DMNPlus.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Pytorch implementation of Dynamic Memory Network Plus "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Introduction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This tutorial shows how to implement the paper [\"Dynamic Memory Network for Textual and Visual Question Answering\"](https://arxiv.org/abs/1603.01417) in Pytorch and the explnation of different modules used in the same. The code has been heavily used from the repository https://github.com/dandelin/Dynamic-memory-networks-plus-Pytorch."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from IPython.display import Image, display\n",
    "Image('Network.png')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's now implement the loader function to load our dataset and all the libraries required for implementation.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "''' This is dataset loader file to load the bAbI dataset. '''\n",
    "import re\n",
    "import numpy as np\n",
    "import os\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch. nn.init as init\n",
    "import torch.nn.functional as F\n",
    "from glob import glob\n",
    "from torch.utils.data import Dataloader\n",
    "from torch.utils.data.dataset import Dataset\n",
    "from torch.utils.data.dataloader import default_collate\n",
    "from torch.autograd import Variable\n",
    "\n",
    "class adict(dict):\n",
    "    def __init__(self, *args, **kargs):\n",
    "        dict.__init__(self, *args, **kargs)\n",
    "        self.__dict__ = self\n",
    "\n",
    "def pad_collate(batch):\n",
    "    max_len_ques = float('-inf')\n",
    "    max_sen_len_context = float('-inf')\n",
    "    max_len_context = float('-inf')\n",
    "\n",
    "    for item in batch:\n",
    "        contexts, ques, _ = item\n",
    "        if len(contexts) > max_len_context:\n",
    "            max_len_context = len(contexts)\n",
    "        if len(ques) > max_len_ques:\n",
    "            max_len_ques = len(ques)\n",
    "        for sen in contexts:\n",
    "            if(len(sen) > max_sen_len_context):\n",
    "                max_sen_len_context = len(sen)\n",
    "    max_len_context = min(max_len_context, 70)\n",
    "    for idx, item in enumerate(batch): # Going through each example in the batch which contains their ow context, question and answer.\n",
    "        context_i, question, answer = item\n",
    "        context_i = context[-max_len_context:] \n",
    "        context = np.zeros((max_len_context, max_sen_len_context))\n",
    "        for i, sen in enumerate(context_i): # going through ith context containing max_len_context sentences and a question\n",
    "            context[i] = np.pad(sen, (0, max_sen_len_context-len(sen)), 'constant', constant_values=0)\n",
    "        question = np.pad(question, (0, max_len_ques-len(question)), 'constant', constant_values=0)\n",
    "        batch[idx] = (context, question, answer)\n",
    "\n",
    "    return default_collate(batch)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "class BabiDataSet(Dataset):\n",
    "    def __init__(self, task_id, mode='train'):\n",
    "        self.mode = mode\n",
    "        self.vocab_path = 'dataset/babi{}_vocab.pkl'.format(task_id)\n",
    "        train_data, test_data = get_train_test(task_id) # Get raw train_data and test_data from babi dataset\n",
    "        self.QA = adict()\n",
    "        self.QA.VOCAB = {'<PAD>': 0, '<EOS>':1}\n",
    "        self.QA.INV_VOCAB = {0:'<PAD>', 1:'<EOS>'}\n",
    "        self.train = self.get_processed_data(train_data)\n",
    "        self.val = [self.train[i][int(9*len(self.train[i])/10):] for i in range(3)] # splitting into 90/10 train/val dataset\n",
    "        self.train = [self.train[i][:int(9*len(self.train[i])/10)] for i in range(3)] # splitting into 90/10 train/val dataset\n",
    "        self.test = self.get_processed_data(test_data)\n",
    "\n",
    "    def set_mode(self, mode):\n",
    "        self.mode = mode \n",
    "\n",
    "    def __len__(self):\n",
    "        if self.mode == 'train':\n",
    "            return len(self.train[0])\n",
    "        elif self.mode == 'val':\n",
    "            return len(self.val[0])\n",
    "        elif self.mode == 'test':\n",
    "            return len(self.test[0])\n",
    "        else:\n",
    "            print (\"Invalid Mode!\")\n",
    "            return\n",
    "\n",
    "    def __getdata__(self, index):\n",
    "        if self.mode == 'train':\n",
    "            contexts, questions, answers = self.train\n",
    "        elif self.mode == 'val':\n",
    "            contexts, questions, answers = self.val\n",
    "        elif self.mode == 'test':\n",
    "            contexts, questions, answers = self.test\n",
    "\n",
    "        return contexts[index], questions[index], answers[index]\n",
    "\n",
    "    def get_processed_data(self, raw_data):\n",
    "        unindexed= get_processed_data(raw_data)\n",
    "        questions=[]\n",
    "        contexts= []\n",
    "        answers= []\n",
    "        for qa in unindexed:\n",
    "            context= [c.lower().split()+ ['<EOS>'] for c in qa['C']]\n",
    "\n",
    "            for con in context:\n",
    "                for token in con:\n",
    "                    self.build_vocab(token)\n",
    "            context= [[self.QA.VOCAB[token] for token in sentence] for sentence in context]\n",
    "            question= qa['Q'].lower().split()+ ['<EOS>']\n",
    "\n",
    "            for token in question:\n",
    "                self.build_vocab(token)\n",
    "            question= [self.QA.VOCAB[token] for token in question]\n",
    "\n",
    "            self.build_vocab(qa['A'].lower())\n",
    "            answer= self.QA.VOCAB[qa['A'].lower()]\n",
    "\n",
    "            contexts.append(context)\n",
    "            questions.append(question)\n",
    "            answers.append(answer)\n",
    "            return (contexts, questions, answers)\n",
    "\n",
    "   \n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_vocab(self, token):\n",
    "    if not token in self.QA.VOCAB:\n",
    "        next_index= len(self.QA.VOCAB)\n",
    "        self.QA.VOCAB[token]= next_index\n",
    "        self.QA.IVOCAB[next_index]= token\n",
    "\n",
    "def get_train_test(task_id):\n",
    "    paths = glob('data/en-10k/qa{}_*'.format(task_id))\n",
    "    for path in paths:\n",
    "        if 'train' in path;\n",
    "            with open(path, 'r') as f:\n",
    "                train = f.read()\n",
    "        elif 'test' in path:\n",
    "            with open(path, 'r') as f:\n",
    "                test = f.read()\n",
    "\n",
    "    return train, test\n",
    "\n",
    "def build_vocab(raw_data):\n",
    "    lowered= raw_data.lower()\n",
    "    tokens= re.findall('[a-zA-Z]+',lowered)\n",
    "    types= set(tokens)\n",
    "    return types"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_unprocessed_data(raw_data):\n",
    "    tasks = []\n",
    "    task = None\n",
    "    data = raw_data.strip().split('\\n')\n",
    "    for i, line in enumerate(data):\n",
    "        id = int(line[0:line.find(' ')])\n",
    "        if id == 1:\n",
    "            task = {\"C\": \"\", \"Q\": \"\", \"A\": \"\", \"S\": \"\"}\n",
    "            counter = 0\n",
    "            id_map = {}\n",
    "\n",
    "        line = line.strip()\n",
    "        line = line.replace('.', ' . ')\n",
    "        line = line[line.find(' ')+1:]\n",
    "        # if not a question\n",
    "        if line.find('?') == -1:\n",
    "            task[\"C\"] += line + '<line>'\n",
    "            id_map[id] = counter\n",
    "            counter += 1\n",
    "        else:\n",
    "            idx = line.find('?')\n",
    "            tmp = line[idx+1:].split('\\t')\n",
    "            task[\"Q\"] = line[:idx]\n",
    "            task[\"A\"] = tmp[1].strip()\n",
    "            task[\"S\"] = [] # Supporting facts\n",
    "            for num in tmp[2].split():\n",
    "                task[\"S\"].append(id_map[int(num.strip())])\n",
    "            tc = task.copy()\n",
    "            tc['C'] = tc['C'].split('<line>')[:-1]\n",
    "            tasks.append(tc)\n",
    "    return tasks\n",
    "\n",
    "\n",
    "if __name__ == '__main__':\n",
    "    dataset_train= BabiDataset(20, is_train= True)\n",
    "    train_loader= DataLoader(dataset_train,batch_size=2, shuffle=True,collate_fn= pad_collate)\n",
    "    for batch_idx, data in enumerate(train_loader):\n",
    "        contexts, questions, answers= data\n",
    "        break\t\t"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This paper is a much improved version of the previous paper titled [\"Dynamic Memory Networks for Natural Language Processing\"](https://arxiv.org/abs/1506.07285). The main contribution of this paper is the improved InputModule for calculating the facts from input sentences keeping in mind the exchange of information between input sentences using a Bidirectional GRU and a improved version of MemoryModule using Attention based GRU model."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The network has four different types of modules namely QuestionModule, InputModule, MmemoryModule and the AnswerModule. Let's understand each of these modules in detail and their implementation."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### QuestionModule"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The QuestionModule is pretty straightforward. It encodes the question of the task into a distributed vector representation. This representation is fed into the episodic memory module and forms the basis or initial state of the memory state."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "class QuestionModule(nn.Module):\n",
    "    def __init__(self, vocab_size, hidden_size):\n",
    "        super(QuestionModule, self).__init__()\n",
    "        self.vocab_size = vocab_size # Size of the vocabulary used in word embedding\n",
    "        self.hidden_size = hidden_size # Size of the hidden state of GRU\n",
    "        self.gru = nn.GRU(hidden_size, hidden_size, batch_first=True)\n",
    "\n",
    "    def forward(self, questions, word_embedding):\n",
    "        # questions.size() = (batch_size, num_tokens)\n",
    "        # word_embedding -> (batch_size, num_tokens, embedding_length)\n",
    "        # self.gru() -> (1, batch_size, hidden_size)\n",
    "\n",
    "        questions = word_embedding(questions) # Word embedding of the question\n",
    "        output, questions = self.gru(questions)\n",
    "        questions = torch.transpose(questions, 0, 1)\n",
    "\n",
    "        return questions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### InputModule "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "InputModule encodes the raw text inputs into distributed vector representation. We replace the single GRU in DMN (first paper) by two different components. The first component is a sentence reader which encodes the words in a sentence into sentence encoding using a specific encoding scheme called Positional Encoder. THe output of the positional encoder are vectors represented as f1, f2, f3,... and these go as the input to the second component called as Input Fusion Layer. The main function of this layer is to allow the interaction between different input sentences to exchange information not only in the forward direction but also in the backward direction i.e., information from future states flowing backwards using a Bidirectional GRU module. Basically input fusion layer allows for distant supporting sentences to have a more direct interaction."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "class InputModule(nn.Module):\n",
    "    def __init__(self, vocab_size, hidden_size):\n",
    "        super(InputModule, self).__init__()\n",
    "        self.vocab_size = vocab_size\n",
    "        self.hidden_size = hidden_size\n",
    "        self.gru = nn.GRU(hidden_size, hidden_size, bidirectional=True, batch_first=True)\n",
    "        for name, param in self.gru.state_dict().items():\n",
    "            if 'weight' in name:\n",
    "                init.xavier_normal(param)\n",
    "        self.dropout = nn.Dropout(0.1)\n",
    "\n",
    "    ''' We will now define the encoding scheme which is positional encoding in the paper \" Dynamic Memory Network \n",
    "    for Textual and Visual Question Answering '''\n",
    "    def positional_encoder(embedded_sentence):\n",
    "        # embedded_sentence.size() = (batch_size, num_sentences, num_tokens, embedding_length)\n",
    "        # l.size() = (num_tokems, embedding_length)\n",
    "        # output.size() = (num_batch, num_sentences, embedding_length)\n",
    "        # The outputs are basically f1, f2, f3,.... which will go into the input fusion layer in the next step to \n",
    "        # add share information between sentences using a BiDirfectional GRU module.\n",
    "\n",
    "        batch_size, num_sentences, num_tokens, embedding_length = embedded_sentence.size()\n",
    "        l = [] # It will be same for all sentences in all batches as num_tokens and embedding_length is same \n",
    "        # for the entire dataset.\n",
    "        for j in range(num_tokens):\n",
    "            x = []\n",
    "            for d in range(embedding_length):\n",
    "                x.append((1 - (j/(num_tokens-1))) - (d/(embedding_length-1)) * (1 - 2*j/(num_tokens-1)))\n",
    "            l.append(x)\n",
    "\n",
    "        l = torch.FloatTensor(l)\n",
    "        l = l.unsqueeze(0) # adding an extra dimension at first place for batch_size\n",
    "        l = l.unsqueeze(1) # adding an extra dimension at sencond place for num_sentences\n",
    "        l = l.expand_as(embedded_sentence) # so that l.size() = (batch_size, num_sentences, num_tokens, embedding_length)\n",
    "\n",
    "        mat = embedded_sentence*Variable(l.cuda())\n",
    "        f_ids = torch.sum(mat, dim=2).squeeze(2) # sum along token dimension\n",
    "\n",
    "        return f_ids\n",
    "\n",
    "\n",
    "    def forward(self, input, word_embedding):\n",
    "        # input.size() = (batch_size, num_sentences, num_tokens)\n",
    "        # word_embedding -> (batch_size, num_sentences, num_tokens, embedding_length)\n",
    "        # positional_encoder(word_embedding(input)) -> (batch_size, num_sentences, embedding_length)\n",
    "        # Now BidirectionalGRU blocks receive their input, the output of the positional encoder and finally give facts\n",
    "        # facts.size() = (batch_size, num_sentences, embedding_length) embedding_length = hidden_size\n",
    "\n",
    "        input = input.view(input.size()[0], -1)\n",
    "        input = word_embedding(input)\n",
    "        input = input.view(input.size()[0], input.size()[1], input.size()[2], -1)\n",
    "        input = self.positional_encoder(input)\n",
    "        input = self.dropout(input)\n",
    "\n",
    "        h0 = Variable(torch.zeros(2, input.size()[0], self.hidden_size).cuda()) # Initializing the initial hidden state (at t=0 time step)\n",
    "        facts, hdn = self.gru(input, h0)\n",
    "        facts = facts[:, :, :hidden_size] + facts[:, :, hidden_size:]\n",
    "\n",
    "        return facts\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### MemoryModule"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "MemoryModule has two separate components, Attention mechanism which gives a contextual vector c(t) which is basically a summary of relevant input for the t_th pass considering question (q) and the previous memory state m(t-1) and the second one is the memory update mechanism which outputs m(t) using c(t) and m(t-1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### AttentionGRUCell"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Attention based GRU is a modification of the original GRU by embedding information from the attention mechanism. In the GRU module, the update gate decides how much of each dimension of the hidden state to retain and how much to update depending upon the input(xi). Since update gate (ui) is calculated using only input(xi) and previous hidden state(hi_1), it certainly lacks any sort of knowledge from the question or the previous memory state. We can use these two to update the hidden state by replacing ui in GRU equation with the gate value(gi_t)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "class AttnGRUCell(nn.Module):\n",
    "    def __init__(self, input_size, hidden_size):\n",
    "        super(AttnGRUCell, self).__init__()\n",
    "        self.input_size = input_size\n",
    "        self.hidden_size = hidden_size\n",
    "        self.Wr = nn.Linear(input_size, hidden_size)\n",
    "        self.Ur = nn.Linear(hidden_size, hidden_size)\n",
    "        self.W = nn.Linear(input_size, hidden_size)\n",
    "        self.U = nn.Linear(hidden_size, hidden_size)\n",
    "\n",
    "        init.xavier_normal(self.Wr.state_dict()['weight'])\n",
    "        init.xavier_normal(self.Ur.state_dict()['weight'])\n",
    "        init.xavier_normal(self.W.state_dict()['weight'])\n",
    "        init.xavier_normal(self.U.state_dict()['weight'])\n",
    "\n",
    "    def forward(self, fact, hi_1, g):\n",
    "        # fact is the final output of InputModule for each sentence and fact.size() = (batch_size, embedding_length)\n",
    "        # hi_1.size() = (batch_size, embedding_length=hidden_size)\n",
    "        # g.size() = (batch_size, )\n",
    "\n",
    "        r_i = F.sigmoid(self.Wr(fact) + self.Ur(hi_1))\n",
    "        h_tilda = F.tanh(self.W(fact) + r*self.U(hi_1))\n",
    "        hi = g*h_tilda + (1 - g)*hi_1\n",
    "\n",
    "        return hi # Returning the next hidden state considering the first fact and so on."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### AttentionGRU"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This class basically update the hidden state using Attention based GRU defined above by iterating over all the sentences. Final hidden state is called the contextual vector which is used to update next memory state."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "class AttnGRU(nn.Module):\n",
    "    def __init__(self, input_size, hidden_size):\n",
    "        super(AttnGRU, self).__init__()\n",
    "        self.input_size = input_size\n",
    "        self.hidden_size = hidden_size\n",
    "        self.AttnGRUCell = AttnGRUCell(input_size, hidden_size)\n",
    "\n",
    "    def forward(self, facts, G):\n",
    "        # facts.size() = (batch_size, num_sentences, embedding_length)\n",
    "        # fact.size() = (batch_size, embedding_length=hidden_size)\n",
    "        # G.size() = (batch_size, num_sentences)\n",
    "        # g.size() = (batch_size, )\n",
    "\n",
    "        h_0 = Variable(torch.zeros(self.hidden_size)).cuda()\n",
    "\n",
    "        for sen in range(facts.size()[1]):\n",
    "            fact = facts[:, sen, :]\n",
    "            g = G[:, sen]\n",
    "            if sen == 0: # Initialization for first sentence only \n",
    "                hi_1 = h_0.unsqueeze(0).expand_as(fact)\n",
    "            hi_1 = self.AttnGRUCell(fact, hi_1, g)\n",
    "        C = hi_1 # Final hidden vector as the contextual vector used for updating memory\n",
    "\n",
    "        return C"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### EpisodicMemoryModule"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The episodic memory module retrieves information from the final input facts by focusing attention on a subset of these facts using gate values. We implement this attention by associating a single scalar value, the attention gate (gi_t) with each fact during each pass. This is calculated by allowing interaction between the fact and both the question and the previous memory state. Once we have the attention gate, we can use an attention mechanism to extract a contextual vector ct which is used to update the memory state."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "class MemoryModule(nn.Module): # Takes facts, question and prev_mem as its and output next_mem\n",
    "    def __init__(self, hidden_size):\n",
    "        super(MemoryModule, self).__init__()\n",
    "        self.hidden_size = hidden_size\n",
    "        self.AttnGRU = AttnGRU(hidden_size, hidden_size)\n",
    "        self.W1 = nn.Linear(4*hidden_size, hidden_size)\n",
    "        self.W2 = nn.Linear(hidden_size, 1)\n",
    "        self.W_mem = nn.Linear(3*hidden_size, hidden_size)\n",
    "\n",
    "        init.xavier_normal(self.W1.state_dict()['weight'])\n",
    "        init.xavier_normal(self.W2.state_dict()['weight'])\n",
    "        init.xavier_normal(self.W_mem.state_dict()['weight'])\n",
    "\n",
    "    def gateMatrix(self, facts, questions, prev_mem):\n",
    "        # facts.size() = (batch_size, num_sentences, embedding_length=hidden_size)\n",
    "        # questions.size() = (batch_size, 1, embedding_length)\n",
    "        # prev_mem.size() = (batch_size, 1, embedding_length)\n",
    "        # z.size() = (batch_size, num_sentences, 4*embedding_length)\n",
    "        # G.size() = (batch_size, num_sentences)\n",
    "\n",
    "        questions = questions.expand_as(facts)\n",
    "        prev_mem = prev_mem.expand_as(facts)\n",
    "\n",
    "        z = torch.cat([facts*questions, facts*prev_mem, torch.abs(facts - questions), torch.abs(facts - prev_mem)], dim=2)\n",
    "        # z.size() = (batch_size, num_sentences, 4*embedding_length)\n",
    "        z = z.view(-1, 4*embedding_length)\n",
    "        Z = self.W2(F.tanh(self.W1(z)))\n",
    "        Z = Z.view(batch_size, -1)\n",
    "        G = F.softmax(Z)\n",
    "\n",
    "        return G\n",
    "\n",
    "    def forward(self, facts, questions, prev_mem):\n",
    "        # questions = questions.unsqueeze(1)\n",
    "        # prev_mem = prev_mem.unsqueeze(1)\n",
    "        G = self.gateMatrix(facts, questions, prev_mem)\n",
    "        C = self.AttnGRU(facts, G)\n",
    "        # Now considering prev_mem, C and question, we will update the memory state as follows\n",
    "        concat = torch.cat([prev_mem.squeeze(1), C, questions.squeeze(1)], dim=1)\n",
    "        next_mem = F.relu(self.W_mem(concat))\n",
    "        next_mem = next_mem.unsqueeze(1)\n",
    "\n",
    "        return next_mem"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### AnswerModule"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The answer module takes in input both the question as well as final memory hidden state to update the the answer. Here we concatenate the final memory state and the question qnd then we can pass it through a linear layer."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "class AnswerModule(nn.Module):\n",
    "    def __init__(self, vocab_size, hidden_size):\n",
    "        super(AnswerModule, self).__init__()\n",
    "        self.vocab_size = vocab_size\n",
    "        self.hidden_size = hidden_size\n",
    "        self.W = nn.Linear(2*hidden_size, vocab_size)\n",
    "        init.xavier_normal(self.W.state_dict()['weight'])\n",
    "        self.dropout = nn.Dropout(0.1)\n",
    "\n",
    "    def forward(self, final_mem, questions):\n",
    "        final_mem = self.dropout(final_mem)\n",
    "        concat = torch.cat([final_mem, questions], dim=2).squeeze(1)\n",
    "        out = self.W(concat) # As per the paper, we are concatenating the final memory state m_T, and the question q and passing \n",
    "        # this resultant vector to a linear layer\n",
    "\n",
    "        return out"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We define the model for the network incorporating the input, question, answer and the episodic memory module. We use the Cross Entropy loss criterion for measuring loss. Vocab size refers to the size of vocabulary used. The output from the input module, question module are used and the representations of memory module are calculated for a fixed number of passes. Predictions are then made using that final representation of memory module and the output from the question module."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "''' We define the model for the network incorporating the input, question, answer and the episodic memory module. We use the Cross Entropy loss criterion for measuring loss'''    \n",
    "class DMN(nn.Module):\n",
    "    def __init__(self, hidden_size, vocab_size, num_pass=3, qa=None):\n",
    "        super(DMN,self).__init__()\n",
    "        self.num_pass= num_pass\n",
    "        self.qa= qa\n",
    "        self.word_embedding= nn.Embedding(vocab_size, hidden_size, padding_index=0, sparse=True)\n",
    "        init.uniform(self.word_embedding.state_dict()['weight'], a= -(3**0.5), b=3**0.5)\n",
    "        self.criterion= nn.CrossEntropyLoss(size_average=False)\n",
    "        \n",
    "        self.input_module= input_module(vocab_size,hidden_size)   ##Vocab size refers to the size of vocabulary used\n",
    "        self.question_module= question_module(vocab_size, hidden_size) \n",
    "        self.memory= episodic_memory(hidden_size)\n",
    "        self.answer_module= answer_module(vocab_size,hidden_size)\n",
    "        \n",
    "    def forward(self, context, questions):\n",
    "        #facts.size()= (batch_size, num_sentences, embedding_length= hidden.size()) \n",
    "        #questions.size() = (batch_size, 1, embedding_length)\n",
    "        facts= self.input_module(context, self.word_embedding)\n",
    "        questions= self.question_module(questions, self.word_embedding)\n",
    "        X= questions\n",
    "        for passes in range(self.num_pass):\n",
    "            X= self.memory(facts, questions, X)\n",
    "        pred= self.answer_module(X, questions)\n",
    "        return pred_id\n",
    "    \n",
    "    '''Total loss to be calculated '''\n",
    "    \n",
    "    def loss(self,context, questions, targets):\n",
    "        output= self.forward(context, questions)\n",
    "        loss= self.criterion(output, targets)\n",
    "        para_loss= 0\n",
    "        for param in self.parameters():\n",
    "            para_loss+= 0.001* torch.sum(param*param)\n",
    "        pred= F.softmax(output)\n",
    "        _, pred_id= torch.max(pred, dim=1)\n",
    "        correct= (pred_id.data == answers.data)\n",
    "        acc= torch.mean(correct.float())   \n",
    "        return loss+para_loss, acc\n",
    "    \n",
    "    def interpret_indexed_tensor(self,var):\n",
    "        if len(var.size()) == 3:\n",
    "            for n, sentences in enumerate(var):\n",
    "                s= ' '.join([self.qa.IVOCAB[elem.data[0]] for elem in sentence])\n",
    "                print (str(n)+'th batch, '+str(i)+'th sentence, '+str(s))\n",
    "                \n",
    "        elif len(var.size()) == 2:\n",
    "            for n, sentence in enumerate(var):\n",
    "                s= ' '.join([self.qa.IVOCAB[elem.data[0]] for elem in sentence])\n",
    "                print (str(n)+'th batch, '+str(s))\n",
    "                \n",
    "        elif len(var.size()) == 1:\n",
    "            for n, token in enumerate(var):\n",
    "                s= self.qa.IVOCAB[token.data[0]]\n",
    "                print (str(n)+'th of batch, '+str(s))\n",
    "        "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We now train, validate and test the network on the BABI dataset provided by facebook. Training is done for 256 epochs and we use Adam optimizer for training. Early stopping criterion is employed for training."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "\n",
    "if __name__ == '__main__':\n",
    "    for itr in range(10):\n",
    "        for task in range(1,21):\n",
    "            dataset= BabiDataset(task)\n",
    "            vocab_size= len(dataset.QA.VOCAB)\n",
    "            hidden_size= 80\n",
    "            \n",
    "            model= DMN(hidden_size, vocab_size, num_pass= 3, qa= dataset.QA)   ##vocab_size denotes the size of word embedding used \n",
    "            early_stop_count= 0\n",
    "            early_stop_flag= False\n",
    "            best_acc= 0\n",
    "            optim= torch.optim.Adam(model.parameters())\n",
    "            \n",
    "            for epoch in range(256):\n",
    "                dataset.set_mode('train')\n",
    "                train_load= DataLoader(dataset, batch_size=100, shuffle= True, collate_fn= pad_collate)  ### Loading the babi dataset\n",
    "                \n",
    "                model.train()                                                       ### training the network\n",
    "                if not early_stop_flag:\n",
    "                    total_acc=0\n",
    "                    count= 0\n",
    "                    for batch_id, data in enumerate(train_load):\n",
    "                        optim.zero_grad()\n",
    "                        context, questions, answers = data\n",
    "                        batch_size= context.size()[0]\n",
    "                        context= Variable(context.long())                           ## context.size() = (batch_size, num_sentences, embedding_length) embedding_length = hidden_size \n",
    "                        questions= Variable(questions.long())                       ## questions.size() = (batch_size, num_tokens)\n",
    "                        answers= Variable(answers)\n",
    "                        \n",
    "                        total_loss, acc= model.loss(context,questions,answers)      ## Loss is calculated and gradients are backpropagated through the layers.\n",
    "                        total_loss.backward()\n",
    "                        total_acc+= acc*batch_size\n",
    "                        count+= batch_size\n",
    "                        \n",
    "                        if batch_id %20 == 0:\n",
    "                            print('training error')\n",
    "                            print ('task '+str(task_id)+',epoch '+str(epoch)+',loss ' +str(loss.data[0])+',total accuracy : '+str(total_acc/cnt))\n",
    "                        optim.step()\n",
    "                    \n",
    "                    '''Validation part'''\n",
    "\n",
    "\n",
    "                    dataset.set_mode('valid')\n",
    "                    valid_load = DataLoader(dataset, batch_size= 100, shuffle= False, collate_fn= pad_collate)    ## Loading the validation data\n",
    "                    \n",
    "                    model.eval()\n",
    "                    total_acc=0\n",
    "                    count=0\n",
    "                    for batch_id, data in enumerate(train_load):\n",
    "                        context, questions, answers = data\n",
    "                        batch_size= context.size()[0]\n",
    "                        context= Variable(context.long())\n",
    "                        questions= Variable(questions.long())\n",
    "                        answers= Variable(answers)\n",
    "                        \n",
    "                        _, acc= model.loss(context,questions,answers)  \n",
    "                        total_loss.backward()\n",
    "                        total_acc+= acc*batch_size\n",
    "                        count+= batch_size\n",
    "                    \n",
    "                    total_acc= total_acc/ count\n",
    "                    \n",
    "                    if total_acc > best_acc:\n",
    "                        best_acc= total_acc\n",
    "                        best_state= model.state_dict()\n",
    "                        early_stop_count= 0\n",
    "                    else:\n",
    "                        early_stop_count+= 1                   \n",
    "                        if early_stop_count > 20:\n",
    "                            early_Stop_flag= True\n",
    "                    \n",
    "                    print ('itr '+str(itr)+',task_id '+str(task_id)+',epoch '+str(epoch)+',total_acc '+str(total_acc))\n",
    "                    \n",
    "                    with open('log.txt', 'a') as fp:\n",
    "                        fp.write('itr '+str(itr)+', task_id '+str(task_id)+',epoch '+str(epoch)+',total_acc '+str(total_acc)+'+\\n ')\n",
    "                    if total_acc == 1.0:\n",
    "                        break\n",
    "                else:\n",
    "                    print('iteration'+str(itr)+'task' +str(task_id)+' Early Stopping at Epoch'  +str(epoch)+'validation accuracy :' +str(best_acc))\n",
    "                    \n",
    "            \n",
    "            \n",
    "            dataset.set_mode('test')\n",
    "            test_load= DataLoader(dataset, batch_size=100, shuffle= False, collate_fn= pad_collate)\n",
    "            \n",
    "            test_acc= 0\n",
    "            count=0\n",
    "            \n",
    "            for batch_id, data in enumerate(test_load):\n",
    "                    context, questions, answers = data\n",
    "                    batch_size= context.size()[0]\n",
    "                    context= Variable(context.long())\n",
    "                    questions= Variable(questions.long())\n",
    "                    answers= Variable(answers)\n",
    "                    \n",
    "                    model.load_state_dict(best_state)\n",
    "                    _, acc= model.loss(context, questions, answers)\n",
    "                    \n",
    "                    test_acc += acc* batch_size \n",
    "                    count += batch_size\n",
    "                    print ('itr '+ str(itr)+'task =' +str(task_id)+ 'Epoch ' +str(epoch)+' test accuracy : '+str(test_acc / count))\n",
    "                    \n",
    "                    \n",
    "                    \n",
    "                    os.makedirs('models',exist_ok= True)\n",
    "                    with open('models/task'+str(task_id)+'_epoch'+str(epoch)+'_run'+str(run)+'_acc'+str(test_acc/cnt)+'.pth', 'wb') as fp:\n",
    "                        torch.save(model.state_dict(), fp)\n",
    "                    with open('log.txt', 'a') as fp:\n",
    "                        fp.write('[itr '+str(itr)+', Task '+str(task_id)+', Epoch '+str(epoch)+'] [Test] Accuracy : '+str(total_acc)+' + \\n')\n",
    "\n",
    "                        "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Conclusion"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We just saw the implementation of DMNPlus for bAbI dataset for 20 different types of task and got a pretty good result. This model can be easily extended for visual question answering as well. Only the input module differs in case of VQA. In this, the module splits the original image into small local regions and considers each region equivalent to a sentence in the input module for text. We use CNN to extract features from each of the small local regions and the final output layer is divided into a number of local regional vectors equivalant to f1, f2, f3,...in case of text. Further process is exactly same as that of textual question answering."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Unlike the previous paper on DMN, this model answers question without providing supporting facts for answering the question. Also DMNPlus can be extended to other modalities such as images for VQA. In future, We will train the model on SQuAD dataset. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.15rc1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}