examples/qlstm_tagger.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "source": [
    "# Hybrid Quantum Long-Short Term Memory NLP Example\n",
    "The following example is adapted from the [official documentation](https://pytorch.org/tutorials/beginner/nlp/sequence_models_tutorial.html) of pytorch, with the purpose of comparing the results of pytorch's classical LSTM and our hybrid quantum implementation.\n",
    "\n",
    "In brief, starting from an input sequence $w_1$, ..., $w_i \\in \\mathcal{V}$ and a tag series $\\{y_j\\}_{j=1,...i} \\in \\mathcal{T}$, being $\\mathcal{V}$ our vocabulary and $\\mathcal{T}$ our tag set, the model outputs a prediction $\\hat{y1}, ..., \\hat{y_M} \\in \\mathcal{T}$ applying the following prediction rule\n",
    "\n",
    "\\begin{equation}\n",
    "\\hat{y_i} = argmax_{j} \\ log \\ Softmax(Ah_i + b)\n",
    "\\end{equation}\n",
    "\n",
    "Please have a look at pytorch's official documentation for further details."
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Defaulting to user installation because normal site-packages is not writeable\r\n",
      "Requirement already satisfied: qlearnkit[pennylane] in /home/mspronesti/Desktop/qlearnkit (0.2.1.dev6+g4c4fa04.d20220927)\r\n",
      "Requirement already satisfied: qiskit-terra==0.20.0 in /home/mspronesti/.local/lib/python3.8/site-packages (from qlearnkit[pennylane]) (0.20.0)\r\n",
      "Requirement already satisfied: qiskit-aer>=0.10.3 in /home/mspronesti/.local/lib/python3.8/site-packages (from qlearnkit[pennylane]) (0.10.4)\r\n",
      "Requirement already satisfied: qiskit-machine-learning>=0.3.1 in /home/mspronesti/.local/lib/python3.8/site-packages (from qlearnkit[pennylane]) (0.3.1)\r\n",
      "Requirement already satisfied: scikit-learn==1.0.2 in /home/mspronesti/.local/lib/python3.8/site-packages (from qlearnkit[pennylane]) (1.0.2)\r\n",
      "Requirement already satisfied: scipy==1.7.3 in /home/mspronesti/.local/lib/python3.8/site-packages (from qlearnkit[pennylane]) (1.7.3)\r\n",
      "Requirement already satisfied: numpy==1.22.0 in /home/mspronesti/.local/lib/python3.8/site-packages (from qlearnkit[pennylane]) (1.22.0)\r\n",
      "Requirement already satisfied: pennylane in /home/mspronesti/.local/lib/python3.8/site-packages (from qlearnkit[pennylane]) (0.20.0)\r\n",
      "Requirement already satisfied: torch in /home/mspronesti/.local/lib/python3.8/site-packages (from qlearnkit[pennylane]) (1.10.2)\r\n",
      "Requirement already satisfied: retworkx>=0.11.0 in /home/mspronesti/.local/lib/python3.8/site-packages (from qiskit-terra==0.20.0->qlearnkit[pennylane]) (0.11.0)\r\n",
      "Requirement already satisfied: psutil>=5 in /home/mspronesti/.local/lib/python3.8/site-packages (from qiskit-terra==0.20.0->qlearnkit[pennylane]) (5.8.0)\r\n",
      "Requirement already satisfied: stevedore>=3.0.0 in /usr/local/lib/python3.8/dist-packages (from qiskit-terra==0.20.0->qlearnkit[pennylane]) (3.5.0)\r\n",
      "Requirement already satisfied: symengine>=0.9 in /home/mspronesti/.local/lib/python3.8/site-packages (from qiskit-terra==0.20.0->qlearnkit[pennylane]) (0.9.2)\r\n",
      "Requirement already satisfied: python-dateutil>=2.8.0 in /home/mspronesti/.local/lib/python3.8/site-packages (from qiskit-terra==0.20.0->qlearnkit[pennylane]) (2.8.2)\r\n",
      "Requirement already satisfied: tweedledum<2.0,>=1.1 in /home/mspronesti/.local/lib/python3.8/site-packages (from qiskit-terra==0.20.0->qlearnkit[pennylane]) (1.1.1)\r\n",
      "Requirement already satisfied: python-constraint>=1.4 in /home/mspronesti/.local/lib/python3.8/site-packages (from qiskit-terra==0.20.0->qlearnkit[pennylane]) (1.4.0)\r\n",
      "Requirement already satisfied: sympy>=1.3 in /home/mspronesti/.local/lib/python3.8/site-packages (from qiskit-terra==0.20.0->qlearnkit[pennylane]) (1.9)\r\n",
      "Requirement already satisfied: dill>=0.3 in /home/mspronesti/.local/lib/python3.8/site-packages (from qiskit-terra==0.20.0->qlearnkit[pennylane]) (0.3.5.1)\r\n",
      "Requirement already satisfied: ply>=3.10 in /home/mspronesti/.local/lib/python3.8/site-packages (from qiskit-terra==0.20.0->qlearnkit[pennylane]) (3.11)\r\n",
      "Requirement already satisfied: joblib>=0.11 in /home/mspronesti/.local/lib/python3.8/site-packages (from scikit-learn==1.0.2->qlearnkit[pennylane]) (1.1.0)\r\n",
      "Requirement already satisfied: threadpoolctl>=2.0.0 in /home/mspronesti/.local/lib/python3.8/site-packages (from scikit-learn==1.0.2->qlearnkit[pennylane]) (3.0.0)\r\n",
      "Requirement already satisfied: fastdtw in /home/mspronesti/.local/lib/python3.8/site-packages (from qiskit-machine-learning>=0.3.1->qlearnkit[pennylane]) (0.3.4)\r\n",
      "Requirement already satisfied: setuptools>=40.1.0 in /home/mspronesti/.local/lib/python3.8/site-packages (from qiskit-machine-learning>=0.3.1->qlearnkit[pennylane]) (59.6.0)\r\n",
      "Requirement already satisfied: networkx in /home/mspronesti/.local/lib/python3.8/site-packages (from pennylane->qlearnkit[pennylane]) (2.6.3)\r\n",
      "Requirement already satisfied: toml in /home/mspronesti/.local/lib/python3.8/site-packages (from pennylane->qlearnkit[pennylane]) (0.10.2)\r\n",
      "Requirement already satisfied: autoray in /home/mspronesti/.local/lib/python3.8/site-packages (from pennylane->qlearnkit[pennylane]) (0.2.5)\r\n",
      "Requirement already satisfied: cachetools in /home/mspronesti/.local/lib/python3.8/site-packages (from pennylane->qlearnkit[pennylane]) (4.2.4)\r\n",
      "Requirement already satisfied: semantic-version==2.6 in /home/mspronesti/.local/lib/python3.8/site-packages (from pennylane->qlearnkit[pennylane]) (2.6.0)\r\n",
      "Requirement already satisfied: appdirs in /home/mspronesti/.local/lib/python3.8/site-packages (from pennylane->qlearnkit[pennylane]) (1.4.4)\r\n",
      "Requirement already satisfied: autograd in /home/mspronesti/.local/lib/python3.8/site-packages (from pennylane->qlearnkit[pennylane]) (1.3)\r\n",
      "Requirement already satisfied: pennylane-lightning>=0.18 in /home/mspronesti/.local/lib/python3.8/site-packages (from pennylane->qlearnkit[pennylane]) (0.20.2)\r\n",
      "Requirement already satisfied: typing-extensions in /home/mspronesti/.local/lib/python3.8/site-packages (from torch->qlearnkit[pennylane]) (4.3.0)\r\n",
      "Requirement already satisfied: ninja in /home/mspronesti/.local/lib/python3.8/site-packages (from pennylane-lightning>=0.18->pennylane->qlearnkit[pennylane]) (1.10.2.3)\r\n",
      "Requirement already satisfied: six>=1.5 in /home/mspronesti/.local/lib/python3.8/site-packages (from python-dateutil>=2.8.0->qiskit-terra==0.20.0->qlearnkit[pennylane]) (1.16.0)\r\n",
      "Requirement already satisfied: pbr!=2.1.0,>=2.0.0 in /usr/local/lib/python3.8/dist-packages (from stevedore>=3.0.0->qiskit-terra==0.20.0->qlearnkit[pennylane]) (5.8.0)\r\n",
      "Requirement already satisfied: mpmath>=0.19 in /home/mspronesti/.local/lib/python3.8/site-packages (from sympy>=1.3->qiskit-terra==0.20.0->qlearnkit[pennylane]) (1.2.1)\r\n",
      "Requirement already satisfied: future>=0.15.2 in /home/mspronesti/.local/lib/python3.8/site-packages (from autograd->pennylane->qlearnkit[pennylane]) (0.18.2)\r\n",
      "Defaulting to user installation because normal site-packages is not writeable\r\n",
      "Requirement already satisfied: matplotlib in /home/mspronesti/.local/lib/python3.8/site-packages (3.4.3)\r\n",
      "Requirement already satisfied: kiwisolver>=1.0.1 in /home/mspronesti/.local/lib/python3.8/site-packages (from matplotlib) (1.3.2)\r\n",
      "Requirement already satisfied: python-dateutil>=2.7 in /home/mspronesti/.local/lib/python3.8/site-packages (from matplotlib) (2.8.2)\r\n",
      "Requirement already satisfied: pyparsing>=2.2.1 in /home/mspronesti/.local/lib/python3.8/site-packages (from matplotlib) (2.4.7)\r\n",
      "Requirement already satisfied: cycler>=0.10 in /home/mspronesti/.local/lib/python3.8/site-packages (from matplotlib) (0.10.0)\r\n",
      "Requirement already satisfied: numpy>=1.16 in /home/mspronesti/.local/lib/python3.8/site-packages (from matplotlib) (1.22.0)\r\n",
      "Requirement already satisfied: pillow>=6.2.0 in /usr/lib/python3/dist-packages (from matplotlib) (7.0.0)\r\n",
      "Requirement already satisfied: six in /home/mspronesti/.local/lib/python3.8/site-packages (from cycler>=0.10->matplotlib) (1.16.0)\r\n"
     ]
    }
   ],
   "source": [
    "!pip install qlearnkit['pennylane']\n",
    "!pip install matplotlib"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "outputs": [],
   "source": [
    "import qlearnkit as ql\n",
    "import numpy as np\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "from matplotlib import pyplot as plt\n"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "Let's first prepare our data"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Vocabulary: {'The': 0, 'dog': 1, 'ate': 2, 'the': 3, 'apple': 4, 'Everybody': 5, 'read': 6, 'that': 7, 'book': 8}\n",
      "Entities: {0: 'DET', 1: 'NN', 2: 'V'}\n"
     ]
    }
   ],
   "source": [
    "tag_to_ix = {\"DET\": 0, \"NN\": 1, \"V\": 2}  # Assign each tag with a unique index\n",
    "ix_to_tag = {i: k for k, i in tag_to_ix.items()}\n",
    "\n",
    "\n",
    "def prepare_sequence(seq, to_ix):\n",
    "    idxs = [to_ix[w] for w in seq]\n",
    "    return torch.tensor(idxs, dtype=torch.long)\n",
    "\n",
    "\n",
    "training_data = [\n",
    "    # Tags are: DET - determiner; NN - noun; V - verb\n",
    "    # For example, the word \"The\" is a determiner\n",
    "    (\"The dog ate the apple\".split(), [\"DET\", \"NN\", \"V\", \"DET\", \"NN\"]),\n",
    "    (\"Everybody read that book\".split(), [\"NN\", \"V\", \"DET\", \"NN\"])\n",
    "]\n",
    "word_to_ix = {}\n",
    "\n",
    "# For each words-list (sentence) and tags-list in each tuple of training_data\n",
    "for sent, tags in training_data:\n",
    "    for word in sent:\n",
    "        if word not in word_to_ix:  # word has not been assigned an index yet\n",
    "            word_to_ix[word] = len(word_to_ix)  # Assign each word with a unique index\n",
    "\n",
    "print(f\"Vocabulary: {word_to_ix}\")\n",
    "print(f\"Entities: {ix_to_tag}\")"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "Now Let's create the Tagger model embedding our LSTM"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "outputs": [],
   "source": [
    "class LSTMTagger(nn.Module):\n",
    "\n",
    "    def __init__(self,\n",
    "                 model,\n",
    "                 embedding_dim,\n",
    "                 hidden_dim,\n",
    "                 vocab_size,\n",
    "                 tagset_size):\n",
    "        super(LSTMTagger, self).__init__()\n",
    "        self.hidden_dim = hidden_dim\n",
    "\n",
    "        self.word_embeddings = nn.Embedding(vocab_size, embedding_dim)\n",
    "\n",
    "        self.model = model\n",
    "\n",
    "\n",
    "        # The linear layer that maps from hidden state space to tag space\n",
    "        self.hidden2tag = nn.Linear(hidden_dim, tagset_size)\n",
    "\n",
    "    def forward(self, sentence):\n",
    "        embeds = self.word_embeddings(sentence)\n",
    "        lstm_out, _ = self.model(embeds.view(len(sentence), 1, -1))\n",
    "        tag_logits = self.hidden2tag(lstm_out.view(len(sentence), -1))\n",
    "        tag_scores = F.log_softmax(tag_logits, dim=1)\n",
    "        return tag_scores"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "Let's set (by hand) some hyperparameters"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "outputs": [],
   "source": [
    "embedding_dim = 8\n",
    "hidden_dim = 6\n",
    "n_layers = 1\n",
    "n_qubits = 4\n",
    "n_epochs = 300\n",
    "device = 'default.qubit'"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "Let's also create an helper trainer that accepting our hyperparameters and a model, to simply call it with our two implementations"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "outputs": [],
   "source": [
    "def trainer(lstm_model, embedding_dim, hidden_dim,\n",
    "            n_epochs, model_label):\n",
    "    # the LSTM Tagger Model\n",
    "    model = LSTMTagger(lstm_model,\n",
    "                   embedding_dim,\n",
    "                   hidden_dim,\n",
    "                   vocab_size=len(word_to_ix),\n",
    "                   tagset_size=len(tag_to_ix))\n",
    "    # loss function and Stochastic Gradient Descend\n",
    "    # as optimizers\n",
    "    loss_function = nn.NLLLoss()\n",
    "    optimizer = optim.SGD(model.parameters(), lr=0.1)\n",
    "\n",
    "    history = {\n",
    "        'loss': [],\n",
    "        'acc': []\n",
    "    }\n",
    "    for epoch in range(n_epochs):\n",
    "        losses = []\n",
    "        preds = []\n",
    "        targets = []\n",
    "        for sentence, tags in training_data:\n",
    "            # Step 1. Remember that Pytorch accumulates gradients.\n",
    "            # We need to clear them out before each instance\n",
    "            model.zero_grad()\n",
    "\n",
    "            # Step 2. Get our inputs ready for the network, that is, turn them into\n",
    "            # Tensors of word indices.\n",
    "            sentence_in = prepare_sequence(sentence, word_to_ix)\n",
    "            labels = prepare_sequence(tags, tag_to_ix)\n",
    "\n",
    "            # Step 3. Run our forward pass.\n",
    "            tag_scores = model(sentence_in)\n",
    "\n",
    "            # Step 4. Compute the loss, gradients, and update the parameters by\n",
    "            #  calling optimizer.step()\n",
    "            loss = loss_function(tag_scores, labels)\n",
    "            loss.backward()\n",
    "            optimizer.step()\n",
    "            losses.append(float(loss))\n",
    "\n",
    "            probs = torch.softmax(tag_scores, dim=-1)\n",
    "            preds.append(probs.argmax(dim=-1))\n",
    "            targets.append(labels)\n",
    "        avg_loss = np.mean(losses)\n",
    "        history['loss'].append(avg_loss)\n",
    "\n",
    "        # print(\"preds\", preds)\n",
    "        preds = torch.cat(preds)\n",
    "        targets = torch.cat(targets)\n",
    "        corrects = (preds == targets)\n",
    "        accuracy = corrects.sum().float() / float(targets.size(0))\n",
    "        history['acc'].append(accuracy)\n",
    "\n",
    "        print(f\"Epoch {epoch + 1} / {n_epochs}: Loss = {avg_loss:.3f} Acc = {accuracy:.2f}\")\n",
    "\n",
    "    with torch.no_grad():\n",
    "        input_sentence = training_data[0][0]\n",
    "        labels = training_data[0][1]\n",
    "        inputs = prepare_sequence(input_sentence, word_to_ix)\n",
    "        tag_scores = model(inputs)\n",
    "\n",
    "        tag_ids = torch.argmax(tag_scores, dim=1).numpy()\n",
    "        tag_labels = [ix_to_tag[k] for k in tag_ids]\n",
    "        print(f\"Sentence:  {input_sentence}\")\n",
    "        print(f\"Labels:    {labels}\")\n",
    "        print(f\"Predicted: {tag_labels}\")\n",
    "\n",
    "    fig, ax1 = plt.subplots()\n",
    "\n",
    "    ax1.set_xlabel(\"Epoch\")\n",
    "    ax1.set_ylabel(\"Loss\")\n",
    "    ax1.plot(history['loss'], label=f\"{model_label} Loss\")\n",
    "\n",
    "    ax2 = ax1.twinx()\n",
    "    ax2.set_ylabel(\"Accuracy\")\n",
    "    ax2.plot(history['acc'], label=f\"{model_label} LSTM Accuracy\", color='tab:red')\n",
    "\n",
    "    plt.title(\"Part-of-Speech Tagger Training\")\n",
    "    plt.ylim(0., 1.5)\n",
    "    plt.legend(loc=\"upper right\")\n",
    "\n",
    "    plt.show()"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "Let's create our Quantum LSTM from qleanrkit"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "outputs": [],
   "source": [
    "import qlearnkit.nn as qnn\n",
    "qlstm = qnn.QLongShortTermMemory(\n",
    "    embedding_dim,\n",
    "    hidden_dim,\n",
    "    n_layers,\n",
    "    n_qubits=n_qubits,\n",
    "    device=device\n",
    ")"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "Let's have a look at its (quantum) architecture\n",
    "\n",
    "**NOTICE:** ignore the following code per se, it has nothing to do with the actual usage of qlearnkit's QLSTM. It's just here to cope with the required parameters of `qml.draw`"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 0: ──H──RY(M0)──RZ(M4)──╭C─────────────────────────────╭X──RX(0.642)──RY(0.351)──RZ(0.687)──┤ ⟨Z⟩ \n",
      " 1: ──H──RY(M1)──RZ(M5)──╰X──╭C───RX(0.138)──RY(0.664)──│───RZ(0.154)────────────────────────┤ ⟨Z⟩ \n",
      " 2: ──H──RY(M2)──RZ(M6)──────╰X──╭C──────────RX(0.493)──│───RY(0.593)──RZ(0.452)─────────────┤ ⟨Z⟩ \n",
      " 3: ──H──RY(M3)──RZ(M7)──────────╰X─────────────────────╰C──RX(0.968)──RY(0.899)──RZ(0.434)──┤ ⟨Z⟩ \n",
      "M0 =\n",
      "tensor([0.4410, 0.2851, 0.0922, 0.2085, 0.4189, 0.5568, 0.3468, 0.4773])\n",
      "M1 =\n",
      "tensor([0.5397, 0.3276, 0.0610, 0.7491, 0.2016, 0.1892, 0.4293, 0.3255])\n",
      "M2 =\n",
      "tensor([0.5904, 0.5577, 0.2555, 0.6175, 0.4668, 0.4356, 0.3120, 0.4006])\n",
      "M3 =\n",
      "tensor([0.6520, 0.0839, 0.3924, 0.0972, 0.4129, 0.7706, 0.1809, 0.6805])\n",
      "M4 =\n",
      "tensor([0.2192, 0.0857, 0.0085, 0.0447, 0.1958, 0.3697, 0.1299, 0.2614])\n",
      "M5 =\n",
      "tensor([0.3446, 0.1150, 0.0037, 0.7129, 0.0418, 0.0367, 0.2065, 0.1134])\n",
      "M6 =\n",
      "tensor([0.4221, 0.3711, 0.0681, 0.4672, 0.2487, 0.2133, 0.1037, 0.1775])\n",
      "M7 =\n",
      "tensor([0.5277, 0.0071, 0.1696, 0.0095, 0.1896, 0.7558, 0.0334, 0.5800])\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# random values to feed into the QNode\n",
    "inputs = torch.rand(embedding_dim, n_qubits)\n",
    "weights = torch.rand(n_layers, n_qubits, 3)\n",
    "import pennylane as qml\n",
    "dev = qml.device(device, wires=n_qubits)\n",
    "circ = qml.QNode(qlstm._construct_vqc, dev)\n",
    "\n",
    "print(qml.draw(circ, expansion_strategy='device', show_all_wires=True)(inputs, weights))"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "Now we can train our hybrid-quantum model"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1 / 300: Loss = 1.216 Acc = 0.22\n",
      "Epoch 2 / 300: Loss = 1.183 Acc = 0.22\n",
      "Epoch 3 / 300: Loss = 1.153 Acc = 0.22\n",
      "Epoch 4 / 300: Loss = 1.126 Acc = 0.22\n",
      "Epoch 5 / 300: Loss = 1.103 Acc = 0.22\n",
      "Epoch 6 / 300: Loss = 1.082 Acc = 0.33\n",
      "Epoch 7 / 300: Loss = 1.064 Acc = 0.44\n",
      "Epoch 8 / 300: Loss = 1.048 Acc = 0.56\n",
      "Epoch 9 / 300: Loss = 1.034 Acc = 0.78\n",
      "Epoch 10 / 300: Loss = 1.021 Acc = 0.67\n",
      "Epoch 11 / 300: Loss = 1.009 Acc = 0.67\n",
      "Epoch 12 / 300: Loss = 0.998 Acc = 0.67\n",
      "Epoch 13 / 300: Loss = 0.988 Acc = 0.67\n",
      "Epoch 14 / 300: Loss = 0.978 Acc = 0.67\n",
      "Epoch 15 / 300: Loss = 0.969 Acc = 0.67\n",
      "Epoch 16 / 300: Loss = 0.959 Acc = 0.67\n",
      "Epoch 17 / 300: Loss = 0.950 Acc = 0.67\n",
      "Epoch 18 / 300: Loss = 0.940 Acc = 0.67\n",
      "Epoch 19 / 300: Loss = 0.930 Acc = 0.67\n",
      "Epoch 20 / 300: Loss = 0.918 Acc = 0.67\n",
      "Epoch 21 / 300: Loss = 0.906 Acc = 0.67\n",
      "Epoch 22 / 300: Loss = 0.893 Acc = 0.67\n",
      "Epoch 23 / 300: Loss = 0.879 Acc = 0.67\n",
      "Epoch 24 / 300: Loss = 0.865 Acc = 0.67\n",
      "Epoch 25 / 300: Loss = 0.853 Acc = 0.78\n",
      "Epoch 26 / 300: Loss = 0.841 Acc = 0.78\n",
      "Epoch 27 / 300: Loss = 0.830 Acc = 0.78\n",
      "Epoch 28 / 300: Loss = 0.820 Acc = 0.78\n",
      "Epoch 29 / 300: Loss = 0.811 Acc = 0.78\n",
      "Epoch 30 / 300: Loss = 0.801 Acc = 0.78\n",
      "Epoch 31 / 300: Loss = 0.792 Acc = 0.78\n",
      "Epoch 32 / 300: Loss = 0.783 Acc = 0.78\n",
      "Epoch 33 / 300: Loss = 0.775 Acc = 0.78\n",
      "Epoch 34 / 300: Loss = 0.766 Acc = 0.78\n",
      "Epoch 35 / 300: Loss = 0.758 Acc = 0.78\n",
      "Epoch 36 / 300: Loss = 0.750 Acc = 0.78\n",
      "Epoch 37 / 300: Loss = 0.742 Acc = 0.78\n",
      "Epoch 38 / 300: Loss = 0.734 Acc = 0.78\n",
      "Epoch 39 / 300: Loss = 0.726 Acc = 0.78\n",
      "Epoch 40 / 300: Loss = 0.719 Acc = 0.78\n",
      "Epoch 41 / 300: Loss = 0.712 Acc = 0.78\n",
      "Epoch 42 / 300: Loss = 0.704 Acc = 0.78\n",
      "Epoch 43 / 300: Loss = 0.697 Acc = 0.78\n",
      "Epoch 44 / 300: Loss = 0.690 Acc = 0.78\n",
      "Epoch 45 / 300: Loss = 0.683 Acc = 0.78\n",
      "Epoch 46 / 300: Loss = 0.677 Acc = 0.78\n",
      "Epoch 47 / 300: Loss = 0.670 Acc = 0.78\n",
      "Epoch 48 / 300: Loss = 0.664 Acc = 0.78\n",
      "Epoch 49 / 300: Loss = 0.657 Acc = 0.78\n",
      "Epoch 50 / 300: Loss = 0.651 Acc = 0.78\n",
      "Epoch 51 / 300: Loss = 0.645 Acc = 0.78\n",
      "Epoch 52 / 300: Loss = 0.639 Acc = 0.78\n",
      "Epoch 53 / 300: Loss = 0.633 Acc = 0.78\n",
      "Epoch 54 / 300: Loss = 0.627 Acc = 0.78\n",
      "Epoch 55 / 300: Loss = 0.621 Acc = 0.78\n",
      "Epoch 56 / 300: Loss = 0.616 Acc = 0.78\n",
      "Epoch 57 / 300: Loss = 0.610 Acc = 0.78\n",
      "Epoch 58 / 300: Loss = 0.605 Acc = 0.78\n",
      "Epoch 59 / 300: Loss = 0.599 Acc = 0.78\n",
      "Epoch 60 / 300: Loss = 0.594 Acc = 0.78\n",
      "Epoch 61 / 300: Loss = 0.589 Acc = 0.78\n",
      "Epoch 62 / 300: Loss = 0.584 Acc = 0.78\n",
      "Epoch 63 / 300: Loss = 0.579 Acc = 0.78\n",
      "Epoch 64 / 300: Loss = 0.574 Acc = 0.78\n",
      "Epoch 65 / 300: Loss = 0.569 Acc = 0.78\n",
      "Epoch 66 / 300: Loss = 0.564 Acc = 0.78\n",
      "Epoch 67 / 300: Loss = 0.560 Acc = 0.78\n",
      "Epoch 68 / 300: Loss = 0.555 Acc = 0.78\n",
      "Epoch 69 / 300: Loss = 0.551 Acc = 0.78\n",
      "Epoch 70 / 300: Loss = 0.546 Acc = 0.78\n",
      "Epoch 71 / 300: Loss = 0.542 Acc = 0.78\n",
      "Epoch 72 / 300: Loss = 0.538 Acc = 0.78\n",
      "Epoch 73 / 300: Loss = 0.533 Acc = 0.78\n",
      "Epoch 74 / 300: Loss = 0.529 Acc = 0.78\n",
      "Epoch 75 / 300: Loss = 0.525 Acc = 0.78\n",
      "Epoch 76 / 300: Loss = 0.521 Acc = 0.78\n",
      "Epoch 77 / 300: Loss = 0.517 Acc = 0.78\n",
      "Epoch 78 / 300: Loss = 0.513 Acc = 0.78\n",
      "Epoch 79 / 300: Loss = 0.510 Acc = 0.78\n",
      "Epoch 80 / 300: Loss = 0.506 Acc = 0.78\n",
      "Epoch 81 / 300: Loss = 0.502 Acc = 0.78\n",
      "Epoch 82 / 300: Loss = 0.499 Acc = 0.78\n",
      "Epoch 83 / 300: Loss = 0.495 Acc = 0.78\n",
      "Epoch 84 / 300: Loss = 0.491 Acc = 0.78\n",
      "Epoch 85 / 300: Loss = 0.488 Acc = 0.78\n",
      "Epoch 86 / 300: Loss = 0.485 Acc = 0.78\n",
      "Epoch 87 / 300: Loss = 0.481 Acc = 0.78\n",
      "Epoch 88 / 300: Loss = 0.478 Acc = 0.78\n",
      "Epoch 89 / 300: Loss = 0.474 Acc = 0.78\n",
      "Epoch 90 / 300: Loss = 0.471 Acc = 0.78\n",
      "Epoch 91 / 300: Loss = 0.468 Acc = 0.78\n",
      "Epoch 92 / 300: Loss = 0.465 Acc = 0.78\n",
      "Epoch 93 / 300: Loss = 0.461 Acc = 0.78\n",
      "Epoch 94 / 300: Loss = 0.458 Acc = 0.78\n",
      "Epoch 95 / 300: Loss = 0.455 Acc = 0.78\n",
      "Epoch 96 / 300: Loss = 0.452 Acc = 0.78\n",
      "Epoch 97 / 300: Loss = 0.449 Acc = 0.78\n",
      "Epoch 98 / 300: Loss = 0.446 Acc = 0.78\n",
      "Epoch 99 / 300: Loss = 0.443 Acc = 0.78\n",
      "Epoch 100 / 300: Loss = 0.440 Acc = 0.78\n",
      "Epoch 101 / 300: Loss = 0.437 Acc = 0.78\n",
      "Epoch 102 / 300: Loss = 0.434 Acc = 0.78\n",
      "Epoch 103 / 300: Loss = 0.431 Acc = 0.78\n",
      "Epoch 104 / 300: Loss = 0.428 Acc = 0.78\n",
      "Epoch 105 / 300: Loss = 0.425 Acc = 0.78\n",
      "Epoch 106 / 300: Loss = 0.422 Acc = 0.78\n",
      "Epoch 107 / 300: Loss = 0.419 Acc = 0.78\n",
      "Epoch 108 / 300: Loss = 0.416 Acc = 0.78\n",
      "Epoch 109 / 300: Loss = 0.413 Acc = 0.78\n",
      "Epoch 110 / 300: Loss = 0.411 Acc = 0.89\n",
      "Epoch 111 / 300: Loss = 0.408 Acc = 0.89\n",
      "Epoch 112 / 300: Loss = 0.405 Acc = 0.89\n",
      "Epoch 113 / 300: Loss = 0.402 Acc = 0.89\n",
      "Epoch 114 / 300: Loss = 0.399 Acc = 0.89\n",
      "Epoch 115 / 300: Loss = 0.396 Acc = 0.89\n",
      "Epoch 116 / 300: Loss = 0.394 Acc = 0.89\n",
      "Epoch 117 / 300: Loss = 0.391 Acc = 1.00\n",
      "Epoch 118 / 300: Loss = 0.388 Acc = 1.00\n",
      "Epoch 119 / 300: Loss = 0.385 Acc = 1.00\n",
      "Epoch 120 / 300: Loss = 0.383 Acc = 1.00\n",
      "Epoch 121 / 300: Loss = 0.380 Acc = 1.00\n",
      "Epoch 122 / 300: Loss = 0.377 Acc = 1.00\n",
      "Epoch 123 / 300: Loss = 0.374 Acc = 1.00\n",
      "Epoch 124 / 300: Loss = 0.372 Acc = 1.00\n",
      "Epoch 125 / 300: Loss = 0.369 Acc = 1.00\n",
      "Epoch 126 / 300: Loss = 0.366 Acc = 1.00\n",
      "Epoch 127 / 300: Loss = 0.364 Acc = 1.00\n",
      "Epoch 128 / 300: Loss = 0.361 Acc = 1.00\n",
      "Epoch 129 / 300: Loss = 0.358 Acc = 1.00\n",
      "Epoch 130 / 300: Loss = 0.355 Acc = 1.00\n",
      "Epoch 131 / 300: Loss = 0.353 Acc = 1.00\n",
      "Epoch 132 / 300: Loss = 0.350 Acc = 1.00\n",
      "Epoch 133 / 300: Loss = 0.347 Acc = 1.00\n",
      "Epoch 134 / 300: Loss = 0.345 Acc = 1.00\n",
      "Epoch 135 / 300: Loss = 0.342 Acc = 1.00\n",
      "Epoch 136 / 300: Loss = 0.339 Acc = 1.00\n",
      "Epoch 137 / 300: Loss = 0.337 Acc = 1.00\n",
      "Epoch 138 / 300: Loss = 0.334 Acc = 1.00\n",
      "Epoch 139 / 300: Loss = 0.331 Acc = 1.00\n",
      "Epoch 140 / 300: Loss = 0.329 Acc = 1.00\n",
      "Epoch 141 / 300: Loss = 0.326 Acc = 1.00\n",
      "Epoch 142 / 300: Loss = 0.323 Acc = 1.00\n",
      "Epoch 143 / 300: Loss = 0.321 Acc = 1.00\n",
      "Epoch 144 / 300: Loss = 0.318 Acc = 1.00\n",
      "Epoch 145 / 300: Loss = 0.315 Acc = 1.00\n",
      "Epoch 146 / 300: Loss = 0.313 Acc = 1.00\n",
      "Epoch 147 / 300: Loss = 0.310 Acc = 1.00\n",
      "Epoch 148 / 300: Loss = 0.307 Acc = 1.00\n",
      "Epoch 149 / 300: Loss = 0.305 Acc = 1.00\n",
      "Epoch 150 / 300: Loss = 0.302 Acc = 1.00\n",
      "Epoch 151 / 300: Loss = 0.299 Acc = 1.00\n",
      "Epoch 152 / 300: Loss = 0.297 Acc = 1.00\n",
      "Epoch 153 / 300: Loss = 0.294 Acc = 1.00\n",
      "Epoch 154 / 300: Loss = 0.292 Acc = 1.00\n",
      "Epoch 155 / 300: Loss = 0.289 Acc = 1.00\n",
      "Epoch 156 / 300: Loss = 0.286 Acc = 1.00\n",
      "Epoch 157 / 300: Loss = 0.284 Acc = 1.00\n",
      "Epoch 158 / 300: Loss = 0.281 Acc = 1.00\n",
      "Epoch 159 / 300: Loss = 0.279 Acc = 1.00\n",
      "Epoch 160 / 300: Loss = 0.276 Acc = 1.00\n",
      "Epoch 161 / 300: Loss = 0.273 Acc = 1.00\n",
      "Epoch 162 / 300: Loss = 0.271 Acc = 1.00\n",
      "Epoch 163 / 300: Loss = 0.268 Acc = 1.00\n",
      "Epoch 164 / 300: Loss = 0.266 Acc = 1.00\n",
      "Epoch 165 / 300: Loss = 0.263 Acc = 1.00\n",
      "Epoch 166 / 300: Loss = 0.261 Acc = 1.00\n",
      "Epoch 167 / 300: Loss = 0.258 Acc = 1.00\n",
      "Epoch 168 / 300: Loss = 0.256 Acc = 1.00\n",
      "Epoch 169 / 300: Loss = 0.253 Acc = 1.00\n",
      "Epoch 170 / 300: Loss = 0.251 Acc = 1.00\n",
      "Epoch 171 / 300: Loss = 0.249 Acc = 1.00\n",
      "Epoch 172 / 300: Loss = 0.246 Acc = 1.00\n",
      "Epoch 173 / 300: Loss = 0.244 Acc = 1.00\n",
      "Epoch 174 / 300: Loss = 0.241 Acc = 1.00\n",
      "Epoch 175 / 300: Loss = 0.239 Acc = 1.00\n",
      "Epoch 176 / 300: Loss = 0.237 Acc = 1.00\n",
      "Epoch 177 / 300: Loss = 0.234 Acc = 1.00\n",
      "Epoch 178 / 300: Loss = 0.232 Acc = 1.00\n",
      "Epoch 179 / 300: Loss = 0.230 Acc = 1.00\n",
      "Epoch 180 / 300: Loss = 0.227 Acc = 1.00\n",
      "Epoch 181 / 300: Loss = 0.225 Acc = 1.00\n",
      "Epoch 182 / 300: Loss = 0.223 Acc = 1.00\n",
      "Epoch 183 / 300: Loss = 0.221 Acc = 1.00\n",
      "Epoch 184 / 300: Loss = 0.219 Acc = 1.00\n",
      "Epoch 185 / 300: Loss = 0.216 Acc = 1.00\n",
      "Epoch 186 / 300: Loss = 0.214 Acc = 1.00\n",
      "Epoch 187 / 300: Loss = 0.212 Acc = 1.00\n",
      "Epoch 188 / 300: Loss = 0.210 Acc = 1.00\n",
      "Epoch 189 / 300: Loss = 0.208 Acc = 1.00\n",
      "Epoch 190 / 300: Loss = 0.206 Acc = 1.00\n",
      "Epoch 191 / 300: Loss = 0.204 Acc = 1.00\n",
      "Epoch 192 / 300: Loss = 0.202 Acc = 1.00\n",
      "Epoch 193 / 300: Loss = 0.200 Acc = 1.00\n",
      "Epoch 194 / 300: Loss = 0.198 Acc = 1.00\n",
      "Epoch 195 / 300: Loss = 0.196 Acc = 1.00\n",
      "Epoch 196 / 300: Loss = 0.194 Acc = 1.00\n",
      "Epoch 197 / 300: Loss = 0.192 Acc = 1.00\n",
      "Epoch 198 / 300: Loss = 0.190 Acc = 1.00\n",
      "Epoch 199 / 300: Loss = 0.188 Acc = 1.00\n",
      "Epoch 200 / 300: Loss = 0.186 Acc = 1.00\n",
      "Epoch 201 / 300: Loss = 0.184 Acc = 1.00\n",
      "Epoch 202 / 300: Loss = 0.183 Acc = 1.00\n",
      "Epoch 203 / 300: Loss = 0.181 Acc = 1.00\n",
      "Epoch 204 / 300: Loss = 0.179 Acc = 1.00\n",
      "Epoch 205 / 300: Loss = 0.177 Acc = 1.00\n",
      "Epoch 206 / 300: Loss = 0.175 Acc = 1.00\n",
      "Epoch 207 / 300: Loss = 0.174 Acc = 1.00\n",
      "Epoch 208 / 300: Loss = 0.172 Acc = 1.00\n",
      "Epoch 209 / 300: Loss = 0.170 Acc = 1.00\n",
      "Epoch 210 / 300: Loss = 0.169 Acc = 1.00\n",
      "Epoch 211 / 300: Loss = 0.167 Acc = 1.00\n",
      "Epoch 212 / 300: Loss = 0.165 Acc = 1.00\n",
      "Epoch 213 / 300: Loss = 0.164 Acc = 1.00\n",
      "Epoch 214 / 300: Loss = 0.162 Acc = 1.00\n",
      "Epoch 215 / 300: Loss = 0.160 Acc = 1.00\n",
      "Epoch 216 / 300: Loss = 0.159 Acc = 1.00\n",
      "Epoch 217 / 300: Loss = 0.157 Acc = 1.00\n",
      "Epoch 218 / 300: Loss = 0.156 Acc = 1.00\n",
      "Epoch 219 / 300: Loss = 0.154 Acc = 1.00\n",
      "Epoch 220 / 300: Loss = 0.153 Acc = 1.00\n",
      "Epoch 221 / 300: Loss = 0.151 Acc = 1.00\n",
      "Epoch 222 / 300: Loss = 0.150 Acc = 1.00\n",
      "Epoch 223 / 300: Loss = 0.149 Acc = 1.00\n",
      "Epoch 224 / 300: Loss = 0.147 Acc = 1.00\n",
      "Epoch 225 / 300: Loss = 0.146 Acc = 1.00\n",
      "Epoch 226 / 300: Loss = 0.144 Acc = 1.00\n",
      "Epoch 227 / 300: Loss = 0.143 Acc = 1.00\n",
      "Epoch 228 / 300: Loss = 0.142 Acc = 1.00\n",
      "Epoch 229 / 300: Loss = 0.140 Acc = 1.00\n",
      "Epoch 230 / 300: Loss = 0.139 Acc = 1.00\n",
      "Epoch 231 / 300: Loss = 0.138 Acc = 1.00\n",
      "Epoch 232 / 300: Loss = 0.136 Acc = 1.00\n",
      "Epoch 233 / 300: Loss = 0.135 Acc = 1.00\n",
      "Epoch 234 / 300: Loss = 0.134 Acc = 1.00\n",
      "Epoch 235 / 300: Loss = 0.133 Acc = 1.00\n",
      "Epoch 236 / 300: Loss = 0.131 Acc = 1.00\n",
      "Epoch 237 / 300: Loss = 0.130 Acc = 1.00\n",
      "Epoch 238 / 300: Loss = 0.129 Acc = 1.00\n",
      "Epoch 239 / 300: Loss = 0.128 Acc = 1.00\n",
      "Epoch 240 / 300: Loss = 0.127 Acc = 1.00\n",
      "Epoch 241 / 300: Loss = 0.125 Acc = 1.00\n",
      "Epoch 242 / 300: Loss = 0.124 Acc = 1.00\n",
      "Epoch 243 / 300: Loss = 0.123 Acc = 1.00\n",
      "Epoch 244 / 300: Loss = 0.122 Acc = 1.00\n",
      "Epoch 245 / 300: Loss = 0.121 Acc = 1.00\n",
      "Epoch 246 / 300: Loss = 0.120 Acc = 1.00\n",
      "Epoch 247 / 300: Loss = 0.119 Acc = 1.00\n",
      "Epoch 248 / 300: Loss = 0.118 Acc = 1.00\n",
      "Epoch 249 / 300: Loss = 0.117 Acc = 1.00\n",
      "Epoch 250 / 300: Loss = 0.116 Acc = 1.00\n",
      "Epoch 251 / 300: Loss = 0.115 Acc = 1.00\n",
      "Epoch 252 / 300: Loss = 0.114 Acc = 1.00\n",
      "Epoch 253 / 300: Loss = 0.113 Acc = 1.00\n",
      "Epoch 254 / 300: Loss = 0.112 Acc = 1.00\n",
      "Epoch 255 / 300: Loss = 0.111 Acc = 1.00\n",
      "Epoch 256 / 300: Loss = 0.110 Acc = 1.00\n",
      "Epoch 257 / 300: Loss = 0.109 Acc = 1.00\n",
      "Epoch 258 / 300: Loss = 0.108 Acc = 1.00\n",
      "Epoch 259 / 300: Loss = 0.107 Acc = 1.00\n",
      "Epoch 260 / 300: Loss = 0.106 Acc = 1.00\n",
      "Epoch 261 / 300: Loss = 0.105 Acc = 1.00\n",
      "Epoch 262 / 300: Loss = 0.104 Acc = 1.00\n",
      "Epoch 263 / 300: Loss = 0.104 Acc = 1.00\n",
      "Epoch 264 / 300: Loss = 0.103 Acc = 1.00\n",
      "Epoch 265 / 300: Loss = 0.102 Acc = 1.00\n",
      "Epoch 266 / 300: Loss = 0.101 Acc = 1.00\n",
      "Epoch 267 / 300: Loss = 0.100 Acc = 1.00\n",
      "Epoch 268 / 300: Loss = 0.099 Acc = 1.00\n",
      "Epoch 269 / 300: Loss = 0.099 Acc = 1.00\n",
      "Epoch 270 / 300: Loss = 0.098 Acc = 1.00\n",
      "Epoch 271 / 300: Loss = 0.097 Acc = 1.00\n",
      "Epoch 272 / 300: Loss = 0.096 Acc = 1.00\n",
      "Epoch 273 / 300: Loss = 0.096 Acc = 1.00\n",
      "Epoch 274 / 300: Loss = 0.095 Acc = 1.00\n",
      "Epoch 275 / 300: Loss = 0.094 Acc = 1.00\n",
      "Epoch 276 / 300: Loss = 0.093 Acc = 1.00\n",
      "Epoch 277 / 300: Loss = 0.093 Acc = 1.00\n",
      "Epoch 278 / 300: Loss = 0.092 Acc = 1.00\n",
      "Epoch 279 / 300: Loss = 0.091 Acc = 1.00\n",
      "Epoch 280 / 300: Loss = 0.090 Acc = 1.00\n",
      "Epoch 281 / 300: Loss = 0.090 Acc = 1.00\n",
      "Epoch 282 / 300: Loss = 0.089 Acc = 1.00\n",
      "Epoch 283 / 300: Loss = 0.088 Acc = 1.00\n",
      "Epoch 284 / 300: Loss = 0.088 Acc = 1.00\n",
      "Epoch 285 / 300: Loss = 0.087 Acc = 1.00\n",
      "Epoch 286 / 300: Loss = 0.086 Acc = 1.00\n",
      "Epoch 287 / 300: Loss = 0.086 Acc = 1.00\n",
      "Epoch 288 / 300: Loss = 0.085 Acc = 1.00\n",
      "Epoch 289 / 300: Loss = 0.084 Acc = 1.00\n",
      "Epoch 290 / 300: Loss = 0.084 Acc = 1.00\n",
      "Epoch 291 / 300: Loss = 0.083 Acc = 1.00\n",
      "Epoch 292 / 300: Loss = 0.083 Acc = 1.00\n",
      "Epoch 293 / 300: Loss = 0.082 Acc = 1.00\n",
      "Epoch 294 / 300: Loss = 0.081 Acc = 1.00\n",
      "Epoch 295 / 300: Loss = 0.081 Acc = 1.00\n",
      "Epoch 296 / 300: Loss = 0.080 Acc = 1.00\n",
      "Epoch 297 / 300: Loss = 0.080 Acc = 1.00\n",
      "Epoch 298 / 300: Loss = 0.079 Acc = 1.00\n",
      "Epoch 299 / 300: Loss = 0.078 Acc = 1.00\n",
      "Epoch 300 / 300: Loss = 0.078 Acc = 1.00\n",
      "Sentence:  ['The', 'dog', 'ate', 'the', 'apple']\n",
      "Labels:    ['DET', 'NN', 'V', 'DET', 'NN']\n",
      "Predicted: ['DET', 'NN', 'V', 'DET', 'NN']\n"
     ]
    },
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 2 Axes>",
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAacAAAEWCAYAAADCeVhIAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/MnkTPAAAACXBIWXMAAAsTAAALEwEAmpwYAABCBUlEQVR4nO3dd3xUZfb48c9JIyEkoYTeQkcg0kFFBWQVLGBXUFFUZC3oWtavuu4iuq67iv5U1ooNKyD2hq4iCIgoqKB0Q03ooaQQCCnn98e9iUNMmYRMZiZz3q/XvLj9njsT5szz3Oc+j6gqxhhjTCAJ83cAxhhjTEmWnIwxxgQcS07GGGMCjiUnY4wxAceSkzHGmIBjyckYY0zAseRkapyInC8iqSKSLSK9/R1PRURkvoiM93ccgU5E5ojIVdW9rQlNlpxCgIhsFpFDbjLYJSLTRaReFY9VHV/UjwITVbWeqv5cyjmuFZG1IpLlxvuZiMQd4zl9TkQud9/jbPf9LvSYz/Z3fKXxjM+N95DH/OWVOZaqnqmqr1b3tiY0WXIKHSNVtR7QB+gH/L0yO4ujuv5e2gKryjjPYOAhYIyqxgHHAbOq6bw+papvugm3HnAmsL1o3l3mV6V9hiXi24r7d+K+3vTYN6Km4zWhzZJTiFHVbcAcoIeINBCRT0Rkj4jsd6dbFW3rlpL+JSLfAjnA68ApwFPuL+unSjuHiISJyN9FZIuI7BaR10QkQUTquCWIcGCFiGwoZff+wHdFJSpV3aeqr6pqlnvs6SLynIh86ZasvhGRth7n7uqu2yci60TkEo91dUTkURHZ6pbInhORGI/154rIchHJFJENIjLCI662IvKte87/iUhiZd53EbnbPWaWiKwWkfM91oWLyGMiki4im0RkoohoUUIQkXYissDd9ysReVpE3vDY/wQRWSwiB0RkhYgM8VhX8jNs72W8Q0QkTUTuEpGdwCte/r2Md6fHicgi9/3e717XmVXcttzrN7WTJacQIyKtgbOAn3E+/1dwSjJtgENAyYQzFpgAxAHjgIX8XiU3sYzTjHNfQ3G+DOsBT6lqrkcJoqeqdihl3++B4SJyv4gMEpE6pWxzOfBPIBFYDrzpXlss8CXwFtAEGA08IyLd3P3+A3QGegEdgZbAJHffAcBrwJ1AfeBUYLPHOS8DrnaPGwX8tYxrL8sGnMSeANwPvCEizd111+GUtHrhlGzPK7HvW8APQCNgMs5nght3S+BT4EGgoRvXuyLS2GN/z89wSyVibuYes627vzd/L54GAutwPqdHgJdERKqwbZnXb2oxVbVXLX/hfMlmAwdwvpyeAWJK2a4XsN9jfj7wQIlt5gPjKzjfXOBGj/kuQB4Q4c4r0LGc/c8EPnbjzQb+HxDurpsOzPTYth5QALQGLgUWljjW88B9gAAHgQ4e604ENnls93gZ8cwH/u4xfyPweQXvwRAgrZz1y4Fz3emvgT97rPuT+x5F4CSBfKCux/o3gDfc6buA10sc+wvgqrI+wwr+Tv7kEf8RILqc7Uv7exnvTo8DUjzW1XWvqVlltq3o+u1Ve19Wcgod56lqfVVtq6o3quohEakrIs+71W+ZwAKgvoiEe+yXWt5BReRv8vsN9OfcxS04+hf6Fpwv2qal7O95Q74NgKrOUdWROL/az8X58vJshFEck6pmA/vcc7YFBrrVWwdE5ABOKasZ0BjnS+9Hj3Wfu8vBSW6lVTMW2ekxnYOTFL0mIle6VYZF5+6BU0rAjd3zffacbgHsU9WcMta3BS4ucc0nA83L2L4y9qjqYY9r8ObvxVPxe+YRf1nvW1nbVnT9ppaym5yh7Q6cUs1AVd0pIr1wqvs8q15Kdlt/1LyqPoTTgMHTdpwvzSJFv353lQxAy2kooKqFwFwR+Rrny7xI66IJcVodNnTPmQp8o6qnlzyWOA0BDgHd1bnvVlIqUFo14zET557YC8AwnPtpBSKynN/f5x1AK49dWntM7wAaikhdjy9oz/WpOCWn68oJoapDD5Tcz5u/l+pW0fWbWspKTqEtDucL+4CINMSp/qrILiq+qT4DuM29kV0PJ3nNUtX8ig4uTqOE0e7Nd3HvBQ0GlnhsdpaInCwiUTj3npaoairwCdBZRMaKSKT76i8ix7mJ7gXgcRFp4p6rpYgMd4/5EnC1iAwTp0FHSxHp6sX74Y1YnC/6Pe55r+boZPs28Bf3nPVxquoAUNUtwDJgsohEiciJwEiPfd8ARorIcHEaVkS7jRk8k111qcrfyzHx4vpNLWXJKbQ9AcQA6Thf/p97sc+TwEVuq6qpZWzzMk7LvgXAJuAwcLOXMe3HaSDwG5CJ8+U7RT2aNePcIL8PpzqvL3AFgDot+s7AaQixHaeq6GGgqFHFXUAKsMStlvoKpySAqv6A0+DhcSAD+IajS39VpqqrgceA73CSezLwrccmLwD/A37BKYl8hlPSLHDXX45zf2wvTsOHWUCue+xUnKrPv+Ekv1ScRh2++L/9BJX/e6kOZV6/qb1E1QYbNMFDRKbjNDSo1HNawcRtRv2cqpaaHEVkFrBWVX1ecglEoX79ocJKTsb4mYjEiMhZIhLhNg2/D3jfY31/EengVjeOwCkpfeCncGtcqF9/qLIGEcb4n+A8+zQL557Op7jPX7maAe/hPOeTBtygpXT7VIuF+vWHJKvWM8YYE3CsWs8YY0zACbpqvbCwMI2Jial4Q2OMMcVycnJUVYOmQBJ0ySkmJoaDBw/6OwxjjAkqInLI3zFURtBkUWOMMaHDkpMxxpiAY8nJGGNMwAm6e07GlJSXl0daWhqHDx+ueGNjarno6GhatWpFZGSkv0M5JpacTNBLS0sjLi6OpKQkyh7LzpjaT1XZu3cvaWlptGvXzt/hHBOr1jNB7/DhwzRq1MgSkwl5IkKjRo1qRS2CJSdTK1hiMsZRW/4vhExyWrszk39/tobs3AqHFDLGGONnPktOIvKyiOwWkZVlrL9cRH4RkV9FZLGI9PRVLABp+w7x/IKNrNuZ6cvTmBBVr97RA/pOnz6diRMnlrvPkCFDWLZsWYXHXrZsGbfcckup65KSkkhPT//D8oyMDK688ko6duxIhw4duPzyy9m/f3+F56qs5cuX89lnn1X7cQHmz5/POeec84fln3zyCb1796Znz55069aN559/nn/961/06tWLXr16ER4eXjw9depUJk+ejIiQkpJSfIwnnngCESnz/U9PTycyMpLnnnvOJ9dmKubLktN0YEQ56zcBg1U1GWc002k+jIXjWsQDsHpHli9PY0y1ys/Pp1+/fkydWta4jqW79tprad++PSkpKWzYsIGOHTsybty4ao/Pl8mpNHl5eUyYMIGPP/6YFStW8PPPPzNkyBDuvfdeli9fzvLly4mJiSmeLkrqycnJzJw5s/g4s2fPpnv37mWeZ/bs2ZxwwgnMmDHDp9eTn281OWXxWXJS1QU4I5WWtX6xqhb9lFsC+GJY6WItEqKJj45g9XYrOZmak5WVRbt27cjLywMgMzPzqPnXX3+dXr160aNHD3744QcAJk+ezNixYxk0aBBjx449qgSxd+9ezjjjDLp378748eMpbVSBlJQUfvzxR/7xj38UL5s0aRIrVqxg3bp1fyiRTJw4kenTpwPwwAMP0L9/f3r06MGECROKjz9kyBDuuusuBgwYQOfOnVm4cCFHjhxh0qRJzJo1i169ejFr1iwmT57Mo48+WnzsHj16sHnzZjZv3kzXrl0ZN24cnTt35vLLL+err75i0KBBdOrUqfjavXk/8/PzadSoEQB16tShS5cuFe533nnn8eGHHwKwYcMGEhISSExMLHP7GTNm8Nhjj7Ft2zbS0tKKl7/22mscf/zx9OzZk7FjxwKwa9cuzj//fHr27EnPnj1ZvHgxmzdvpkePHsX7Pfroo0yePLn4vbz11lvp168fTz75JB9//DEDBw6kd+/e/OlPf2LXrl0AZGdnc/XVV5OcnMzxxx/Pu+++y8svv8ytt95afNwXXniB2267zav3riIV1XZ5bNdfRPJF5KJqOXEZAqUp+bXAnLJWisgEYAJAVFRUlU4gInRrEc+aHZacarOdDz1E7pq11XrMOsd1pdnf/lbuNocOHaJXr17F8/v27WPUqFHExcUxZMgQPv30U8477zxmzpzJBRdcUPwMSk5ODsuXL2fBggVcc801rFzpfC+sXr2aRYsWERMTw/z584uPe//993PyySczadIkPv30U1566aU/xLJ69eri6q0i4eHh9O7dmzVr1lC/fv0yr2PixIlMmuQMJTV27Fg++eQTRo4cCTi/8n/44Qc+++wz7r//fr766iseeOABli1bxlNPPQVQ/AVcmpSUFGbPns3LL79M//79eeutt1i0aBEfffQRDz30EB988EF5bzEADRs2ZNSoUbRt25Zhw4ZxzjnnMGbMGMLCyv+dHR8fT+vWrVm5ciUffvghl156Ka+88kqp26amprJjxw4GDBjAJZdcwqxZs7jjjjtYtWoVDz74IIsXLyYxMZF9+5zf3rfccguDBw/m/fffp6CggOzs7AqrUI8cOVJcpbh//36WLFmCiPDiiy/yyCOP8Nhjj/HPf/6ThIQEfv311+LtIiMj+de//sWUKVOIjIzklVde4fnnn6/wffPSdOAp4LWyNhCRcOBh4H/VddKy+L1BhIgMxUlOd5W1japOU9V+qtovIqLq+fS45vGs25lFQaGNYWWql2dV0vLly3nggQeK140fP774i/CVV17h6quvLl43ZswYAE499VQyMzM5cOAAAKNGjaK03vcXLFjAFVdcAcDZZ59NgwYNqvU65s2bx8CBA0lOTubrr79m1apVxesuuOACAPr27cvmzZsrfex27dqRnJxMWFgY3bt3Z9iwYYgIycnJlTreiy++yNy5cxkwYACPPvoo11xzjVf7jR49mpkzZ/LBBx9w/vnnl7ndrFmzuOSSS4r3Kara+/rrr7n44ouLS1wNGzYsXn7DDTcAzo+AhISECmO59NJLi6fT0tIYPnw4ycnJTJkypfg9/+qrr7jpppuKt2vQoAH16tXjtNNO45NPPmHt2rXk5eWRnJzs1fVXpKLaLtfNwLvA7mo5aTn8WnISkeOBF4EzVXWvr893XPN4DuUVsGXvQdo3rlfxDiboVFTC8YdBgwaxefNm5s+fT0FBwVHVPSWb/RbNx8bGVvl83bp1Y/ny5RQWFhaXKAoLC1mxYgV9+vRh69atFBYWFm9f9EzM4cOHufHGG1m2bBmtW7dm8uTJRz0vU6dOHcD5Ai7rXklERESpx/bcHyAsLKx4PiwsrNL3XpKTk0lOTmbs2LG0a9euuFqyPOeccw533nkn/fr1Iz4+vsztZsyYwc6dO3nzzTcB2L59O7/99lul4ivvfYCjP9+bb76Z22+/nVGjRjF//vxyS5/g/Nh56KGH6Nq161E/dLwJS0Q8W4BMU1Wv7/WLSEvgfGAo0L8yJ64Kv5WcRKQNztDLY1V1fU2cs1vzokYRVrVnataVV17JZZdd9ocvk1mzZgGwaNEiEhISKvzVfeqpp/LWW28BMGfOnFKrjzp27Ejv3r158MEHi5c9+OCDDBs2jDZt2tC2bVtWr15Nbm4uBw4cYO7cucDvX6CJiYlkZ2fzzjvvVHhdcXFxZGX93sgoKSmJn376CYCffvqJTZs2VXiMysjOzj6qmnP58uW0bdvWq33r1q3Lww8/zL333lvmNuvXryc7O5tt27YV3yu75557mDFjBqeddhqzZ89m717nd3RRtd6wYcN49tlnASgoKCAjI4OmTZuye/du9u7dS25uLp988kmZ58zIyKBly5YAvPrqq8XLTz/9dJ5++uni+aLPeuDAgaSmpvLWW28Vl7y9lF9UA+W+KtsI7QngLlUtrGjD6uDLpuQzgO+ALiKSJiLXisj1InK9u8kkoBHwjIgsL5HRfaJT03pEhInddzI1rqgpd8kvk+joaHr37s31119f6v2jku677z4WLFhA9+7dee+992jTpk2p27388sv89ttvdOjQgcaNG7NkyZLiZtGtW7fmkksuoUePHlxyySX07t0bgPr163PdddfRo0cPhg8fTv/+Ff84Hjp0aPE9rlmzZnHhhReyb98+unfvzlNPPUXnzp0rPEZ55s6dS6tWrYpfP//8M4888ghdunShV69e3HfffV6VmoqMHj2aPn36lLl+xowZf6jyu/DCC5kxYwbdu3fn3nvvZfDgwfTs2ZPbb78dgCeffJJ58+aRnJxM3759Wb16NZGRkUyaNIkBAwZw+umn07Vr1zLPOXnyZC6++GL69u17VCONv//97+zfv58ePXrQs2dP5s2bV7zukksuYdCgQdVerVuBfsBMEdkMXITz3X2er04mpbX2CWSxsbF6LIMNDn98AS0bxPDyOJ+XSk0NWbNmDccdd5y/wyjXO++8w4cffsjrr79e4+det24dZ599NlOnTuWss86q8fOb6nfOOedw2223MWzYsFLXl/Z/QkRyVLXc+mIRSQI+UdUeFWw33d2u4uJ1FQVKa70ac1zzOL7fVNE9P2Oqz80338ycOXNq9HkgT126dDnqAVQTvA4cOMCAAQPo2bNnmYmpqtzariFAooikAfcBkQCqWuNPI4dccurWIp4Plm9n/8EjNIitWrN0Yyrjv//9r79DMLVE/fr1Wb/eN7foVdXrG1iqOs4nQXjwe1Pymta9hXPD+ddtGX6OxFSnYKueNsZXasv/hZBLTsmtEhCB5akH/B2KqSbR0dHs3bu31vynNKaqisZzio6O9ncoxyzkqvXioyPp1KQeP2+t/k4wjX+0atWKtLQ09uzZ4+9QjPG7opFwg13IJSeA3q0b8MXqnahqrRn7JJRFRkYG/aifxpijhVy1HkDvNvU5kJPHpvSqN0k3xhjjOyGZnPolOX1iLd1sTcqNMSYQhWRy6tA4lsR6dViy0ZKTMcYEopBMTiLCwPYNWbLRWngZY0wgCsnkBHBC+0bsyDjMlr05/g7FGGNMCSGbnE7q4IykuTAl3c+RGGOMKSlkk1P7xFhaN4zhm3U+HzPLGGNMJYVschIRhnRuwuINe8nNL/B3OMYYYzyEbHICGNKlMTlHCqzVnjHGBJiQTk6DOiZSNyqcz1fu9HcoxhhjPIR0coqODGdo1yZ8uXonBYXWpNwYYwJFSCcngDN7NCM9+wjfb9zr71CMMca4Qj45DevalLg6Ecz+Mc3foRhjjHGFfHKKiQpnVK8WfPbrDjIO5fk7HGOMMVhyAuDS/q3JzS/koxXb/R2KMcYYLDkBkNwyga7N4nh7aaq/QzHGGIMlJ8B5IHd0/9b8ui2Dldsy/B2OMcbUOBF5WUR2i8jKMtZfLiK/iMivIrJYRHr6Mh5LTq7ze7ciNiqcFxZu9HcoxhjjD9OBEeWs3wQMVtVk4J/ANF8GY8nJlVA3kstPaMvHK7az1XoqN8aEGFVdAJTZXY6qLlbV/e7sEqCVL+Ox5OTh2pPbEREWxnMLNvg7FGOMqW4RIrLM4zXhGI51LTCnugIrjc+Skxf1lyIiU0Ukxa3H7OOrWLzVND6aC/u24p1laezKPOzvcIwxpjrlq2o/j1eVquVEZChOcrqresM7mi9LTtMpv/7yTKCT+5oAPOvDWLx2/eD2FKoyde5v/g7FGGMCiogcD7wInKuqPu1Wx2fJqaL6S+Bc4DV1LAHqi0hzX8XjrbaNYrlsYBtmLk0lZXe2v8MxxpiAICJtgPeAsaq63tfn8+c9p5aA54NFae6yPxCRCUX1pPn5+T4P7C/DOhETGc5/5qz1+bmMMSYQiMgM4Dugi4ikici1InK9iFzvbjIJaAQ8IyLLRWSZL+OJ8OXBq4tbNzoNIDY21ufdhzeqV4cbhnRgyhfrWLwhnZM6JPr6lMYY41eqOqaC9eOB8TUUjl9LTtuA1h7zrdxlAeHak9vRsn4MD3y8mvyCQn+HY4wxIcWfyekj4Eq31d4JQIaq7vBjPEeJjgznH+ccx9qdWbyxZIu/wzHGmJDiy6bkFdVffgZsBFKAF4AbfRVLVQ3v3oxTOiXy2JfrSc/O9Xc4xhgTMkQ1uEaAjY2N1YMHD9bY+VJ2ZzPiiQVc2KcVD190fI2d1xhjqpOI5KhqrL/j8Jb1EFGBjk3qcfWgJN7+MZXlqQf8HY4xxoQES05euGVYJxLr1eG+D1dSWBhcJU1jjAlGlpy8EBcdyT1ndmVFWgbv2HDuxhjjc5acvHR+75b0a9uAhz9fa8O5G2OMj1ly8pKIMHlUd/blHOHxL33ec4cxxoQ0S06V0KNlApcPbMPrS7awdmemv8Mxxphay5JTJd1xehfioiO478NVBFszfGOMCRaWnCqpQWwUdw7vwveb9vHprwHToYUxxtQqlpyqYHT/NnRqUo+n522w0pMxxviAJacqCA8Trju1PWt2ZPJtik/H2zLGmJBkyamKzu3VgsZxdXh+wQZ/h2KMMbWOJacqqhMRzriTklj4WzprdljLPWOMqU6WnI7BFQPbUjcqnBcXbvJ3KMYYU6tYcjoGCXUjubBPKz5esd2G1DDGmGpkyekYXXVSEkcKCpnx/VZ/h2KMMbWGJadj1LFJPU7t3JjXl2whz4ZzN8aYamHJqRpcfVISu7Ny+cweyjXGBCkReVlEdovIyjLWi4hMFZEUEflFRPr4Mh5LTtVgcOfGtG8cy3PfbLSHco0xwWo6MKKc9WcCndzXBOBZXwYT4cuDh4qwMOHGIR356+wVzF2zmz91a+rvkEJG3rZtbBl3NYU5Of4OxRifa3jF5STecINPjq2qC0QkqZxNzgVeU+cX+BIRqS8izVXVJ1VGlpyqybm9WvDk3PX8d14Kw45rgoj4O6SQcHj9evJSU4kbPpzwBvX9HY4xPhXVseOx7B4hIss85qep6rRK7N8SSPWYT3OXWXIKZJHhYdw4pCP3vPcrC39L59TOjf0dUkgozMoCoMlttxKVlOTfYIwJbPmq2s/fQXjL7jlVowv7tKJl/Rge/d86Cgvt3lNNKMh0klNYfLyfIzGm1tsGtPaYb+Uu8wlLTtUoKiKMO87ozC9pGXz8y3Z/hxMSCjIzAAiPi/NzJMbUeh8BV7qt9k4AMnx1vwksOVW783q1pEfLeB75fB2H8wr8HU6tV5iZhdSti0RG+jsUY4KaiMwAvgO6iEiaiFwrIteLyPXuJp8BG4EU4AXgRl/GY/ecqllYmHDvWd0Y88ISXlq0iZuGHtMNTFOBgqxMKzUZUw1UdUwF6xW4qYbC8W3JSURGiMg696Gtu0tZ30ZE5onIz+5DXWf5Mp6acmKHRgzv3pT/fv0bqfusibMvFWZmER5vycmY2sZnyUlEwoGncR7c6gaMEZFuJTb7O/C2qvYGRgPP+CqemnbfyO6EiTDpw5X2YK4PFWRlERZnjSGMqW18WXIaAKSo6kZVPQLMxHmIy5MCRd8sCUCtaUXQon4Mt5/emXnr9jBn5U5/h1NrFWZatZ4xtZEvk1NZD2x5mgxcISJpODfbbvZhPDVu3ElJdGsez/0fryLrcJ6/w6mVCjIzrRm5MbWQv1vrjQGmq2or4CzgdRH5Q0wiMkFElonIsvz8/BoPsqoiwsN46IJkdmflMuWLdf4Op1YqyMoi3JKTMbWOL5OTNw9sXQu8DaCq3wHRQGLJA6nqNFXtp6r9IiKCq4Fhr9b1uerEJF77bguLN6T7O5xaRQsLKczKIswaRBhT6/gyOS0FOolIOxGJwmnw8FGJbbYCwwBE5Dic5LTHhzH5xV0jutIuMZY7Z/9i1XvVqDAnBwoLCbcGEcbUOj5LTqqaD0wEvgDW4LTKWyUiD4jIKHezO4DrRGQFMAMYp7WwaVtMVDiPXtyTHRmH+Nena/wdTq1RmJkJYE3JjamFfFpHpqqf4TR08Fw2yWN6NTDIlzEEir5tG/DnwR14dv4GhndvxtCuTfwdUtArcDt9tabkxtQ+wXUDJ8jd+qdOzFu7m/979xc+u+UUGsfV8XdIQeHw+vXs+Mc/0Lyjq0Q15xBgJSdjaiN/t9YLKXUiwnlydG8yD+Vxx+wV1nO5l3KWLePwil+IaNiIyKbNil9R7doRf845RCcn+ztEY0wJIjKytNbX3rKSUw3r0iyOSSO7ce/7K3lh4Ub+PLiDv0MKeIXusBitnnmasKgoP0djjPHSpcATIvIu8LKqrq3MzlZy8oPLBrThzB7NmPLFOpanHvB3OAGvIDMTiY62xGRMEFHVK4DewAZguoh85z6z6lU9vCUnPxAR/nPB8TSNj+bmGT+Rac3Ly1VoPY8bE5RUNRN4B6f7uubA+cBPIlJhb0CWnPwkoW4kU8f0ZvuBw9zz7q/WOWw5CjKzrIsiY4KMiIwSkfeB+UAkMEBVzwR64jxGVC5LTn7Ut20D/npGFz79dQcvLdrk73AClpWcjAlKFwKPq2qyqk5R1d0AqpqD0ztQuSw5+dn1g9szvHtT/j1nLd9t2OvvcAKSU3Ky5GRMkJkM/FA0IyIxIpIEoKpzK9rZkpOfiQiPXtyTpEZ1mfjWT+zIOOTvkAKOM9qtVesZE2RmA4Ue8wXuMq9YcgoAcdGRPD+2H7n5hVz/xk/k5hf4O6SAUpiRSXiCJSdjgkyEO5YfAO60101uLTkFiI5N6vHoxT1ZkXqA+z5cZQ0kXKpqo90aE5z2ePSjioicC3g9NIMlpwAyokczJg7tyMylqdZAwqU5OVBQYF0UGVMDRGSEiKwTkRQRubuU9W1EZJ6I/Cwiv4jIWeUc7nrgbyKyVURSgbuAP3sbi/UQEWBuP70zKbuz+ddna0hqFMufujX1d0h+9XvnrpacjPElEQkHngZOxxm5fKmIfOR20F3k7zgjTDwrIt1wOvZOKu14qroBOEFE6rnz2ZWJx6uSk4jEFvWRJCKd3fbrkZU5kfFOWJjw+KW96NEigVtm/syq7Rn+DsmvCoqHxbBqPWN8bACQoqob3ftDM4FzS2yjQNF/xgRge3kHFJGzgRuB20VkkohMKm97T95W6y0AokWkJfA/YCww3duTmMqJiQrnxav6kRATyfhXl7E787C/Q/KbQis5GVNdIkRkmcdrQon1LYFUj/k0d5mnycAVIpKGU2oqs6cHEXkOp3+9mwEBLgbaehust8lJ3AenLgCeUdWLge7ensRUXtP4aF68qh8Zh/K49tVlHMzN93dIflGQUVRySvBzJMYEvXxV7efxmlaFY4wBpqtqK+As4PVyeh4/SVWvBPar6v3AiUBnb0/k7T0nEZETgcv5/cnecG9PEgwKjxxh9yNTSLzpRiIaNKjSMdKnvcChn3+utpjigTezc1mx4ABffR5Fr9b1CROptuMHg/xduwAbs8mYGrANaO0x38pd5ulaYASAqn4nItFAIrC7lOMVVfnkiEgLYC9O/3pe8TY53QrcA7zvDrXeHpjn7UmCQe669ex/4w3q9ulN/FnlNUAp294XX0QiIohoVn2NGBKA4+vksS09gy0HD9CqQYxTQA4hsaeeQmRzr/+mjTFVsxToJCLtcJLSaOCyEttsBYbh9DJ+HBAN7CnjeB+LSH1gCvATzv2qF7wNxqvkpKrfAN8AuEW4dFW9xduTBAM9kgs4XeVUaf/CQgqzski84Xoa31L9b81381KY8sU6rh6UxKRzuiEhVoIyxviWquaLyETgC5yasZfdwsgDwDJV/Qinw9YXROQ2nGQzTkt5KNPNE3NV9QDwroh8AkSrqtctvLxKTiLyFk6b9QKc7BovIk+q6hRvTxToCg87JdCCrMyq7X/wIKj67GHRG4d0ID07l1e+3UxivTrcNLSjT85jjAldqvoZTkMHz2WTPKZXA4O8OE6hiDyNM54TqpoL5FYmFm8bRHRzx+U4D5gDtMNpsVdraK7Ty0ZhFUtOhcVNnn1zb0RE+MfZ3Ti3VwumfLGOVxdv9sl5jDGmmswVkQulitU83t5zinSfazoPeEpV80SkVvWv83u1XtVKTkX7+bLJc1iY00lszpEC7vtoFZHhYVw2sI3PzmeMMcfgz8DtQL6IHMa5W66q6lX1krclp+eBzUAssEBE2gJV+xYPUJrrJKfCKlbrFd2r8nWT58jwMJ66rDdDuzTmb+//ytvLUiveyRhjapiqxqlqmKpGqWq8O+/1fQ9vG0RMBaZ6LNoiIkMrG2wgK8w9tgYRRUmtJpo814kI59kr+nLda8u4691fiAwXzu/dyufnNcYYb4nIqaUtV9UF3uzvbYOIBOA+oOhk3wAPALWmb52ie05VbRBRlNRqajjx6Mhwpo3txzXTl3LH2ysoLIQL+1qCMsYEjDs9pqNxukf6ETjNm529rdZ7GcgCLnFfmcAr3scY+IruOVW5QURRyakGu9mJiQrnpXH9OLFDI+6YvYI3lmypsXMbY0x5VHWkx+t0oAew39v9vU1OHVT1PrdDwI1uVxTtK9qpou7X3W0uEZHVIrLKbbLuF79X61Wx5JSRCSI13gdc3agIXrqqP6d1bcLfP1jJiws31uj5jTHGS2nAcd5u7G1rvUMicrKqLgIQkUFAueOJe9P9uoh0wul5YpCq7heRJt4GXt1+b0qeiapW+iHXgqwswurVQ8Jqfois6MhwnruiL7fNWs6Dn67hYG4BtwzraA/qGmP8RkT+i/OgLjgFoV44PUV4xdvkdD3wmnvvCZyi2VUV7FPc/bobaFH3655jg1wHPK2q+wFUtbT+mWqEug/hal4empuLREdXav/CzMwardIrKSoijCdH9yI6MpzHv1rP/pwj/OOcboSHWYIyxvjFMo/pfGCGqn7r7c7ettZbAfQUkXh3PlNEbgV+KWe30rpfH1him84AIvItTncZk1X185IHcrt2nwAQFeX1EPSVUnjk94eXCzIzCatkcirIyqqxxhBliQgPY8pFx9MwNpIXFm5iV+ZhHr/USVjGGFPD3gEOq2oBOLVpIlLXHeGiQpWqg1LVTLenCHAerjpWEUAnYAhOV+wvuB0FljzvtKJu3iMifDN4b1G1Hvze20NlFGRm+LXkVCQsTLj37G78/ezjmLNyJ1e+9AMZOXn+DssYE3rmAjEe8zHAV97ufCw3SCqqL/Km+/U04CNVzVPVTcB6nGRV44oewoWqPetUmJlFWELgjNY6/pT2/HdMb5anHuCi5xaTus+rHyvGGFNdoj2HZnen63q787EUQyrqvsib7tc/wCkxvSIiiTjVfH5pbqYe1Xp7n3+ejObNKrV/3rZtRHcPrPEXR/ZsQWK9Ovz59WWc+/S3PHdFXwa0a+jvsIwxoeGgiPRR1Z8ARKQvFTSk81RuchKRLEpPQsLRxbU/8LL79S+AM0RkNU6P53eq6l5vg69Ohbm5RLZpg4SHc2jlSli5slL7S0wMdfv391F0VXdih0Z8cNMgxr+6jMtfXMKD5/Xg0v7WH58xxuduBWaLyHacnNEMZ9h2r0gpQ3EEtNjYWD148GC1H3fLVePQ/HyS3nyj2o8dCDJy8pg44ycW/pbOtSe3454zuxIRXvPN3o0x/iEiOaoaW8PnjAS6uLPrVNXrG+D27eTS3FzC6tTxdxg+k1A3klfG9WfcSUm8tGgTY1/6gd1Zhyve0RhjqkBEbgJiVXWlqq4E6onIjd7ub8nJVXgkF6nFyQmcpuaTR3Xn0Yt78nPqfs6euojvN/qlFtUYU/td546EC4D7POt13u5sycmluUdqfXIqclHfVnxw0yDi6kRw2Yvf89w3GygsDK7qXWNMwAv3HGjQ7TXI6wdVLTm59PBhwur45gHfQNS1WTwfThzEiB7N+M+ctUx4fRkHco5UvKMxxnjnc2CWiAwTkWHADJyR1L1iyclVeCQXiQqNklORuOhInhrTm8kju/HN+j2MeGIhi35L93dYxpja4S7ga5zu764HfqWCVt6eLDm5Qqlaz5OIMG5QO967YRCxdcK54qXveeDj1RzOK/B3aMaYGladI0moaiHwPc4o6gNwxnFa420svukLKAhpbi4SQtV6JSW3SuCTm0/hP3PW8PK3m1iUsocnLu1NtxaB0+uFMcZ3qmskCRHpjNO5whggHZgFoKqVGj3dSk6Aqtb6puTeiIkK5/5zezD96v7sz8njvKe/5el5KeQVFPo7NGOM7xWPJKGqR4CikSQ8eTOSxFqcUtI5qnqyqv4Xp5OFSrHkhDNMBhBy95zKMqRLE7649VT+1K0JU75Yx7lPfcvKbRn+DssYc2wiRGSZx2tCifWljSTRssQ2nYHOIvKtiCwRkRGlnOcCYAcwT0RecBtDVHrsHktO/N7pq0RbcirSMDaKZy7vy3NX9GVPdi7nPv0tD3++1u5FGRO88otGd3Bf06pwjApHklDVD1R1NNAVmIfTjVETEXlWRM7w9kSWnPg9OYV6tV5pRvRoxle3DeaiPq14dv4GznpyIUvswV1jaqNqHUlCVQ+q6luqOtI91s84Lfi8YskJj5KTVeuVKqFuJA9fdDxvXDuQvMJCRk9bwq0zf2Z3pnV/ZEwtUjyShIhE4Ywk8VGJbT7AKTVRmZEkVHW/Oy7fMG+DseSE0yM5EJJNySvj5E6JfHnbYG4Z1onPVu7ktMe+4aVFm8i3BhPGBD1VzQeKRpJYA7xdNJKEiIxyN/sC2OuOJDEPH44kYb2SA4fXrGHT+RfQ8r9TiT/99Go9dm21Of0gkz9exfx1e+jaLI4Hzu1hY0UZE8D80Sv5sbCSE1B4yKmeCouO9nMkwSMpMZZXxvXn+bF9yTqczyXPf8dfZv7MtgNejyVmjDFlsodwgcKsTADC4+2B08oQEYZ3b8apnRrzzPwUpi3YyJyVO7n25HbcOKQDcdGR/g7RGBOkrOQEFGRmARAWZ8mpKmKiwrnjjC58/dchnJ3cnGfnb2DIlPm8vmSL3Y8yxlSJJSegoLjkFOfnSIJby/oxPH5pLz6aOIgOTerxjw9WMuLJhXy9dhfBdm/TGONflpyAwkwnOYVZtV61OL5VfWZNOIFpY/tSUKhcM30Zl73wPT9t3e/v0IwxQcKSE061nkRHExYVuh2/VjcR4YzuzfjfbacyeWQ31u/K4oJnFjP+1aWs2ZHp7/CMMQHOkhNOg4jwOKvS84XI8DDGDWrHgv8byp3Du/D9pn2cNXUhN8/4mY17sv0dnjEmQFlywik5WZWeb8XWieCmoR1Z9H+nceOQDny1ehenP76Au975xZqfG2P+wB7CBbZecw2FOYdImjmjWo9ryrYnK5en56Xw1vdbAbhsYBtuGNKBpvH2rJkxvhBsD+FacgI2XXgR4Y0a0mZaVTrpNcdi24FDTP3qN975KY3wMGF0/9bcMKQDzRO8Hs3ZGOOFYEtOVq0HFGRlEW7POPlFy/oxPHzR8cy7YwgX9G7JW99vZfAj87n3/V9J25/j7/CMMX5iyQmnKXl4giUnf2rTqC7/ufB45t85hIv6teLtZakMmTKfu9/9ha17LUkZE2p8mpxEZISIrBORFBG5u5ztLhQRFZF+voynNKpKQVaW9Q4RIFo1qMtD5yfzzZ1DuWxgG977aRtDH5vPnbNXsDm9eqtzjTGBy2fJSUTCgaeBM4FuwBgR6VbKdnHAX4DvfRVLeTQnBwoKrHeIANOifgwPnNuDBf83lCtPbMtHK7Zz2mPzuWXGz/aclDEhwJclpwFAiqpuVNUjwEzg3FK2+yfwMOCXkesKsor61bPkFIiaJURz38juLLxrKNed0p65a3Zx5pMLuWb6UpZu3ufv8IwxPuLL5NQSSPWYT3OXFRORPkBrVf20vAOJyAQRWSYiy/Lz86s1yIIM65E8GDSJi+aes45j8d3DuOP0zixPPcDFz33Hxc8tZt7a3dZ3nzG1jN8aRIhIGPD/gDsq2tYd3refqvaLiKjeUT5suIzgklA3kpuHdWLRXUO5b2Q3tu0/xNXTl3LW1EV8tGK79YJuTC3hy+S0DWjtMd/KXVYkDugBzBeRzcAJwEc13SjChssITnWjIrh6UDvm3zmUKRcdz5H8Am6Z8TPD/t83vPn9Fg7nFfg7RGPMMfBlcloKdBKRdiISBYwGPipaqaoZqpqoqkmqmgQsAUap6jIfxvQHhTZcRlCLigjj4n6t+fK2wTx3RV/qx0Ry7/srOfWReTz/zQayc6u3GtgYUzN8lpxUNR+YCHwBrAHeVtVVIvKAiIzy1Xkrq7jkZNV6QS0sTBjRoxkf3DSIN8cPpHPTOP49Zy0n/XsuU75Yy+4sv7S3McZUUch3X7TnmWdIn/pfuv76CxJpw4rXJitSD/Ds/A18sXonkeFhXNinJeNPaU+HxvX8HZoxNc6b7otEZATwJBAOvKiq/yljuwuBd4D+vqrtqt7WBUGoMCMTqVvXElMt1LN1fZ4b25dN6Qd5ceFGZv+YxsylqZx+XFP+PLg9fds29HeIxgQMj2dTT8dpXb1URD5S1dUltquRZ1NDvvuigqwsa6lXy7VLjOVf5yez+O7TuHloR37YvI8Ln/2Oi55dzP9W7aSwMLhqD4zxkYB6NjXkk5MNNBg6EuvV4fYzurD47tOYPLIbOzMPM+H1H/nT498w84et1sLP1HYRRc+Luq8JJdZX27Op1SHkq/VsoMHQUzcqgnGD2nHFCW35bOVOpi3YwN3v/cpjX65n3ElJXDGwLQl1rZrX1Dr5qlrlR3U8nk0dV20RlSPkS04FVnIKWRHhYYzq2YKPJ57Mm+MHclzzeKZ8sY6T/jOXf36y2kboNaEmoJ5NDfmSU2FmFmEdO/o7DONHIsKgjokM6pjI6u2ZvLBwI9MXb+bVxZsZ2bMF153Snm4trHRtar3iZ1NxktJo4LKilaqaASQWzYvIfOCvvmqtZyWnzEwbaNAU69Yinscv7cWC/xvKVScl8b9VOzlr6kLGvvQ936akWx9+ptYKtGdTQ/o5Jy0sZG33HiTecD2Nb7mlWo5papeMnDze+H4Lr3y7mfTsXLq3iGfCqe05K7k5keEh/9vOBBEbpj2IFB48CKrWr54pU0LdSG4a2pFFdw3lPxckcyivgL/MXM7gR+YxbcEGMg/n+TtEY2ql0E5OmdavnvFOdGQ4owe04avbBvPSVf1o2yiWhz5by4kPzeWBj1eTus+GkjemOoV0gwgbaNBUVliYMOy4pgw7rikrt2Xw4sKNvPbdZqYv3sSZPZpz7Snt6NOmgb/DNCbohXZysoEGzTHo0TKBJ0b35q4zu/Lq4i289f0WPv11B33a1Oe6U9pzRvdmhIeJv8M0JiiFdrWeDTRoqkHzhBjuPrMr390zjMkju5GefYQb3vyJIY/O4+VFm2zYDmOqIKSTkw2XYapTbB2n54l5fx3Cc1f0oWlcNA98spoT/z2Xf3+2hh0Z9lCvMd4K6Wq94pKT3XMy1Sg8TBjRozkjejTn5637eXHRJl5YuJGXFm3i7OObM/7k9iS3SvB3mMYEtJBOTsUlp3o2vo/xjd5tGvD0ZQ1I3ZfD9MWbmbU0lQ+Xb6dv2wZcdVISZ/ZoZs9LGVOKkH4Id+dDD5Hx3vt0Wba0Wo5nTEUyD+cxe1kar323mS17c2gSV4fLB7ZlzMDWNImL9nd4phYLtodwQzo5bb/rbg4u/YFOX39dLcczxluFhco36/cwffFmvlm/h8hw4ezk5lx1UhK9rSm68YFgS06hXa2XlUV4vNX9m5oXFiYM7dqEoV2bsHFPNq99t4V3fkzjg+Xb6dkqgatOSuLs45tTJyLc36Ea4xchXXLacsVYEKHt669Vy/GMORbZufm891Mary7ezIY9B0msF8WYAW0YM6ANLerH+Ds8E+SCreQU0slp47nnEdmqFa2ffqpajmdMdVBVFqWk8+rizcxduxsBhnRpwpgBbRjapTER1oDCVEGwJacQr9bLJNqakZsAIyKc0qkxp3RqTOq+HN5elsqspalc99oymsVHc0n/1lzavzUtrTRlarGQLjmt69efhAvOp9nf/lYtxzPGV/IKCvl67W5m/LCVb9bvAWBI58ZcNrCtlaaMV6zkFCS0oIDC7GxrEGGCQmR4GMO7N2N492ak7c9h1tLfS1NN4+twQZ9WXNinFR2b2DN7pnYI2ZJTwYEDrD/hRJr+7R4aXnllNURmTM3Kd0tTM5em8s36PRQUKr1a1+fCvq0YdXwLEupG+jtEE0Cs5BQkfh8uw/rVM8EpIjyMM7o344zuzdiddZgPf97OOz+m8Y8PVvLPj1dzeremXNi3Jad2smo/E3x8mpxEZATwJBAOvKiq/ymx/nZgPJAP7AGuUdUtvoilIDub/J07i+dzN2wEbKBBUzs0iYvmulPbM/6Udqzansk7P6bx4fJtfPrrDhLr1WFkz+aM6tmCXq3rI2LDeJjA57NqPREJB9YDpwNpwFJgjKqu9thmKPC9quaIyA3AEFW9tLzjVrVaL/Pzz9l2621/WJ40cwYxvXpV+njGBLoj+YXMW7ebd39MY/66PRwpKKRVgxhG9mzBqJ4t6NoszhJVCAm2aj1fJqcTgcmqOtydvwdAVf9dxva9gadUdVB5x61qcsrbsYNDy5cftSwsNpbYk09GwqzKw9RuGYfy+N+qnXz8yw6+TUmnoFDp2KQeI49vwciezWnf2BpS1HbeJKdAqu3yZXK6CBihquPd+bHAQFWdWMb2TwE7VfXBUtZNACYAREVF9c3NzfVJzMaEgvTsXOas3MnHK7bzw6Z9AHRrHu+0BuzRlC5NrURVG1WUnHxV21XleAMhOYnIFcBEYLCqlpt5qvM5J2NC3Y6MQ3z6yw7mrNzJT1v3owpJjeoy3G1o0bt1fcJsqPlawYvk5JParqryZYOIbUBrj/lW7rKjiMifgHvxIjEZY6pX84QYxp/SnvGntGd31mG+XL2Lz1fu5KVFm3h+wUaaxNXhjO5NOb1bMwa2a0h0pHVEG8QiRGSZx/w0VZ3mMd8SSPWYTwMGlnO8a4E51RjfUXyZnJYCnUSkHU5SGg1c5rmBm3mfxylh7fZhLMaYCjSJi+bygW25fGBbMnLy+HrdLr5YuYt3f9zGG0u2EhMZzqCOiQzt2pihXZpYZ7TBJ19V+1XHgdzarn7A4Oo4Xml8lpxUNV9EJgJf4Nxce1lVV4nIA8AyVf0ImALUA2a7ddxbVXWUr2IyxngnoW4k5/duxfm9W3HoSAHfbUxn3to9fL12N1+t2QVA12ZxzrAfXZrQp019e5Yq+AVUbVfI9hBhjKk8VSVldzbz1u3m67W7WbZ5P/mFSlx0BCe2b8TJnRI5qUMiHRrHWqOKAOPFPacInAYRw3CS0lLgMlVd5bFNb+AdnNqu33waryUnY0xVZR7O49vf0vlm/R4WpaSTtv8QAM3ioxnUMZGTOzViUIdEmsTbEPT+5mVT8rOAJ/i9tutfnrVdIvIVkAzscHfxWW2XJSdjTLXZujeHRSnpfJuSzrcb0jmQkwdApyb1GNCuIQPaNaR/UkO7X+UH9hCuj1lyMiY4FBYqq3dksiglncUb9vLTlv1k5+YD0LJ+DAPbNaS/m6ysGtD3LDn5mCUnY4JTfkEha3dm8cOmfSzdvI8fNu1j78EjADSKjaJfUgN6t2lAr9b1SW6ZQGydkO2X2icsOfmYJSdjagdVZWP6QZZu2scPm/exbPN+tu7LASBMoHPTOHq1ru+82tSnU5M4wu2B4Cqz5ORjlpyMqb32ZueyIu0Ay7ceYHlaBitSD5BxyLlvVTcqnOSWCfRomUD3FvF0axFPh8b1iLQm7F6x5ORjlpyMCR2qyqb0g0clrLU7MsnNLwQgKiKMLk3j6N4ivjhhdW0Wb1WCpbDk5GOWnIwJbfkFhWxMP8jq7Zms2p7B6h2ZrNqeWdwyUATaNYqlS7M4OjWNo3PTenRuGkdSo1iiIkK3lGXJyccsORljSlJVdmQcdhOWk7R+253Nlr0HKXS/4iLChKTEWDo3rUenJnF0dhNX2xBJWpacfMySkzHGW4fzCtiwJ5vfdmXz2+4s1u/K5rddWWzZl0PRV194mNCqQQztEmNJahRL+8bOv+0SY2lRP6bWNMKw5ORjlpyMMcfKM2lt2JPNpvSDbEo/yOb0gxw8UlC8XVR4GG0a1aVdopOs2jaqS+sGdWndsC4t6kdTJyJ4emm35ORjlpyMMb6iquzJyi1OVpv2HmTTnoNs3nuQzXtzOOI2xADn3lbTuGhaNYihdcO6zr8N6hbPN0uIDqiWhJacfMySkzHGHwoLlZ2Zh0ndl0Pa/kOk7nf/ded3ZBwqvr8FzrNazRNiaJ4QTbOEaPffo+cb16tTY725W3LyMUtOxphAlFdQyM4MJ3kVJa6ipLUz4zA7Mg4XN4EvEibOOFq/J69omsVH0zQ+miZxdWgSX4fGcdHER0ccc/dOlpx8zJKTMSYYqSoHcvLYmXm4OFntzDjk/JvpzO84cOioe15F6kSE0SS+DleekMR1p7av0vmDLTnZk2rGGFMDRIQGsVE0iI3iuObxZW6XdTiP3Vm57M7MZXfWYfZk5brzh2kSX6cGI/YvKzkZY0wICLaSU+A0JTHGGGNclpyMMcYEHEtOxhhjAo4lJ2OMMQHHkpMxxpiAY8nJGGMMACIyQkTWiUiKiNxdyvo6IjLLXf+9iCT5KhZLTsYYYxCRcOBp4EygGzBGRLqV2OxaYL+qdgQeBx72VTyWnIwxxgAMAFJUdaOqHgFmAueW2OZc4FV3+h1gmBxrv0plCLoeInJyclREDlVx9wggvzrj8SO7lsBk1xKY7FogRkSWecxPU9VpHvMtgVSP+TRgYIljFG+jqvkikgE0AtKrEE+5gi45qWqVS3siskxV+1VnPP5i1xKY7FoCk11L8LFqPWOMMQDbgNYe863cZaVuIyIRQAKw1xfBWHIyxhgDsBToJCLtRCQKGA18VGKbj4Cr3OmLgK/VRx20Bl213jGaVvEmQcOuJTDZtQQmu5YKuPeQJgJfAOHAy6q6SkQeAJap6kfAS8DrIpIC7MNJYD4RdL2SG2OMqf2sWs8YY0zAseRkjDEm4IRMcqqoW45AJyKbReRXEVle9KyCiDQUkS9F5Df33wb+jrM0IvKyiOwWkZUey0qNXRxT3c/pFxHp47/I/6iMa5ksItvcz2a5iJzlse4e91rWichw/0T9RyLSWkTmichqEVklIn9xlwfd51LOtQTj5xItIj+IyAr3Wu53l7dzuwtKcbsPinKX11h3QjVOVWv9C+fm3gagPRAFrAC6+TuuSl7DZiCxxLJHgLvd6buBh/0dZxmxnwr0AVZWFDtwFjAHEOAE4Ht/x+/FtUwG/lrKtt3cv7U6QDv3bzDc39fgxtYc6ONOxwHr3XiD7nMp51qC8XMRoJ47HQl8777fbwOj3eXPATe40zcCz7nTo4FZ/r6G6nqFSsnJm245gpFnVyKvAuf5L5SyqeoCnJY9nsqK/VzgNXUsAeqLSPMaCdQLZVxLWc4FZqpqrqpuAlJw/hb9TlV3qOpP7nQWsAbn6f+g+1zKuZayBPLnoqqa7c5Gui8FTsPpLgj++LnUSHdCNS1UklNp3XKU98cbiBT4n4j8KCIT3GVNVXWHO70TaOqf0KqkrNiD9bOa6FZ3vexRvRoU1+JWBfXG+ZUe1J9LiWuBIPxcRCRcRJYDu4EvcUp2B1S1qMsiz3iP6k4IKOpOKOiFSnKqDU5W1T44PQbfJCKneq5Up1wflM8FBHPsrmeBDkAvYAfwmF+jqQQRqQe8C9yqqpme64LtcynlWoLyc1HVAlXthdNDwwCgq38j8o9QSU7edMsR0FR1m/vvbuB9nD/aXUVVK+6/u/0XYaWVFXvQfVaqusv9QikEXuD3KqKAvhYRicT5Mn9TVd9zFwfl51LatQTr51JEVQ8A84ATcapRizpN8Iy3xroTqmmhkpy86ZYjYIlIrIjEFU0DZwArOborkauAD/0TYZWUFftHwJVu67ATgAyPaqaAVOLey/k4nw041zLabVHVDugE/FDT8ZXGvS/xErBGVf+fx6qg+1zKupYg/Vwai0h9dzoGOB3nHto8nO6C4I+fS410J1Tj/N0io6ZeOK2N1uPU397r73gqGXt7nNZFK4BVRfHj1C3PBX4DvgIa+jvWMuKfgVOtkodTX35tWbHjtFZ62v2cfgX6+Tt+L67ldTfWX3C+LJp7bH+vey3rgDP9Hb9HXCfjVNn9Aix3X2cF4+dSzrUE4+dyPPCzG/NKYJK7vD1OAk0BZgN13OXR7nyKu769v6+hul7WfZExxpiAEyrVesYYY4KIJSdjjDEBx5KTMcaYgGPJyRhjTMCx5GSMMSbgWHIypgQRKfDoyXq5VGMv9iKS5NmjuTGmdKE2TLsx3jikTvcxxhg/sZKTMV4SZ0ytR8QZV+sHEenoLk8Ska/dDkbnikgbd3lTEXnfHZtnhYic5B4qXERecMfr+Z/bE4AxxoMlJ2P+KKZEtd6lHusyVDUZeAp4wl32X+BVVT0eeBOY6i6fCnyjqj1xxoBa5S7vBDytqt2BA8CFPr0aY4KQ9RBhTAkikq2q9UpZvhk4TVU3uh2N7lTVRiKSjtM1Tp67fIeqJorIHqCVquZ6HCMJ+FJVO7nzdwGRqvpgDVyaMUHDSk7GVI6WMV0ZuR7TBdi9X2P+wJKTMZVzqce/37nTi3F6uge4HFjoTs8FboDiAeQSaipIY4Kd/WIz5o9i3JFIi3yuqkXNyRuIyC84pZ8x7rKbgVdE5E5gD3C1u/wvwDQRuRanhHQDTo/mxpgK2D0nY7zk3nPqp6rp/o7FmNrOqvWMMcYEHCs5GWOMCThWcjLGGBNwLDkZY4wJOJacjDHGBBxLTsYYYwKOJSdjjDEB5/8DpEfspoSENI8AAAAASUVORK5CYII=\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "trainer(qlstm, embedding_dim, hidden_dim, n_epochs, model_label='Hybrid Quantum')"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "Let's compare it with the classical version"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1 / 300: Loss = 1.119 Acc = 0.33\n",
      "Epoch 2 / 300: Loss = 1.108 Acc = 0.33\n",
      "Epoch 3 / 300: Loss = 1.098 Acc = 0.33\n",
      "Epoch 4 / 300: Loss = 1.089 Acc = 0.44\n",
      "Epoch 5 / 300: Loss = 1.081 Acc = 0.44\n",
      "Epoch 6 / 300: Loss = 1.073 Acc = 0.44\n",
      "Epoch 7 / 300: Loss = 1.067 Acc = 0.44\n",
      "Epoch 8 / 300: Loss = 1.061 Acc = 0.44\n",
      "Epoch 9 / 300: Loss = 1.055 Acc = 0.44\n",
      "Epoch 10 / 300: Loss = 1.051 Acc = 0.44\n",
      "Epoch 11 / 300: Loss = 1.046 Acc = 0.44\n",
      "Epoch 12 / 300: Loss = 1.042 Acc = 0.56\n",
      "Epoch 13 / 300: Loss = 1.038 Acc = 0.44\n",
      "Epoch 14 / 300: Loss = 1.034 Acc = 0.44\n",
      "Epoch 15 / 300: Loss = 1.031 Acc = 0.44\n",
      "Epoch 16 / 300: Loss = 1.028 Acc = 0.44\n",
      "Epoch 17 / 300: Loss = 1.025 Acc = 0.44\n",
      "Epoch 18 / 300: Loss = 1.022 Acc = 0.44\n",
      "Epoch 19 / 300: Loss = 1.019 Acc = 0.44\n",
      "Epoch 20 / 300: Loss = 1.016 Acc = 0.44\n",
      "Epoch 21 / 300: Loss = 1.013 Acc = 0.44\n",
      "Epoch 22 / 300: Loss = 1.011 Acc = 0.44\n",
      "Epoch 23 / 300: Loss = 1.008 Acc = 0.44\n",
      "Epoch 24 / 300: Loss = 1.005 Acc = 0.44\n",
      "Epoch 25 / 300: Loss = 1.003 Acc = 0.44\n",
      "Epoch 26 / 300: Loss = 1.000 Acc = 0.44\n",
      "Epoch 27 / 300: Loss = 0.998 Acc = 0.44\n",
      "Epoch 28 / 300: Loss = 0.995 Acc = 0.44\n",
      "Epoch 29 / 300: Loss = 0.992 Acc = 0.44\n",
      "Epoch 30 / 300: Loss = 0.990 Acc = 0.44\n",
      "Epoch 31 / 300: Loss = 0.987 Acc = 0.44\n",
      "Epoch 32 / 300: Loss = 0.984 Acc = 0.44\n",
      "Epoch 33 / 300: Loss = 0.981 Acc = 0.44\n",
      "Epoch 34 / 300: Loss = 0.978 Acc = 0.44\n",
      "Epoch 35 / 300: Loss = 0.976 Acc = 0.44\n",
      "Epoch 36 / 300: Loss = 0.973 Acc = 0.44\n",
      "Epoch 37 / 300: Loss = 0.970 Acc = 0.44\n",
      "Epoch 38 / 300: Loss = 0.967 Acc = 0.44\n",
      "Epoch 39 / 300: Loss = 0.963 Acc = 0.44\n",
      "Epoch 40 / 300: Loss = 0.960 Acc = 0.44\n",
      "Epoch 41 / 300: Loss = 0.957 Acc = 0.44\n",
      "Epoch 42 / 300: Loss = 0.954 Acc = 0.44\n",
      "Epoch 43 / 300: Loss = 0.950 Acc = 0.56\n",
      "Epoch 44 / 300: Loss = 0.947 Acc = 0.56\n",
      "Epoch 45 / 300: Loss = 0.943 Acc = 0.56\n",
      "Epoch 46 / 300: Loss = 0.940 Acc = 0.56\n",
      "Epoch 47 / 300: Loss = 0.936 Acc = 0.56\n",
      "Epoch 48 / 300: Loss = 0.932 Acc = 0.67\n",
      "Epoch 49 / 300: Loss = 0.929 Acc = 0.67\n",
      "Epoch 50 / 300: Loss = 0.925 Acc = 0.67\n",
      "Epoch 51 / 300: Loss = 0.921 Acc = 0.67\n",
      "Epoch 52 / 300: Loss = 0.917 Acc = 0.67\n",
      "Epoch 53 / 300: Loss = 0.913 Acc = 0.67\n",
      "Epoch 54 / 300: Loss = 0.909 Acc = 0.67\n",
      "Epoch 55 / 300: Loss = 0.904 Acc = 0.67\n",
      "Epoch 56 / 300: Loss = 0.900 Acc = 0.67\n",
      "Epoch 57 / 300: Loss = 0.896 Acc = 0.67\n",
      "Epoch 58 / 300: Loss = 0.891 Acc = 0.67\n",
      "Epoch 59 / 300: Loss = 0.887 Acc = 0.67\n",
      "Epoch 60 / 300: Loss = 0.882 Acc = 0.67\n",
      "Epoch 61 / 300: Loss = 0.877 Acc = 0.67\n",
      "Epoch 62 / 300: Loss = 0.872 Acc = 0.67\n",
      "Epoch 63 / 300: Loss = 0.868 Acc = 0.67\n",
      "Epoch 64 / 300: Loss = 0.863 Acc = 0.67\n",
      "Epoch 65 / 300: Loss = 0.858 Acc = 0.67\n",
      "Epoch 66 / 300: Loss = 0.853 Acc = 0.67\n",
      "Epoch 67 / 300: Loss = 0.847 Acc = 0.67\n",
      "Epoch 68 / 300: Loss = 0.842 Acc = 0.67\n",
      "Epoch 69 / 300: Loss = 0.837 Acc = 0.67\n",
      "Epoch 70 / 300: Loss = 0.831 Acc = 0.67\n",
      "Epoch 71 / 300: Loss = 0.826 Acc = 0.67\n",
      "Epoch 72 / 300: Loss = 0.821 Acc = 0.67\n",
      "Epoch 73 / 300: Loss = 0.815 Acc = 0.67\n",
      "Epoch 74 / 300: Loss = 0.809 Acc = 0.67\n",
      "Epoch 75 / 300: Loss = 0.804 Acc = 0.67\n",
      "Epoch 76 / 300: Loss = 0.798 Acc = 0.67\n",
      "Epoch 77 / 300: Loss = 0.792 Acc = 0.67\n",
      "Epoch 78 / 300: Loss = 0.786 Acc = 0.67\n",
      "Epoch 79 / 300: Loss = 0.780 Acc = 0.67\n",
      "Epoch 80 / 300: Loss = 0.774 Acc = 0.67\n",
      "Epoch 81 / 300: Loss = 0.768 Acc = 0.67\n",
      "Epoch 82 / 300: Loss = 0.762 Acc = 0.67\n",
      "Epoch 83 / 300: Loss = 0.756 Acc = 0.67\n",
      "Epoch 84 / 300: Loss = 0.750 Acc = 0.67\n",
      "Epoch 85 / 300: Loss = 0.743 Acc = 0.67\n",
      "Epoch 86 / 300: Loss = 0.737 Acc = 0.67\n",
      "Epoch 87 / 300: Loss = 0.730 Acc = 0.67\n",
      "Epoch 88 / 300: Loss = 0.724 Acc = 0.67\n",
      "Epoch 89 / 300: Loss = 0.717 Acc = 0.67\n",
      "Epoch 90 / 300: Loss = 0.711 Acc = 0.67\n",
      "Epoch 91 / 300: Loss = 0.704 Acc = 0.67\n",
      "Epoch 92 / 300: Loss = 0.697 Acc = 0.67\n",
      "Epoch 93 / 300: Loss = 0.690 Acc = 0.67\n",
      "Epoch 94 / 300: Loss = 0.683 Acc = 0.67\n",
      "Epoch 95 / 300: Loss = 0.676 Acc = 0.67\n",
      "Epoch 96 / 300: Loss = 0.669 Acc = 0.67\n",
      "Epoch 97 / 300: Loss = 0.662 Acc = 0.78\n",
      "Epoch 98 / 300: Loss = 0.655 Acc = 0.78\n",
      "Epoch 99 / 300: Loss = 0.648 Acc = 0.78\n",
      "Epoch 100 / 300: Loss = 0.640 Acc = 0.78\n",
      "Epoch 101 / 300: Loss = 0.633 Acc = 0.78\n",
      "Epoch 102 / 300: Loss = 0.626 Acc = 0.78\n",
      "Epoch 103 / 300: Loss = 0.618 Acc = 0.78\n",
      "Epoch 104 / 300: Loss = 0.611 Acc = 0.78\n",
      "Epoch 105 / 300: Loss = 0.604 Acc = 0.78\n",
      "Epoch 106 / 300: Loss = 0.596 Acc = 0.78\n",
      "Epoch 107 / 300: Loss = 0.589 Acc = 0.78\n",
      "Epoch 108 / 300: Loss = 0.581 Acc = 0.78\n",
      "Epoch 109 / 300: Loss = 0.574 Acc = 0.78\n",
      "Epoch 110 / 300: Loss = 0.567 Acc = 0.78\n",
      "Epoch 111 / 300: Loss = 0.559 Acc = 0.78\n",
      "Epoch 112 / 300: Loss = 0.552 Acc = 0.78\n",
      "Epoch 113 / 300: Loss = 0.544 Acc = 0.78\n",
      "Epoch 114 / 300: Loss = 0.537 Acc = 0.78\n",
      "Epoch 115 / 300: Loss = 0.530 Acc = 0.78\n",
      "Epoch 116 / 300: Loss = 0.523 Acc = 0.89\n",
      "Epoch 117 / 300: Loss = 0.515 Acc = 0.89\n",
      "Epoch 118 / 300: Loss = 0.508 Acc = 0.89\n",
      "Epoch 119 / 300: Loss = 0.501 Acc = 0.89\n",
      "Epoch 120 / 300: Loss = 0.494 Acc = 0.89\n",
      "Epoch 121 / 300: Loss = 0.487 Acc = 0.89\n",
      "Epoch 122 / 300: Loss = 0.480 Acc = 0.89\n",
      "Epoch 123 / 300: Loss = 0.473 Acc = 0.89\n",
      "Epoch 124 / 300: Loss = 0.466 Acc = 0.89\n",
      "Epoch 125 / 300: Loss = 0.459 Acc = 0.89\n",
      "Epoch 126 / 300: Loss = 0.452 Acc = 0.89\n",
      "Epoch 127 / 300: Loss = 0.445 Acc = 0.89\n",
      "Epoch 128 / 300: Loss = 0.438 Acc = 0.89\n",
      "Epoch 129 / 300: Loss = 0.431 Acc = 0.89\n",
      "Epoch 130 / 300: Loss = 0.425 Acc = 0.89\n",
      "Epoch 131 / 300: Loss = 0.418 Acc = 0.89\n",
      "Epoch 132 / 300: Loss = 0.412 Acc = 0.89\n",
      "Epoch 133 / 300: Loss = 0.405 Acc = 0.89\n",
      "Epoch 134 / 300: Loss = 0.399 Acc = 0.89\n",
      "Epoch 135 / 300: Loss = 0.392 Acc = 0.89\n",
      "Epoch 136 / 300: Loss = 0.386 Acc = 0.89\n",
      "Epoch 137 / 300: Loss = 0.380 Acc = 0.89\n",
      "Epoch 138 / 300: Loss = 0.373 Acc = 0.89\n",
      "Epoch 139 / 300: Loss = 0.367 Acc = 1.00\n",
      "Epoch 140 / 300: Loss = 0.361 Acc = 1.00\n",
      "Epoch 141 / 300: Loss = 0.355 Acc = 1.00\n",
      "Epoch 142 / 300: Loss = 0.349 Acc = 1.00\n",
      "Epoch 143 / 300: Loss = 0.343 Acc = 1.00\n",
      "Epoch 144 / 300: Loss = 0.338 Acc = 1.00\n",
      "Epoch 145 / 300: Loss = 0.332 Acc = 1.00\n",
      "Epoch 146 / 300: Loss = 0.326 Acc = 1.00\n",
      "Epoch 147 / 300: Loss = 0.321 Acc = 1.00\n",
      "Epoch 148 / 300: Loss = 0.315 Acc = 1.00\n",
      "Epoch 149 / 300: Loss = 0.310 Acc = 1.00\n",
      "Epoch 150 / 300: Loss = 0.305 Acc = 1.00\n",
      "Epoch 151 / 300: Loss = 0.299 Acc = 1.00\n",
      "Epoch 152 / 300: Loss = 0.294 Acc = 1.00\n",
      "Epoch 153 / 300: Loss = 0.289 Acc = 1.00\n",
      "Epoch 154 / 300: Loss = 0.284 Acc = 1.00\n",
      "Epoch 155 / 300: Loss = 0.279 Acc = 1.00\n",
      "Epoch 156 / 300: Loss = 0.274 Acc = 1.00\n",
      "Epoch 157 / 300: Loss = 0.269 Acc = 1.00\n",
      "Epoch 158 / 300: Loss = 0.265 Acc = 1.00\n",
      "Epoch 159 / 300: Loss = 0.260 Acc = 1.00\n",
      "Epoch 160 / 300: Loss = 0.256 Acc = 1.00\n",
      "Epoch 161 / 300: Loss = 0.251 Acc = 1.00\n",
      "Epoch 162 / 300: Loss = 0.247 Acc = 1.00\n",
      "Epoch 163 / 300: Loss = 0.242 Acc = 1.00\n",
      "Epoch 164 / 300: Loss = 0.238 Acc = 1.00\n",
      "Epoch 165 / 300: Loss = 0.234 Acc = 1.00\n",
      "Epoch 166 / 300: Loss = 0.230 Acc = 1.00\n",
      "Epoch 167 / 300: Loss = 0.226 Acc = 1.00\n",
      "Epoch 168 / 300: Loss = 0.222 Acc = 1.00\n",
      "Epoch 169 / 300: Loss = 0.218 Acc = 1.00\n",
      "Epoch 170 / 300: Loss = 0.214 Acc = 1.00\n",
      "Epoch 171 / 300: Loss = 0.211 Acc = 1.00\n",
      "Epoch 172 / 300: Loss = 0.207 Acc = 1.00\n",
      "Epoch 173 / 300: Loss = 0.204 Acc = 1.00\n",
      "Epoch 174 / 300: Loss = 0.200 Acc = 1.00\n",
      "Epoch 175 / 300: Loss = 0.197 Acc = 1.00\n",
      "Epoch 176 / 300: Loss = 0.193 Acc = 1.00\n",
      "Epoch 177 / 300: Loss = 0.190 Acc = 1.00\n",
      "Epoch 178 / 300: Loss = 0.187 Acc = 1.00\n",
      "Epoch 179 / 300: Loss = 0.184 Acc = 1.00\n",
      "Epoch 180 / 300: Loss = 0.181 Acc = 1.00\n",
      "Epoch 181 / 300: Loss = 0.178 Acc = 1.00\n",
      "Epoch 182 / 300: Loss = 0.175 Acc = 1.00\n",
      "Epoch 183 / 300: Loss = 0.172 Acc = 1.00\n",
      "Epoch 184 / 300: Loss = 0.169 Acc = 1.00\n",
      "Epoch 185 / 300: Loss = 0.166 Acc = 1.00\n",
      "Epoch 186 / 300: Loss = 0.164 Acc = 1.00\n",
      "Epoch 187 / 300: Loss = 0.161 Acc = 1.00\n",
      "Epoch 188 / 300: Loss = 0.158 Acc = 1.00\n",
      "Epoch 189 / 300: Loss = 0.156 Acc = 1.00\n",
      "Epoch 190 / 300: Loss = 0.153 Acc = 1.00\n",
      "Epoch 191 / 300: Loss = 0.151 Acc = 1.00\n",
      "Epoch 192 / 300: Loss = 0.148 Acc = 1.00\n",
      "Epoch 193 / 300: Loss = 0.146 Acc = 1.00\n",
      "Epoch 194 / 300: Loss = 0.144 Acc = 1.00\n",
      "Epoch 195 / 300: Loss = 0.142 Acc = 1.00\n",
      "Epoch 196 / 300: Loss = 0.139 Acc = 1.00\n",
      "Epoch 197 / 300: Loss = 0.137 Acc = 1.00\n",
      "Epoch 198 / 300: Loss = 0.135 Acc = 1.00\n",
      "Epoch 199 / 300: Loss = 0.133 Acc = 1.00\n",
      "Epoch 200 / 300: Loss = 0.131 Acc = 1.00\n",
      "Epoch 201 / 300: Loss = 0.129 Acc = 1.00\n",
      "Epoch 202 / 300: Loss = 0.127 Acc = 1.00\n",
      "Epoch 203 / 300: Loss = 0.125 Acc = 1.00\n",
      "Epoch 204 / 300: Loss = 0.123 Acc = 1.00\n",
      "Epoch 205 / 300: Loss = 0.122 Acc = 1.00\n",
      "Epoch 206 / 300: Loss = 0.120 Acc = 1.00\n",
      "Epoch 207 / 300: Loss = 0.118 Acc = 1.00\n",
      "Epoch 208 / 300: Loss = 0.116 Acc = 1.00\n",
      "Epoch 209 / 300: Loss = 0.115 Acc = 1.00\n",
      "Epoch 210 / 300: Loss = 0.113 Acc = 1.00\n",
      "Epoch 211 / 300: Loss = 0.111 Acc = 1.00\n",
      "Epoch 212 / 300: Loss = 0.110 Acc = 1.00\n",
      "Epoch 213 / 300: Loss = 0.108 Acc = 1.00\n",
      "Epoch 214 / 300: Loss = 0.107 Acc = 1.00\n",
      "Epoch 215 / 300: Loss = 0.105 Acc = 1.00\n",
      "Epoch 216 / 300: Loss = 0.104 Acc = 1.00\n",
      "Epoch 217 / 300: Loss = 0.102 Acc = 1.00\n",
      "Epoch 218 / 300: Loss = 0.101 Acc = 1.00\n",
      "Epoch 219 / 300: Loss = 0.100 Acc = 1.00\n",
      "Epoch 220 / 300: Loss = 0.098 Acc = 1.00\n",
      "Epoch 221 / 300: Loss = 0.097 Acc = 1.00\n",
      "Epoch 222 / 300: Loss = 0.096 Acc = 1.00\n",
      "Epoch 223 / 300: Loss = 0.094 Acc = 1.00\n",
      "Epoch 224 / 300: Loss = 0.093 Acc = 1.00\n",
      "Epoch 225 / 300: Loss = 0.092 Acc = 1.00\n",
      "Epoch 226 / 300: Loss = 0.091 Acc = 1.00\n",
      "Epoch 227 / 300: Loss = 0.090 Acc = 1.00\n",
      "Epoch 228 / 300: Loss = 0.089 Acc = 1.00\n",
      "Epoch 229 / 300: Loss = 0.087 Acc = 1.00\n",
      "Epoch 230 / 300: Loss = 0.086 Acc = 1.00\n",
      "Epoch 231 / 300: Loss = 0.085 Acc = 1.00\n",
      "Epoch 232 / 300: Loss = 0.084 Acc = 1.00\n",
      "Epoch 233 / 300: Loss = 0.083 Acc = 1.00\n",
      "Epoch 234 / 300: Loss = 0.082 Acc = 1.00\n",
      "Epoch 235 / 300: Loss = 0.081 Acc = 1.00\n",
      "Epoch 236 / 300: Loss = 0.080 Acc = 1.00\n",
      "Epoch 237 / 300: Loss = 0.079 Acc = 1.00\n",
      "Epoch 238 / 300: Loss = 0.078 Acc = 1.00\n",
      "Epoch 239 / 300: Loss = 0.077 Acc = 1.00\n",
      "Epoch 240 / 300: Loss = 0.076 Acc = 1.00\n",
      "Epoch 241 / 300: Loss = 0.076 Acc = 1.00\n",
      "Epoch 242 / 300: Loss = 0.075 Acc = 1.00\n",
      "Epoch 243 / 300: Loss = 0.074 Acc = 1.00\n",
      "Epoch 244 / 300: Loss = 0.073 Acc = 1.00\n",
      "Epoch 245 / 300: Loss = 0.072 Acc = 1.00\n",
      "Epoch 246 / 300: Loss = 0.071 Acc = 1.00\n",
      "Epoch 247 / 300: Loss = 0.071 Acc = 1.00\n",
      "Epoch 248 / 300: Loss = 0.070 Acc = 1.00\n",
      "Epoch 249 / 300: Loss = 0.069 Acc = 1.00\n",
      "Epoch 250 / 300: Loss = 0.068 Acc = 1.00\n",
      "Epoch 251 / 300: Loss = 0.067 Acc = 1.00\n",
      "Epoch 252 / 300: Loss = 0.067 Acc = 1.00\n",
      "Epoch 253 / 300: Loss = 0.066 Acc = 1.00\n",
      "Epoch 254 / 300: Loss = 0.065 Acc = 1.00\n",
      "Epoch 255 / 300: Loss = 0.065 Acc = 1.00\n",
      "Epoch 256 / 300: Loss = 0.064 Acc = 1.00\n",
      "Epoch 257 / 300: Loss = 0.063 Acc = 1.00\n",
      "Epoch 258 / 300: Loss = 0.063 Acc = 1.00\n",
      "Epoch 259 / 300: Loss = 0.062 Acc = 1.00\n",
      "Epoch 260 / 300: Loss = 0.061 Acc = 1.00\n",
      "Epoch 261 / 300: Loss = 0.061 Acc = 1.00\n",
      "Epoch 262 / 300: Loss = 0.060 Acc = 1.00\n",
      "Epoch 263 / 300: Loss = 0.059 Acc = 1.00\n",
      "Epoch 264 / 300: Loss = 0.059 Acc = 1.00\n",
      "Epoch 265 / 300: Loss = 0.058 Acc = 1.00\n",
      "Epoch 266 / 300: Loss = 0.058 Acc = 1.00\n",
      "Epoch 267 / 300: Loss = 0.057 Acc = 1.00\n",
      "Epoch 268 / 300: Loss = 0.057 Acc = 1.00\n",
      "Epoch 269 / 300: Loss = 0.056 Acc = 1.00\n",
      "Epoch 270 / 300: Loss = 0.055 Acc = 1.00\n",
      "Epoch 271 / 300: Loss = 0.055 Acc = 1.00\n",
      "Epoch 272 / 300: Loss = 0.054 Acc = 1.00\n",
      "Epoch 273 / 300: Loss = 0.054 Acc = 1.00\n",
      "Epoch 274 / 300: Loss = 0.053 Acc = 1.00\n",
      "Epoch 275 / 300: Loss = 0.053 Acc = 1.00\n",
      "Epoch 276 / 300: Loss = 0.052 Acc = 1.00\n",
      "Epoch 277 / 300: Loss = 0.052 Acc = 1.00\n",
      "Epoch 278 / 300: Loss = 0.051 Acc = 1.00\n",
      "Epoch 279 / 300: Loss = 0.051 Acc = 1.00\n",
      "Epoch 280 / 300: Loss = 0.050 Acc = 1.00\n",
      "Epoch 281 / 300: Loss = 0.050 Acc = 1.00\n",
      "Epoch 282 / 300: Loss = 0.050 Acc = 1.00\n",
      "Epoch 283 / 300: Loss = 0.049 Acc = 1.00\n",
      "Epoch 284 / 300: Loss = 0.049 Acc = 1.00\n",
      "Epoch 285 / 300: Loss = 0.048 Acc = 1.00\n",
      "Epoch 286 / 300: Loss = 0.048 Acc = 1.00\n",
      "Epoch 287 / 300: Loss = 0.047 Acc = 1.00\n",
      "Epoch 288 / 300: Loss = 0.047 Acc = 1.00\n",
      "Epoch 289 / 300: Loss = 0.047 Acc = 1.00\n",
      "Epoch 290 / 300: Loss = 0.046 Acc = 1.00\n",
      "Epoch 291 / 300: Loss = 0.046 Acc = 1.00\n",
      "Epoch 292 / 300: Loss = 0.045 Acc = 1.00\n",
      "Epoch 293 / 300: Loss = 0.045 Acc = 1.00\n",
      "Epoch 294 / 300: Loss = 0.045 Acc = 1.00\n",
      "Epoch 295 / 300: Loss = 0.044 Acc = 1.00\n",
      "Epoch 296 / 300: Loss = 0.044 Acc = 1.00\n",
      "Epoch 297 / 300: Loss = 0.044 Acc = 1.00\n",
      "Epoch 298 / 300: Loss = 0.043 Acc = 1.00\n",
      "Epoch 299 / 300: Loss = 0.043 Acc = 1.00\n",
      "Epoch 300 / 300: Loss = 0.043 Acc = 1.00\n",
      "Sentence:  ['The', 'dog', 'ate', 'the', 'apple']\n",
      "Labels:    ['DET', 'NN', 'V', 'DET', 'NN']\n",
      "Predicted: ['DET', 'NN', 'V', 'DET', 'NN']\n"
     ]
    },
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 2 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "clstm = nn.LSTM(embedding_dim, hidden_dim, num_layers=n_layers)\n",
    "trainer(clstm, embedding_dim, hidden_dim, n_epochs,model_label='Classical')"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "In case you want to try with a different backend simulator you could also install `pennylane-qiskit` and play around with `qiskit.aer` or `qiskit.qasm`.\n",
    "Be aware that it's going to be way slower because of qiskit simulators."
   ],
   "metadata": {
    "collapsed": false
   }
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}