acc: 0.8538 - val_loss: 0.4828 - val_acc: 0.8310\n",
            "Epoch 7/30\n",
            "10000/10000 [==============================] - 1s 130us/step - loss: 0.3558 - acc: 0.8785 - val_loss: 0.3736 - val_acc: 0.8620\n",
            "Epoch 8/30\n",
            "10000/10000 [==============================] - 1s 128us/step - loss: 0.3175 - acc: 0.8924 - val_loss: 0.3516 - val_acc: 0.8790\n",
            "Epoch 9/30\n",
            "10000/10000 [==============================] - 1s 129us/step - loss: 0.2859 - acc: 0.9073 - val_loss: 0.3680 - val_acc: 0.8770\n",
            "Epoch 10/30\n",
            "10000/10000 [==============================] - 1s 132us/step - loss: 0.2665 - acc: 0.9189 - val_loss: 0.4269 - val_acc: 0.8630\n",
            "Epoch 11/30\n",
            "10000/10000 [==============================] - 1s 129us/step - loss: 0.2476 - acc: 0.9243 - val_loss: 0.2957 - val_acc: 0.9050\n",
            "Epoch 12/30\n",
            "10000/10000 [==============================] - 1s 129us/step - loss: 0.2273 - acc: 0.9291 - val_loss: 0.3986 - val_acc: 0.8700\n",
            "Epoch 13/30\n",
            "10000/10000 [==============================] - 1s 130us/step - loss: 0.2236 - acc: 0.9325 - val_loss: 0.4251 - val_acc: 0.8710\n",
            "Epoch 14/30\n",
            "10000/10000 [==============================] - 1s 131us/step - loss: 0.2061 - acc: 0.9388 - val_loss: 0.3526 - val_acc: 0.8910\n",
            "Epoch 15/30\n",
            "10000/10000 [==============================] - 1s 129us/step - loss: 0.2042 - acc: 0.9404 - val_loss: 0.2917 - val_acc: 0.9080\n",
            "Epoch 16/30\n",
            "10000/10000 [==============================] - 1s 130us/step - loss: 0.2047 - acc: 0.9433 - val_loss: 0.3104 - val_acc: 0.9020\n",
            "Epoch 17/30\n",
            "10000/10000 [==============================] - 1s 142us/step - loss: 0.1968 - acc: 0.9442 - val_loss: 0.3222 - val_acc: 0.9050\n",
            "Epoch 18/30\n",
            "10000/10000 [==============================] - 1s 149us/step - loss: 0.1798 - acc: 0.9492 - val_loss: 0.3173 - val_acc: 0.9070\n",
            "Epoch 19/30\n",
            "10000/10000 [==============================] - 2s 151us/step - loss: 0.1830 - acc: 0.9527 - val_loss: 0.3341 - val_acc: 0.9040\n",
            "Epoch 20/30\n",
            "10000/10000 [==============================] - 1s 149us/step - loss: 0.1849 - acc: 0.9508 - val_loss: 0.2349 - val_acc: 0.9210\n",
            "Epoch 21/30\n",
            "10000/10000 [==============================] - 1s 150us/step - loss: 0.1770 - acc: 0.9535 - val_loss: 0.3019 - val_acc: 0.9200\n",
            "Epoch 22/30\n",
            "10000/10000 [==============================] - 1s 150us/step - loss: 0.1636 - acc: 0.9566 - val_loss: 0.3096 - val_acc: 0.9120\n",
            "Epoch 23/30\n",
            "10000/10000 [==============================] - 1s 148us/step - loss: 0.1700 - acc: 0.9551 - val_loss: 0.3264 - val_acc: 0.9030\n",
            "Epoch 24/30\n",
            "10000/10000 [==============================] - 1s 132us/step - loss: 0.1510 - acc: 0.9588 - val_loss: 0.2691 - val_acc: 0.9260\n",
            "Epoch 25/30\n",
            "10000/10000 [==============================] - 1s 129us/step - loss: 0.1553 - acc: 0.9568 - val_loss: 0.2966 - val_acc: 0.9160\n",
            "Epoch 26/30\n",
            "10000/10000 [==============================] - 1s 130us/step - loss: 0.1511 - acc: 0.9593 - val_loss: 0.3529 - val_acc: 0.9020\n",
            "Epoch 27/30\n",
            "10000/10000 [==============================] - 1s 130us/step - loss: 0.1580 - acc: 0.9592 - val_loss: 0.3585 - val_acc: 0.9060\n",
            "Epoch 28/30\n",
            "10000/10000 [==============================] - 1s 130us/step - loss: 0.1514 - acc: 0.9619 - val_loss: 0.3807 - val_acc: 0.9110\n",
            "Epoch 29/30\n",
            "10000/10000 [==============================] - 1s 129us/step - loss: 0.1548 - acc: 0.9616 - val_loss: 0.3590 - val_acc: 0.9020\n",
            "Epoch 30/30\n",
            "10000/10000 [==============================] - 1s 129us/step - loss: 0.1416 - acc: 0.9640 - val_loss: 0.3006 - val_acc: 0.9240\n",
            "story:\n",
            "\n",
            "0.00000 \t 0.00000 \t 0 Mary went back to the hallway .\n",
            "0.00000 \t 0.00000 \t 1 John took the milk there .\n",
            "0.00000 \t 0.00000 \t 2 John moved to the kitchen .\n",
            "0.00000 \t 0.00000 \t 3 Sandra took the apple there .\n",
            "0.00000 \t 0.00005 \t 5 Daniel went back to the hallway .\n",
            "0.00000 \t 0.00000 \t 6 John dropped the milk .\n",
            "0.00000 \t 0.00137 \t 8 Mary went to the kitchen .\n",
            "0.00000 \t 0.00315 \t 9 Mary moved to the garden .\n",
            "0.00000 \t 0.99543 \t 11 Sandra journeyed to the bedroom .\n",
            "1.00000 \t 0.00000 \t 12 Sandra left the apple .\n",
            "0.00000 \t 0.00000 \t 14 Daniel journeyed to the bedroom .\n",
            "0.00000 \t 0.00000 \t 15 John got the milk there .\n",
            "question: Where is the apple ?\n",
            "answer: bedroom\n",
            "prediction: bedroom\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD8CAYAAACMwORRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nx5d0CA8AAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD9CAYAAABHnDf0AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "v4N19hMPbHn7",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}