Neural_Turing_Machine_Implementation.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Neural_Turing_Machine_Implementation.ipynb",
      "version": "0.3.2",
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "metadata": {
        "id": "0DlbXn79fsnL",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Neural Turing Machine: Implementation\n",
        "---\n",
        "In this file we are going to implement both a feedforward (Feedforward-NTM) and LSTM controller (LSTM-NTM). We will also implement an LSTM for the same task (LSTM). This will be the baseline to compare the performance of the NTM models with. This section is very much the programming related part of our implementation and it is where all the codes and primary results reside. We will analyze/use these results in the file titled **Neural Turing Machine: Reports and Conclusions **."
      ]
    },
    {
      "metadata": {
        "id": "hRL0tD89Tmyn",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "##  Importing\n",
        "Below we are importing all the needed modules to accomplish our task."
      ]
    },
    {
      "metadata": {
        "id": "YETIgdbzGsvG",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "from keras.layers.core import Dense\n",
        "from keras.layers.recurrent import Recurrent, GRU, LSTM\n",
        "from keras.models import Sequential\n",
        "from keras.optimizers import Adam, SGD\n",
        "from keras.initializers import RandomNormal\n",
        "import keras\n",
        "from keras.layers import LSTM, Activation\n",
        "import matplotlib.pyplot as plt\n",
        "from keras import backend as K\n",
        "import tensorflow as tf\n",
        "import warnings\n",
        "from keras.initializers import Orthogonal, Zeros, Constant\n",
        "from keras.engine.topology import InputSpec \n",
        "from keras.activations import get as get_activations\n",
        "from keras.activations import softmax, tanh, sigmoid, hard_sigmoid\n",
        "from datetime import datetime\n",
        "from keras.callbacks import TensorBoard, ModelCheckpoint, TerminateOnNaN \n",
        "from keras.layers.core import Activation\n",
        "from keras.layers.wrappers import TimeDistributed\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "-Ufv6IQPi9sY",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "## Initializations"
      ]
    },
    {
      "metadata": {
        "id": "Jw-AvcM1dtb6",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "In this part we define some variables and initialize them. These variables are:\n",
        "\n",
        "*    **input and output dimensions**: The input data are $8$-bit binary data followed by a flag bit.\n",
        "\n",
        "*    **hidden units** : The number of hidden units is equal to $100$."
      ]
    },
    {
      "metadata": {
        "id": "Z-MPDaM1i8dQ",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "input_dim = output_dim = 9\n",
        "units = 100  "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "F9yf8IJYecqQ",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "## Creating Random Input Data for The Copy Task\n",
        "\n",
        "This function is a generator of random input binary data. It generates two matrices, input data and output data.\n",
        "\n",
        "The input data is a series of $X$s where each $X$ is an $8$-bit random binary data followed by a flag bit. The length of this serie is randomly generated and varies from 1 to 20. However, the function always generates two matrices of size $41\\times 9$ as input and outputs. The structure of the input and output matrices is as follows:\n",
        "\n",
        "*    **Input data structure**: The first $20$ rows indicate a random binary sequence. If the length of the sequence is less than $20$, the rest of the first $20$ rows are filled with the flag sequence.\n",
        "The $21$st row is the flag, showing that the input data is finished and the network should start generating its output.\n",
        "The last $20$ rows are filled with flag because the network is producing its output (copying the input) while seeing these rows.\n",
        "\n",
        "*    **Output data structure**: The first $20$ rows are filled with flag sequences because the network is reading the input and saving it, so its output does not matter. The $21$st row is the flag, showing that the input data is finished and the network should start generating its output. The last $20$ rows are the data sequence which the network should produce. If the length of the sequence was less than $20$, the last rows are filled with the flag sequence.\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "id": "Qeu65BAzeWfQ",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "#Creating random input data for the copy task\n",
        "def get_sample(batch_size=128, in_bits=8, max_size=20, min_size=1):\n",
        "    in_bits = in_bits +1  #One flag point at the end.\n",
        "    l = max_size*2 +1\n",
        "    inp_mat = np.zeros((batch_size, l, in_bits))\n",
        "    out_mat = np.zeros((batch_size, l, in_bits)) \n",
        "    while True:\n",
        "        # make flag       \n",
        "        flag = np.zeros((1, in_bits))\n",
        "        flag[0, in_bits-1] = 1\n",
        "        \n",
        "        for i in range(batch_size):\n",
        "          #creating the actual sequence\n",
        "          ts = np.random.randint(low=min_size, high=max_size+1) #random sequence length\n",
        "          #print(ts)\n",
        "          actual_sequence = np.random.uniform(size=(ts, in_bits)) > 0.5\n",
        "          actual_sequence[:,in_bits-1] = 0\n",
        "          \n",
        "          zero_pad_begin = np.repeat(flag, max_size-ts, axis=0)  #flag_padding , max_size-ts          \n",
        "          zero_pad_end = np.repeat(flag, max_size, axis=0)  #flag_padding , max_size\n",
        "         \n",
        "          inp = np.concatenate( (zero_pad_begin , actual_sequence , flag, zero_pad_end) ,axis=0)\n",
        "          \n",
        "          out = np.concatenate( (zero_pad_end , flag , actual_sequence , zero_pad_begin) ,axis=0)\n",
        "          \n",
        "          inp_mat[i,:,:] = inp\n",
        "          out_mat[i,:,:] = out\n",
        "          \n",
        "        yield inp_mat, out_mat\n",
        "        \n",
        "inp, out = next(get_sample(batch_size=2, max_size=6))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "ILWdTSmPehga",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "## Training And Testing Functions\n",
        "\n",
        "\n",
        "We define some functions here for training and testing a model.\n",
        "\n",
        "\n",
        "*   **test_model** : This function tests a given model with the provided sequence length. \n",
        "\n",
        "*   **train_model**: this function trains the given model with the provided parameters.\n",
        "\n",
        "*   **test_with_sequence_lengths**: This function tests a model (using the **test_model** function) for different input sequence length from $10$ to $100$.\n",
        "\n",
        "*   **load_and_test_model**: This function is defined for some tests. It tests a previously trained and saved model."
      ]
    },
    {
      "metadata": {
        "id": "tJQRCmvTePGK",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "LOG_PATH_BASE=\"logs/\"     #for tensorboard callbacks\n",
        "\n",
        "def test_model(model, sequence_length=None, verboose=False, test_batch_size=0):#rewritten\n",
        "    input_dim = model.input_dim\n",
        "    output_dim = model.output_dim\n",
        "    if test_batch_size == 0:\n",
        "      batch_size = model.batch_size\n",
        "    else:\n",
        "      batch_size = test_batch_size\n",
        "      \n",
        "    I, y_true = next(get_sample(batch_size=batch_size))\n",
        "\n",
        "    #print(\"testing the model with data: shape on input is {0} and shape of output is {1}\".format(I.shape, y_true.shape))\n",
        "   \n",
        "    l = model.evaluate(x=I, y=y_true, batch_size=batch_size, verbose=1)\n",
        "    print(l)\n",
        "      \n",
        "    res = model.predict(I, batch_size=batch_size)\n",
        "    y_pred = np.asarray(res)\n",
        "\n",
        "    y_pred = (y_pred > 0.5).astype('float64')  # wherever Y>0.5 turn it to one, else turn it to zero\n",
        "    x = y_true == y_pred  #comparing only the parts with data, not zero paddings\n",
        "    acc = x.mean() * 100\n",
        "\n",
        "    if verboose:\n",
        "        print(\"the overall accuracy for sequence_length {0} was: {1}\".format(sequence_length, x.mean()))\n",
        "   \n",
        "    return acc\n",
        "\n",
        "\n",
        "def train_model(model, epochs=10, min_size=5, max_size=20, callbacks=None, verboose=False):\n",
        "    input_dim = model.input_dim\n",
        "    output_dim = model.output_dim\n",
        "    batch_size = model.batch_size\n",
        "\n",
        "    sample_generator = get_sample(batch_size=batch_size)\n",
        "    if verboose:\n",
        "        for j in range(epochs):\n",
        "            history = model.fit_generator(sample_generator, steps_per_epoch=10, epochs=j+1, callbacks=callbacks, initial_epoch=j, nb_worker=1, max_q_size=10)\n",
        "            print(\"currently at epoch {0}\".format(j+1))\n",
        "            for i in [5,10,20,40]:\n",
        "                test_model(model, sequence_length=i, verboose=True)\n",
        "    else:\n",
        "        history = model.fit_generator(sample_generator, steps_per_epoch=10, epochs=epochs, callbacks=callbacks, nb_worker=1, max_q_size=10)\n",
        "        plt.plot(history.history['loss'], label='train')\n",
        "        plt.legend()\n",
        "        plt.show()\n",
        "        \n",
        "    #saving the history loss\n",
        "    los = history.history['loss']\n",
        "    np_loss_history = np.array(los)\n",
        "    np.savetxt(\"loss_history_ntm_\"+model.name+\".txt\", np_loss_history, delimiter=\",\")\n",
        "    \n",
        "    model.save(\"final_model_\"+model.name+\".hdf5\") \n",
        "    print(\"done training\")\n",
        "\n",
        "\n",
        "def test_with_sequence_lengths(model, testrange=[10,20,30,40,50,60,70,80,90,100], epochs=100, verboose=True):\n",
        "    ts = datetime.now().strftime(\"%Y-%m-%d_%H:%M:%S\")\n",
        "    log_path = LOG_PATH_BASE + ts + \"_-_\" + model.name \n",
        "    tensorboard = TensorBoard(log_dir=log_path,\n",
        "                                write_graph=False, \n",
        "                                #histogram_freq = 1,\n",
        "                                histogram_freq=0,\n",
        "                                write_images=True,\n",
        "                                batch_size = model.batch_size,\n",
        "                                write_grads=True)\n",
        "    model_saver =  ModelCheckpoint(log_path + \"/model.ckpt.{epoch:04d}.hdf5\", monitor='loss', period=1)\n",
        "    callbacks = [tensorboard, TerminateOnNaN(), model_saver]\n",
        "\n",
        "    train_model(model, epochs=epochs, callbacks=callbacks, verboose=verboose)\n",
        "\n",
        "    for i in testrange:\n",
        "        acc = test_model(model, sequence_length=i, verboose=verboose)\n",
        "        print(\"the accuracy for length {0} was: {1}%\".format(i,acc))\n",
        "    return\n",
        "\n",
        "#defined for some tests with the previously saved model\n",
        "def load_and_test_model(modelname, testrange=[10,20,40,80], verboose=False):\n",
        "    from keras.models import load_model\n",
        "    model = load_model(modelname)\n",
        "    model.input_dim = 9\n",
        "    model.output_dim = 9\n",
        "    model.batch_size = 100\n",
        "    for i in testrange:\n",
        "        acc = test_model(model, sequence_length=i, verboose=verboose)\n",
        "        print(\"the accuracy for length {0} was: {1}%\".format(i,acc))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "OHByXYMwev20",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "## LSTM Model Generation\n",
        "\n",
        "This function generates an **LSTM** model with the desired dimensions."
      ]
    },
    {
      "metadata": {
        "id": "h9MzcryjH8uU",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def gen_model_lstm(input_dim=10, output_dim=8, batch_size=100, sgd='', loss='binary_crossentropy'):\n",
        "    model_LSTM = Sequential()\n",
        "    model_LSTM.name = \"LSTM\"\n",
        "    model_LSTM.batch_size = batch_size\n",
        "    model_LSTM.input_dim = input_dim\n",
        "    model_LSTM.output_dim = output_dim\n",
        "\n",
        "    model_LSTM.add(LSTM(input_shape=(None, input_dim), units=units, return_sequences=True,kernel_initializer='glorot_normal'))\n",
        "    model_LSTM.add(LSTM(units=output_dim, return_sequences=True))\n",
        "    \n",
        "    model_LSTM.add(Activation(activation_func))\n",
        "    \n",
        "    model_LSTM.compile(loss=loss, optimizer=sgd, metrics = ['binary_accuracy','categorical_accuracy','mse'] , sample_weight_mode=\"temporal\") \n",
        "\n",
        "    return model_LSTM"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "nDuirDvjezEM",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "## NTM\n",
        "\n",
        "let's first define some helper functions.\n",
        "\n",
        "\n",
        "*   **gen_shift_tensor**: Shifting function: Generates shifting tensor.<br>\n",
        "This function first creates an identity matrix and then shifts it multiple times. \n",
        "The input **n_shifts** of this function indicates what kind of shifts we have. \n",
        "When the **shift_n** is $n$, we have a shift range of $\\left\\{-\\frac{n}{2},\\dots,-1,0,1,\\dots,\\frac{n}{2}\\right\\}$ i.e. there is a value $v$ from this range and a corresponding identity matrix $I$ that has been shifted $v$ times. <br>\n",
        "Each of these shifted matrices when multiplied by another matrix, will shift that matix with the corresponding shift number.\n",
        "Then this function returns a tensor with **n_shifts** of rotated versions of the identity matrix. \n",
        "Therefore, when this tensor is multiplied by a vector, the results are n_shifts shifted versions of that vector.\n",
        "\n",
        "The inputs to this function should be the number of slots in the memory, and the shift range.\n",
        "*   **normalize_weights**: equation $[9]$ of the paper\n",
        "*   **similarity_by_cosine_distance**: equation $[6]$ of the paper\n",
        "*   **controller_read_head_dim**: Calculates the dimension of the read head. Read head consists of :\n",
        "\n",
        "  * **key** vector with size: Memory_width\n",
        "\n",
        "  * **beta** with size: $1$\n",
        "\n",
        "  * **g** with size: $1$\n",
        "\n",
        "  * **shift** vector with size: shift_range\n",
        "\n",
        "  * **gamma** with size: $1$\n",
        "\n",
        "*   **controller_write_head_dim**: Calculates the dimension of the write head. Write head consists of :\n",
        "\n",
        "  *  **key** vector with size: Memory_width\n",
        "\n",
        "  *  **beta** with size: $1$\n",
        "\n",
        "  * **g** with size: $1$\n",
        "\n",
        "  * **shift** vector with size: shift_range\n",
        "\n",
        "  * **gamma** with size: $1$\n",
        "\n",
        "  * **add** vector with size: Memory_width\n",
        "\n",
        "  * **erase** vector with size: Memory_width\n",
        "\n",
        "*   **controller_input_output_shape**: calculates the controller input and output dimensions. \n",
        "\n",
        "  * **output size** : output dimension + read and write heads (we have only one read head and only one write head).\n",
        "\n",
        "  * **input size**  : input dimension + read_vector \n",
        "\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "id": "EJqeq5GJ_pIo",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def gen_shift_tensor(leng, n_shifts):        \n",
        "    eye = np.eye(leng)\n",
        "    shifts = range(n_shifts//2, -n_shifts//2, -1)\n",
        "    C = np.asarray([np.roll(eye, s, axis=1) for s in shifts])\n",
        "    return K.variable(C.astype(K.floatx()))  # a tensor with dimensions (n_shifts, leng, leng)\n",
        "\n",
        "def normalize_weights(w):\n",
        "    return w / (K.sum(w, axis=1, keepdims=True))\n",
        "\n",
        "def similarity_by_cosine_distance(M, k):\n",
        "    nk = K.l2_normalize(k, axis=-1)\n",
        "    nM = K.l2_normalize(M, axis=-1)\n",
        "    cosine_distance = K.batch_dot(nM, nk)\n",
        "    return cosine_distance\n",
        "\n",
        "def controller_read_head_dim(Memory_width, shift_range):    \n",
        "    return (Memory_width + 1 + 1 + shift_range + 1)\n",
        "\n",
        "def controller_write_head_dim(Memory_width, shift_range):\n",
        "    controller_read_head_emitting_dim = controller_read_head_dim(Memory_width, shift_range)    \n",
        "    return controller_read_head_emitting_dim + 2*Memory_width\n",
        "  \n",
        "def controller_input_output_shape(input_dim, output_dim, Memory_width, Memory_height, shift_range, read_heads, write_heads):\n",
        "    controller_read_head_emitting_dim = controller_read_head_dim(Memory_width, shift_range)\n",
        "    controller_write_head_emitting_dim = controller_write_head_dim(Memory_width, shift_range)\n",
        "    \n",
        "    controller_output_dim = (output_dim + \n",
        "            read_heads * controller_read_head_emitting_dim + \n",
        "            write_heads * controller_write_head_emitting_dim)\n",
        "    controller_input_dim = input_dim + read_heads * Memory_width\n",
        "\n",
        "    return controller_input_dim, controller_output_dim  "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "CK12-g2k_plA",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Now we will implement the **NTM** layer as a **Keras recurrent layer**. The new **NeuralTuringMachine class** extends the keras Recurrent class, and we define its methods below. In this way, we can consider it as a normal recurrent layer and add it easily to our model.\n",
        "\n",
        "Some of the fields of this class are as follows:\n",
        "\n",
        "\n",
        "\n",
        "*   **shift_range**: The available shifts will be from minus half of shift_range up to half of shift_range\n",
        "*   **Memory_height** and **Memory_width**: The memory size in Memory_height by Memory_width ($N$ by $M$ in the paper)\n",
        "*   **controller_model**: This is a Keras model that will be the controller of the **NTM** network.\n",
        "\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "id": "JHoxDkxLTTjf",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "class NeuralTuringMachine(Recurrent):   \n",
        "    def __init__(self, units, \n",
        "                        Memory_height=128,\n",
        "                        Memory_width=20,\n",
        "                        shift_range=3, #-1,0,1\n",
        "                        controller_model=None,\n",
        "                        read_heads=1,\n",
        "                        write_heads=1,\n",
        "                        activation='sigmoid',\n",
        "                        batch_size=100,                 \n",
        "                        stateful=False,\n",
        "                        **kwargs):\n",
        "        self.output_dim = units\n",
        "        self.units = units\n",
        "        self.Memory_height = Memory_height\n",
        "        self.Memory_width = Memory_width\n",
        "        self.shift_range = shift_range\n",
        "        self.controller = controller_model\n",
        "        self.activation = get_activations(activation)\n",
        "        self.read_heads = read_heads\n",
        "        self.write_heads = write_heads\n",
        "        self.batch_size = batch_size\n",
        "\n",
        "        try:\n",
        "            if controller.state.stateful:\n",
        "                self.controller_with_state = True \n",
        "        except:\n",
        "            self.controller_with_state = False\n",
        "\n",
        "        self.controller_read_head_emitting_dim = controller_read_head_dim(Memory_width, shift_range)\n",
        "        self.controller_write_head_emitting_dim = controller_write_head_dim(Memory_width, shift_range)\n",
        "\n",
        "        super(NeuralTuringMachine, self).__init__(**kwargs)\n",
        "\n",
        "    def build(self, input_shape):\n",
        "        if self.controller is None:\n",
        "          return False\n",
        "        \n",
        "        bs, input_length, input_dim = input_shape\n",
        "\n",
        "        self.controller_input_dim, self.controller_output_dim = controller_input_output_shape(\n",
        "                input_dim, self.units, self.Memory_width, self.Memory_height, self.shift_range, self.read_heads,\n",
        "                self.write_heads)\n",
        "            \n",
        "        self.shift_tensor = gen_shift_tensor(self.Memory_height, self.shift_range)\n",
        "        \n",
        "        self.trainable_weights = self.controller.trainable_weights   \n",
        "        self.states = [None, None, None, None]\n",
        "        self.state_spec = [InputSpec(shape=(None, self.output_dim)),              # old_ntm_output\n",
        "                            InputSpec(shape=(None, self.Memory_height, self.Memory_width)),          # Memory\n",
        "                            InputSpec(shape=(None, self.read_heads, self.Memory_height)),   # weights_read\n",
        "                            InputSpec(shape=(None, self.write_heads, self.Memory_height))]  # weights_write\n",
        "\n",
        "        super(NeuralTuringMachine, self).build(input_shape)\n",
        "\n",
        "\n",
        "    def get_initial_state(self, X):\n",
        "        init_old_ntm_output = K.ones((self.batch_size, self.output_dim), name=\"init_old_ntm_output\")*0.42 \n",
        "        init_M = K.ones((self.batch_size, self.Memory_height , self.Memory_width), name='main_memory')*0.042\n",
        "        init_wr = np.zeros((self.batch_size, self.read_heads, self.Memory_height))\n",
        "        init_wr[:,:,0] = 1\n",
        "        init_wr = K.variable(init_wr, name=\"init_weights_read\")\n",
        "        init_ww = np.zeros((self.batch_size, self.write_heads, self.Memory_height))\n",
        "        init_ww[:,:,0] = 1\n",
        "        init_ww = K.variable(init_ww, name=\"init_weights_write\")\n",
        "        return [init_old_ntm_output, init_M, init_wr, init_ww]\n",
        "\n",
        "    # Equation (2) of the paper\n",
        "    def _read_from_memory(self, weights, M):        \n",
        "        return K.sum((weights[:, :, None]*M),axis=1)\n",
        "\n",
        "    # Equation (3) of the paper\n",
        "    def _write_to_memory_erase(self, M, w, e):        \n",
        "        M_tilda = M * (1 - w[:, :, None]*e[:, None, :])\n",
        "        return M_tilda\n",
        "\n",
        "    # Equation (4) of the paper    \n",
        "    def _write_to_memory_add(self, M_tilda, w, a):\n",
        "        M_out = M_tilda + w[:, :, None]*a[:, None, :]\n",
        "        return M_out\n",
        "\n",
        "    # Figure 2 operations to produce the weight vector\n",
        "    def _get_weight_vector(self, M, w_tm1, k, beta, g, s, gamma):\n",
        "        ## Content-based addressing\n",
        "        num = beta * similarity_by_cosine_distance(M, k)\n",
        "        w_c  = K.softmax(num) # It turns out that equation (5) is just softmax.\n",
        "        ## Location-based addressing\n",
        "        # Equation 7:\n",
        "        w_g = (g * w_c) + (1-g)*w_tm1\n",
        "        # Equation 8:        \n",
        "        C_s = K.sum(K.repeat_elements(self.shift_tensor[None, :, :, :], self.batch_size, axis=0) * s[:,:,None,None], axis=1)\n",
        "        w_tilda = K.batch_dot(C_s, w_g)\n",
        "        # Equation 9:\n",
        "        w_out = normalize_weights(w_tilda ** gamma)\n",
        "\n",
        "        return w_out\n",
        "\n",
        "    def _run_controller(self, inputs, read_vector):\n",
        "        controller_input = K.concatenate([inputs, read_vector])\n",
        "        if self.controller_with_state or len(self.controller.input_shape) == 3:\n",
        "            controller_input = controller_input[:,None,:]\n",
        "            controller_output = self.controller.call(controller_input)\n",
        "            if self.controller.output_shape == 3:\n",
        "                controller_output = controller_output[:,0,:]\n",
        "        else:\n",
        "            controller_output = self.controller.call(controller_input)\n",
        "\n",
        "        return controller_output\n",
        "      \n",
        "    \"\"\" \n",
        "    takes the controller output and splits it into three parts:\n",
        "          ntm_output, read and write addressing \n",
        "    \n",
        "    returns : tensor ntm_output \n",
        "              list controller_instructions_read  : (k, beta, g, shift, gamma)\n",
        "              list controller_instructions_write : (k, beta, g, shift, gamma, add vector, erase vector)\n",
        "    \"\"\"\n",
        "    def _split_and_apply_activations(self, controller_output):        \n",
        "        # splitting\n",
        "        ntm_output, controller_instructions_read, controller_instructions_write = tf.split(\n",
        "                    controller_output,\n",
        "                    np.asarray([self.output_dim,\n",
        "                                self.read_heads * self.controller_read_head_emitting_dim,\n",
        "                                self.write_heads * self.controller_write_head_emitting_dim]),\n",
        "                    axis=1)\n",
        "\n",
        "        controller_instructions_read = tf.split(controller_instructions_read, self.read_heads, axis=1)\n",
        "        controller_instructions_write = tf.split(controller_instructions_write, self.write_heads, axis=1)\n",
        "\n",
        "        controller_instructions_read = [\n",
        "                tf.split(single_head_data, np.asarray([self.Memory_width, 1, 1, 3, 1]), axis=1) for \n",
        "                single_head_data in controller_instructions_read]\n",
        "        \n",
        "        controller_instructions_write = [\n",
        "                tf.split(single_head_data, np.asarray([self.Memory_width, 1, 1, 3, 1, self.Memory_width, self.Memory_width]), axis=1) for \n",
        "                single_head_data in controller_instructions_write]\n",
        "        \n",
        "        #activation\n",
        "        ntm_output = self.activation(ntm_output)\n",
        "\n",
        "        controller_instructions_read = [(k, tf.exp(beta), sigmoid(g), softmax(shift), 1 + tf.log(tf.exp(gamma)+1)) for\n",
        "                (k, beta, g, shift, gamma) in controller_instructions_read]\n",
        "        controller_instructions_write = [\n",
        "                (k, tf.exp(beta), sigmoid(g), softmax(shift), 1 + tf.log(tf.exp(gamma)+1), hard_sigmoid(erase_vector), tanh(add_vector))  for \n",
        "                (k, beta, g, shift, gamma, erase_vector, add_vector) in controller_instructions_write]\n",
        "       \n",
        "        return (ntm_output, controller_instructions_read, controller_instructions_write)\n",
        "\n",
        "\n",
        "    @property\n",
        "    def output_shape(self):\n",
        "        input_shape = self.input_shape\n",
        "        if self.return_sequences:\n",
        "            return input_shape[0], input_shape[1], self.output_dim\n",
        "        else:\n",
        "            return input_shape[0], self.output_dim\n",
        "\n",
        "    def step(self, layer_input, states):\n",
        "        _, M, weights_read_tm1, weights_write_tm1 = states[:4]\n",
        "        weights_read_tm1 = K.reshape(weights_read_tm1,\n",
        "                (self.batch_size, self.read_heads, self.Memory_height))\n",
        "        weights_write_tm1 = K.reshape(weights_write_tm1, \n",
        "                (self.batch_size, self.write_heads, self.Memory_height))\n",
        "        # Printing the read and write weights for  visualization purpose (Does not work in Google Colaboratory and Jupiter notebook)\n",
        "        for iii in range(0,128):\n",
        "                weights_write_tm1 = tf.Print(weights_write_tm1, [weights_write_tm1[0,0,iii]], message=\"weights_write_tm1\")   \n",
        "        weights_write_tm1 = tf.Print(weights_write_tm1, [], message=\"-----\")\n",
        "        for iii in range(0,128):\n",
        "                weights_read_tm1 = tf.Print(weights_read_tm1, [weights_read_tm1[0,0,iii]], message=\"weights_read_tm1\")   \n",
        "        weights_read_tm1 = tf.Print(weights_read_tm1, [], message=\"DONE-----\")\n",
        "        #Done printing\n",
        "        \n",
        "        memory_read_input = K.concatenate([self._read_from_memory(weights_read_tm1[:,i], M) for \n",
        "                i in range(self.read_heads)])\n",
        "\n",
        "        controller_output = self._run_controller(layer_input, memory_read_input)\n",
        "\n",
        "        ntm_output, controller_instructions_read, controller_instructions_write = \\\n",
        "                self._split_and_apply_activations(controller_output)\n",
        "\n",
        "        # addressing:\n",
        "        weights_write = []\n",
        "        for i in range(self.write_heads):\n",
        "            write_head = controller_instructions_write[i]\n",
        "            old_weight_vector = weights_write_tm1[:,i]\n",
        "            weight_vector = self._get_weight_vector(M, old_weight_vector, *tuple(write_head[:5]))\n",
        "            weights_write.append(weight_vector)\n",
        "            \n",
        "        # erasing:\n",
        "        for i in range(self.write_heads):\n",
        "            M = self._write_to_memory_erase(M, weights_write[i], controller_instructions_write[i][5])\n",
        "        \n",
        "        # adding:\n",
        "        for i in range(self.write_heads):\n",
        "            M = self._write_to_memory_add(M, weights_write[i], controller_instructions_write[i][6])\n",
        "\n",
        "        weights_read = []\n",
        "        for i in range(self.read_heads):\n",
        "            read_head = controller_instructions_read[i]\n",
        "            old_weight_vector = weights_read_tm1[:,i]\n",
        "            weight_vector = self._get_weight_vector(M, old_weight_vector, *read_head)\n",
        "            weights_read.append(weight_vector)\n",
        "        \n",
        "        #with sess.as_default():\n",
        "        #  self.wr = weights_read[0].eval()\n",
        "        #  self.ww = weights_write[0].eval()\n",
        "                \n",
        "        return ntm_output, [ntm_output, M, K.stack(weights_read, axis=1), K.stack(weights_write, axis=1)] \n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "E1Js5E95e_0I",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "## NTM Model Generation\n",
        "Below function will generate an **NTM** model using the define **NTM** class mentioned in the previous section."
      ]
    },
    {
      "metadata": {
        "id": "MTG-1jCaTWs_",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "Memory_height = 128\n",
        "Memory_width = 20\n",
        "learning_rate = 5e-4\n",
        "clipnorm = 10\n",
        "\n",
        "def gen_model(input_dim, batch_size, output_dim,\n",
        "                Memory_height=Memory_height,\n",
        "                Memory_width=Memory_width,\n",
        "                controller_model=None,\n",
        "                activation=\"sigmoid\",\n",
        "                read_heads = 1,\n",
        "                write_heads = 1):\n",
        "\n",
        "    model = Sequential()\n",
        "    model.name = \"NTM_-_\" + controller_model.name\n",
        "    model.batch_size = batch_size\n",
        "    model.input_dim = input_dim\n",
        "    model.output_dim = output_dim\n",
        "    \n",
        "    ntm = NeuralTuringMachine(output_dim, Memory_height=Memory_height, Memory_width=Memory_width, shift_range=3,\n",
        "              controller_model=controller_model,\n",
        "              activation=activation,\n",
        "              read_heads = read_heads,\n",
        "              write_heads = write_heads,\n",
        "              return_sequences=True,input_shape=(None, input_dim),\n",
        "              batch_size = batch_size)\n",
        "    model.add(ntm)\n",
        "    \n",
        "    sgd = Adam(lr=learning_rate, clipnorm=clipnorm)\n",
        "    model.compile(loss='binary_crossentropy', optimizer=sgd, metrics = ['binary_accuracy','categorical_accuracy','mse'], sample_weight_mode=\"temporal\")\n",
        "\n",
        "    return model"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "p0l3Zny5fDrw",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "##  Running NTM with LSTM\n",
        "Here we will create an **LSTM-NTM** model with the hyper-parameters specified below as well as the chart in the **Report** section under the **Hyper-parameter Values**. We will train the model for $200$ epochs. We then test our model on a sequence of different length of inputs. The result of which will be used in the **Report** section under **Generalization of The Model to Longer Sequences**."
      ]
    },
    {
      "metadata": {
        "id": "SYcQQ2TATbzv",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 8356
        },
        "outputId": "b5c2aae0-b25f-4a24-b4b3-432a37abf60c"
      },
      "cell_type": "code",
      "source": [
        "batch_size = 100\n",
        "read_heads = 1\n",
        "write_heads = 1\n",
        "epochs = 200\n",
        "verboose = 0\n",
        "clipnorm = 10 \n",
        "sameInit = RandomNormal(seed=0) # bias_initializer\n",
        "\n",
        "lr = 0.01 # 5e-4   #0.001 was also good\n",
        "#sgd = Adam(lr=lr, clipnorm=clipnorm, decay=0.5, amsgrad=True)\n",
        "sgd = SGD(lr=lr, clipnorm=clipnorm, decay=0.5)\n",
        "\n",
        "\n",
        "controller_input_dim, controller_output_dim = controller_input_output_shape(input_dim, output_dim, 20, 128, 3, read_heads,\n",
        "            write_heads)\n",
        "controller = Sequential()\n",
        "controller.name=\"NTM_LSTM\"\n",
        "     \n",
        "controller.add(LSTM(units=units,\n",
        "                                kernel_initializer='random_normal', \n",
        "                                bias_initializer='random_normal',\n",
        "                                activation='linear',\n",
        "                                stateful=True,\n",
        "                                implementation=1,   # best for gpu. other ones also might not work.\n",
        "                                batch_input_shape=(batch_size, None, controller_input_dim)))\n",
        "        \n",
        "controller.add(Dense(units=controller_output_dim,\n",
        "                                kernel_initializer=sameInit,\n",
        "                                bias_initializer=sameInit,\n",
        "                                activation='softmax'))\n",
        "        \n",
        "controller.compile(loss='binary_crossentropy', optimizer=sgd, metrics = ['binary_accuracy','categorical_accuracy','mse'], sample_weight_mode=\"temporal\")\n",
        "print(controller.summary())    \n",
        "model_ntm_lstm = gen_model(input_dim=input_dim, output_dim=output_dim, batch_size=batch_size,\n",
        "                                    controller_model=controller, read_heads=read_heads, write_heads=write_heads,\n",
        "                                    activation=\"softmax\")\n",
        "print(model_ntm_lstm.summary())\n",
        "print(\"model built, starting the copy experiment\")\n",
        "\n",
        "test_with_sequence_lengths(model_ntm_lstm, epochs=epochs, verboose=verboose)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "lstm_1 (LSTM)                (100, 100)                52000     \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (100, 101)                10201     \n",
            "=================================================================\n",
            "Total params: 62,201\n",
            "Trainable params: 62,201\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "None\n",
            "****\n",
            "Tensor(\"neural_turing_machine_1/mul_10:0\", shape=(100, 128, 20), dtype=float32)\n",
            "****\n",
            "Tensor(\"neural_turing_machine_1/while/mul_8:0\", shape=(100, 128, 20), dtype=float32)\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "neural_turing_machine_1 (Neu (None, None, 9)           62201     \n",
            "=================================================================\n",
            "Total params: 62,201\n",
            "Trainable params: 62,201\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "None\n",
            "model built, starting the copy experiment\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:65: UserWarning: The semantics of the Keras 2 argument `steps_per_epoch` is not the same as the Keras 1 argument `samples_per_epoch`. `steps_per_epoch` is the number of batches to draw from the generator at each epoch. Basically steps_per_epoch = samples_per_epoch/batch_size. Similarly `nb_val_samples`->`validation_steps` and `val_samples`->`steps` arguments have changed. Update your method calls accordingly.\n",
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:65: UserWarning: Update your `fit_generator` call to the Keras 2 API: `fit_generator(<generator..., steps_per_epoch=10, epochs=200, callbacks=[<keras.ca..., workers=1, max_queue_size=10)`\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n",
            "10/10 [==============================] - 10s 1s/step - loss: 0.5325 - binary_accuracy: 0.8006 - categorical_accuracy: 0.0662 - mean_squared_error: 0.1675\n",
            "Epoch 2/200\n",
            "10/10 [==============================] - 9s 933ms/step - loss: 0.5255 - binary_accuracy: 0.8039 - categorical_accuracy: 0.0647 - mean_squared_error: 0.1648\n",
            "Epoch 3/200\n",
            "10/10 [==============================] - 9s 939ms/step - loss: 0.5281 - binary_accuracy: 0.8027 - categorical_accuracy: 0.1421 - mean_squared_error: 0.1658\n",
            "Epoch 4/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.5249 - binary_accuracy: 0.8042 - categorical_accuracy: 0.2727 - mean_squared_error: 0.1646\n",
            "Epoch 5/200\n",
            "10/10 [==============================] - 9s 940ms/step - loss: 0.5262 - binary_accuracy: 0.8035 - categorical_accuracy: 0.5443 - mean_squared_error: 0.1651\n",
            "Epoch 6/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.5259 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7415 - mean_squared_error: 0.1650\n",
            "Epoch 7/200\n",
            "10/10 [==============================] - 9s 934ms/step - loss: 0.5252 - binary_accuracy: 0.8040 - categorical_accuracy: 0.7455 - mean_squared_error: 0.1648\n",
            "Epoch 8/200\n",
            "10/10 [==============================] - 9s 939ms/step - loss: 0.5254 - binary_accuracy: 0.8038 - categorical_accuracy: 0.7424 - mean_squared_error: 0.1649\n",
            "Epoch 9/200\n",
            "10/10 [==============================] - 9s 934ms/step - loss: 0.5181 - binary_accuracy: 0.8073 - categorical_accuracy: 0.7540 - mean_squared_error: 0.1622\n",
            "Epoch 10/200\n",
            "10/10 [==============================] - 9s 931ms/step - loss: 0.5215 - binary_accuracy: 0.8056 - categorical_accuracy: 0.7500 - mean_squared_error: 0.1634\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 11/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.5260 - binary_accuracy: 0.8033 - categorical_accuracy: 0.7444 - mean_squared_error: 0.1652\n",
            "Epoch 12/200\n",
            "10/10 [==============================] - 9s 933ms/step - loss: 0.5234 - binary_accuracy: 0.8043 - categorical_accuracy: 0.7437 - mean_squared_error: 0.1643\n",
            "Epoch 13/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.5236 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7444 - mean_squared_error: 0.1645\n",
            "Epoch 14/200\n",
            "10/10 [==============================] - 9s 929ms/step - loss: 0.5235 - binary_accuracy: 0.8028 - categorical_accuracy: 0.7412 - mean_squared_error: 0.1648\n",
            "Epoch 15/200\n",
            "10/10 [==============================] - 9s 938ms/step - loss: 0.5153 - binary_accuracy: 0.8047 - categorical_accuracy: 0.7466 - mean_squared_error: 0.1623\n",
            "Epoch 16/200\n",
            "10/10 [==============================] - 9s 933ms/step - loss: 0.5111 - binary_accuracy: 0.8036 - categorical_accuracy: 0.7415 - mean_squared_error: 0.1615\n",
            "Epoch 17/200\n",
            "10/10 [==============================] - 9s 934ms/step - loss: 0.5066 - binary_accuracy: 0.8025 - categorical_accuracy: 0.7422 - mean_squared_error: 0.1606\n",
            "Epoch 18/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4975 - binary_accuracy: 0.8044 - categorical_accuracy: 0.7459 - mean_squared_error: 0.1578\n",
            "Epoch 19/200\n",
            "10/10 [==============================] - 9s 937ms/step - loss: 0.4962 - binary_accuracy: 0.8025 - categorical_accuracy: 0.7397 - mean_squared_error: 0.1579\n",
            "Epoch 20/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 0.4930 - binary_accuracy: 0.8017 - categorical_accuracy: 0.7377 - mean_squared_error: 0.1573"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 9s 926ms/step - loss: 0.4921 - binary_accuracy: 0.8020 - categorical_accuracy: 0.7391 - mean_squared_error: 0.1570\n",
            "Epoch 21/200\n",
            "10/10 [==============================] - 9s 926ms/step - loss: 0.4832 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7436 - mean_squared_error: 0.1542\n",
            "Epoch 22/200\n",
            "10/10 [==============================] - 9s 931ms/step - loss: 0.4768 - binary_accuracy: 0.8042 - categorical_accuracy: 0.7448 - mean_squared_error: 0.1522\n",
            "Epoch 23/200\n",
            "10/10 [==============================] - 9s 929ms/step - loss: 0.4735 - binary_accuracy: 0.8040 - categorical_accuracy: 0.7436 - mean_squared_error: 0.1510\n",
            "Epoch 24/200\n",
            "10/10 [==============================] - 9s 935ms/step - loss: 0.4732 - binary_accuracy: 0.8030 - categorical_accuracy: 0.7429 - mean_squared_error: 0.1509\n",
            "Epoch 25/200\n",
            "10/10 [==============================] - 9s 933ms/step - loss: 0.4628 - binary_accuracy: 0.8066 - categorical_accuracy: 0.7526 - mean_squared_error: 0.1471\n",
            "Epoch 26/200\n",
            "10/10 [==============================] - 9s 931ms/step - loss: 0.4688 - binary_accuracy: 0.8038 - categorical_accuracy: 0.7462 - mean_squared_error: 0.1492\n",
            "Epoch 27/200\n",
            "10/10 [==============================] - 9s 941ms/step - loss: 0.4700 - binary_accuracy: 0.8033 - categorical_accuracy: 0.7420 - mean_squared_error: 0.1495\n",
            "Epoch 28/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4641 - binary_accuracy: 0.8055 - categorical_accuracy: 0.7477 - mean_squared_error: 0.1473\n",
            "Epoch 29/200\n",
            " 3/10 [========>.....................] - ETA: 6s - loss: 0.4698 - binary_accuracy: 0.8030 - categorical_accuracy: 0.7445 - mean_squared_error: 0.1493"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 9s 943ms/step - loss: 0.4688 - binary_accuracy: 0.8036 - categorical_accuracy: 0.7431 - mean_squared_error: 0.1489\n",
            "Epoch 30/200\n",
            "10/10 [==============================] - 9s 942ms/step - loss: 0.4724 - binary_accuracy: 0.8020 - categorical_accuracy: 0.7409 - mean_squared_error: 0.1502\n",
            "Epoch 31/200\n",
            "10/10 [==============================] - 9s 937ms/step - loss: 0.4682 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7436 - mean_squared_error: 0.1487\n",
            "Epoch 32/200\n",
            "10/10 [==============================] - 9s 931ms/step - loss: 0.4844 - binary_accuracy: 0.7973 - categorical_accuracy: 0.7265 - mean_squared_error: 0.1543\n",
            "Epoch 33/200\n",
            "10/10 [==============================] - 9s 928ms/step - loss: 0.4697 - binary_accuracy: 0.8030 - categorical_accuracy: 0.7420 - mean_squared_error: 0.1491\n",
            "Epoch 34/200\n",
            "10/10 [==============================] - 9s 930ms/step - loss: 0.4630 - binary_accuracy: 0.8055 - categorical_accuracy: 0.7509 - mean_squared_error: 0.1468\n",
            "Epoch 35/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4704 - binary_accuracy: 0.8027 - categorical_accuracy: 0.7415 - mean_squared_error: 0.1494\n",
            "Epoch 36/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4658 - binary_accuracy: 0.8045 - categorical_accuracy: 0.7452 - mean_squared_error: 0.1477\n",
            "Epoch 37/200\n",
            "10/10 [==============================] - 9s 934ms/step - loss: 0.4719 - binary_accuracy: 0.8020 - categorical_accuracy: 0.7404 - mean_squared_error: 0.1499\n",
            "Epoch 38/200\n",
            "10/10 [==============================] - 9s 938ms/step - loss: 0.4673 - binary_accuracy: 0.8038 - categorical_accuracy: 0.7450 - mean_squared_error: 0.1483\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 39/200\n",
            "10/10 [==============================] - 9s 935ms/step - loss: 0.4642 - binary_accuracy: 0.8051 - categorical_accuracy: 0.7465 - mean_squared_error: 0.1472\n",
            "Epoch 40/200\n",
            "10/10 [==============================] - 9s 934ms/step - loss: 0.4627 - binary_accuracy: 0.8056 - categorical_accuracy: 0.7495 - mean_squared_error: 0.1466\n",
            "Epoch 41/200\n",
            "10/10 [==============================] - 9s 934ms/step - loss: 0.4723 - binary_accuracy: 0.8019 - categorical_accuracy: 0.7399 - mean_squared_error: 0.1499\n",
            "Epoch 42/200\n",
            "10/10 [==============================] - 9s 933ms/step - loss: 0.4680 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7445 - mean_squared_error: 0.1485\n",
            "Epoch 43/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4641 - binary_accuracy: 0.8050 - categorical_accuracy: 0.7472 - mean_squared_error: 0.1471\n",
            "Epoch 44/200\n",
            "10/10 [==============================] - 9s 937ms/step - loss: 0.4696 - binary_accuracy: 0.8030 - categorical_accuracy: 0.7412 - mean_squared_error: 0.1490\n",
            "Epoch 45/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4659 - binary_accuracy: 0.8043 - categorical_accuracy: 0.7460 - mean_squared_error: 0.1477\n",
            "Epoch 46/200\n",
            "10/10 [==============================] - 9s 931ms/step - loss: 0.4730 - binary_accuracy: 0.8016 - categorical_accuracy: 0.7378 - mean_squared_error: 0.1502\n",
            "Epoch 47/200\n",
            "10/10 [==============================] - 9s 937ms/step - loss: 0.4671 - binary_accuracy: 0.8038 - categorical_accuracy: 0.7443 - mean_squared_error: 0.1481\n",
            "Epoch 48/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 0.4678 - binary_accuracy: 0.8036 - categorical_accuracy: 0.7432 - mean_squared_error: 0.1483"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 9s 938ms/step - loss: 0.4687 - binary_accuracy: 0.8033 - categorical_accuracy: 0.7413 - mean_squared_error: 0.1486\n",
            "Epoch 49/200\n",
            "10/10 [==============================] - 9s 937ms/step - loss: 0.4713 - binary_accuracy: 0.8022 - categorical_accuracy: 0.7404 - mean_squared_error: 0.1496\n",
            "Epoch 50/200\n",
            "10/10 [==============================] - 9s 935ms/step - loss: 0.4686 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7434 - mean_squared_error: 0.1486\n",
            "Epoch 51/200\n",
            "10/10 [==============================] - 9s 939ms/step - loss: 0.4729 - binary_accuracy: 0.8016 - categorical_accuracy: 0.7379 - mean_squared_error: 0.1501\n",
            "Epoch 52/200\n",
            "10/10 [==============================] - 9s 938ms/step - loss: 0.4565 - binary_accuracy: 0.8079 - categorical_accuracy: 0.7563 - mean_squared_error: 0.1444\n",
            "Epoch 53/200\n",
            "10/10 [==============================] - 9s 939ms/step - loss: 0.4660 - binary_accuracy: 0.8043 - categorical_accuracy: 0.7455 - mean_squared_error: 0.1477\n",
            "Epoch 54/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4649 - binary_accuracy: 0.8047 - categorical_accuracy: 0.7456 - mean_squared_error: 0.1473\n",
            "Epoch 55/200\n",
            "10/10 [==============================] - 9s 930ms/step - loss: 0.4622 - binary_accuracy: 0.8058 - categorical_accuracy: 0.7477 - mean_squared_error: 0.1464\n",
            "Epoch 56/200\n",
            "10/10 [==============================] - 9s 934ms/step - loss: 0.4712 - binary_accuracy: 0.8021 - categorical_accuracy: 0.7420 - mean_squared_error: 0.1495\n",
            "Epoch 57/200\n",
            " 3/10 [========>.....................] - ETA: 6s - loss: 0.4733 - binary_accuracy: 0.8013 - categorical_accuracy: 0.7410 - mean_squared_error: 0.1503"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 9s 941ms/step - loss: 0.4678 - binary_accuracy: 0.8036 - categorical_accuracy: 0.7434 - mean_squared_error: 0.1483\n",
            "Epoch 58/200\n",
            "10/10 [==============================] - 9s 940ms/step - loss: 0.4683 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7436 - mean_squared_error: 0.1485\n",
            "Epoch 59/200\n",
            "10/10 [==============================] - 9s 931ms/step - loss: 0.4715 - binary_accuracy: 0.8022 - categorical_accuracy: 0.7396 - mean_squared_error: 0.1496\n",
            "Epoch 60/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4671 - binary_accuracy: 0.8038 - categorical_accuracy: 0.7438 - mean_squared_error: 0.1481\n",
            "Epoch 61/200\n",
            "10/10 [==============================] - 9s 931ms/step - loss: 0.4704 - binary_accuracy: 0.8025 - categorical_accuracy: 0.7419 - mean_squared_error: 0.1492\n",
            "Epoch 62/200\n",
            "10/10 [==============================] - 9s 949ms/step - loss: 0.4735 - binary_accuracy: 0.8014 - categorical_accuracy: 0.7378 - mean_squared_error: 0.1503\n",
            "Epoch 63/200\n",
            "10/10 [==============================] - 9s 937ms/step - loss: 0.4648 - binary_accuracy: 0.8046 - categorical_accuracy: 0.7474 - mean_squared_error: 0.1473\n",
            "Epoch 64/200\n",
            "10/10 [==============================] - 9s 939ms/step - loss: 0.4625 - binary_accuracy: 0.8054 - categorical_accuracy: 0.7517 - mean_squared_error: 0.1465\n",
            "Epoch 65/200\n",
            "10/10 [==============================] - 9s 928ms/step - loss: 0.4656 - binary_accuracy: 0.8044 - categorical_accuracy: 0.7456 - mean_squared_error: 0.1476\n",
            "Epoch 66/200\n",
            "10/10 [==============================] - 9s 938ms/step - loss: 0.4676 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7446 - mean_squared_error: 0.1483\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 67/200\n",
            "10/10 [==============================] - 9s 938ms/step - loss: 0.4782 - binary_accuracy: 0.7996 - categorical_accuracy: 0.7321 - mean_squared_error: 0.1519\n",
            "Epoch 68/200\n",
            "10/10 [==============================] - 9s 938ms/step - loss: 0.4741 - binary_accuracy: 0.8010 - categorical_accuracy: 0.7384 - mean_squared_error: 0.1505\n",
            "Epoch 69/200\n",
            "10/10 [==============================] - 9s 940ms/step - loss: 0.4652 - binary_accuracy: 0.8045 - categorical_accuracy: 0.7460 - mean_squared_error: 0.1474\n",
            "Epoch 70/200\n",
            "10/10 [==============================] - 9s 928ms/step - loss: 0.4720 - binary_accuracy: 0.8021 - categorical_accuracy: 0.7366 - mean_squared_error: 0.1497\n",
            "Epoch 71/200\n",
            "10/10 [==============================] - 9s 931ms/step - loss: 0.4669 - binary_accuracy: 0.8038 - categorical_accuracy: 0.7454 - mean_squared_error: 0.1480\n",
            "Epoch 72/200\n",
            "10/10 [==============================] - 9s 933ms/step - loss: 0.4698 - binary_accuracy: 0.8028 - categorical_accuracy: 0.7412 - mean_squared_error: 0.1490\n",
            "Epoch 73/200\n",
            "10/10 [==============================] - 9s 935ms/step - loss: 0.4629 - binary_accuracy: 0.8053 - categorical_accuracy: 0.7495 - mean_squared_error: 0.1466\n",
            "Epoch 74/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4656 - binary_accuracy: 0.8044 - categorical_accuracy: 0.7456 - mean_squared_error: 0.1475\n",
            "Epoch 75/200\n",
            "10/10 [==============================] - 9s 935ms/step - loss: 0.4715 - binary_accuracy: 0.8021 - categorical_accuracy: 0.7397 - mean_squared_error: 0.1496\n",
            "Epoch 76/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 0.4685 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7433 - mean_squared_error: 0.1486"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 9s 934ms/step - loss: 0.4678 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7436 - mean_squared_error: 0.1483\n",
            "Epoch 77/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4723 - binary_accuracy: 0.8018 - categorical_accuracy: 0.7388 - mean_squared_error: 0.1498\n",
            "Epoch 78/200\n",
            "10/10 [==============================] - 9s 933ms/step - loss: 0.4696 - binary_accuracy: 0.8029 - categorical_accuracy: 0.7409 - mean_squared_error: 0.1489\n",
            "Epoch 79/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4701 - binary_accuracy: 0.8026 - categorical_accuracy: 0.7422 - mean_squared_error: 0.1491\n",
            "Epoch 80/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4668 - binary_accuracy: 0.8040 - categorical_accuracy: 0.7439 - mean_squared_error: 0.1479\n",
            "Epoch 81/200\n",
            "10/10 [==============================] - 9s 931ms/step - loss: 0.4798 - binary_accuracy: 0.7990 - categorical_accuracy: 0.7309 - mean_squared_error: 0.1524\n",
            "Epoch 82/200\n",
            "10/10 [==============================] - 9s 937ms/step - loss: 0.4653 - binary_accuracy: 0.8043 - categorical_accuracy: 0.7483 - mean_squared_error: 0.1474\n",
            "Epoch 83/200\n",
            "10/10 [==============================] - 9s 937ms/step - loss: 0.4686 - binary_accuracy: 0.8031 - categorical_accuracy: 0.7439 - mean_squared_error: 0.1486\n",
            "Epoch 84/200\n",
            "10/10 [==============================] - 9s 925ms/step - loss: 0.4709 - binary_accuracy: 0.8023 - categorical_accuracy: 0.7406 - mean_squared_error: 0.1494\n",
            "Epoch 85/200\n",
            " 3/10 [========>.....................] - ETA: 6s - loss: 0.4705 - binary_accuracy: 0.8024 - categorical_accuracy: 0.7427 - mean_squared_error: 0.1492"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4716 - binary_accuracy: 0.8020 - categorical_accuracy: 0.7405 - mean_squared_error: 0.1496\n",
            "Epoch 86/200\n",
            "10/10 [==============================] - 9s 928ms/step - loss: 0.4733 - binary_accuracy: 0.8014 - categorical_accuracy: 0.7385 - mean_squared_error: 0.1502\n",
            "Epoch 87/200\n",
            "10/10 [==============================] - 9s 939ms/step - loss: 0.4650 - binary_accuracy: 0.8045 - categorical_accuracy: 0.7468 - mean_squared_error: 0.1473\n",
            "Epoch 88/200\n",
            "10/10 [==============================] - 9s 941ms/step - loss: 0.4735 - binary_accuracy: 0.8013 - categorical_accuracy: 0.7376 - mean_squared_error: 0.1503\n",
            "Epoch 89/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4682 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7419 - mean_squared_error: 0.1484\n",
            "Epoch 90/200\n",
            "10/10 [==============================] - 9s 943ms/step - loss: 0.4688 - binary_accuracy: 0.8031 - categorical_accuracy: 0.7425 - mean_squared_error: 0.1486\n",
            "Epoch 91/200\n",
            "10/10 [==============================] - 9s 933ms/step - loss: 0.4660 - binary_accuracy: 0.8042 - categorical_accuracy: 0.7453 - mean_squared_error: 0.1477\n",
            "Epoch 92/200\n",
            "10/10 [==============================] - 9s 935ms/step - loss: 0.4748 - binary_accuracy: 0.8008 - categorical_accuracy: 0.7375 - mean_squared_error: 0.1507\n",
            "Epoch 93/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4679 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7446 - mean_squared_error: 0.1483\n",
            "Epoch 94/200\n",
            "10/10 [==============================] - 9s 943ms/step - loss: 0.4694 - binary_accuracy: 0.8029 - categorical_accuracy: 0.7420 - mean_squared_error: 0.1488\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 95/200\n",
            "10/10 [==============================] - 9s 938ms/step - loss: 0.4770 - binary_accuracy: 0.8000 - categorical_accuracy: 0.7342 - mean_squared_error: 0.1515\n",
            "Epoch 96/200\n",
            "10/10 [==============================] - 9s 930ms/step - loss: 0.4650 - binary_accuracy: 0.8045 - categorical_accuracy: 0.7472 - mean_squared_error: 0.1473\n",
            "Epoch 97/200\n",
            "10/10 [==============================] - 9s 930ms/step - loss: 0.4659 - binary_accuracy: 0.8041 - categorical_accuracy: 0.7468 - mean_squared_error: 0.1477\n",
            "Epoch 98/200\n",
            "10/10 [==============================] - 9s 935ms/step - loss: 0.4657 - binary_accuracy: 0.8043 - categorical_accuracy: 0.7458 - mean_squared_error: 0.1476\n",
            "Epoch 99/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4667 - binary_accuracy: 0.8039 - categorical_accuracy: 0.7450 - mean_squared_error: 0.1479\n",
            "Epoch 100/200\n",
            "10/10 [==============================] - 9s 929ms/step - loss: 0.4580 - binary_accuracy: 0.8073 - categorical_accuracy: 0.7525 - mean_squared_error: 0.1449\n",
            "Epoch 101/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4756 - binary_accuracy: 0.8005 - categorical_accuracy: 0.7361 - mean_squared_error: 0.1510\n",
            "Epoch 102/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4673 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7447 - mean_squared_error: 0.1481\n",
            "Epoch 103/200\n",
            "10/10 [==============================] - 9s 934ms/step - loss: 0.4706 - binary_accuracy: 0.8024 - categorical_accuracy: 0.7403 - mean_squared_error: 0.1492\n",
            "Epoch 104/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 0.4682 - binary_accuracy: 0.8033 - categorical_accuracy: 0.7437 - mean_squared_error: 0.1484"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 9s 933ms/step - loss: 0.4666 - binary_accuracy: 0.8040 - categorical_accuracy: 0.7449 - mean_squared_error: 0.1479\n",
            "Epoch 105/200\n",
            "10/10 [==============================] - 9s 935ms/step - loss: 0.4662 - binary_accuracy: 0.8041 - categorical_accuracy: 0.7452 - mean_squared_error: 0.1477\n",
            "Epoch 106/200\n",
            "10/10 [==============================] - 9s 935ms/step - loss: 0.4684 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7438 - mean_squared_error: 0.1485\n",
            "Epoch 107/200\n",
            "10/10 [==============================] - 9s 934ms/step - loss: 0.4602 - binary_accuracy: 0.8064 - categorical_accuracy: 0.7517 - mean_squared_error: 0.1457\n",
            "Epoch 108/200\n",
            "10/10 [==============================] - 9s 939ms/step - loss: 0.4636 - binary_accuracy: 0.8051 - categorical_accuracy: 0.7482 - mean_squared_error: 0.1468\n",
            "Epoch 109/200\n",
            "10/10 [==============================] - 9s 934ms/step - loss: 0.4668 - binary_accuracy: 0.8039 - categorical_accuracy: 0.7442 - mean_squared_error: 0.1479\n",
            "Epoch 110/200\n",
            "10/10 [==============================] - 9s 931ms/step - loss: 0.4674 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7439 - mean_squared_error: 0.1481\n",
            "Epoch 111/200\n",
            "10/10 [==============================] - 9s 934ms/step - loss: 0.4633 - binary_accuracy: 0.8052 - categorical_accuracy: 0.7486 - mean_squared_error: 0.1467\n",
            "Epoch 112/200\n",
            "10/10 [==============================] - 9s 937ms/step - loss: 0.4683 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7421 - mean_squared_error: 0.1484\n",
            "Epoch 113/200\n",
            " 3/10 [========>.....................] - ETA: 6s - loss: 0.4678 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7447 - mean_squared_error: 0.1483"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 9s 937ms/step - loss: 0.4647 - binary_accuracy: 0.8046 - categorical_accuracy: 0.7475 - mean_squared_error: 0.1472\n",
            "Epoch 114/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4737 - binary_accuracy: 0.8012 - categorical_accuracy: 0.7380 - mean_squared_error: 0.1503\n",
            "Epoch 115/200\n",
            "10/10 [==============================] - 9s 939ms/step - loss: 0.4707 - binary_accuracy: 0.8025 - categorical_accuracy: 0.7385 - mean_squared_error: 0.1493\n",
            "Epoch 116/200\n",
            "10/10 [==============================] - 9s 941ms/step - loss: 0.4660 - binary_accuracy: 0.8041 - categorical_accuracy: 0.7464 - mean_squared_error: 0.1477\n",
            "Epoch 117/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4746 - binary_accuracy: 0.8011 - categorical_accuracy: 0.7339 - mean_squared_error: 0.1506\n",
            "Epoch 118/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4668 - binary_accuracy: 0.8040 - categorical_accuracy: 0.7425 - mean_squared_error: 0.1479\n",
            "Epoch 119/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4719 - binary_accuracy: 0.8019 - categorical_accuracy: 0.7405 - mean_squared_error: 0.1497\n",
            "Epoch 120/200\n",
            "10/10 [==============================] - 9s 933ms/step - loss: 0.4706 - binary_accuracy: 0.8024 - categorical_accuracy: 0.7404 - mean_squared_error: 0.1492\n",
            "Epoch 121/200\n",
            "10/10 [==============================] - 9s 931ms/step - loss: 0.4563 - binary_accuracy: 0.8078 - categorical_accuracy: 0.7560 - mean_squared_error: 0.1443\n",
            "Epoch 122/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4615 - binary_accuracy: 0.8058 - categorical_accuracy: 0.7503 - mean_squared_error: 0.1461\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 123/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4672 - binary_accuracy: 0.8036 - categorical_accuracy: 0.7464 - mean_squared_error: 0.1481\n",
            "Epoch 124/200\n",
            "10/10 [==============================] - 9s 935ms/step - loss: 0.4642 - binary_accuracy: 0.8049 - categorical_accuracy: 0.7466 - mean_squared_error: 0.1470\n",
            "Epoch 125/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4731 - binary_accuracy: 0.8013 - categorical_accuracy: 0.7400 - mean_squared_error: 0.1501\n",
            "Epoch 126/200\n",
            "10/10 [==============================] - 9s 946ms/step - loss: 0.4662 - binary_accuracy: 0.8040 - categorical_accuracy: 0.7462 - mean_squared_error: 0.1477\n",
            "Epoch 127/200\n",
            "10/10 [==============================] - 9s 940ms/step - loss: 0.4720 - binary_accuracy: 0.8020 - categorical_accuracy: 0.7378 - mean_squared_error: 0.1497\n",
            "Epoch 128/200\n",
            "10/10 [==============================] - 9s 934ms/step - loss: 0.4726 - binary_accuracy: 0.8016 - categorical_accuracy: 0.7389 - mean_squared_error: 0.1499\n",
            "Epoch 129/200\n",
            "10/10 [==============================] - 9s 933ms/step - loss: 0.4553 - binary_accuracy: 0.8082 - categorical_accuracy: 0.7571 - mean_squared_error: 0.1440\n",
            "Epoch 130/200\n",
            "10/10 [==============================] - 9s 933ms/step - loss: 0.4651 - binary_accuracy: 0.8045 - categorical_accuracy: 0.7460 - mean_squared_error: 0.1474\n",
            "Epoch 131/200\n",
            "10/10 [==============================] - 9s 946ms/step - loss: 0.4728 - binary_accuracy: 0.8015 - categorical_accuracy: 0.7399 - mean_squared_error: 0.1500\n",
            "Epoch 132/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 0.4600 - binary_accuracy: 0.8064 - categorical_accuracy: 0.7525 - mean_squared_error: 0.1456"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 9s 934ms/step - loss: 0.4606 - binary_accuracy: 0.8062 - categorical_accuracy: 0.7519 - mean_squared_error: 0.1458\n",
            "Epoch 133/200\n",
            "10/10 [==============================] - 9s 938ms/step - loss: 0.4685 - binary_accuracy: 0.8033 - categorical_accuracy: 0.7412 - mean_squared_error: 0.1485\n",
            "Epoch 134/200\n",
            "10/10 [==============================] - 9s 934ms/step - loss: 0.4628 - binary_accuracy: 0.8054 - categorical_accuracy: 0.7487 - mean_squared_error: 0.1465\n",
            "Epoch 135/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4711 - binary_accuracy: 0.8023 - categorical_accuracy: 0.7392 - mean_squared_error: 0.1494\n",
            "Epoch 136/200\n",
            "10/10 [==============================] - 9s 939ms/step - loss: 0.4697 - binary_accuracy: 0.8026 - categorical_accuracy: 0.7433 - mean_squared_error: 0.1490\n",
            "Epoch 137/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4668 - binary_accuracy: 0.8039 - categorical_accuracy: 0.7444 - mean_squared_error: 0.1479\n",
            "Epoch 138/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4692 - binary_accuracy: 0.8029 - categorical_accuracy: 0.7428 - mean_squared_error: 0.1488\n",
            "Epoch 139/200\n",
            "10/10 [==============================] - 9s 930ms/step - loss: 0.4640 - binary_accuracy: 0.8049 - categorical_accuracy: 0.7476 - mean_squared_error: 0.1469\n",
            "Epoch 140/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4690 - binary_accuracy: 0.8030 - categorical_accuracy: 0.7427 - mean_squared_error: 0.1487\n",
            "Epoch 141/200\n",
            " 3/10 [========>.....................] - ETA: 6s - loss: 0.4574 - binary_accuracy: 0.8075 - categorical_accuracy: 0.7522 - mean_squared_error: 0.1447"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 9s 927ms/step - loss: 0.4661 - binary_accuracy: 0.8040 - categorical_accuracy: 0.7463 - mean_squared_error: 0.1477\n",
            "Epoch 142/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4645 - binary_accuracy: 0.8048 - categorical_accuracy: 0.7453 - mean_squared_error: 0.1471\n",
            "Epoch 143/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4680 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7425 - mean_squared_error: 0.1483\n",
            "Epoch 144/200\n",
            "10/10 [==============================] - 9s 939ms/step - loss: 0.4686 - binary_accuracy: 0.8031 - categorical_accuracy: 0.7438 - mean_squared_error: 0.1486\n",
            "Epoch 145/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4645 - binary_accuracy: 0.8047 - categorical_accuracy: 0.7480 - mean_squared_error: 0.1471\n",
            "Epoch 146/200\n",
            "10/10 [==============================] - 9s 934ms/step - loss: 0.4629 - binary_accuracy: 0.8053 - categorical_accuracy: 0.7494 - mean_squared_error: 0.1466\n",
            "Epoch 147/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4649 - binary_accuracy: 0.8046 - categorical_accuracy: 0.7459 - mean_squared_error: 0.1472\n",
            "Epoch 148/200\n",
            "10/10 [==============================] - 9s 931ms/step - loss: 0.4660 - binary_accuracy: 0.8041 - categorical_accuracy: 0.7462 - mean_squared_error: 0.1476\n",
            "Epoch 149/200\n",
            "10/10 [==============================] - 9s 939ms/step - loss: 0.4626 - binary_accuracy: 0.8054 - categorical_accuracy: 0.7492 - mean_squared_error: 0.1465\n",
            "Epoch 150/200\n",
            "10/10 [==============================] - 9s 928ms/step - loss: 0.4658 - binary_accuracy: 0.8042 - categorical_accuracy: 0.7469 - mean_squared_error: 0.1476\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 151/200\n",
            "10/10 [==============================] - 9s 939ms/step - loss: 0.4706 - binary_accuracy: 0.8025 - categorical_accuracy: 0.7394 - mean_squared_error: 0.1492\n",
            "Epoch 152/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4784 - binary_accuracy: 0.7995 - categorical_accuracy: 0.7322 - mean_squared_error: 0.1519\n",
            "Epoch 153/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4700 - binary_accuracy: 0.8027 - categorical_accuracy: 0.7403 - mean_squared_error: 0.1490\n",
            "Epoch 154/200\n",
            "10/10 [==============================] - 9s 930ms/step - loss: 0.4655 - binary_accuracy: 0.8044 - categorical_accuracy: 0.7461 - mean_squared_error: 0.1475\n",
            "Epoch 155/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4643 - binary_accuracy: 0.8048 - categorical_accuracy: 0.7475 - mean_squared_error: 0.1471\n",
            "Epoch 156/200\n",
            "10/10 [==============================] - 9s 938ms/step - loss: 0.4680 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7412 - mean_squared_error: 0.1483\n",
            "Epoch 157/200\n",
            "10/10 [==============================] - 9s 940ms/step - loss: 0.4656 - binary_accuracy: 0.8044 - categorical_accuracy: 0.7453 - mean_squared_error: 0.1475\n",
            "Epoch 158/200\n",
            "10/10 [==============================] - 10s 951ms/step - loss: 0.4724 - binary_accuracy: 0.8017 - categorical_accuracy: 0.7394 - mean_squared_error: 0.1499\n",
            "Epoch 159/200\n",
            "10/10 [==============================] - 9s 940ms/step - loss: 0.4677 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7433 - mean_squared_error: 0.1482\n",
            "Epoch 160/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 0.4682 - binary_accuracy: 0.8033 - categorical_accuracy: 0.7445 - mean_squared_error: 0.1484"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 9s 935ms/step - loss: 0.4696 - binary_accuracy: 0.8027 - categorical_accuracy: 0.7427 - mean_squared_error: 0.1489\n",
            "Epoch 161/200\n",
            "10/10 [==============================] - 9s 935ms/step - loss: 0.4737 - binary_accuracy: 0.8012 - categorical_accuracy: 0.7373 - mean_squared_error: 0.1503\n",
            "Epoch 162/200\n",
            "10/10 [==============================] - 9s 931ms/step - loss: 0.4710 - binary_accuracy: 0.8023 - categorical_accuracy: 0.7395 - mean_squared_error: 0.1494\n",
            "Epoch 163/200\n",
            "10/10 [==============================] - 9s 931ms/step - loss: 0.4693 - binary_accuracy: 0.8030 - categorical_accuracy: 0.7414 - mean_squared_error: 0.1488\n",
            "Epoch 164/200\n",
            "10/10 [==============================] - 9s 933ms/step - loss: 0.4657 - binary_accuracy: 0.8042 - categorical_accuracy: 0.7463 - mean_squared_error: 0.1475\n",
            "Epoch 165/200\n",
            "10/10 [==============================] - 9s 941ms/step - loss: 0.4672 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7438 - mean_squared_error: 0.1481\n",
            "Epoch 166/200\n",
            "10/10 [==============================] - 9s 929ms/step - loss: 0.4710 - binary_accuracy: 0.8022 - categorical_accuracy: 0.7409 - mean_squared_error: 0.1494\n",
            "Epoch 167/200\n",
            "10/10 [==============================] - 9s 933ms/step - loss: 0.4640 - binary_accuracy: 0.8048 - categorical_accuracy: 0.7496 - mean_squared_error: 0.1470\n",
            "Epoch 168/200\n",
            "10/10 [==============================] - 9s 936ms/step - loss: 0.4689 - binary_accuracy: 0.8031 - categorical_accuracy: 0.7411 - mean_squared_error: 0.1486\n",
            "Epoch 169/200\n",
            " 3/10 [========>.....................] - ETA: 6s - loss: 0.4600 - binary_accuracy: 0.8065 - categorical_accuracy: 0.7514 - mean_squared_error: 0.1456"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 9s 921ms/step - loss: 0.4687 - binary_accuracy: 0.8031 - categorical_accuracy: 0.7428 - mean_squared_error: 0.1486\n",
            "Epoch 170/200\n",
            "10/10 [==============================] - 9s 937ms/step - loss: 0.4647 - binary_accuracy: 0.8046 - categorical_accuracy: 0.7479 - mean_squared_error: 0.1472\n",
            "Epoch 171/200\n",
            "10/10 [==============================] - 9s 931ms/step - loss: 0.4751 - binary_accuracy: 0.8008 - categorical_accuracy: 0.7348 - mean_squared_error: 0.1508\n",
            "Epoch 172/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4601 - binary_accuracy: 0.8064 - categorical_accuracy: 0.7524 - mean_squared_error: 0.1456\n",
            "Epoch 173/200\n",
            "10/10 [==============================] - 9s 926ms/step - loss: 0.4655 - binary_accuracy: 0.8044 - categorical_accuracy: 0.7461 - mean_squared_error: 0.1475\n",
            "Epoch 174/200\n",
            "10/10 [==============================] - 9s 934ms/step - loss: 0.4618 - binary_accuracy: 0.8056 - categorical_accuracy: 0.7515 - mean_squared_error: 0.1462\n",
            "Epoch 175/200\n",
            "10/10 [==============================] - 9s 934ms/step - loss: 0.4682 - binary_accuracy: 0.8033 - categorical_accuracy: 0.7430 - mean_squared_error: 0.1484\n",
            "Epoch 176/200\n",
            "10/10 [==============================] - 9s 931ms/step - loss: 0.4644 - binary_accuracy: 0.8047 - categorical_accuracy: 0.7486 - mean_squared_error: 0.1471\n",
            "Epoch 177/200\n",
            "10/10 [==============================] - 9s 929ms/step - loss: 0.4677 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7434 - mean_squared_error: 0.1482\n",
            "Epoch 178/200\n",
            "10/10 [==============================] - 9s 926ms/step - loss: 0.4719 - binary_accuracy: 0.8019 - categorical_accuracy: 0.7396 - mean_squared_error: 0.1497\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 179/200\n",
            "10/10 [==============================] - 9s 938ms/step - loss: 0.4699 - binary_accuracy: 0.8027 - categorical_accuracy: 0.7407 - mean_squared_error: 0.1490\n",
            "Epoch 180/200\n",
            "10/10 [==============================] - 9s 917ms/step - loss: 0.4675 - binary_accuracy: 0.8036 - categorical_accuracy: 0.7436 - mean_squared_error: 0.1481\n",
            "Epoch 181/200\n",
            "10/10 [==============================] - 8s 830ms/step - loss: 0.4597 - binary_accuracy: 0.8066 - categorical_accuracy: 0.7519 - mean_squared_error: 0.1455\n",
            "Epoch 182/200\n",
            "10/10 [==============================] - 8s 845ms/step - loss: 0.4735 - binary_accuracy: 0.8014 - categorical_accuracy: 0.7358 - mean_squared_error: 0.1502\n",
            "Epoch 183/200\n",
            "10/10 [==============================] - 9s 946ms/step - loss: 0.4655 - binary_accuracy: 0.8043 - categorical_accuracy: 0.7466 - mean_squared_error: 0.1475\n",
            "Epoch 184/200\n",
            "10/10 [==============================] - 10s 968ms/step - loss: 0.4656 - binary_accuracy: 0.8044 - categorical_accuracy: 0.7445 - mean_squared_error: 0.1475\n",
            "Epoch 185/200\n",
            "10/10 [==============================] - 10s 953ms/step - loss: 0.4659 - binary_accuracy: 0.8042 - categorical_accuracy: 0.7458 - mean_squared_error: 0.1476\n",
            "Epoch 186/200\n",
            "10/10 [==============================] - 9s 943ms/step - loss: 0.4671 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7445 - mean_squared_error: 0.1480\n",
            "Epoch 187/200\n",
            "10/10 [==============================] - 9s 935ms/step - loss: 0.4645 - binary_accuracy: 0.8046 - categorical_accuracy: 0.7481 - mean_squared_error: 0.1471\n",
            "Epoch 188/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 0.4677 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7436 - mean_squared_error: 0.1482"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 9s 934ms/step - loss: 0.4685 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7428 - mean_squared_error: 0.1485\n",
            "Epoch 189/200\n",
            "10/10 [==============================] - 9s 938ms/step - loss: 0.4643 - binary_accuracy: 0.8048 - categorical_accuracy: 0.7468 - mean_squared_error: 0.1471\n",
            "Epoch 190/200\n",
            "10/10 [==============================] - 9s 947ms/step - loss: 0.4712 - binary_accuracy: 0.8021 - categorical_accuracy: 0.7412 - mean_squared_error: 0.1494\n",
            "Epoch 191/200\n",
            "10/10 [==============================] - 9s 940ms/step - loss: 0.4600 - binary_accuracy: 0.8064 - categorical_accuracy: 0.7530 - mean_squared_error: 0.1456\n",
            "Epoch 192/200\n",
            "10/10 [==============================] - 9s 937ms/step - loss: 0.4611 - binary_accuracy: 0.8059 - categorical_accuracy: 0.7517 - mean_squared_error: 0.1460\n",
            "Epoch 193/200\n",
            "10/10 [==============================] - 9s 938ms/step - loss: 0.4640 - binary_accuracy: 0.8049 - categorical_accuracy: 0.7475 - mean_squared_error: 0.1470\n",
            "Epoch 194/200\n",
            "10/10 [==============================] - 9s 931ms/step - loss: 0.4678 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7441 - mean_squared_error: 0.1483\n",
            "Epoch 195/200\n",
            "10/10 [==============================] - 9s 937ms/step - loss: 0.4715 - binary_accuracy: 0.8021 - categorical_accuracy: 0.7398 - mean_squared_error: 0.1495\n",
            "Epoch 196/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4670 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7457 - mean_squared_error: 0.1480\n",
            "Epoch 197/200\n",
            " 3/10 [========>.....................] - ETA: 6s - loss: 0.4653 - binary_accuracy: 0.8042 - categorical_accuracy: 0.7493 - mean_squared_error: 0.1474"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 9s 943ms/step - loss: 0.4703 - binary_accuracy: 0.8024 - categorical_accuracy: 0.7423 - mean_squared_error: 0.1492\n",
            "Epoch 198/200\n",
            "10/10 [==============================] - 9s 937ms/step - loss: 0.4738 - binary_accuracy: 0.8013 - categorical_accuracy: 0.7364 - mean_squared_error: 0.1503\n",
            "Epoch 199/200\n",
            "10/10 [==============================] - 9s 932ms/step - loss: 0.4704 - binary_accuracy: 0.8024 - categorical_accuracy: 0.7412 - mean_squared_error: 0.1492\n",
            "Epoch 200/200\n",
            "10/10 [==============================] - 9s 940ms/step - loss: 0.4620 - binary_accuracy: 0.8057 - categorical_accuracy: 0.7491 - mean_squared_error: 0.1463\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAecAAAFKCAYAAAAnj5dkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7fc248249e10>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "done training\n",
            "100/100 [==============================] - 0s 4ms/step\n",
            "[0.4676644504070282, 0.8031436800956726, 0.7502439022064209, 0.14829477667808533]\n",
            "the accuracy for length 10 was: 80.31436314363144%\n",
            "100/100 [==============================] - 0s 3ms/step\n",
            "[0.47557684779167175, 0.8002710342407227, 0.7397561073303223, 0.15100136399269104]\n",
            "the accuracy for length 20 was: 80.02710027100271%\n",
            "100/100 [==============================] - 0s 3ms/step\n",
            "[0.47057998180389404, 0.8024119138717651, 0.7407317161560059, 0.1492253690958023]\n",
            "the accuracy for length 30 was: 80.24119241192412%\n",
            "100/100 [==============================] - 0s 3ms/step\n",
            "[0.4813690483570099, 0.798184335231781, 0.7319512367248535, 0.15298040211200714]\n",
            "the accuracy for length 40 was: 79.81842818428184%\n",
            "100/100 [==============================] - 0s 3ms/step\n",
            "[0.4754076600074768, 0.8006775379180908, 0.7341463565826416, 0.15087386965751648]\n",
            "the accuracy for length 50 was: 80.06775067750678%\n",
            "100/100 [==============================] - 0s 3ms/step\n",
            "[0.4674258232116699, 0.8036314249038696, 0.7436585426330566, 0.1481315791606903]\n",
            "the accuracy for length 60 was: 80.36314363143632%\n",
            "100/100 [==============================] - 0s 3ms/step\n",
            "[0.46534302830696106, 0.8043361306190491, 0.7473170757293701, 0.14742964506149292]\n",
            "the accuracy for length 70 was: 80.43360433604336%\n",
            "100/100 [==============================] - 0s 3ms/step\n",
            "[0.48868247866630554, 0.7954742312431335, 0.7231707572937012, 0.15549297630786896]\n",
            "the accuracy for length 80 was: 79.54742547425474%\n",
            "100/100 [==============================] - 0s 3ms/step\n",
            "[0.47620561718940735, 0.8004336357116699, 0.7321951389312744, 0.15113680064678192]\n",
            "the accuracy for length 90 was: 80.04336043360433%\n",
            "100/100 [==============================] - 0s 3ms/step\n",
            "[0.4720529317855835, 0.8018699288368225, 0.7390244007110596, 0.14973154664039612]\n",
            "the accuracy for length 100 was: 80.1869918699187%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "CZQhGZ_tfQPk",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "##  Running NTM with Feedforward\n",
        "Here we will create a **Feedforward-NTM** model with the hyper-parameters specified below as well as the chart in the **Report** section under the **Hyper-parameter Values**. We will train the model for $200$ epochs. We then test our model on a sequence of different length of inputs. The result of which will be used in the **Report** section under **Generalization of The Model to Longer Sequences**."
      ]
    },
    {
      "metadata": {
        "id": "Plx6FjbUd9FS",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 8113
        },
        "outputId": "d21283d1-d94e-435c-cec5-153350ac5cb7"
      },
      "cell_type": "code",
      "source": [
        "batch_size = 100\n",
        "read_heads = 1\n",
        "write_heads = 1\n",
        "epochs = 200\n",
        "verboose = 0\n",
        "clipnorm = 10  \n",
        "sameInit = RandomNormal(seed=0) # bias_initializer\n",
        "\n",
        "lr = 0.01 # 5e-4   #0.001 was good\n",
        "#sgd = Adam(lr=lr, clipnorm=clipnorm, decay=0.5, amsgrad=True)\n",
        "sgd = SGD(lr=lr, clipnorm=clipnorm, decay=0.5)\n",
        "\n",
        "controller_input_dim, controller_output_dim = controller_input_output_shape(input_dim, output_dim, 20, 128, 3, read_heads,\n",
        "            write_heads)\n",
        "\n",
        "controller = Sequential()\n",
        "controller.name= \"feedforward\"\n",
        "    \n",
        "controller.add(Dense(units=units,\n",
        "                                kernel_initializer=sameInit,\n",
        "                                bias_initializer=sameInit,\n",
        "                                activation='softmax',\n",
        "                                input_dim=controller_input_dim))\n",
        "        \n",
        "controller.add(Dense(units=controller_output_dim,\n",
        "                                kernel_initializer=sameInit,\n",
        "                                bias_initializer=sameInit,\n",
        "                                activation='softmax'))\n",
        "    \n",
        "        \n",
        "controller.compile(loss='binary_crossentropy', optimizer=sgd, metrics = ['binary_accuracy','categorical_accuracy','mse'], sample_weight_mode=\"temporal\")\n",
        "model_ntm_feedforward = gen_model(input_dim=input_dim, output_dim=output_dim, batch_size=batch_size,\n",
        "                                    controller_model=controller, read_heads=read_heads, write_heads=write_heads,\n",
        "                                    activation=\"softmax\")\n",
        "\n",
        "print(model_ntm_feedforward.summary())\n",
        "print(\"model built, starting the copy experiment\")\n",
        "test_with_sequence_lengths(model_ntm_feedforward, epochs=epochs, verboose=verboose)\n",
        "\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Tensor(\"neural_turing_machine_2/mul_10:0\", shape=(100, 128, 20), dtype=float32)\n",
            "Tensor(\"neural_turing_machine_2/while/mul_8:0\", shape=(100, 128, 20), dtype=float32)\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "neural_turing_machine_2 (Neu (None, None, 9)           13201     \n",
            "=================================================================\n",
            "Total params: 13,201\n",
            "Trainable params: 13,201\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "None\n",
            "model built, starting the copy experiment\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:44: UserWarning: The semantics of the Keras 2 argument `steps_per_epoch` is not the same as the Keras 1 argument `samples_per_epoch`. `steps_per_epoch` is the number of batches to draw from the generator at each epoch. Basically steps_per_epoch = samples_per_epoch/batch_size. Similarly `nb_val_samples`->`validation_steps` and `val_samples`->`steps` arguments have changed. Update your method calls accordingly.\n",
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:44: UserWarning: Update your `fit_generator` call to the Keras 2 API: `fit_generator(<generator..., steps_per_epoch=10, epochs=200, callbacks=[<keras.ca..., workers=1, max_queue_size=10)`\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n",
            "10/10 [==============================] - 3s 341ms/step - loss: 0.5256 - binary_accuracy: 0.8039 - categorical_accuracy: 0.0635 - mean_squared_error: 0.1649\n",
            "Epoch 2/200\n",
            "10/10 [==============================] - 3s 304ms/step - loss: 0.5252 - binary_accuracy: 0.8041 - categorical_accuracy: 0.0629 - mean_squared_error: 0.1647\n",
            "Epoch 3/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5249 - binary_accuracy: 0.8042 - categorical_accuracy: 0.0630 - mean_squared_error: 0.1646\n",
            "Epoch 4/200\n",
            "10/10 [==============================] - 3s 299ms/step - loss: 0.5284 - binary_accuracy: 0.8025 - categorical_accuracy: 0.0636 - mean_squared_error: 0.1659\n",
            "Epoch 5/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5232 - binary_accuracy: 0.8050 - categorical_accuracy: 0.0641 - mean_squared_error: 0.1640\n",
            "Epoch 6/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.5237 - binary_accuracy: 0.8048 - categorical_accuracy: 0.0639 - mean_squared_error: 0.1642\n",
            "Epoch 7/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.5278 - binary_accuracy: 0.8028 - categorical_accuracy: 0.0660 - mean_squared_error: 0.1657\n",
            "Epoch 8/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5258 - binary_accuracy: 0.8038 - categorical_accuracy: 0.1564 - mean_squared_error: 0.1650\n",
            "Epoch 9/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5241 - binary_accuracy: 0.8046 - categorical_accuracy: 0.7469 - mean_squared_error: 0.1643\n",
            "Epoch 10/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5330 - binary_accuracy: 0.8003 - categorical_accuracy: 0.7330 - mean_squared_error: 0.1677\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 11/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5246 - binary_accuracy: 0.8043 - categorical_accuracy: 0.7460 - mean_squared_error: 0.1645\n",
            "Epoch 12/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5343 - binary_accuracy: 0.7996 - categorical_accuracy: 0.7365 - mean_squared_error: 0.1682\n",
            "Epoch 13/200\n",
            "10/10 [==============================] - 3s 305ms/step - loss: 0.5247 - binary_accuracy: 0.8043 - categorical_accuracy: 0.7453 - mean_squared_error: 0.1646\n",
            "Epoch 14/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.5248 - binary_accuracy: 0.8042 - categorical_accuracy: 0.7467 - mean_squared_error: 0.1646\n",
            "Epoch 15/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.5308 - binary_accuracy: 0.8013 - categorical_accuracy: 0.7388 - mean_squared_error: 0.1668\n",
            "Epoch 16/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5279 - binary_accuracy: 0.8027 - categorical_accuracy: 0.7403 - mean_squared_error: 0.1658\n",
            "Epoch 17/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.5261 - binary_accuracy: 0.8036 - categorical_accuracy: 0.7441 - mean_squared_error: 0.1651\n",
            "Epoch 18/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5323 - binary_accuracy: 0.8006 - categorical_accuracy: 0.7364 - mean_squared_error: 0.1674\n",
            "Epoch 19/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5273 - binary_accuracy: 0.8029 - categorical_accuracy: 0.7417 - mean_squared_error: 0.1656\n",
            "Epoch 20/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 0.5255 - binary_accuracy: 0.8038 - categorical_accuracy: 0.7449 - mean_squared_error: 0.1649"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 3s 300ms/step - loss: 0.5254 - binary_accuracy: 0.8039 - categorical_accuracy: 0.7451 - mean_squared_error: 0.1648\n",
            "Epoch 21/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5257 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7430 - mean_squared_error: 0.1650\n",
            "Epoch 22/200\n",
            "10/10 [==============================] - 3s 299ms/step - loss: 0.5275 - binary_accuracy: 0.8028 - categorical_accuracy: 0.7423 - mean_squared_error: 0.1657\n",
            "Epoch 23/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5244 - binary_accuracy: 0.8043 - categorical_accuracy: 0.7460 - mean_squared_error: 0.1645\n",
            "Epoch 24/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5212 - binary_accuracy: 0.8058 - categorical_accuracy: 0.7492 - mean_squared_error: 0.1633\n",
            "Epoch 25/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5253 - binary_accuracy: 0.8039 - categorical_accuracy: 0.7449 - mean_squared_error: 0.1648\n",
            "Epoch 26/200\n",
            "10/10 [==============================] - 3s 305ms/step - loss: 0.5292 - binary_accuracy: 0.8020 - categorical_accuracy: 0.7389 - mean_squared_error: 0.1663\n",
            "Epoch 27/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5302 - binary_accuracy: 0.8015 - categorical_accuracy: 0.7378 - mean_squared_error: 0.1667\n",
            "Epoch 28/200\n",
            "10/10 [==============================] - 3s 304ms/step - loss: 0.5312 - binary_accuracy: 0.8010 - categorical_accuracy: 0.7363 - mean_squared_error: 0.1670\n",
            "Epoch 29/200\n",
            " 3/10 [========>.....................] - ETA: 2s - loss: 0.5226 - binary_accuracy: 0.8051 - categorical_accuracy: 0.7493 - mean_squared_error: 0.1638"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5221 - binary_accuracy: 0.8054 - categorical_accuracy: 0.7505 - mean_squared_error: 0.1636\n",
            "Epoch 30/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.5214 - binary_accuracy: 0.8057 - categorical_accuracy: 0.7496 - mean_squared_error: 0.1634\n",
            "Epoch 31/200\n",
            "10/10 [==============================] - 3s 299ms/step - loss: 0.5239 - binary_accuracy: 0.8044 - categorical_accuracy: 0.7453 - mean_squared_error: 0.1643\n",
            "Epoch 32/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.5286 - binary_accuracy: 0.8022 - categorical_accuracy: 0.7420 - mean_squared_error: 0.1661\n",
            "Epoch 33/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5298 - binary_accuracy: 0.8016 - categorical_accuracy: 0.7385 - mean_squared_error: 0.1665\n",
            "Epoch 34/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5242 - binary_accuracy: 0.8043 - categorical_accuracy: 0.7426 - mean_squared_error: 0.1644\n",
            "Epoch 35/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5299 - binary_accuracy: 0.8015 - categorical_accuracy: 0.7394 - mean_squared_error: 0.1666\n",
            "Epoch 36/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5265 - binary_accuracy: 0.8031 - categorical_accuracy: 0.7413 - mean_squared_error: 0.1653\n",
            "Epoch 37/200\n",
            "10/10 [==============================] - 3s 305ms/step - loss: 0.5213 - binary_accuracy: 0.8056 - categorical_accuracy: 0.7515 - mean_squared_error: 0.1634\n",
            "Epoch 38/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5245 - binary_accuracy: 0.8041 - categorical_accuracy: 0.7449 - mean_squared_error: 0.1646\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 39/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.5268 - binary_accuracy: 0.8030 - categorical_accuracy: 0.7445 - mean_squared_error: 0.1655\n",
            "Epoch 40/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5272 - binary_accuracy: 0.8028 - categorical_accuracy: 0.7405 - mean_squared_error: 0.1656\n",
            "Epoch 41/200\n",
            "10/10 [==============================] - 3s 305ms/step - loss: 0.5263 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7427 - mean_squared_error: 0.1653\n",
            "Epoch 42/200\n",
            "10/10 [==============================] - 3s 305ms/step - loss: 0.5293 - binary_accuracy: 0.8018 - categorical_accuracy: 0.7398 - mean_squared_error: 0.1664\n",
            "Epoch 43/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5291 - binary_accuracy: 0.8018 - categorical_accuracy: 0.7391 - mean_squared_error: 0.1663\n",
            "Epoch 44/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5226 - binary_accuracy: 0.8049 - categorical_accuracy: 0.7457 - mean_squared_error: 0.1639\n",
            "Epoch 45/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5276 - binary_accuracy: 0.8025 - categorical_accuracy: 0.7420 - mean_squared_error: 0.1658\n",
            "Epoch 46/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5250 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7445 - mean_squared_error: 0.1648\n",
            "Epoch 47/200\n",
            "10/10 [==============================] - 3s 304ms/step - loss: 0.5245 - binary_accuracy: 0.8039 - categorical_accuracy: 0.7447 - mean_squared_error: 0.1647\n",
            "Epoch 48/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 0.5272 - binary_accuracy: 0.8026 - categorical_accuracy: 0.7431 - mean_squared_error: 0.1657"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 3s 304ms/step - loss: 0.5272 - binary_accuracy: 0.8026 - categorical_accuracy: 0.7428 - mean_squared_error: 0.1657\n",
            "Epoch 49/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5228 - binary_accuracy: 0.8047 - categorical_accuracy: 0.7472 - mean_squared_error: 0.1640\n",
            "Epoch 50/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5254 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7435 - mean_squared_error: 0.1650\n",
            "Epoch 51/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5277 - binary_accuracy: 0.8023 - categorical_accuracy: 0.7409 - mean_squared_error: 0.1659\n",
            "Epoch 52/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5246 - binary_accuracy: 0.8038 - categorical_accuracy: 0.7425 - mean_squared_error: 0.1647\n",
            "Epoch 53/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5324 - binary_accuracy: 0.8000 - categorical_accuracy: 0.7342 - mean_squared_error: 0.1676\n",
            "Epoch 54/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5299 - binary_accuracy: 0.8012 - categorical_accuracy: 0.7359 - mean_squared_error: 0.1667\n",
            "Epoch 55/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.5201 - binary_accuracy: 0.8059 - categorical_accuracy: 0.7506 - mean_squared_error: 0.1631\n",
            "Epoch 56/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5251 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7422 - mean_squared_error: 0.1649\n",
            "Epoch 57/200\n",
            " 3/10 [========>.....................] - ETA: 2s - loss: 0.5277 - binary_accuracy: 0.8021 - categorical_accuracy: 0.7398 - mean_squared_error: 0.1659"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5191 - binary_accuracy: 0.8063 - categorical_accuracy: 0.7537 - mean_squared_error: 0.1627\n",
            "Epoch 58/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5255 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7439 - mean_squared_error: 0.1651\n",
            "Epoch 59/200\n",
            "10/10 [==============================] - 3s 307ms/step - loss: 0.5264 - binary_accuracy: 0.8027 - categorical_accuracy: 0.7428 - mean_squared_error: 0.1655\n",
            "Epoch 60/200\n",
            "10/10 [==============================] - 3s 297ms/step - loss: 0.5266 - binary_accuracy: 0.8026 - categorical_accuracy: 0.7420 - mean_squared_error: 0.1656\n",
            "Epoch 61/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.5195 - binary_accuracy: 0.8059 - categorical_accuracy: 0.7509 - mean_squared_error: 0.1629\n",
            "Epoch 62/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5270 - binary_accuracy: 0.8023 - categorical_accuracy: 0.7409 - mean_squared_error: 0.1657\n",
            "Epoch 63/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5187 - binary_accuracy: 0.8062 - categorical_accuracy: 0.7500 - mean_squared_error: 0.1627\n",
            "Epoch 64/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5249 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7424 - mean_squared_error: 0.1650\n",
            "Epoch 65/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5272 - binary_accuracy: 0.8021 - categorical_accuracy: 0.7415 - mean_squared_error: 0.1658\n",
            "Epoch 66/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5293 - binary_accuracy: 0.8011 - categorical_accuracy: 0.7387 - mean_squared_error: 0.1666\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 67/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5222 - binary_accuracy: 0.8044 - categorical_accuracy: 0.7472 - mean_squared_error: 0.1640\n",
            "Epoch 68/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5256 - binary_accuracy: 0.8028 - categorical_accuracy: 0.7405 - mean_squared_error: 0.1653\n",
            "Epoch 69/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5234 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7432 - mean_squared_error: 0.1645\n",
            "Epoch 70/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5247 - binary_accuracy: 0.8031 - categorical_accuracy: 0.7430 - mean_squared_error: 0.1650\n",
            "Epoch 71/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5231 - binary_accuracy: 0.8038 - categorical_accuracy: 0.7466 - mean_squared_error: 0.1644\n",
            "Epoch 72/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5279 - binary_accuracy: 0.8014 - categorical_accuracy: 0.7392 - mean_squared_error: 0.1662\n",
            "Epoch 73/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5229 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7450 - mean_squared_error: 0.1644\n",
            "Epoch 74/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5237 - binary_accuracy: 0.8033 - categorical_accuracy: 0.7445 - mean_squared_error: 0.1647\n",
            "Epoch 75/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5221 - binary_accuracy: 0.8040 - categorical_accuracy: 0.7466 - mean_squared_error: 0.1641\n",
            "Epoch 76/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 0.5227 - binary_accuracy: 0.8036 - categorical_accuracy: 0.7442 - mean_squared_error: 0.1644"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 3s 302ms/step - loss: 0.5232 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7435 - mean_squared_error: 0.1646\n",
            "Epoch 77/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5207 - binary_accuracy: 0.8045 - categorical_accuracy: 0.7478 - mean_squared_error: 0.1637\n",
            "Epoch 78/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5243 - binary_accuracy: 0.8027 - categorical_accuracy: 0.7413 - mean_squared_error: 0.1650\n",
            "Epoch 79/200\n",
            "10/10 [==============================] - 3s 304ms/step - loss: 0.5177 - binary_accuracy: 0.8058 - categorical_accuracy: 0.7517 - mean_squared_error: 0.1626\n",
            "Epoch 80/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5215 - binary_accuracy: 0.8039 - categorical_accuracy: 0.7464 - mean_squared_error: 0.1640\n",
            "Epoch 81/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5238 - binary_accuracy: 0.8027 - categorical_accuracy: 0.7403 - mean_squared_error: 0.1649\n",
            "Epoch 82/200\n",
            "10/10 [==============================] - 3s 304ms/step - loss: 0.5206 - binary_accuracy: 0.8041 - categorical_accuracy: 0.7452 - mean_squared_error: 0.1638\n",
            "Epoch 83/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.5216 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7421 - mean_squared_error: 0.1642\n",
            "Epoch 84/200\n",
            "10/10 [==============================] - 3s 306ms/step - loss: 0.5184 - binary_accuracy: 0.8050 - categorical_accuracy: 0.7484 - mean_squared_error: 0.1630\n",
            "Epoch 85/200\n",
            " 3/10 [========>.....................] - ETA: 2s - loss: 0.5297 - binary_accuracy: 0.7995 - categorical_accuracy: 0.7315 - mean_squared_error: 0.1672"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5246 - binary_accuracy: 0.8019 - categorical_accuracy: 0.7381 - mean_squared_error: 0.1654\n",
            "Epoch 86/200\n",
            "10/10 [==============================] - 3s 305ms/step - loss: 0.5165 - binary_accuracy: 0.8056 - categorical_accuracy: 0.7488 - mean_squared_error: 0.1624\n",
            "Epoch 87/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.5238 - binary_accuracy: 0.8020 - categorical_accuracy: 0.7405 - mean_squared_error: 0.1652\n",
            "Epoch 88/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.5152 - binary_accuracy: 0.8060 - categorical_accuracy: 0.7498 - mean_squared_error: 0.1620\n",
            "Epoch 89/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5234 - binary_accuracy: 0.8019 - categorical_accuracy: 0.7399 - mean_squared_error: 0.1651\n",
            "Epoch 90/200\n",
            "10/10 [==============================] - 3s 299ms/step - loss: 0.5201 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7432 - mean_squared_error: 0.1639\n",
            "Epoch 91/200\n",
            "10/10 [==============================] - 3s 299ms/step - loss: 0.5222 - binary_accuracy: 0.8022 - categorical_accuracy: 0.7405 - mean_squared_error: 0.1648\n",
            "Epoch 92/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5209 - binary_accuracy: 0.8027 - categorical_accuracy: 0.7431 - mean_squared_error: 0.1643\n",
            "Epoch 93/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5183 - binary_accuracy: 0.8038 - categorical_accuracy: 0.7466 - mean_squared_error: 0.1634\n",
            "Epoch 94/200\n",
            "10/10 [==============================] - 3s 299ms/step - loss: 0.5156 - binary_accuracy: 0.8048 - categorical_accuracy: 0.7491 - mean_squared_error: 0.1625\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 95/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5173 - binary_accuracy: 0.8039 - categorical_accuracy: 0.7445 - mean_squared_error: 0.1632\n",
            "Epoch 96/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5184 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7441 - mean_squared_error: 0.1636\n",
            "Epoch 97/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5215 - binary_accuracy: 0.8015 - categorical_accuracy: 0.7389 - mean_squared_error: 0.1649\n",
            "Epoch 98/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5169 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7431 - mean_squared_error: 0.1632\n",
            "Epoch 99/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5132 - binary_accuracy: 0.8051 - categorical_accuracy: 0.7474 - mean_squared_error: 0.1619\n",
            "Epoch 100/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5200 - binary_accuracy: 0.8016 - categorical_accuracy: 0.7416 - mean_squared_error: 0.1645\n",
            "Epoch 101/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5118 - binary_accuracy: 0.8053 - categorical_accuracy: 0.7480 - mean_squared_error: 0.1615\n",
            "Epoch 102/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5173 - binary_accuracy: 0.8024 - categorical_accuracy: 0.7429 - mean_squared_error: 0.1636\n",
            "Epoch 103/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5184 - binary_accuracy: 0.8017 - categorical_accuracy: 0.7393 - mean_squared_error: 0.1641\n",
            "Epoch 104/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 0.5139 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7472 - mean_squared_error: 0.1625"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 3s 300ms/step - loss: 0.5137 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7472 - mean_squared_error: 0.1624\n",
            "Epoch 105/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5097 - binary_accuracy: 0.8053 - categorical_accuracy: 0.7480 - mean_squared_error: 0.1610\n",
            "Epoch 106/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5135 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7431 - mean_squared_error: 0.1625\n",
            "Epoch 107/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5120 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7452 - mean_squared_error: 0.1620\n",
            "Epoch 108/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5150 - binary_accuracy: 0.8021 - categorical_accuracy: 0.7369 - mean_squared_error: 0.1632\n",
            "Epoch 109/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5103 - binary_accuracy: 0.8039 - categorical_accuracy: 0.7465 - mean_squared_error: 0.1616\n",
            "Epoch 110/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5107 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7434 - mean_squared_error: 0.1618\n",
            "Epoch 111/200\n",
            "10/10 [==============================] - 3s 299ms/step - loss: 0.5070 - binary_accuracy: 0.8049 - categorical_accuracy: 0.7483 - mean_squared_error: 0.1605\n",
            "Epoch 112/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.5120 - binary_accuracy: 0.8023 - categorical_accuracy: 0.7397 - mean_squared_error: 0.1624\n",
            "Epoch 113/200\n",
            " 3/10 [========>.....................] - ETA: 2s - loss: 0.5134 - binary_accuracy: 0.8015 - categorical_accuracy: 0.7367 - mean_squared_error: 0.1630"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5121 - binary_accuracy: 0.8020 - categorical_accuracy: 0.7392 - mean_squared_error: 0.1625\n",
            "Epoch 114/200\n",
            "10/10 [==============================] - 3s 304ms/step - loss: 0.5112 - binary_accuracy: 0.8021 - categorical_accuracy: 0.7397 - mean_squared_error: 0.1623\n",
            "Epoch 115/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5103 - binary_accuracy: 0.8023 - categorical_accuracy: 0.7397 - mean_squared_error: 0.1620\n",
            "Epoch 116/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5045 - binary_accuracy: 0.8046 - categorical_accuracy: 0.7470 - mean_squared_error: 0.1600\n",
            "Epoch 117/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.5098 - binary_accuracy: 0.8019 - categorical_accuracy: 0.7387 - mean_squared_error: 0.1620\n",
            "Epoch 118/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.5071 - binary_accuracy: 0.8028 - categorical_accuracy: 0.7428 - mean_squared_error: 0.1611\n",
            "Epoch 119/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.5070 - binary_accuracy: 0.8025 - categorical_accuracy: 0.7430 - mean_squared_error: 0.1611\n",
            "Epoch 120/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.5062 - binary_accuracy: 0.8026 - categorical_accuracy: 0.7415 - mean_squared_error: 0.1609\n",
            "Epoch 121/200\n",
            "10/10 [==============================] - 3s 306ms/step - loss: 0.5041 - binary_accuracy: 0.8033 - categorical_accuracy: 0.7435 - mean_squared_error: 0.1602\n",
            "Epoch 122/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4996 - binary_accuracy: 0.8050 - categorical_accuracy: 0.7481 - mean_squared_error: 0.1587\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 123/200\n",
            "10/10 [==============================] - 3s 298ms/step - loss: 0.4961 - binary_accuracy: 0.8063 - categorical_accuracy: 0.7503 - mean_squared_error: 0.1574\n",
            "Epoch 124/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4977 - binary_accuracy: 0.8052 - categorical_accuracy: 0.7505 - mean_squared_error: 0.1581\n",
            "Epoch 125/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.5023 - binary_accuracy: 0.8029 - categorical_accuracy: 0.7418 - mean_squared_error: 0.1598\n",
            "Epoch 126/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4909 - binary_accuracy: 0.8077 - categorical_accuracy: 0.7555 - mean_squared_error: 0.1558\n",
            "Epoch 127/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.4954 - binary_accuracy: 0.8053 - categorical_accuracy: 0.7489 - mean_squared_error: 0.1575\n",
            "Epoch 128/200\n",
            "10/10 [==============================] - 3s 304ms/step - loss: 0.4911 - binary_accuracy: 0.8070 - categorical_accuracy: 0.7520 - mean_squared_error: 0.1560\n",
            "Epoch 129/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.4956 - binary_accuracy: 0.8047 - categorical_accuracy: 0.7465 - mean_squared_error: 0.1577\n",
            "Epoch 130/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.4997 - binary_accuracy: 0.8026 - categorical_accuracy: 0.7390 - mean_squared_error: 0.1592\n",
            "Epoch 131/200\n",
            "10/10 [==============================] - 3s 299ms/step - loss: 0.4961 - binary_accuracy: 0.8038 - categorical_accuracy: 0.7458 - mean_squared_error: 0.1580\n",
            "Epoch 132/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 0.4972 - binary_accuracy: 0.8031 - categorical_accuracy: 0.7424 - mean_squared_error: 0.1584"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 3s 302ms/step - loss: 0.4960 - binary_accuracy: 0.8036 - categorical_accuracy: 0.7440 - mean_squared_error: 0.1580\n",
            "Epoch 133/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.4955 - binary_accuracy: 0.8036 - categorical_accuracy: 0.7437 - mean_squared_error: 0.1579\n",
            "Epoch 134/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4988 - binary_accuracy: 0.8019 - categorical_accuracy: 0.7391 - mean_squared_error: 0.1591\n",
            "Epoch 135/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.5026 - binary_accuracy: 0.7999 - categorical_accuracy: 0.7331 - mean_squared_error: 0.1605\n",
            "Epoch 136/200\n",
            "10/10 [==============================] - 3s 297ms/step - loss: 0.4894 - binary_accuracy: 0.8055 - categorical_accuracy: 0.7492 - mean_squared_error: 0.1558\n",
            "Epoch 137/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.4904 - binary_accuracy: 0.8048 - categorical_accuracy: 0.7448 - mean_squared_error: 0.1562\n",
            "Epoch 138/200\n",
            "10/10 [==============================] - 3s 305ms/step - loss: 0.4906 - binary_accuracy: 0.8044 - categorical_accuracy: 0.7470 - mean_squared_error: 0.1563\n",
            "Epoch 139/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.4961 - binary_accuracy: 0.8018 - categorical_accuracy: 0.7391 - mean_squared_error: 0.1583\n",
            "Epoch 140/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4924 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7427 - mean_squared_error: 0.1570\n",
            "Epoch 141/200\n",
            " 3/10 [========>.....................] - ETA: 2s - loss: 0.5025 - binary_accuracy: 0.7987 - categorical_accuracy: 0.7299 - mean_squared_error: 0.1606"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4929 - binary_accuracy: 0.8028 - categorical_accuracy: 0.7415 - mean_squared_error: 0.1572\n",
            "Epoch 142/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.4846 - binary_accuracy: 0.8061 - categorical_accuracy: 0.7518 - mean_squared_error: 0.1543\n",
            "Epoch 143/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.4957 - binary_accuracy: 0.8011 - categorical_accuracy: 0.7363 - mean_squared_error: 0.1583\n",
            "Epoch 144/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.4846 - binary_accuracy: 0.8057 - categorical_accuracy: 0.7495 - mean_squared_error: 0.1543\n",
            "Epoch 145/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4956 - binary_accuracy: 0.8007 - categorical_accuracy: 0.7359 - mean_squared_error: 0.1583\n",
            "Epoch 146/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4895 - binary_accuracy: 0.8031 - categorical_accuracy: 0.7430 - mean_squared_error: 0.1561\n",
            "Epoch 147/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.4806 - binary_accuracy: 0.8068 - categorical_accuracy: 0.7530 - mean_squared_error: 0.1530\n",
            "Epoch 148/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.4895 - binary_accuracy: 0.8028 - categorical_accuracy: 0.7415 - mean_squared_error: 0.1562\n",
            "Epoch 149/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.4907 - binary_accuracy: 0.8021 - categorical_accuracy: 0.7389 - mean_squared_error: 0.1566\n",
            "Epoch 150/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.4870 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7442 - mean_squared_error: 0.1554\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 151/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.4842 - binary_accuracy: 0.8045 - categorical_accuracy: 0.7463 - mean_squared_error: 0.1544\n",
            "Epoch 152/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4862 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7453 - mean_squared_error: 0.1551\n",
            "Epoch 153/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.4828 - binary_accuracy: 0.8047 - categorical_accuracy: 0.7462 - mean_squared_error: 0.1539\n",
            "Epoch 154/200\n",
            "10/10 [==============================] - 3s 299ms/step - loss: 0.4856 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7444 - mean_squared_error: 0.1549\n",
            "Epoch 155/200\n",
            "10/10 [==============================] - 3s 305ms/step - loss: 0.4834 - binary_accuracy: 0.8042 - categorical_accuracy: 0.7450 - mean_squared_error: 0.1541\n",
            "Epoch 156/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.4880 - binary_accuracy: 0.8020 - categorical_accuracy: 0.7406 - mean_squared_error: 0.1558\n",
            "Epoch 157/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4815 - binary_accuracy: 0.8046 - categorical_accuracy: 0.7485 - mean_squared_error: 0.1535\n",
            "Epoch 158/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.4842 - binary_accuracy: 0.8033 - categorical_accuracy: 0.7443 - mean_squared_error: 0.1545\n",
            "Epoch 159/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.4790 - binary_accuracy: 0.8054 - categorical_accuracy: 0.7484 - mean_squared_error: 0.1526\n",
            "Epoch 160/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 0.4807 - binary_accuracy: 0.8046 - categorical_accuracy: 0.7463 - mean_squared_error: 0.1532"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 3s 302ms/step - loss: 0.4818 - binary_accuracy: 0.8041 - categorical_accuracy: 0.7454 - mean_squared_error: 0.1536\n",
            "Epoch 161/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4826 - binary_accuracy: 0.8036 - categorical_accuracy: 0.7453 - mean_squared_error: 0.1539\n",
            "Epoch 162/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4869 - binary_accuracy: 0.8017 - categorical_accuracy: 0.7402 - mean_squared_error: 0.1554\n",
            "Epoch 163/200\n",
            "10/10 [==============================] - 3s 307ms/step - loss: 0.4770 - binary_accuracy: 0.8057 - categorical_accuracy: 0.7498 - mean_squared_error: 0.1520\n",
            "Epoch 164/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.4812 - binary_accuracy: 0.8038 - categorical_accuracy: 0.7453 - mean_squared_error: 0.1534\n",
            "Epoch 165/200\n",
            "10/10 [==============================] - 3s 304ms/step - loss: 0.4760 - binary_accuracy: 0.8059 - categorical_accuracy: 0.7510 - mean_squared_error: 0.1516\n",
            "Epoch 166/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.4822 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7431 - mean_squared_error: 0.1538\n",
            "Epoch 167/200\n",
            "10/10 [==============================] - 3s 304ms/step - loss: 0.4824 - binary_accuracy: 0.8031 - categorical_accuracy: 0.7405 - mean_squared_error: 0.1538\n",
            "Epoch 168/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.4827 - binary_accuracy: 0.8027 - categorical_accuracy: 0.7425 - mean_squared_error: 0.1540\n",
            "Epoch 169/200\n",
            " 3/10 [========>.....................] - ETA: 2s - loss: 0.4691 - binary_accuracy: 0.8083 - categorical_accuracy: 0.7563 - mean_squared_error: 0.1492"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4760 - binary_accuracy: 0.8054 - categorical_accuracy: 0.7497 - mean_squared_error: 0.1516\n",
            "Epoch 170/200\n",
            "10/10 [==============================] - 3s 299ms/step - loss: 0.4812 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7424 - mean_squared_error: 0.1534\n",
            "Epoch 171/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.4791 - binary_accuracy: 0.8040 - categorical_accuracy: 0.7443 - mean_squared_error: 0.1527\n",
            "Epoch 172/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.4707 - binary_accuracy: 0.8073 - categorical_accuracy: 0.7531 - mean_squared_error: 0.1498\n",
            "Epoch 173/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.4793 - binary_accuracy: 0.8036 - categorical_accuracy: 0.7453 - mean_squared_error: 0.1528\n",
            "Epoch 174/200\n",
            "10/10 [==============================] - 3s 299ms/step - loss: 0.4829 - binary_accuracy: 0.8021 - categorical_accuracy: 0.7397 - mean_squared_error: 0.1541\n",
            "Epoch 175/200\n",
            "10/10 [==============================] - 3s 299ms/step - loss: 0.4764 - binary_accuracy: 0.8046 - categorical_accuracy: 0.7474 - mean_squared_error: 0.1518\n",
            "Epoch 176/200\n",
            "10/10 [==============================] - 3s 298ms/step - loss: 0.4772 - binary_accuracy: 0.8042 - categorical_accuracy: 0.7460 - mean_squared_error: 0.1521\n",
            "Epoch 177/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4787 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7435 - mean_squared_error: 0.1526\n",
            "Epoch 178/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4802 - binary_accuracy: 0.8028 - categorical_accuracy: 0.7433 - mean_squared_error: 0.1531\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 179/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.4809 - binary_accuracy: 0.8025 - categorical_accuracy: 0.7404 - mean_squared_error: 0.1533\n",
            "Epoch 180/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.4783 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7442 - mean_squared_error: 0.1525\n",
            "Epoch 181/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.4770 - binary_accuracy: 0.8039 - categorical_accuracy: 0.7447 - mean_squared_error: 0.1520\n",
            "Epoch 182/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4735 - binary_accuracy: 0.8052 - categorical_accuracy: 0.7500 - mean_squared_error: 0.1508\n",
            "Epoch 183/200\n",
            "10/10 [==============================] - 3s 299ms/step - loss: 0.4754 - binary_accuracy: 0.8044 - categorical_accuracy: 0.7468 - mean_squared_error: 0.1514\n",
            "Epoch 184/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.4744 - binary_accuracy: 0.8047 - categorical_accuracy: 0.7475 - mean_squared_error: 0.1511\n",
            "Epoch 185/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.4837 - binary_accuracy: 0.8008 - categorical_accuracy: 0.7375 - mean_squared_error: 0.1544\n",
            "Epoch 186/200\n",
            "10/10 [==============================] - 3s 301ms/step - loss: 0.4768 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7439 - mean_squared_error: 0.1519\n",
            "Epoch 187/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.4739 - binary_accuracy: 0.8047 - categorical_accuracy: 0.7479 - mean_squared_error: 0.1509\n",
            "Epoch 188/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 0.4820 - binary_accuracy: 0.8015 - categorical_accuracy: 0.7368 - mean_squared_error: 0.1537"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 3s 300ms/step - loss: 0.4835 - binary_accuracy: 0.8008 - categorical_accuracy: 0.7355 - mean_squared_error: 0.1542\n",
            "Epoch 189/200\n",
            "10/10 [==============================] - 3s 299ms/step - loss: 0.4680 - binary_accuracy: 0.8070 - categorical_accuracy: 0.7543 - mean_squared_error: 0.1488\n",
            "Epoch 190/200\n",
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4769 - binary_accuracy: 0.8033 - categorical_accuracy: 0.7440 - mean_squared_error: 0.1519\n",
            "Epoch 191/200\n",
            "10/10 [==============================] - 3s 304ms/step - loss: 0.4797 - binary_accuracy: 0.8021 - categorical_accuracy: 0.7397 - mean_squared_error: 0.1529\n",
            "Epoch 192/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.4756 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7453 - mean_squared_error: 0.1515\n",
            "Epoch 193/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.4726 - binary_accuracy: 0.8049 - categorical_accuracy: 0.7471 - mean_squared_error: 0.1504\n",
            "Epoch 194/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.4767 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7431 - mean_squared_error: 0.1518\n",
            "Epoch 195/200\n",
            "10/10 [==============================] - 3s 303ms/step - loss: 0.4791 - binary_accuracy: 0.8021 - categorical_accuracy: 0.7423 - mean_squared_error: 0.1527\n",
            "Epoch 196/200\n",
            "10/10 [==============================] - 3s 304ms/step - loss: 0.4724 - binary_accuracy: 0.8048 - categorical_accuracy: 0.7478 - mean_squared_error: 0.1503\n",
            "Epoch 197/200\n",
            " 3/10 [========>.....................] - ETA: 2s - loss: 0.4879 - binary_accuracy: 0.7985 - categorical_accuracy: 0.7309 - mean_squared_error: 0.1558"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 3s 302ms/step - loss: 0.4840 - binary_accuracy: 0.8000 - categorical_accuracy: 0.7351 - mean_squared_error: 0.1544\n",
            "Epoch 198/200\n",
            "10/10 [==============================] - 3s 305ms/step - loss: 0.4743 - binary_accuracy: 0.8039 - categorical_accuracy: 0.7464 - mean_squared_error: 0.1510\n",
            "Epoch 199/200\n",
            "10/10 [==============================] - 3s 300ms/step - loss: 0.4772 - binary_accuracy: 0.8027 - categorical_accuracy: 0.7415 - mean_squared_error: 0.1520\n",
            "Epoch 200/200\n",
            "10/10 [==============================] - 3s 305ms/step - loss: 0.4702 - binary_accuracy: 0.8054 - categorical_accuracy: 0.7502 - mean_squared_error: 0.1496\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAecAAAFKCAYAAAAnj5dkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f94f41d28d0>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "done training\n",
            "100/100 [==============================] - 0s 3ms/step\n",
            "[0.48443305492401123, 0.7999187111854553, 0.7297561168670654, 0.1544853001832962]\n",
            "the accuracy for length 10 was: 79.9918699186992%\n",
            "100/100 [==============================] - 0s 1ms/step\n",
            "[0.45721930265426636, 0.810569167137146, 0.7651219367980957, 0.14502696692943573]\n",
            "the accuracy for length 20 was: 81.0569105691057%\n",
            "100/100 [==============================] - 0s 1ms/step\n",
            "[0.4843052923679352, 0.7998916506767273, 0.7317073345184326, 0.15446029603481293]\n",
            "the accuracy for length 30 was: 79.98915989159892%\n",
            "100/100 [==============================] - 0s 1ms/step\n",
            "[0.46028557419776917, 0.8093224763870239, 0.761951208114624, 0.1461029052734375]\n",
            "the accuracy for length 40 was: 80.93224932249322%\n",
            "100/100 [==============================] - 0s 1ms/step\n",
            "[0.4692862629890442, 0.805691123008728, 0.752439022064209, 0.14925652742385864]\n",
            "the accuracy for length 50 was: 80.5691056910569%\n",
            "100/100 [==============================] - 0s 1ms/step\n",
            "[0.4760558307170868, 0.8030081987380981, 0.7448780536651611, 0.15162017941474915]\n",
            "the accuracy for length 60 was: 80.30081300813008%\n",
            "100/100 [==============================] - 0s 1ms/step\n",
            "[0.4707658886909485, 0.8051761984825134, 0.7495121955871582, 0.14975754916667938]\n",
            "the accuracy for length 70 was: 80.51761517615176%\n",
            "100/100 [==============================] - 0s 1ms/step\n",
            "[0.4643636643886566, 0.8076423406600952, 0.7585365772247314, 0.14754125475883484]\n",
            "the accuracy for length 80 was: 80.76422764227642%\n",
            "100/100 [==============================] - 0s 1ms/step\n",
            "[0.4763076901435852, 0.8031707406044006, 0.7385365962982178, 0.151642844080925]\n",
            "the accuracy for length 90 was: 80.31707317073172%\n",
            "100/100 [==============================] - 0s 1ms/step\n",
            "[0.48148486018180847, 0.8008130192756653, 0.7392683029174805, 0.15352383255958557]\n",
            "the accuracy for length 100 was: 80.08130081300813%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "TQp0Z2H8fVCg",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "##  Running LSTM\n",
        "Here we will create a vanilla **LSTM** model with the hyper-parameters specified below as well as the chart in the **Report** section under the **Hyper-parameter Values**. We will train the model for $200$ epochs. We then test our model on a sequence of different length of inputs. The result of which will be used in the **Report** section under **Generalization of The Model to Longer Sequences**."
      ]
    },
    {
      "metadata": {
        "id": "70Mp_JtteEDG",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 8165
        },
        "outputId": "40ac72fd-0a5e-4ebf-f9a1-039390485086"
      },
      "cell_type": "code",
      "source": [
        "# not bad : batch_size = 25    , lr = 5e-5 , sgd = 'adam', activation_func = 'softmax'  try this with 2000 epochs later.\n",
        "# not bad : batch_size = 25    , lr = 5e-5 , sgd = 'adam', activation_func = 'softmax'    try this with 2000 epochs later.\n",
        "# not bad : batch_size = 50 sgd = 'adam' activation_func = 'softmax'\n",
        "\n",
        "batch_size = 100 \n",
        "epochs = 200\n",
        "verboose = 0\n",
        "lr = 0.001 # 5e-4\n",
        "clipnorm = 10  \n",
        "activation_func = 'softmax' #'hard_sigmoid'\n",
        "\n",
        "sameInit = RandomUniform(minval=0.8, maxval=0.99, seed=0) # bias_initializer\n",
        "\n",
        "sgd = Adam(lr=lr, clipnorm=clipnorm, decay=0.5, amsgrad=True)\n",
        "#sgd = SGD(lr=lr)\n",
        "#sgd = 'adam'\n",
        "\n",
        "loss='categorical_crossentropy'\n",
        "model = gen_model_lstm(input_dim=input_dim, output_dim=output_dim, batch_size=batch_size, sgd=sgd, loss=loss)\n",
        "\n",
        "print(model.summary())\n",
        "print(\"model built, starting the copy experiment\")\n",
        "test_with_sequence_lengths(model, epochs=epochs, verboose=verboose)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "lstm_2 (LSTM)                (None, None, 100)         44000     \n",
            "_________________________________________________________________\n",
            "lstm_3 (LSTM)                (None, None, 9)           3960      \n",
            "_________________________________________________________________\n",
            "activation_1 (Activation)    (None, None, 9)           0         \n",
            "=================================================================\n",
            "Total params: 47,960\n",
            "Trainable params: 47,960\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "None\n",
            "model built, starting the copy experiment\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:65: UserWarning: The semantics of the Keras 2 argument `steps_per_epoch` is not the same as the Keras 1 argument `samples_per_epoch`. `steps_per_epoch` is the number of batches to draw from the generator at each epoch. Basically steps_per_epoch = samples_per_epoch/batch_size. Similarly `nb_val_samples`->`validation_steps` and `val_samples`->`steps` arguments have changed. Update your method calls accordingly.\n",
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:65: UserWarning: Update your `fit_generator` call to the Keras 2 API: `fit_generator(<generator..., steps_per_epoch=10, epochs=200, callbacks=[<keras.ca..., workers=1, max_queue_size=10)`\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Variable *= will be deprecated. Use variable.assign_mul if you want assignment to the variable value or 'x = x * y' if you want a new python Tensor object.\n",
            "Epoch 1/200\n",
            "10/10 [==============================] - 3s 280ms/step - loss: 3.7648 - binary_accuracy: 0.8058 - categorical_accuracy: 0.5268 - mean_squared_error: 0.1613\n",
            "Epoch 2/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.7514 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7434 - mean_squared_error: 0.1612\n",
            "Epoch 3/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.6955 - binary_accuracy: 0.8049 - categorical_accuracy: 0.7466 - mean_squared_error: 0.1591\n",
            "Epoch 4/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.6772 - binary_accuracy: 0.8049 - categorical_accuracy: 0.7470 - mean_squared_error: 0.1584\n",
            "Epoch 5/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.6968 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7417 - mean_squared_error: 0.1591\n",
            "Epoch 6/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.6733 - binary_accuracy: 0.8041 - categorical_accuracy: 0.7444 - mean_squared_error: 0.1582\n",
            "Epoch 7/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.7168 - binary_accuracy: 0.8018 - categorical_accuracy: 0.7385 - mean_squared_error: 0.1598\n",
            "Epoch 8/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.7453 - binary_accuracy: 0.8003 - categorical_accuracy: 0.7337 - mean_squared_error: 0.1609\n",
            "Epoch 9/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.6817 - binary_accuracy: 0.8029 - categorical_accuracy: 0.7413 - mean_squared_error: 0.1585\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 10/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.6888 - binary_accuracy: 0.8023 - categorical_accuracy: 0.7408 - mean_squared_error: 0.1587\n",
            "Epoch 11/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.6767 - binary_accuracy: 0.8027 - categorical_accuracy: 0.7411 - mean_squared_error: 0.1582\n",
            "Epoch 12/200\n",
            "10/10 [==============================] - 2s 229ms/step - loss: 3.7072 - binary_accuracy: 0.8011 - categorical_accuracy: 0.7383 - mean_squared_error: 0.1593\n",
            "Epoch 13/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.7567 - binary_accuracy: 0.7989 - categorical_accuracy: 0.7290 - mean_squared_error: 0.1612\n",
            "Epoch 14/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.6415 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7438 - mean_squared_error: 0.1568\n",
            "Epoch 15/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.6192 - binary_accuracy: 0.8045 - categorical_accuracy: 0.7467 - mean_squared_error: 0.1559\n",
            "Epoch 16/200\n",
            "10/10 [==============================] - 2s 229ms/step - loss: 3.6476 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7454 - mean_squared_error: 0.1570\n",
            "Epoch 17/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.6626 - binary_accuracy: 0.8024 - categorical_accuracy: 0.7400 - mean_squared_error: 0.1575\n",
            "Epoch 18/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.6291 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7438 - mean_squared_error: 0.1562\n",
            "Epoch 19/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 3.6332 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7433 - mean_squared_error: 0.1564"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 2s 227ms/step - loss: 3.6458 - binary_accuracy: 0.8029 - categorical_accuracy: 0.7422 - mean_squared_error: 0.1568\n",
            "Epoch 20/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5783 - binary_accuracy: 0.8057 - categorical_accuracy: 0.7495 - mean_squared_error: 0.1542\n",
            "Epoch 21/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.6248 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7440 - mean_squared_error: 0.1560\n",
            "Epoch 22/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5894 - binary_accuracy: 0.8050 - categorical_accuracy: 0.7493 - mean_squared_error: 0.1546\n",
            "Epoch 23/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.6418 - binary_accuracy: 0.8027 - categorical_accuracy: 0.7416 - mean_squared_error: 0.1566\n",
            "Epoch 24/200\n",
            "10/10 [==============================] - 2s 230ms/step - loss: 3.6445 - binary_accuracy: 0.8025 - categorical_accuracy: 0.7412 - mean_squared_error: 0.1567\n",
            "Epoch 25/200\n",
            "10/10 [==============================] - 2s 231ms/step - loss: 3.6071 - binary_accuracy: 0.8040 - categorical_accuracy: 0.7449 - mean_squared_error: 0.1552\n",
            "Epoch 26/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.6034 - binary_accuracy: 0.8041 - categorical_accuracy: 0.7456 - mean_squared_error: 0.1551\n",
            "Epoch 27/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.6437 - binary_accuracy: 0.8023 - categorical_accuracy: 0.7401 - mean_squared_error: 0.1566\n",
            "Epoch 28/200\n",
            " 3/10 [========>.....................] - ETA: 1s - loss: 3.6532 - binary_accuracy: 0.8018 - categorical_accuracy: 0.7393 - mean_squared_error: 0.1570"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 2s 228ms/step - loss: 3.6307 - binary_accuracy: 0.8029 - categorical_accuracy: 0.7400 - mean_squared_error: 0.1561\n",
            "Epoch 29/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5726 - binary_accuracy: 0.8052 - categorical_accuracy: 0.7495 - mean_squared_error: 0.1539\n",
            "Epoch 30/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5062 - binary_accuracy: 0.8081 - categorical_accuracy: 0.7540 - mean_squared_error: 0.1513\n",
            "Epoch 31/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.6120 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7431 - mean_squared_error: 0.1553\n",
            "Epoch 32/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.5746 - binary_accuracy: 0.8049 - categorical_accuracy: 0.7473 - mean_squared_error: 0.1539\n",
            "Epoch 33/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5982 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7447 - mean_squared_error: 0.1548\n",
            "Epoch 34/200\n",
            "10/10 [==============================] - 2s 229ms/step - loss: 3.5586 - binary_accuracy: 0.8053 - categorical_accuracy: 0.7497 - mean_squared_error: 0.1533\n",
            "Epoch 35/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.6028 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7425 - mean_squared_error: 0.1549\n",
            "Epoch 36/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5682 - binary_accuracy: 0.8050 - categorical_accuracy: 0.7458 - mean_squared_error: 0.1536\n",
            "Epoch 37/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5591 - binary_accuracy: 0.8052 - categorical_accuracy: 0.7496 - mean_squared_error: 0.1532\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 38/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.6371 - binary_accuracy: 0.8019 - categorical_accuracy: 0.7390 - mean_squared_error: 0.1562\n",
            "Epoch 39/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.6352 - binary_accuracy: 0.8018 - categorical_accuracy: 0.7393 - mean_squared_error: 0.1561\n",
            "Epoch 40/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.5967 - binary_accuracy: 0.8033 - categorical_accuracy: 0.7449 - mean_squared_error: 0.1546\n",
            "Epoch 41/200\n",
            "10/10 [==============================] - 2s 222ms/step - loss: 3.5606 - binary_accuracy: 0.8048 - categorical_accuracy: 0.7481 - mean_squared_error: 0.1532\n",
            "Epoch 42/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.6319 - binary_accuracy: 0.8019 - categorical_accuracy: 0.7379 - mean_squared_error: 0.1560\n",
            "Epoch 43/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.5767 - binary_accuracy: 0.8042 - categorical_accuracy: 0.7442 - mean_squared_error: 0.1538\n",
            "Epoch 44/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.6513 - binary_accuracy: 0.8009 - categorical_accuracy: 0.7353 - mean_squared_error: 0.1567\n",
            "Epoch 45/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.5748 - binary_accuracy: 0.8041 - categorical_accuracy: 0.7438 - mean_squared_error: 0.1537\n",
            "Epoch 46/200\n",
            "10/10 [==============================] - 2s 220ms/step - loss: 3.5915 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7430 - mean_squared_error: 0.1544\n",
            "Epoch 47/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 3.5974 - binary_accuracy: 0.8030 - categorical_accuracy: 0.7417 - mean_squared_error: 0.1546"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 2s 228ms/step - loss: 3.5852 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7433 - mean_squared_error: 0.1541\n",
            "Epoch 48/200\n",
            "10/10 [==============================] - 2s 229ms/step - loss: 3.6002 - binary_accuracy: 0.8029 - categorical_accuracy: 0.7409 - mean_squared_error: 0.1547\n",
            "Epoch 49/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.6802 - binary_accuracy: 0.7994 - categorical_accuracy: 0.7330 - mean_squared_error: 0.1577\n",
            "Epoch 50/200\n",
            "10/10 [==============================] - 2s 229ms/step - loss: 3.5915 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7441 - mean_squared_error: 0.1543\n",
            "Epoch 51/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5691 - binary_accuracy: 0.8040 - categorical_accuracy: 0.7452 - mean_squared_error: 0.1535\n",
            "Epoch 52/200\n",
            "10/10 [==============================] - 2s 229ms/step - loss: 3.5726 - binary_accuracy: 0.8038 - categorical_accuracy: 0.7441 - mean_squared_error: 0.1536\n",
            "Epoch 53/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.6504 - binary_accuracy: 0.8006 - categorical_accuracy: 0.7345 - mean_squared_error: 0.1566\n",
            "Epoch 54/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5151 - binary_accuracy: 0.8061 - categorical_accuracy: 0.7535 - mean_squared_error: 0.1514\n",
            "Epoch 55/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5679 - binary_accuracy: 0.8042 - categorical_accuracy: 0.7418 - mean_squared_error: 0.1534\n",
            "Epoch 56/200\n",
            " 3/10 [========>.....................] - ETA: 1s - loss: 3.6112 - binary_accuracy: 0.8020 - categorical_accuracy: 0.7412 - mean_squared_error: 0.1551"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5903 - binary_accuracy: 0.8029 - categorical_accuracy: 0.7430 - mean_squared_error: 0.1543\n",
            "Epoch 57/200\n",
            "10/10 [==============================] - 2s 229ms/step - loss: 3.5437 - binary_accuracy: 0.8048 - categorical_accuracy: 0.7474 - mean_squared_error: 0.1525\n",
            "Epoch 58/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.6006 - binary_accuracy: 0.8024 - categorical_accuracy: 0.7405 - mean_squared_error: 0.1546\n",
            "Epoch 59/200\n",
            "10/10 [==============================] - 2s 220ms/step - loss: 3.5058 - binary_accuracy: 0.8062 - categorical_accuracy: 0.7518 - mean_squared_error: 0.1510\n",
            "Epoch 60/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.5291 - binary_accuracy: 0.8052 - categorical_accuracy: 0.7483 - mean_squared_error: 0.1519\n",
            "Epoch 61/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.5638 - binary_accuracy: 0.8038 - categorical_accuracy: 0.7424 - mean_squared_error: 0.1532\n",
            "Epoch 62/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5935 - binary_accuracy: 0.8026 - categorical_accuracy: 0.7414 - mean_squared_error: 0.1543\n",
            "Epoch 63/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.6063 - binary_accuracy: 0.8020 - categorical_accuracy: 0.7402 - mean_squared_error: 0.1548\n",
            "Epoch 64/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5268 - binary_accuracy: 0.8052 - categorical_accuracy: 0.7491 - mean_squared_error: 0.1518\n",
            "Epoch 65/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.6468 - binary_accuracy: 0.8004 - categorical_accuracy: 0.7345 - mean_squared_error: 0.1563\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 66/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5538 - binary_accuracy: 0.8040 - categorical_accuracy: 0.7457 - mean_squared_error: 0.1528\n",
            "Epoch 67/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5900 - binary_accuracy: 0.8025 - categorical_accuracy: 0.7399 - mean_squared_error: 0.1542\n",
            "Epoch 68/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5360 - binary_accuracy: 0.8049 - categorical_accuracy: 0.7447 - mean_squared_error: 0.1521\n",
            "Epoch 69/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5625 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7449 - mean_squared_error: 0.1531\n",
            "Epoch 70/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5343 - binary_accuracy: 0.8047 - categorical_accuracy: 0.7459 - mean_squared_error: 0.1520\n",
            "Epoch 71/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5203 - binary_accuracy: 0.8053 - categorical_accuracy: 0.7487 - mean_squared_error: 0.1515\n",
            "Epoch 72/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5278 - binary_accuracy: 0.8049 - categorical_accuracy: 0.7483 - mean_squared_error: 0.1517\n",
            "Epoch 73/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5594 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7431 - mean_squared_error: 0.1530\n",
            "Epoch 74/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5656 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7426 - mean_squared_error: 0.1532\n",
            "Epoch 75/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 3.5957 - binary_accuracy: 0.8018 - categorical_accuracy: 0.7412 - mean_squared_error: 0.1543"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 2s 227ms/step - loss: 3.5804 - binary_accuracy: 0.8025 - categorical_accuracy: 0.7424 - mean_squared_error: 0.1537\n",
            "Epoch 76/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5309 - binary_accuracy: 0.8048 - categorical_accuracy: 0.7449 - mean_squared_error: 0.1518\n",
            "Epoch 77/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.6029 - binary_accuracy: 0.8015 - categorical_accuracy: 0.7376 - mean_squared_error: 0.1546\n",
            "Epoch 78/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5297 - binary_accuracy: 0.8046 - categorical_accuracy: 0.7450 - mean_squared_error: 0.1518\n",
            "Epoch 79/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5273 - binary_accuracy: 0.8046 - categorical_accuracy: 0.7465 - mean_squared_error: 0.1517\n",
            "Epoch 80/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5446 - binary_accuracy: 0.8039 - categorical_accuracy: 0.7454 - mean_squared_error: 0.1523\n",
            "Epoch 81/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5136 - binary_accuracy: 0.8050 - categorical_accuracy: 0.7491 - mean_squared_error: 0.1512\n",
            "Epoch 82/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5017 - binary_accuracy: 0.8057 - categorical_accuracy: 0.7505 - mean_squared_error: 0.1507\n",
            "Epoch 83/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5464 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7451 - mean_squared_error: 0.1524\n",
            "Epoch 84/200\n",
            " 3/10 [========>.....................] - ETA: 1s - loss: 3.5691 - binary_accuracy: 0.8028 - categorical_accuracy: 0.7380 - mean_squared_error: 0.1533"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 2s 224ms/step - loss: 3.5115 - binary_accuracy: 0.8051 - categorical_accuracy: 0.7471 - mean_squared_error: 0.1511\n",
            "Epoch 85/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.5160 - binary_accuracy: 0.8048 - categorical_accuracy: 0.7472 - mean_squared_error: 0.1512\n",
            "Epoch 86/200\n",
            "10/10 [==============================] - 2s 223ms/step - loss: 3.5439 - binary_accuracy: 0.8036 - categorical_accuracy: 0.7433 - mean_squared_error: 0.1523\n",
            "Epoch 87/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5864 - binary_accuracy: 0.8019 - categorical_accuracy: 0.7387 - mean_squared_error: 0.1539\n",
            "Epoch 88/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5550 - binary_accuracy: 0.8031 - categorical_accuracy: 0.7430 - mean_squared_error: 0.1527\n",
            "Epoch 89/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5484 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7428 - mean_squared_error: 0.1524\n",
            "Epoch 90/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.5311 - binary_accuracy: 0.8041 - categorical_accuracy: 0.7470 - mean_squared_error: 0.1518\n",
            "Epoch 91/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5514 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7420 - mean_squared_error: 0.1526\n",
            "Epoch 92/200\n",
            "10/10 [==============================] - 2s 229ms/step - loss: 3.5542 - binary_accuracy: 0.8033 - categorical_accuracy: 0.7421 - mean_squared_error: 0.1526\n",
            "Epoch 93/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5137 - binary_accuracy: 0.8046 - categorical_accuracy: 0.7474 - mean_squared_error: 0.1511\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 94/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.5740 - binary_accuracy: 0.8022 - categorical_accuracy: 0.7377 - mean_squared_error: 0.1534\n",
            "Epoch 95/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5548 - binary_accuracy: 0.8030 - categorical_accuracy: 0.7433 - mean_squared_error: 0.1527\n",
            "Epoch 96/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.5727 - binary_accuracy: 0.8022 - categorical_accuracy: 0.7421 - mean_squared_error: 0.1533\n",
            "Epoch 97/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.5625 - binary_accuracy: 0.8026 - categorical_accuracy: 0.7410 - mean_squared_error: 0.1529\n",
            "Epoch 98/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.6313 - binary_accuracy: 0.7995 - categorical_accuracy: 0.7349 - mean_squared_error: 0.1556\n",
            "Epoch 99/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.5794 - binary_accuracy: 0.8019 - categorical_accuracy: 0.7397 - mean_squared_error: 0.1536\n",
            "Epoch 100/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5396 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7422 - mean_squared_error: 0.1520\n",
            "Epoch 101/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.6462 - binary_accuracy: 0.7987 - categorical_accuracy: 0.7332 - mean_squared_error: 0.1562\n",
            "Epoch 102/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.4834 - binary_accuracy: 0.8057 - categorical_accuracy: 0.7495 - mean_squared_error: 0.1499\n",
            "Epoch 103/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 3.5271 - binary_accuracy: 0.8038 - categorical_accuracy: 0.7454 - mean_squared_error: 0.1516"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 2s 224ms/step - loss: 3.5237 - binary_accuracy: 0.8039 - categorical_accuracy: 0.7459 - mean_squared_error: 0.1514\n",
            "Epoch 104/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.4952 - binary_accuracy: 0.8052 - categorical_accuracy: 0.7492 - mean_squared_error: 0.1503\n",
            "Epoch 105/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5889 - binary_accuracy: 0.8011 - categorical_accuracy: 0.7390 - mean_squared_error: 0.1539\n",
            "Epoch 106/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5861 - binary_accuracy: 0.8015 - categorical_accuracy: 0.7349 - mean_squared_error: 0.1538\n",
            "Epoch 107/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5227 - binary_accuracy: 0.8041 - categorical_accuracy: 0.7449 - mean_squared_error: 0.1514\n",
            "Epoch 108/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.5720 - binary_accuracy: 0.8017 - categorical_accuracy: 0.7394 - mean_squared_error: 0.1533\n",
            "Epoch 109/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.5649 - binary_accuracy: 0.8020 - categorical_accuracy: 0.7395 - mean_squared_error: 0.1530\n",
            "Epoch 110/200\n",
            "10/10 [==============================] - 2s 223ms/step - loss: 3.5889 - binary_accuracy: 0.8010 - categorical_accuracy: 0.7361 - mean_squared_error: 0.1539\n",
            "Epoch 111/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.4904 - binary_accuracy: 0.8054 - categorical_accuracy: 0.7481 - mean_squared_error: 0.1501\n",
            "Epoch 112/200\n",
            " 3/10 [========>.....................] - ETA: 1s - loss: 3.5601 - binary_accuracy: 0.8022 - categorical_accuracy: 0.7385 - mean_squared_error: 0.1528"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5706 - binary_accuracy: 0.8019 - categorical_accuracy: 0.7388 - mean_squared_error: 0.1532\n",
            "Epoch 113/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5584 - binary_accuracy: 0.8023 - categorical_accuracy: 0.7405 - mean_squared_error: 0.1527\n",
            "Epoch 114/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5113 - binary_accuracy: 0.8042 - categorical_accuracy: 0.7450 - mean_squared_error: 0.1509\n",
            "Epoch 115/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5001 - binary_accuracy: 0.8046 - categorical_accuracy: 0.7471 - mean_squared_error: 0.1505\n",
            "Epoch 116/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5070 - binary_accuracy: 0.8044 - categorical_accuracy: 0.7452 - mean_squared_error: 0.1507\n",
            "Epoch 117/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5426 - binary_accuracy: 0.8031 - categorical_accuracy: 0.7428 - mean_squared_error: 0.1521\n",
            "Epoch 118/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5785 - binary_accuracy: 0.8012 - categorical_accuracy: 0.7384 - mean_squared_error: 0.1535\n",
            "Epoch 119/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5226 - binary_accuracy: 0.8036 - categorical_accuracy: 0.7434 - mean_squared_error: 0.1513\n",
            "Epoch 120/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.4665 - binary_accuracy: 0.8060 - categorical_accuracy: 0.7502 - mean_squared_error: 0.1492\n",
            "Epoch 121/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5319 - binary_accuracy: 0.8033 - categorical_accuracy: 0.7425 - mean_squared_error: 0.1517\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 122/200\n",
            "10/10 [==============================] - 2s 222ms/step - loss: 3.5124 - binary_accuracy: 0.8040 - categorical_accuracy: 0.7429 - mean_squared_error: 0.1509\n",
            "Epoch 123/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.4986 - binary_accuracy: 0.8045 - categorical_accuracy: 0.7469 - mean_squared_error: 0.1504\n",
            "Epoch 124/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5136 - binary_accuracy: 0.8040 - categorical_accuracy: 0.7433 - mean_squared_error: 0.1510\n",
            "Epoch 125/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5189 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7432 - mean_squared_error: 0.1511\n",
            "Epoch 126/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5462 - binary_accuracy: 0.8030 - categorical_accuracy: 0.7424 - mean_squared_error: 0.1522\n",
            "Epoch 127/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5471 - binary_accuracy: 0.8022 - categorical_accuracy: 0.7413 - mean_squared_error: 0.1523\n",
            "Epoch 128/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5306 - binary_accuracy: 0.8033 - categorical_accuracy: 0.7428 - mean_squared_error: 0.1516\n",
            "Epoch 129/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5053 - binary_accuracy: 0.8040 - categorical_accuracy: 0.7454 - mean_squared_error: 0.1506\n",
            "Epoch 130/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5427 - binary_accuracy: 0.8025 - categorical_accuracy: 0.7422 - mean_squared_error: 0.1521\n",
            "Epoch 131/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 3.5363 - binary_accuracy: 0.8027 - categorical_accuracy: 0.7397 - mean_squared_error: 0.1518"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 2s 226ms/step - loss: 3.5328 - binary_accuracy: 0.8029 - categorical_accuracy: 0.7405 - mean_squared_error: 0.1517\n",
            "Epoch 132/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5291 - binary_accuracy: 0.8031 - categorical_accuracy: 0.7424 - mean_squared_error: 0.1515\n",
            "Epoch 133/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.4772 - binary_accuracy: 0.8051 - categorical_accuracy: 0.7479 - mean_squared_error: 0.1495\n",
            "Epoch 134/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5444 - binary_accuracy: 0.8022 - categorical_accuracy: 0.7401 - mean_squared_error: 0.1521\n",
            "Epoch 135/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.4895 - binary_accuracy: 0.8045 - categorical_accuracy: 0.7462 - mean_squared_error: 0.1500\n",
            "Epoch 136/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5377 - binary_accuracy: 0.8025 - categorical_accuracy: 0.7416 - mean_squared_error: 0.1518\n",
            "Epoch 137/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.4479 - binary_accuracy: 0.8062 - categorical_accuracy: 0.7511 - mean_squared_error: 0.1484\n",
            "Epoch 138/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.4826 - binary_accuracy: 0.8047 - categorical_accuracy: 0.7462 - mean_squared_error: 0.1497\n",
            "Epoch 139/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.4574 - binary_accuracy: 0.8057 - categorical_accuracy: 0.7500 - mean_squared_error: 0.1487\n",
            "Epoch 140/200\n",
            " 3/10 [========>.....................] - ETA: 1s - loss: 3.5588 - binary_accuracy: 0.8016 - categorical_accuracy: 0.7376 - mean_squared_error: 0.1526"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5096 - binary_accuracy: 0.8036 - categorical_accuracy: 0.7438 - mean_squared_error: 0.1507\n",
            "Epoch 141/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5107 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7435 - mean_squared_error: 0.1508\n",
            "Epoch 142/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5794 - binary_accuracy: 0.8009 - categorical_accuracy: 0.7363 - mean_squared_error: 0.1534\n",
            "Epoch 143/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5197 - binary_accuracy: 0.8031 - categorical_accuracy: 0.7421 - mean_squared_error: 0.1511\n",
            "Epoch 144/200\n",
            "10/10 [==============================] - 2s 223ms/step - loss: 3.4217 - binary_accuracy: 0.8070 - categorical_accuracy: 0.7534 - mean_squared_error: 0.1473\n",
            "Epoch 145/200\n",
            "10/10 [==============================] - 2s 229ms/step - loss: 3.5788 - binary_accuracy: 0.8006 - categorical_accuracy: 0.7354 - mean_squared_error: 0.1534\n",
            "Epoch 146/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.4677 - binary_accuracy: 0.8052 - categorical_accuracy: 0.7481 - mean_squared_error: 0.1491\n",
            "Epoch 147/200\n",
            "10/10 [==============================] - 2s 223ms/step - loss: 3.5282 - binary_accuracy: 0.8028 - categorical_accuracy: 0.7404 - mean_squared_error: 0.1514\n",
            "Epoch 148/200\n",
            "10/10 [==============================] - 2s 223ms/step - loss: 3.4780 - binary_accuracy: 0.8048 - categorical_accuracy: 0.7444 - mean_squared_error: 0.1495\n",
            "Epoch 149/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.5393 - binary_accuracy: 0.8025 - categorical_accuracy: 0.7417 - mean_squared_error: 0.1518\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 150/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5382 - binary_accuracy: 0.8026 - categorical_accuracy: 0.7376 - mean_squared_error: 0.1518\n",
            "Epoch 151/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.5185 - binary_accuracy: 0.8030 - categorical_accuracy: 0.7415 - mean_squared_error: 0.1510\n",
            "Epoch 152/200\n",
            "10/10 [==============================] - 2s 223ms/step - loss: 3.5504 - binary_accuracy: 0.8022 - categorical_accuracy: 0.7396 - mean_squared_error: 0.1523\n",
            "Epoch 153/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.4525 - binary_accuracy: 0.8056 - categorical_accuracy: 0.7498 - mean_squared_error: 0.1485\n",
            "Epoch 154/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.4630 - binary_accuracy: 0.8056 - categorical_accuracy: 0.7478 - mean_squared_error: 0.1488\n",
            "Epoch 155/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5458 - binary_accuracy: 0.8019 - categorical_accuracy: 0.7374 - mean_squared_error: 0.1521\n",
            "Epoch 156/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.5042 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7430 - mean_squared_error: 0.1505\n",
            "Epoch 157/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5026 - binary_accuracy: 0.8034 - categorical_accuracy: 0.7433 - mean_squared_error: 0.1504\n",
            "Epoch 158/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.4889 - binary_accuracy: 0.8041 - categorical_accuracy: 0.7450 - mean_squared_error: 0.1498\n",
            "Epoch 159/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 3.5526 - binary_accuracy: 0.8014 - categorical_accuracy: 0.7367 - mean_squared_error: 0.1523"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 2s 227ms/step - loss: 3.5597 - binary_accuracy: 0.8012 - categorical_accuracy: 0.7357 - mean_squared_error: 0.1526\n",
            "Epoch 160/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5353 - binary_accuracy: 0.8020 - categorical_accuracy: 0.7398 - mean_squared_error: 0.1516\n",
            "Epoch 161/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.4880 - binary_accuracy: 0.8041 - categorical_accuracy: 0.7452 - mean_squared_error: 0.1498\n",
            "Epoch 162/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.4512 - binary_accuracy: 0.8055 - categorical_accuracy: 0.7492 - mean_squared_error: 0.1484\n",
            "Epoch 163/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5241 - binary_accuracy: 0.8027 - categorical_accuracy: 0.7394 - mean_squared_error: 0.1512\n",
            "Epoch 164/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.4767 - binary_accuracy: 0.8047 - categorical_accuracy: 0.7460 - mean_squared_error: 0.1494\n",
            "Epoch 165/200\n",
            "10/10 [==============================] - 2s 222ms/step - loss: 3.4830 - binary_accuracy: 0.8042 - categorical_accuracy: 0.7450 - mean_squared_error: 0.1496\n",
            "Epoch 166/200\n",
            "10/10 [==============================] - 2s 222ms/step - loss: 3.4907 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7448 - mean_squared_error: 0.1499\n",
            "Epoch 167/200\n",
            "10/10 [==============================] - 2s 220ms/step - loss: 3.5532 - binary_accuracy: 0.8010 - categorical_accuracy: 0.7380 - mean_squared_error: 0.1523\n",
            "Epoch 168/200\n",
            " 3/10 [========>.....................] - ETA: 1s - loss: 3.5272 - binary_accuracy: 0.8020 - categorical_accuracy: 0.7413 - mean_squared_error: 0.1513"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 2s 225ms/step - loss: 3.4841 - binary_accuracy: 0.8041 - categorical_accuracy: 0.7453 - mean_squared_error: 0.1496\n",
            "Epoch 169/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5220 - binary_accuracy: 0.8025 - categorical_accuracy: 0.7402 - mean_squared_error: 0.1511\n",
            "Epoch 170/200\n",
            "10/10 [==============================] - 2s 223ms/step - loss: 3.5329 - binary_accuracy: 0.8019 - categorical_accuracy: 0.7384 - mean_squared_error: 0.1515\n",
            "Epoch 171/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.4989 - binary_accuracy: 0.8032 - categorical_accuracy: 0.7428 - mean_squared_error: 0.1502\n",
            "Epoch 172/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.5126 - binary_accuracy: 0.8029 - categorical_accuracy: 0.7409 - mean_squared_error: 0.1507\n",
            "Epoch 173/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.5340 - binary_accuracy: 0.8019 - categorical_accuracy: 0.7397 - mean_squared_error: 0.1515\n",
            "Epoch 174/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.4767 - binary_accuracy: 0.8044 - categorical_accuracy: 0.7438 - mean_squared_error: 0.1493\n",
            "Epoch 175/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.4820 - binary_accuracy: 0.8040 - categorical_accuracy: 0.7449 - mean_squared_error: 0.1495\n",
            "Epoch 176/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.4995 - binary_accuracy: 0.8033 - categorical_accuracy: 0.7421 - mean_squared_error: 0.1502\n",
            "Epoch 177/200\n",
            "10/10 [==============================] - 2s 229ms/step - loss: 3.4639 - binary_accuracy: 0.8047 - categorical_accuracy: 0.7469 - mean_squared_error: 0.1488\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 178/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.5184 - binary_accuracy: 0.8027 - categorical_accuracy: 0.7394 - mean_squared_error: 0.1509\n",
            "Epoch 179/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5049 - binary_accuracy: 0.8031 - categorical_accuracy: 0.7423 - mean_squared_error: 0.1504\n",
            "Epoch 180/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.4908 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7440 - mean_squared_error: 0.1498\n",
            "Epoch 181/200\n",
            "10/10 [==============================] - 2s 227ms/step - loss: 3.4721 - binary_accuracy: 0.8042 - categorical_accuracy: 0.7449 - mean_squared_error: 0.1491\n",
            "Epoch 182/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.4658 - binary_accuracy: 0.8046 - categorical_accuracy: 0.7466 - mean_squared_error: 0.1489\n",
            "Epoch 183/200\n",
            "10/10 [==============================] - 2s 229ms/step - loss: 3.5394 - binary_accuracy: 0.8017 - categorical_accuracy: 0.7370 - mean_squared_error: 0.1517\n",
            "Epoch 184/200\n",
            "10/10 [==============================] - 2s 230ms/step - loss: 3.5878 - binary_accuracy: 0.8001 - categorical_accuracy: 0.7311 - mean_squared_error: 0.1535\n",
            "Epoch 185/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.4746 - binary_accuracy: 0.8041 - categorical_accuracy: 0.7453 - mean_squared_error: 0.1492\n",
            "Epoch 186/200\n",
            "10/10 [==============================] - 2s 229ms/step - loss: 3.5121 - binary_accuracy: 0.8025 - categorical_accuracy: 0.7402 - mean_squared_error: 0.1507\n",
            "Epoch 187/200\n",
            " 9/10 [==========================>...] - ETA: 0s - loss: 3.4573 - binary_accuracy: 0.8049 - categorical_accuracy: 0.7436 - mean_squared_error: 0.1485"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r10/10 [==============================] - 2s 225ms/step - loss: 3.4508 - binary_accuracy: 0.8052 - categorical_accuracy: 0.7445 - mean_squared_error: 0.1483\n",
            "Epoch 188/200\n",
            "10/10 [==============================] - 2s 228ms/step - loss: 3.4379 - binary_accuracy: 0.8062 - categorical_accuracy: 0.7473 - mean_squared_error: 0.1477\n",
            "Epoch 189/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.4463 - binary_accuracy: 0.8057 - categorical_accuracy: 0.7481 - mean_squared_error: 0.1480\n",
            "Epoch 190/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.4342 - binary_accuracy: 0.8057 - categorical_accuracy: 0.7490 - mean_squared_error: 0.1476\n",
            "Epoch 191/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.4024 - binary_accuracy: 0.8071 - categorical_accuracy: 0.7525 - mean_squared_error: 0.1464\n",
            "Epoch 192/200\n",
            "10/10 [==============================] - 2s 223ms/step - loss: 3.4090 - binary_accuracy: 0.8070 - categorical_accuracy: 0.7526 - mean_squared_error: 0.1466\n",
            "Epoch 193/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.5501 - binary_accuracy: 0.8009 - categorical_accuracy: 0.7337 - mean_squared_error: 0.1521\n",
            "Epoch 194/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.4686 - binary_accuracy: 0.8041 - categorical_accuracy: 0.7443 - mean_squared_error: 0.1489\n",
            "Epoch 195/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.5804 - binary_accuracy: 0.8000 - categorical_accuracy: 0.7335 - mean_squared_error: 0.1532\n",
            "Epoch 196/200\n",
            " 3/10 [========>.....................] - ETA: 1s - loss: 3.4909 - binary_accuracy: 0.8035 - categorical_accuracy: 0.7401 - mean_squared_error: 0.1498"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "10/10 [==============================] - 2s 226ms/step - loss: 3.4859 - binary_accuracy: 0.8037 - categorical_accuracy: 0.7403 - mean_squared_error: 0.1496\n",
            "Epoch 197/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.4235 - binary_accuracy: 0.8062 - categorical_accuracy: 0.7493 - mean_squared_error: 0.1471\n",
            "Epoch 198/200\n",
            "10/10 [==============================] - 2s 226ms/step - loss: 3.5445 - binary_accuracy: 0.8010 - categorical_accuracy: 0.7349 - mean_squared_error: 0.1519\n",
            "Epoch 199/200\n",
            "10/10 [==============================] - 2s 224ms/step - loss: 3.5225 - binary_accuracy: 0.8021 - categorical_accuracy: 0.7365 - mean_squared_error: 0.1510\n",
            "Epoch 200/200\n",
            "10/10 [==============================] - 2s 225ms/step - loss: 3.4818 - binary_accuracy: 0.8036 - categorical_accuracy: 0.7404 - mean_squared_error: 0.1494\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAecAAAFKCAYAAAAnj5dkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nkeL3JALNY5EAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7fc21575e8d0>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "done training\n",
            "100/100 [==============================] - 0s 4ms/step\n",
            "[3.6384315490722656, 0.7969105839729309, 0.726097583770752, 0.15552771091461182]\n",
            "the accuracy for length 10 was: 79.6910569105691%\n",
            "100/100 [==============================] - 0s 556us/step\n",
            "[3.4163522720336914, 0.8067209124565125, 0.7424390316009521, 0.14689578115940094]\n",
            "the accuracy for length 20 was: 80.67208672086721%\n",
            "100/100 [==============================] - 0s 568us/step\n",
            "[3.5294976234436035, 0.8017615675926208, 0.7334146499633789, 0.15134131908416748]\n",
            "the accuracy for length 30 was: 80.17615176151762%\n",
            "100/100 [==============================] - 0s 663us/step\n",
            "[3.4582035541534424, 0.8046070337295532, 0.7439024448394775, 0.14849230647087097]\n",
            "the accuracy for length 40 was: 80.46070460704607%\n",
            "100/100 [==============================] - 0s 637us/step\n",
            "[3.3071625232696533, 0.8103252649307251, 0.7687804698944092, 0.14257843792438507]\n",
            "the accuracy for length 50 was: 81.03252032520325%\n",
            "100/100 [==============================] - 0s 638us/step\n",
            "[3.598930835723877, 0.7988618016242981, 0.7243902683258057, 0.15400083363056183]\n",
            "the accuracy for length 60 was: 79.88617886178861%\n",
            "100/100 [==============================] - 0s 671us/step\n",
            "[3.393568992614746, 0.8070732951164246, 0.754878044128418, 0.14594316482543945]\n",
            "the accuracy for length 70 was: 80.70731707317074%\n",
            "100/100 [==============================] - 0s 674us/step\n",
            "[3.2569894790649414, 0.8129540085792542, 0.7702438831329346, 0.1405734419822693]\n",
            "the accuracy for length 80 was: 81.29539295392954%\n",
            "100/100 [==============================] - 0s 728us/step\n",
            "[3.30767560005188, 0.8106775283813477, 0.7641463279724121, 0.1425677388906479]\n",
            "the accuracy for length 90 was: 81.06775067750678%\n",
            "100/100 [==============================] - 0s 667us/step\n",
            "[3.6528210639953613, 0.7966125011444092, 0.7229268550872803, 0.15604816377162933]\n",
            "the accuracy for length 100 was: 79.66124661246613%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "EMAVaxHKfXLA",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "##  Plot Generation For Further Analysis\n",
        "In this section we are going to produce a few more graphs that we will us later on in the **Discussion** section. The loss data for these graphs comes from the sections called **Running NTM with LSTM**, **Running NTM with Feedforward**, and **Running LSTM**."
      ]
    },
    {
      "metadata": {
        "id": "dz5RkzQmHcnC",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 677
        },
        "outputId": "ee323ec0-7734-4553-9d74-1663230e9a5a"
      },
      "cell_type": "code",
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "LSTM_arr = np.loadtxt(\"loss_history_ntm_LSTM.txt\",  delimiter=\",\")\n",
        "NTM_feedforward_arr = np.loadtxt(\"loss_history_ntm_NTM_-_feedforward.txt\",  delimiter=\",\")\n",
        "NTM_LSTM_arr = np.loadtxt(\"loss_history_ntm_NTM_-_NTM_LSTM.txt\",  delimiter=\",\")\n",
        "\n",
        "\n",
        "plt.plot(NTM_feedforward_arr, label='NTM_feedforward')\n",
        "plt.plot(NTM_LSTM_arr, label='NTM_LSTM')\n",
        "plt.plot(LSTM_arr, label='LSTM')\n",
        "\n",
        "plt.legend()\n",
        "plt.show()\n",
        "\n",
        "\n",
        "plt.plot(NTM_feedforward_arr, label='NTM_feedforward')\n",
        "plt.plot(NTM_LSTM_arr, label='NTM_LSTM')\n",
        "\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeEAAAFKCAYAAAAqkecjAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n4EBMAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7fc21345c048>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAecAAAFKCAYAAAAnj5dkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4yLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvNQv5yAAAIABJREFUeJzsvXlgHPV9/v/M7ux9ale7Oiyf8oVt\nDNg4QIwNGEMd0jSGtjFHUieQhjQkNA3wTeqkuN82pV9SpwlJSKBuaPKDNDEQmuY2V0KAGGwO3zay\n5Uu3VtJq793ZOX5/zLGz0l6SdyVZer/+wdLOjD6zEvPs874+jCRJEgiCIAiCmDIYJnsBBEEQBEHk\nQ+JMEARBEFMMEmeCIAiCmGKQOBMEQRDEFIPEmSAIgiCmGCTOBEEQBDHFYCd7ASqhUKyq16ursyMc\nTlb1mpMF3cvUhO5lakL3MjWhexlNIOAq+tq0dc4sa5zsJVQNupepCd3L1ITuZWpC9zI2pq04EwRB\nEMSFCokzQRAEQUwxSJwJgiAIYopB4kwQBEEQUwwSZ4IgCIKYYpA4EwRBEMQUg8SZIAiCIKYYJM4E\nQRAEMcWYMhPCCIIgiJlHT083tmzZjCee+BEWLlwEAPj1r3+B/v4+vPXWXnAch9OnT2HJkqUAgK98\n5Z/w1a8+iLlz5+GBB7Zp1/npT3fhG9/4N7z22lsV/dzvfOebWLCgFTfd9KHq31QVIHEmCIIgJpV5\n8+bjsce+jR07vqV9LxhswHe+8x/o6enGV77yRXznO/+Rd05b23vgeR4sK8vYa6/9AX5//YSuu5bM\nCHFOpnnEUhwa6uyTvRSNI2eG4LGb0RJ0TvZSCIIg8PTLJ7HveH9Vr7lmaRAf2bCw7HFLllyEdDqN\nt9/eh9Wr11R07WXLlmPv3jfw/vdfjb6+XrAsC5PJVPKc3bt/jR/96IcIBBpgsViwYEErBEHA1772\nL+ju7gLP8/jkJz+N1avX4LOf/RTWrLkC77zzFoaHh/Hww9+A0+nEgw9+CYCIRCKFL3zhi1iyZCke\nf/xRHDy4H6Io4JZbPoIbbthU0T2UYkbknH/y8gn843/tQ4YTJnspGt/56SE89ULbZC+DIAhiSvCp\nT30G//Ef34UkSRUdf+211+PFF3cDAF566QWsX39dyeMlScLjjz+KRx75Hh5++N/R1dUBAHjhhd/C\n76/Ht7/9OP71X7+Ob33r69o5DocDjzzyPVx55fvxhz+8jLff3otAIIgnn3wSDz74zwiHh3DgwLvo\n6+vFo4/uxCOPPIYf/vAJZDLpcb4LOWaEc+4ZTCDDCYilOFjMtsleDnhBRCYrIJbkJnspBEEQAICP\nbFhYkcutFbNnz8HixUvx0kvPV3T8JZdchocf/ioymTReeeVlPPzwN/DDH36/6PGRSAR2uwN1dT4A\nwMUXXwIAOHz4IA4ceBcHD+4HAGQyGWSzWe1nAEAwGEQkEsGGDSuxc+f38OCDD+KKK9bhyivfj6ee\n+gGOHDmEz372UwAASRIxMDCAWbNaxvdGKMwIcY4mZBFMZ6aGc87yIgAgPYWcPEEQxGTziU98El/4\nwudwyy1/qeWSi2EwGLBmzZX46U+fgdVqg9frLXm8JEkwGBjta1GUn8Msa8Jf/dWdBUPRRqMx7/z6\n+nr84Ac/Rnv7EfzgB0/iyJFDsNvt+NM//TA+9rFPjOVWyzIjwtoRRZxTHD/JK5HhsrIopzJTYz0E\nQRBTAZ/Pj3XrrsH//u9zFR1/3XXX46mnfoBrr91Q9liPx4N4PI5YLAae53Ho0AEAwLJlK/Daa68A\nAMLhITz++KNFr7Fv35vYt+9NXH311fi7v3sAx48fxbJlK/D6669CFEVkMhl84xtfq2jt5Zj2zjnN\n8eCy8iekqSKGGZ1zFiUJBoYpcwZBEMTM4LbbPoaf/eynFR176aWrYDabcc01pfPNgOy077zzU/js\nZz+FpqYmLFjQCgDYsGEj3nlnHz796TshCALuvPNTRa/R0jIb//RP/4BnnvkReF7EXXfdjYsvvgSX\nXbYad9/9CQASbr75LytaezkYqdLse40JhWJVu5YoifDXOxEeTKIvnMTfP/4GAODuP1uOK5Y1VO3n\njJfOUBwPfn8vAODRv1sPm6X0Z6RAwFXV92cyoXuZmtC9TE3oXqYm1bqXQMBV9LVp6Zx/fPw5nI6d\nwZfX3Kflm4Gp45zVnDMgr6mcOBMEQRCVcfToYXz3u98a9f3rr78RN9/8F5OwovExLVUhwkXRE+9H\nWsggEteJ8xTLOQNT5wMDQRDEdGDZshWjBpZciEzLgjCXSR7sEecSiCb1zrm61dHhWAYPfPd17D8x\nMKbzMlmdc56iFduxJIcHvvtHvH6oZ7KXQhAEMeOYluLsNDsAALFsPN85V9mlnugcxmA0g6NnhsZ0\nnt45p6u0piwvYih6/o3vKmd7YxiMpnFkjPdGEARBnD/TUpxdZtU5x/Occ7WEUGVQEcPheGZM5+Xl\nnKvknH/9xll86fE9GBhOVeV6A8q9DUaqJ/jny9vv9eNT//Z79A0lJ3spBEEQNWV6irMS1h7lnKsc\nQlaFazg+tklfGb76OedzfTHwgoRTPdGqXE+9t2q68fPlRGcEvCCivTsy2UshCIKoKRUVhD300EM4\ncOAAGIbBtm3bsHLlSu21DRs2oLGxUZuksmPHDrjdbnzpS1/C4OAgMpkMPvOZz+C668r3oVULLazN\nJRBNGmE0MBBEqeph7Zw4j805c9n8au2qrEUR0c5QAu+7qHrXC8c4CKIIo2HyP8epUZCBKeTmCYI4\nPyZyy8h/+Zd/xLXXXo+1a9flff/ll1/Erl0/gslkQjKZxG23fRQ33LAJf/u3fwNBEHDu3Fl4vV64\n3R6sXr0G11yzFh//+Mfx3HO/Qn19AAAgCAJuvvkmfPjDt+Cuu+4+7/elrDjv3bsXZ8+exa5du9De\n3o5t27Zh165decfs3LkTDodD+/rXv/41VqxYgb/+679GV1cX7rzzzgkVZ60gLBtHJG6F22FGIp2t\nvjjrwtqSJIGpcJjIWKq1D7YPgjsxgMsXld4KbSgqf0DoCsUrWkM51A8eoiQhEufgc1urct3zQW2L\nI3EmiOnFZG4ZyXEcHn30m3jyyV2w2x0YHh7Gffd9DtdcswGPPPI9AKNF/dSpo2hsbMJLLz2PLVvu\nAAC8885bsFqr95wsK8579uzBxo0bAQCtra2IRCKIx+NwOotvdXjTTTdp/+7p6UFDw8QO/sg55zgi\nCTdaAg6IolT9sLYizrwgIZHm4bSV3q5MhdPlnMvN19718gn0DCbxj59YgzkNhRvWuayAeEoe1N5Z\nLXHWhbMHo+kpJc5TKQ9OENOF507+Eu/2H6rqNS8LXoxbFv5p2eMmasvIQmQyGaTTKWQyHOx2B7xe\nL77//SfLnve+912Fl156QRPnl156Hu9731Vj/vnFKCvOAwMDWL58ufa1z+dDKBTKE+ft27ejq6sL\nq1evxn333ac5yFtvvRW9vb147LHHyi6krs4OljWWPa4SPLwFAJASU+AFEQGfHVlBRCLNIxBwob1z\nGG8c7sWtNy6B0VD56MxzvVH83++/iftuX4U5je681izGxJac9qKHNemGqTNM0fO4rKAVP728vxsP\nfPTygsfp3XJoOA2n23Zeg014QcRwLBeq56TiaxwP471WPCVHGYbjXFXXcz5MlXVUA7qXqclE3Yu9\nyzym52FF17SZ89Zf6F4yGQesVhM+//nP4Ytf/CL+5E+ug8tl1Y7PZBxgWUPeuWYzi82bP4Snn34a\nH/7wB/Dznz+ND37wA/je975X8v2yWk3weGyj1nTbbbfh9ttvwbp167Bu3TrcdNNNeS545HmnTgEt\nLY3o7j6HZHIIzc3NOHHiOLZu3Yru7u6q/M7G/AQfOe3z3nvvxbp16+DxeHDPPfdg9+7d2LRJ3t3j\nJz/5CY4dO4YHHngAP//5z0uGfcPh6lbgWlkLBmLDyr8NMLMGhFJZhEIx7Hr+OPYc6cO8oAOLZ5fe\nyUTP83vOoH8oief3nMG6lU15r50+NwQHW9kf9rDOlYaj6aJj4M72xiAqb/er+7vwwSvmIOAdveXl\nSaXdiQEgAThwvBetzZ6K1lKIgeEURAlwWFkk0jzOdIYRml359Tr647BZjKj3jF7reMfeiaKESEL+\nwNAfTqKvL5q3w8xkQOMIpyZ0L+Nj06wbsWnWjVW/rrr+YvcyNJRAOp2Fw+HH/PmL8JOf/BQcx2nn\nDg0lwPNi3rkcx2Pu3CV499396OwM4Ve/+g0efvgbEIRHS75f6XQWkUhq1DEf/egncf31N+HNN/+I\nZ575KR577HE88cRTsFisRc9LJDK4+urr8PTTz2HRoiVYuXIV4vEMEolMxb+zUiJetsonGAxiYCA3\nZKO/vx+BQED7evPmzfD7/WBZFuvXr0dbWxsOHz6Mnh55eMVFF10EQRAwNDSx/bIeiwvxbEL+t9MM\nq5kFx4uyK1Sqq09151c2Hz8bxj8+sbdogdeJTlnsz/REtbBvk98OAAiPoSis0j5nNUS9fIEfkgQ8\nv7ej4HHqWhbMcgMAukKJitdS6nqLWuQPLmo+uxz94SS+89whbH9iL773s8Nlj//NG2fxr0+9jUwF\n6YZYKgv1c6EgSmMuwiMIYurziU98Ek899UPwfPn6oLFuGVmKTCaNpqZmbN78F/jWtx6Dz+fH0aNH\nyp53zTXX4fe/fxm/+92LuPba68f98wtRVpzXrl2L3bt3AwCOHDmCYDCohbRjsRjuuusu7VPOvn37\nsGjRIrz11lt44oknAMhh8WQyibq6uqouvBxuqwspIQlAgttu1sK8aU7QHuynRrTkvN0Wwrn+OI6f\nDY+6niCKaO+SxfxsXxx9Q3I/sepQx9JOVWlBmCrOd2xaCr/bilcPdoMXxFHHhRXxXLnAn3deMcqJ\noVpwpUYVBitsp/rGMwfxTlsIANA7VL7f+t2TAzjRGcEv95wpe2wskf/+UlEYQUw/arllZDH27XsT\nDzzwee0DQSaTQSwWQ2NjU5kzAb+/Hi6XC8ePH8PFF68se/xYKBvWXrVqFZYvX45bb70VDMNg+/bt\neO655+ByuXDDDTdg/fr12LJlCywWC5YtW4ZNmzYhk8ngy1/+Mm6//Xak02k8+OCDMExwK47b4oQI\nETDy8DgtsFnkPG8qw2u9z+0jnHNIGeDRX2CQR0d/HBlFVHlBxMF2OZrQOsuN1w715Dm5oWga33zm\nIDZe3oL1lzSPupZaEMYaDSWL1Dr7ZZFd0OzB0rlevH6oFwORNBp99rzjhmKyUK1Y4Mf/vHq6pHM+\ndmYIO36yH1/YcimWz/cVPEYV45aAAzYLW5E484KIvqEkWme5YWaNOHY2DC4rwGwqXkeQUIrYfvvm\nOay9uGnUfemJKG1UHqcZkTgnF4XNLrssgiAuMGq1ZaTK449/Bz/+sVzwNW/eAtx//5fQ1nYcf/M3\nd8JqtSGbzeIjH7kNTU2jn92FuPba63HmzOmqa1xFOef7778/7+ulS5dq/966dSu2bt2a97rVasXX\nv/71Kixv/HgsciyfMXFw202wmeVbjSY4JBW3Go5lMKSrRFbFORQeLc4nOmSXvXi2F20dw3ivQw5x\nt86SnbMq+KIk4fu/OobOUBwHTg4UFmdF5D0OUxnnnIDfbYXDZtKEq3coOUrEBhXn3Ox3IOC1lnTO\n7d1RSAD+eLi3uDgrrtTvscLvtlQkzup76nFYYFEEOZLgCubIVRKprNaD/t8vtOELWy4teqxaqb2g\nyY13TwxoE8yKIUkS3jzWh1fe7cbmdfOxZM7ERm4IgqiMpqZmfPnL/6h9bbfb8YtfPJ/3+sjqabWt\nymAw4Gc/+432/Wef/UXJn6X/OXruuGMr7rhja8HXCp13xRVXYMGCZQCAP//zj2jfv+mmD5X8+WNh\nWu5KBchhbQBgWA4n0wdxnD0EYCl6R4x+PNUdhc9thShJCA3LD/y+As5ZzTf/yZrZaOsYhiTJzre5\n3gGjgdGc84tvdeKYEhYfihXOi3K8CNbIwG41YSBSOPwbTXKIJDhc0iqHqhvqZEEuNLpyKJqGw8rC\nYjaiJeDEuycGEElw8DjMo44NK2s6dGoQoigVLKpSQ8Y+txU+txWdoQSSaR52a/E/l2RaFmeHldVa\nyiJxWZz/69fHEEtmce9f5MI+kiS3n81rcoFhGBw+PYRYkoPLPnrNgE6cm2VxHizyvgFAIp3FN585\noKUh/ve10/g/t1dXnPvDSSR5CfYKiwAJgpgYaMvIKY5bcc4wcXhn6C0MSn1gbLM1cZ7f5MbpnihO\ndUdx+dIghmMZLZ870jlLkoQTnRF4nGasXOiH2WQAlxXhd1tgYBh4nRYMx2UX/uzv2+Gym2AwMJoQ\njoTLCjCzRtjMRqQzAkRJgmFEJXuXEtJuCcr5fdUtq+L80tudePNYH+77yKUYimYQrJMd6uygLM5n\ne2NYqQi7HnVN8VQW7d0RrehLz2A0DZfdBIvJCL8SVRiKpmG3Fu9tT6TlELXDaoLHKbeyqR9Y3j0x\ngHgqi3gqC7WUMJURIIgSnFYTvC4LTnZGEE9ly4rzvCa56K1Uzvn1gz1o74riklY/okkOx88No384\nhWAJF1+M0z1RmFkDZgXy7/0/f3kM4XgG//Y37x/zNQmCqB20ZeQUx22RH6Z2dwZ9yX4AgMEeRc+g\nLG6XLvSDYaDNaQ7p3HIkweUVTYWGU4gkOCxu8cJoMGCuMgzE75GFy+uS86B7j/WDF0R8+Or5aPLZ\nEU1weZtcqHC8CLPJAJuFhYTCBVqdSt64RREFVXzVDxevHujGyc4IXjvUg0xWgM8lC+KCZlm8Rha7\nqaj5aQA4cHJw1OuiJGEomtZE2eeWr1sutK06Z7uV1Ry7+j6qA1LO6OZ+xxUxd9pMcFhlp60eVwhV\nnAMeK9x2U0lxfuu9EBgG+PhNF2HDqhYAwB/HsfUlL4jY8ZN3sfOXR0e9NhRLIxxNj2otJAiCqAbT\nVpyNoiwuZl8IEuQHKGOPoWdQFr0Gnx2z6p040xsDL4haEZjZJL8l+qKw0z1yz5qaX56vuDdVwLxO\nCwRRwu/3d4FhgMuXBFGnNNEXavnRnLNSQV4o79yh5I1bAg5lXUb43Rb0hVPIZAVNvH/9xlkAgE/5\noLBAqR4f2SamEo5l4HNbYGINOHBy9D7U0QQHXpBQr1xP/QBSTpxzzpmF16mKcyYvN6zflEMtBnPY\nTHDYWOV7+e9De3cEP/jNcWR5QSsIczvM8HtsGIykIRYQxnAsg5NdESyZ7YXHYcblS4KwmI14/VBP\nweNL0dEfRyojaOkOPck0D1HKn/ZGEARRLaatOA8Pyw/ilCknQAZ7DP1KyNrrtKB1lhtZXsS5vrj2\nAF6qFA7160LbSUV4VEeoulPVzXqVMG5/OIUls71wO8ya4ywU2uaysnO2quJcwDmf7Y3BaGDQoCv+\navDZEY5l8N65YU1o1OurztlpMyFYZ8Op7ugoMcryAmLJLBrq7Fg2tw5dA4m8iAEAnFZEXR0gon4A\nOXxqqGiYXn6PVOesD2tzebnhMz26Bn6dODsV56wKvMoL+zrwhwPdOHx6CNEEJ79nZhb1HisEUcrb\ncSyr7PSltnKtXhIEAFjMRqxZGsRgNKPVAlRKm1L0l8rwedENQRS1savlxq8SBEGMh2krzqGQ+tCU\nBcpmtMNgj0IQZafjcZpx0VxZiI+eGdJEavk8uYJZL1pppbparUJevSSAj924GNdeNgsANKcIAGuW\nyqKgimWhLRc5Xm4xspnl640cRNIViqOjP47l831gjblfkSrUbxzpBQAsm5crctLPvm5tdiOZ4dE3\nlIQoSVpLVlgRszqXBSsXygPij5zJDYfJ8iKe/t1JMAxw5XJ5HnpzvQNOmwn7Tw7gge/+EXsO9466\nHwBI6ArCtLC22vKkcKonqoWB1RC202aCw1Y4rK26/xMdEUQTHNxKPlp19ceVdq0nn38Pn/76K/jx\niyew91gfGACrFucG5Vx9sdyvuO9YX8G1F+NEZy41oI+A6Me2prnqbqZCEAQBTGNx7uzJuaoGewBz\nHHPAmLKAKQOwHDrSJ7F4jhsMcuJsNDBYMkcukOrXjRNVXZNFCXkbDQZct6pFy5WqzlkvCmpYe6Tb\nFEUJvCDBzOqdc/4D/jUlP6qKiopaFKa6wy0bFsHEymvy68RZH9r++Wun8eATe9HWMYyw8kGhzmXB\nvEY5b949kOuJfuGtDvSFU9hwWYu2yYbDasLDn74Kf/UnSyBBwiv7uwq827nogt1qgtVshNlkQCSe\nC2t7HGZEExwGlAhFnjgrVeB65xxNclpe+b2OMGLJrCb6FykfSnb+8ijue/R1/O6dLhgYBi+81YET\nnREsbPGgTvlwBAALZ3ngsLI4fHqo4hyxXAQ4rH2t/z0mdetMZ8g5EwRRfaatOJ/qjAGC/NCf75mL\nZocsdAZ7DNaFh/Bfx57Cd488hqbZHE50RtAzmES9x6q1LOlzzqlsBmzjaYAt/CD2KkKweLZXC+mq\n4jCynYpTwq/5zjl3XV4QsedwLxxWFpcszN/+TBVnjhfhdpjREnDgfRcFR4W/1bD7gZMD2L1PHvl5\nonM4LwSuXkstkIvEM/jF62fgspuwef38vJ9rs7C49rJZaPTZca4/XjB3q3fODMPA67Agksg559VL\n5A8tbR3hvOOdutYrfc5ZXzx2picGQZTgVsR5xXw/vvyx1Vgx34dEmsf1q1vw9c+uxWrlg9FVKxrz\n1mYwMFg2z4ehaEa733L0DiURS2a1Knq9c07qIh2ZLIkzQRDVZ1q2UsWSHELhFNywIYsYFrjnwsRY\ngR7A6O8B4w7BYbKjK94DNPVCil+CVLgRrbPcsJiN8DjNeTnnLr4Npjnv4Vh0Py7CDaN+3oImNy6a\nW4cPXDlH+15dkZwzl5XD6mbWoBWE6R/2h04NIprM4vrVLZorVtELcGuzGwzD4GM3LsGfXjUvr6d5\ndtAJE2vAW++FtO+d7YtrQlPntsJmkQu3epUCuSNnhpDJCvizq+dpEYGRzG10oWcwiVA4lbcWIL9a\nG5DTBie7IggNp2E0MLhsUQAvv9OFE+fCWNzk0pyzo0hYWw1pzwo4tIln+jar1lkefGHLpchwAizK\nh5zP3Lyi4JAWAFgx34d9x/tx+PQQmusdo14fiRrSXjavDodPD+XNTlfvFaCwNkEQtWFaOuezfXLh\nkd0oP6Tne+ZivkduqWHruwEAdyz9C/ztZXfDxJhgbj0Ig7cP2bqTePLo06ivM2Ewmtb6nhOS/KDu\nSnUW/Hk2C4sHbrsMK+bn+opdNhNYo2FUzlmdDmY25aq19Tnn1w/JOd2RIW0AqHdbtS3dVHdsNhlH\nCSVrzLV7eZ1m2C0sOvpi2gYWaj68ye/AYDSDDCfgbK+cly7U96wyJyhfU31/9eTC2qo4WyBJcsVz\nncuC+U1yCuGEUmSVSI1updKHtdUK+RvX5GZ0ugsMVVGFGQAYhkGT31Fw9zN1Gtrh04NIZXj85y+P\n4vCp0a1kKmpIe81Fcg3BcCyXJskXZ3LOBEFUn+kpzr3yg32ZdwUuC1yMRkcQ9XYfJF4WDrPowsX1\ny7C4rhWfXPExgJFgWfwuzhn34o3et2CrkyeAqSHZtCRfryNxruKcJcMw8Lkso52z0nqjD2vrq7VP\ndA6j3mPF3MbRW4kZDIxWIb5AaecqxkKl7euDV83DnAYn+sIpdCsuuU4T59xI0LO9UTAMMDtQfNDI\n3Ab5tXN9o8eDJtI8rGYjjMp8WdXJ84KIeo8VdiuLRr8dJzuHIUlSnnM2sQZYTEYtrC1JEk73RFHv\nsWLV4gBUqS008axSfG4rZtU70HZuGE/86hj+eLgXL749+sPW3mN9+OFvj2P/iQHYLaz2gStcJKx9\nPuKs/1viBRGHTw2Oud2LIIjpybQW55sWXoNPXvwxGBgDGIaBISML1mzmYhgY+dZXBJagMbYWQsyL\nOdaFAACbU5kUpuSdM4x8vQSfQCg1uje4GHUui9I3nOuF1XLO+oIw5WGfyvCIJbMlN4BobZaLm+aV\nEedNV87BJz6wFNde1qwVd7V1DIM1GrQcb5NfDu92DyRwtj+OJr8jz4mOZLZynXMFnHMindUKu4D8\nCna1WK3Z70AyzSOa4JBIZzVRBgCHjdUEOxRJI57KYkGzGw6rSZvOVcg5j4Xl833geBFvKwV1XSNm\nkP/mjbN47H+P4JX93Uikebz/4kZ4HGYYGAbDsSJh7RKz0UvxxpFefO6br6JDqaT/xetn8O9PH8C+\nY/3juh5BENOLaSnOs4NOrFnWkFexCwC2+DwIET8WO1bkff9Dy65Cfeh6XDvnSgDyZhlArmgpa8w9\nxNuHz1S8jjq3BRLyi4m0nLPJoNvGUv456oeBUptF3HHjYnz1r6/Uzi2G227Gukua8yaaCaIEn8ui\nhX1V53ygfQAZTtCccTGcNhP8bivO9sVGRRASaR52Xa7a48i99+ogk6BPvq++cArxVL6YO6wmLayt\n9lqrw16WKhX0vhG/z7GyYoEc2nbbTZjb4MJgNKMJ7Utvd+KZ37fD57bgK391OR677xrcvnExDAYG\nHqd5REGYrlp7nAVhrx3qQTLD47lX2pHK8HhJcfGHSoTaCYKYOUxLcf7Q2vl48K4rR+Ue3VwruPfW\nwO/KF6HLFgfw1U9egaBLFgGekcPZKY5HMpuCaOAgcbLAtEfOVLwOn0udS60XZ9U568LaSrW2Ogil\nlDhbTMYxh3dn60RX/4FFdc77T8jRAFXESzG30YVYMpu3fzUviMhwQlnnrN+8I57iNQcPyMKf5gTw\ngojTPfni/GdXz8dff2iZlmcfL8vm+vCh98/D5z9yCZbOlX/XXQNxxJIcfvziCbgdZtx/62VY0OzO\n2+pSnZ2ufiAplHP+0fNt+Pv/eEMLSx89M4RdL5+AKI4OU6c5Hu+dk3PaB9oH8cPfHtdC5UfOVN7u\nRRDE9GVainMx7Mqezt4i4uYyy2LFQxbTdEbAYFoe0mGMNcBqtODUGMRZFUJ93lmfcx4Z1s45Zyuq\nSZPfrlV+q1XkgCygVrNRW1OhPPdI5ihCry8KU4VF75z1IWjVOTco+fKewSRSmXxxzvU689r88FnK\n6FKnzYSrljcWLPQaCwYDg5tArc0AAAAgAElEQVTXL8C8Rrc2s7wzlNAmrl2/albBlILXaQYv5PLk\nhXLOp3oi6BtKaj3xv3unC7v3duBc/+gUwLEzYQiihCWz5Q8Ie4/1w2I2YsUCHyJxDt0j2r0iCY4E\nmyBmGDNKnFUxVHuRR+I0yWKQhSySaY7HQEoWZ1ZwYb5nLvqSIRwdfA+/Of0S3urbj0im8AxrIBeG\nzRPnbC7nbGYNMDCMNoQkFCkf1h4PRoNBm9Gtd85ydXNOjOZU4JznFMg7j2yjAnKDWYDcRK+g4pzV\nGdsOvThrvc5ZhIZTcFjZoi1d1SAnznHNxRbb83nkh6xCrVRqMZsq1urvVN22Us9BJXT959e0aiH7\n6y6dhTXKyNGjp3NT2872xvB3334Nbx6tfLpZe1ek5KhVgiCmPjNKnK9a3ogrlzWgwVdY/CxGC1jG\niIyoinPOOZslJ1o98wAAjx74Pn55ejf+68h/Y9vrX8VbffsLXk91qfpRoDnnLBepuewmLexdSc55\nvKiiqobaVRp9smg31NnK5rEBaJPF3joeyrWaqTtM6cTUaTfBaGDAIDdaVHXqZ3oVcbbqnXOu1zk0\nnEZ9Dd4DPU1+OxhG3przvY4wzKxBC6OPRBVnNe+cN4REEWPVVatirYp0e1f+7mCSJOFg+yAcVhYL\nmt2448YlWHtxI266ai6WKaNjj+pGqqobtVQ6F3w4nsH/+9E7eOb3Jys6niCIqcmMEudViwP41J8t\n19p9RsIwDJxmJ1KiHFZMZXLO2Qo3LguuRL3Nj1XBlbhz+R24fs56AMDZaEfB6wW8NphNBvzu3S48\n9r+HMRzPaBOlzKwcYp/f5EY4lkE4lkFoOA2nzVSRSI4V9cE/Mq+sOudKQtqA7IivXtmEzlAc//OH\nUwAKO2cDw8DvtsLvsWrzwRmGQVO9QyuKG5lzBuQKal4Qx7X38lgwm4xoqLPjTF8MnaEEWmd5Rg19\nUfHqNvIAgFSah01JkaQ5AaIoaYKt1g+o4nxyhDh3hRIIxzJYscAPg4HBrHoH7vrgMrnYzmNFg8+O\n4x3D2gcftc2uMzS6fa0QbR3DEEQpr7qcIIgLj2k5Iex8cJoc6E/KBVJpTkA8KYcg7YwbjY4g/u9V\nX9SOnetuwUvn/oBEtvBISIfVhPu2XKpsyCC3yKgOVi04ap3lxv6TAzjROYzBSAqzg5WJ5Fi5fEkA\n//7ZtXnhZiAnyq3KPO5KuH3jIrR1DOO3b55TRmjmtovU8zebV4w6t7neqVVjF8o5qyFvtZ+7lrQE\nHFp+W52pXgjviLB2Ip2F02aCIEhIc/yIHHSuLQ4ABiJpRBIcPA4zRFHC88o41ZUL/CjE8nl1ePmd\nLpzqjmLxbK/WqtUVSkAUJRgMpfPuuZ20aDgKQVzIzCjnXAlOkwOcyAGMgBTHYzA1BClrgo0dXaTl\nMMmuM8kXn9e8qMWLr2y9HA4ri3N98bycM5AbFvLWeyHwglT1YjAVhmFGCTMgj7W879ZLcd2qWRVf\ny2pm8akPLQcA/Pz1M3nbReqZ2+ga5cibA7nRmeo+zvK/5XPVbSVrEdofSYtu4MrSIvlmQO+cc2Ft\nu9UEq4VFmhO0aWcACm4l2d4VQSYr4Ls/O4zXDvWgyW/HpYvy56arqKF1NZytOmeOF9EXLj8XvK0j\nopxHY0UJ4kKGnPMInErFNmPmkOKyGMqEIWWcBYdzWI1WGBhDUeesYmDkjSnO9sa0h7bqnOc1umFg\nGBw4Kbv1iRAlPQzDaNtkjoUFzW60BJ041RPVXOdI51wI/VzrQmFtdYrZRLwP6nATU4l8MwDUOXPO\nmRdEcFkRdkWY05yQNxM8zfGQJNlRGw0MBFHCya4IXjvYg/0nB7B0jhf33HJx0dSFmhrQwuM6V94Z\nSmjtb4VIprPaYJXxDkchCGJqQM55BC6T/MC22kQk+TgESYCYsWnbRephGAZ21oZENjXqtZEE62wQ\nREnbFcmsXM9iNmJ20ImsUig20eJ8Pixs8SDLi1qx0kjnXIim+pxbzS8Ik0VJ7Riqdc4ZAGYHZaFb\nWCLfDAA2ixEWkxHDsYyubYyFTXXO6XznzPEiJEn+AMMw8oCT/ScHcNHcOnxhy6Ulq9DtIzZD0Ttg\ndZpYMU52RaA2XCUprE0QFzQkziNwKO1UZhuPNOQQq5Sx5w2lyD/ejkQ2UfA1ParYqIU9+uu1zsq5\ntgtJnBe1yCF5teipIuccKOyc9W1VrJEZNd2tFgTr7Pj4B5Ziy4aFJY9jGAb1Xiv6hlNaCNtuUcWZ\nH+GcBc21ep0WzA7IH7x8bgvu/vByrTiuGKqjTimpAv12op39cUiShP0nBvI+EKioIW3WaAAviNoH\nPoIgLjxInEeghrVZC4+MQRVnmzYDeiQOkx1JPlV2SIS6c5RaVGTWOTU17wxUfwBJLVk0Sw5nq7du\nr0CcvU4LrEqKIL8gLPfveo+tbOFTtVh/SXNF/d2z6h3IcILmXlXnLEn5fexpjtdSF1azESsX+mE2\nGfCZzRfDbS8/2a2Yc3ZYWXT0x/HqwR5866cH8cK+0R0CbZ3DYBhgyWyPtpZakuWFUfPJCYKoDiTO\nI3Apzpk1Z8GzcuWwlHKWFGdREpEW0gVfVxlZfay2UgHy3sQAYDQwo/qQpzJ+jzXP4VYizgzDoNEn\n9xjrj9dvgjEVowdq8dgJxZ3arbmWN3X3MkB2ujlxZnHzugX45ueurnj0aC7nnGvNMrEGzGt0YTCa\nxrO/bweQPxK2P5zEr/acwZmeKOYEXahT/oZSNc47/+aNc/iH7+9F/3D5tA5BEGOjooKwhx56CAcO\nHADDMNi2bRtWrlypvbZhwwY0NjbCaJQfrDt27EBDQwO+9rWv4e233wbP87j77rtx44031uYOqozT\nLD+EDSYOYGRxFkuIs52VHXEim4SNLS4q6lxpFbMuh13vsSLgtcJuMU2YY6wWi1o82HusP2+7yHLc\nev0iDEbTo0K8DhuLTFaYkHzzWFFHibYp+zyrYW0AGNQJZYrjNVG0mo1gGAZWc+V1l+qxqnNOczxs\nZiNagk4cORPWQujRpNxzHUlw+PLONyGIEgwMg6tXNmlV3bVup1K30YzEM1Pyd0YQFzJlnxp79+7F\n2bNnsWvXLrS3t2Pbtm3YtWtX3jE7d+6Ew5HLJb7xxhs4ceIEdu3ahXA4jJtvvvnCEWfFOcPEwWCK\nw8a4kBJMRbdSVNupEtkk6m2Fe1cBOYTrsLJIpOUqXr0wMQyDL96+6oITZkBuFdt7rL+ifLPK4tmF\ne4qdVnlaWmACepzHijbuUx/WtqrirHPOXM45j2eYjMHAwGo25m0jajWz2s+v91gxHM8gllS21wyn\nIIgS3ndREB+9cQmcNpM2HKbWzlltC1SHyhAEUT3KWp09e/Zg48aNAIDW1lZEIhHE46XzTGvWrMEj\njzwCAHC73UilUhCEC6N6VM05Z41RMOYM7JLc/2ouUK0N5ItzOdTQdqHKYJ/bWrAPeaqj5ssrqdQu\nh1oUNhVdmN9jhcVs1Kqh7RZWy53nhbU5Qcv1WkvsjV0Ku5XNiTMnwGox4uJWP5bM9uITH1gKl92M\nmOacZfe6oNmj5fC1orIa55wziihz49w2kyCI4pQV54GBAdTV5QY0+Hw+hEKhvGO2b9+O2267DTt2\n7IAkSTAajbDbZdF69tlnsX79ei3sPdWxszYwYJA0yPfIcrL4lMo5A0CyInGWjy1W+X0hMjvoRLDO\nVtF2k+VQxXkqOmcDI4/aVHHocs6ZrADWaIDRwCDN8drgkPGKs83CIpnmIYoSMpwAm5mF227GF+9Y\nhYvm+eC2m/PC2gDythHVRovWOKytinKGJ3EmiGoz5rjbyKrke++9F+vWrYPH48E999yD3bt3Y9Om\nTQCAF198Ec8++yyeeOKJstetq7ODZasrWoHA+ATDZXEgmpGjAwZODsE2BFwFr9eUlkPZjFUs+/Pm\nz/LizaN9sFnYMa9tvPcyETz+9xvlTS4q3NKx2L385cbFaG3x4pKlDee9PWQtWDi7DqeU0aOzmtwY\nSuT2tHY7zOCyAnhBAqt8+GoIFv6bKYfbYUHPQAIOt/whxeOy5l3HX2fD2b4YXG4beEl+n+bO8mrH\nNNTL/zWajGP6+WNdq6T8jixW85T7+5xq6zkf6F6mJrW+l7LiHAwGMTAwoH3d39+PQCCgfb1582bt\n3+vXr0dbWxs2bdqEV199FY899hj+8z//Ey5X+ZsIVzCacCwEAi6EQqP30q0EO5sT52RYqXxNZgpe\nj0/KD6i+8FDZn+e0yIEK1sCMaW3ncy9TjVL3EnCa8SeXt2BgYGq25/hdOXeaTnJazhmQ3SoDCfFk\nFgPKvG4uxY3r92YyMhAloP2MPNfdwEh517Eo9Qqnzw2hR/m+mOW1Y7Kcko8eSlT888fzNxZXPpwM\njuHnTAQz5f+XCw26l8LXKUbZsPbatWuxe/duAMCRI0cQDAbhdMrFKbFYDHfddRc4Tv6fdN++fVi0\naBFisRi+9rWv4fHHH4fXW3xDgamK2k4liQbEhuUHcvGwtnxsZTlnOaxdahoVMXVp0YW17VY2rwrb\naTXBamZH9DmPbzqu2us8pBSa2UZcx+2Qw//RZBYRZacsj1MX1jbnjwCtFRyfm/tNEER1Kfv0WLVq\nFZYvX45bb70VDMNg+/bteO655+ByuXDDDTdg/fr12LJlCywWC5YtW4ZNmzbh6aefRjgcxuc//3nt\nOg8//DCam5trejPVQq3YllJOJFLyA6i4OMuhx7EUhE2nnPNMYlZQ/lDKGhmYWUNeNbbDZgIviEhz\nglaIdT45ZwAYUoabWC3511GHmUSTHCIJDqyR0QRdf3ytq7UzWrU25ZwJotpU9NH+/vvvz/t66dKl\n2r+3bt2KrVu35r2+ZcsWbNmypQrLmxzUXmcxmQs5lHXOJXamUnHZTPjgVXMrmkhFTD3cdrNceMXI\n7W/6sLbDKvdoC6Ic2gYA6zj35VYHkRRzzk677JxjSQ7RRAYehzkvR2+foGptTqvWJudMENWGdqUq\ngFOpwK5EnM0GE1jGWJFzZhgGf35Na3UWSUwKd33wIvCiXBSpd85Om0kbHKIO5zhf56z2T4/sl9ac\nc0J2ziP3AFfD6bWu1s5QtTZB1AwS5wK0eufDYrAiHcntuWsq0ufMMIw8X7sCcSYufFYsyA2a0Tta\nh80Eq9LeNBzPgGHy56ePBS2sHS0s8m6lbapvKAVekPLaqPTHJ2sY1hYlSdtYg+NInAmi2lBlUgEu\n8i3G/1n5JUhpObxtNhlgKNHa4zA5KnLOxPRiZFhbdazRBAermR13O5jap6zmnEc6Z5fSD67ucKYv\nBgNyU8ZquaezPs+coYIwgqg6JM5F0LuVYiFtFbvJhhSfhijRQ2omMTKsrf7NSFJOYMeD3SKLb1gJ\na490zi4lrN09IG9VOtI5q2sbq3N+fl8Hntz9Xtkd1oD8PDMVhBFE9SFxLoI+ZFlOnB0mByRISPK0\nO89MgjUatBnpDqspT0TH20YF5Aq61Balkc7ZYjbCYjJqr7sLiLPVbNRauirld+924XfvdmnbYpYi\noxNkEmeCqD4kzkUwmwxQo5LFNr1QcbCVt1MR0wtVkB02U54gj7cYDBjtugtdy2XPzTIv5JztFnk+\ntyRJ+N7PDuNXe86U/JmSJGFYCaP/8XBv2TXqBZn6nAmi+pA4F0He6k9+KFbinIHK5msT0wv1b0Qf\n1gYA23mI88hNRArtbqWGtgHA4xi9YYrVwkIQJQxG09h3vB9vHu0v+TOTaV5zw28c6YUglhbcDIW1\nCaKmkDiXQHVC5cW58p2piOmF5pytbJ6Ink9Ye6RzHtnnDABunXN2OwvknJV1qbPA1d2rijEYyaVk\nosksjpweKnl8nnOmPmeCqDokziWo1DnbxzAljJheNPkdCHitMJuMI3LO43fOFpNR6w5gmMLbleY5\nZ3vhgjAAON0ji3M8mQUvyCIqihJEMb/oa0DZ9vKSVrlVrFxoO5NXrU3OmSCqDfU5l0B9wBXby1ll\nLFPCiOnFnR+8CIIgC11+znn8/2sxDAObxYhEmoetSEuWS5mvbTUbC9ZEqH+7qnOWAMSSWdS5LPh/\nP3oHDiuLv/3LS7TjhxTnfNniAPrCKbx7YgCCKMJoKPy3r88zk3MmiOpDzrkEFeecqSBsxmIxGbVx\nm3nO+TxaqYCcuBZryVKnhBUqBtOv5UxvbuecSCIDXhDR3h3Bex3DeS1Tg4pz9rksmNfoQpYXEY4W\nD4VzVK1NEDWFxLkEtgpzzh6LGwAQTg/XfE3E1KVaYW0g105VbD63Wq1dTJzV87M6hzsc5zAcz0CS\ngDQnIJ7Kaq+p4ux1WVDvlbdJDSnfK4Q+rC2IEnhBBC+I2mAUgiDODxLnEmjOucyD1metAwMGA6nB\niVgWMUWpVlgbyG1+UagYDMg550I9zkBhUY8mOG0kKACEhnPiq4pzncuCeo8cCRoYLt63r4ay1Yh7\nlhfx2sEePPj9vTjZFSl6HkEQlUHiXIJKq7VZAwuf1UviPMOppnO2ac658HX8HtndBry2kucDQINP\n7iYYjme0na4AIKQT38FoCmbWALuFRcBTuXN2KqNEuayAvrCc1jlF4kwQ5w2JcwnUB2M5cQaAepsf\nES4GTuBqvSxiimIwMFrxYKHe5LGgiXMR59zkd+D+Wy/FTVfNLXJ+7m/2ojleAEAkwWk7XQEjxDmS\nhtdlAcMwqFcEfyBSyjnni3OGF5FIy+NCu5SxogRBjB8S5xJUGtYGZHEGgIFU6f5QYnqjimm1cs6l\nhpksm+eDY8TAEhV9OHzp3DoAQCQ+Mqwtiy8viIjEM6hzysNMfG4LDAyDgeHyzllt6eKyAhJKDrub\nxJkgzhsS5xKoD55iD0A9AUWcQxTantGoYnq+Oedctfb4rqPPOS9q8cJoYBBJZAo6Z7VIrM4ti7PR\nYIDPbUGopHOWc85qYRqXFZFUnHP3YKKizTMIgigO9TmX4IqLGmA2GbCy1V/22JxzJnGeyVTLOefC\n2uO7jl3XiuV1muF2mBGJc8hwAqxmI2wWVisIG47JqRjVOQNAvceK4+eGkeUFmNjRa9Ccsy7nnEjL\nzjmVERCOZeBzW8e1doIgyDmXxGI24spljTCx5d8mEmcCyInp+eactWrtcV5HzTk3+R1gGAYehxnD\ncQ6D0Qz8bisCHiuGYmnwgohwXA51e116cVbzzoVD21rOWQ1r84KWcwYo70wQ5wuJc5UI2HwAKKw9\n0/G6LGCNBk1cx8v8JjfMJgPmN7nHdb7VzOIj1y3E5nXzAcj90LwgIpXh4XNbEfDaIEnAUDSNsLIb\nVZ5zVnqdi4nzaOcsas4ZALpCJM4EcT5QWLtKWFkrnCYHOecZzu0bF+EDV8ypqMK/FLODTjx237Xn\ndY1NV8zR/u3RbY7hd1vgVYQ4NJzWtoqs0znnQJleZ44XwRoNWqQgkc6Cy4rwOmWHTkVhBHF+kHOu\nIgGbH4PpMASRxhnOVFx2M+Y0uCZ7GaPQbytZpzhnQC4KU8PaenEuNyUskxVgMRlgVj6EqO57fpMb\nRgNDYW2COE9InKtIvc0PURIRztAQBmJqMdI5a+I7nMJgNA2GyZ82Vm5KGJcVYDYZtb5uVZzdDjMa\n/Xaq2CaI84TEuYpQURgxVdE7Z7/OOb96sAcnOyOY2+gGa8w9DjxOM1ijoYRzFhVxVpyz4r4dVhNm\n1TuQ4YS8ti2CIMYGiXMVoV5nYqqid84+txUehxkm1oB4KouA14q///iavOMNDIN6jzXPOWd5QZvB\nzWUFWFgDLGx+WNthZdFcL2+hej5557ffC2k/iyBmIiTOVYScMzFV8SohawZybplhGFzS6kfrLDe2\nfXQ1muudo85pqLMhkea1YSX/3+73sG3nG4gmOXBZEWZzLqytFpU5bCY0+WVx7h0qHBIXRQm/e6cz\nr7pbT99QEo/+zyE894dT53XPBHEhU1G19kMPPYQDBw6AYRhs27YNK1eu1F7bsGEDGhsbYTTKn6B3\n7NiBhoYGtLW14TOf+Qw+/vGP46Mf/WhtVj/FIHEmpiqqc3Yr4WoA+MzNF0OSJDDq1lIjWLU4gAPt\ng3jjSC82rG7Bm0f7wQsiOvriECUJFjZXEKb2ONstLIJ1cshc3QhjJIdODeLJ59uQ4gTcdOXo2eDq\ntpO9Q1RURsxcyorz3r17cfbsWezatQvt7e3Ytm0bdu3alXfMzp074XA4tK+TyST++Z//GVdddVX1\nVzyFcZudMBvNFNYmphwm1ojZQSca6vJ3sSomzABw+dIgnnqhDX883AuHzQRekEd2dvTL4mk2GWEe\nMaDHYTOhoU7eBatvqLA494WVsaGxTMHXuwfl8/rDxceHEsR0p2xYe8+ePdi4cSMAoLW1FZFIBPF4\n6Q3VzWYzdu7ciWAwWJ1VXiAwDIN6qw8DqUGqVCWmHP+w9XLc/eHlFR9vs7BYtTiAvnAKP3/ttPZ9\n1dladAVhKg4rC4vZiDqXpag4q7tdRZOFd3DrGZQdcyLNI54qHPomiOlOWec8MDCA5ctz/0P7fD6E\nQiE4nbkc1fbt29HV1YXVq1fjvvvuA8uyYNmxzTepq7ODLTDD93wIBCa+33SWtwHdXb2wuAGPtXo/\nfzLupVbQvUxNCt3LB9bOx5tH+xBNZjG/2Y3T3VEtl+xxW9HS7Mk7fvYsLwJ1dsxucOHgyQG4vfZR\nA1liKTkEns6KBX+mvkI8C2Zc7/F0/71cqNC9VM6YJ4SNdIT33nsv1q1bB4/Hg3vuuQe7d+/Gpk2b\nxryQcJH81HgJBFwIhWJVvWYluA3yw+p45zks8BTea3esTNa91AK6l6lJsXuZVSdXdkcSHDa9bw52\n/uIozvVFAQACLyA8lADDAOpjIZ3IIMQL8Ck57qNt/WgJ5hebdSvOeyiSGvUzRUlCR2/ue++dHkCd\nbWyPqZnwe7kQoXspfJ1ilA1rB4NBDAwMaF/39/cjEAhoX2/evBl+vx8sy2L9+vVoa2s7z+Ve2FBR\nGDGdMBoM2LxuPi5dWI9LF9Yj4LWCF2QltpiMYBhGC20bGEYb59ngk/POvSNC25Ik6cLao0PWQ5E0\nOF6EU5nZTXlnYqZSVpzXrl2L3bt3AwCOHDmCYDCohbRjsRjuuusucJycO9q3bx8WLVpUw+VOfajX\nmZhuXHPpLNz7FythYg0IenMFZaooW5SiMIeN1QrMVHEeWbGdzPBIZeTxtvFkFuKISJxaDKZu0xoi\ncSZmKGXjRatWrcLy5ctx6623gmEYbN++Hc899xxcLhduuOEGrF+/Hlu2bIHFYsGyZcuwadMmHD58\nGA8//DC6urrAsix2796Nb3/72/B6vRNxT5MKOWdiOhOsswOQ/7ZVUZZFOgu71aQd11jEOQ8M5/LJ\noiQhmeZht7L4zRtncfnSoFYMtmKBD28c6UNfkfGhBDHdqSiZc//99+d9vXTpUu3fW7duxdatW/Ne\nX7FiBZ588skqLO/Cw2f1wsAYSJyJaUlQ14plVkLYqoN26rbJrPdYYWAYrW1KRd2CkgEgAYgmOPQM\nJvDTV07hnbYBtATklszZASfqPVYKaxMzFpoQVmWMBiN8Fi+FtYlpSUAX1lZHd6q9znrnzBoNqPda\nR7VTqfnmJmXEZyzJaRPITvdE8eaxPhgYBg0+O4J1NkQTHFIZvnY3RBBTFBLnGlBv8yPGxZHmCw9Z\nIIgLlTznbMp3zo4RVdWNPjtiyWzemE7VOc9vlKtUY8lsXqiby4oI1NnAGg0I1OW2tSSImQaJcw2o\nt8t558H00CSvhCCqS73HCnWomMWk5pyVgjCLKe/Y3KSwnLiqm1nMb3YDkAeRqIK9Yr4PANDsl89r\nUFw6hbaJmQiJcw2gim1iusIaDfC75b2gc9XahZ1zg2/0jO1QJAWbxagVjEUTnBbq/vgHlmJxiwdX\nLm8EoBafAf3knIkZyJiHkBDloYptYjoT8NowEElrk79U56zPOQOyywZyoWy5xzmNgMcGt10eUhJL\nZTEQScPjNMPntuJLH12tna9toFFkDChBTGfIOdeAeqscnhtIUVibmH7Ma3TBaGC0na60nLM1/7O+\n6rDVUHYizSPDCaj3WOFStrAcjmUwFM0g4MnfkAOQxZk1Mtosb1GS8MgzB/DrN87W5sYIYgpBzrkG\nuC1ysUucK71BCEFciPzZ1fNx9comeJ0WAIBZDWuPcM5+xTkPRmVxVsPX9R4rnDYWDIBzyvaTqsvW\nwxoNaAk40dEfBy+I6A+ncKB9EPFUtuBWkwQxnSDnXANsrOwCUny6zJEEceFhMRnR5M9tEet2yKLs\nc1vyjrOaWThtJs05h5Sq7HqPFUaDAQ6bSRNufwFxBoD5TW7wgoSO/jhOdkUAyFPGCGK6Q865BpgM\nLEwGE5I85cqI6c/Gy2djUYsXs0dscAHIoe2ewQQkSULPgDz9q1ERdpfdpG0JWcg5A8C8JhfwLnCm\nJ4rTPfJGA8k0iTMx/SHnXCPsrA3JLFWZEtMfi8mIxbO92lxtPX6PFRwvIpbMolsZzam2SqlFYQBQ\n7x2dcwZk5wwAp3tiJZ3zyBndBHGhQ+JcI+wmG4W1iRmPVhQWTaN3MAkza4BPcckuey5HXcw5N/sd\nMJsMOHJmSJvTneVFZHlRO0aSJHztv9/FN585UKvbIIgJh8S5RthYG5J8CqIklj+YIKYpai45NJxC\n71ASjT47DIrDViu2GQA+V2FxNhgYzG1wIRzLn7anH+nZ0R9HW8cw2jqGa3AHBDE5kDjXCDtrgwQJ\nGYGb7KUQxKShOucTHRFwvKjN1AZyYW2vywITW/xRpIa2gZzD1oe29x7rBwCkOYHmcBPTBhLnGqFW\nbFPemZjJqGJ66LQ8kKdJyTcDubB2sZC2yrwmuTWRYXIjPtWiMEmSsPdYn3bscJzm2RPTAxLnGmE3\nKeLMkzgTMxc1rK3Ox272j3bO5cRZdc6zA07UuVXnLFd5n+qJYiCShlqKNhwjcSamByTONcLOyg+R\nFIkzMYNxWFlYlH2fgQw+yR8AACAASURBVHznrI7nbAmMbsHSE/Ta8Kfvn4vN6xfAbpG7P1MZAQCw\nTwlpX7KwHgAQHqNzPnY2jB/+9jgEkWpDiKkFiXONsLPknAmCYRjUK25X3adZZU6DC/+w9XJsvLyl\n7DVuWd+KSxfWw66MCE0q21C+0xaCw8pi7cVNADCqcKwcrx7oxiv7u9EVSozpPIKoNSTONcJmkh9C\nKco5EzMcNbSt7tOsZ36TGybWWOi0gtgU55zM8BBEEYORNGYFnFpofDg2tgLMhJK7HquoE0StIXGu\nEWpYm5wzMdNRK7abdSHt8aKGtZNpHvFkFhIAt8OMOpc8OnSsYW3VgZM4E1MNEucaYdfma5M4EzMb\n1Tnr53GPFy2sneERScgu2eMww2k3wWhgxiyyqnMeInEmphgkzjXCRjlnggAAtDbL1dZL53jP+1q5\ngjAeUUWc3Q4zDAwDr9My5laqnHOmaX7E1II2vqgRWisV5ZyJGc6SOXX49ufXjdpScjzYdGFtvXMG\ngDqXBae6oxDEyuZsS5KkOWdqwSKmGuSca4SNwtoEoVENYQYAq9kIhpHD2nrnDMiTxkRJwnCFLpjL\nipqQU1ibmGqQONcIq9ECBgyFtQmiijAMA7uFRaqQc3bKRWHq/tHlSCghbYAKwoipB4lzjWAYRt42\nknamIoiqYreyec5ZH9YGKhdn/b7QNJebmGpUlHN+6KGHcODAATAMg23btmHlypXaaxs2bEBjYyOM\nRrlXcceOHWhoaCh5zkzBZrIhlU1O9jIIYlphs7DoC6c05+zSNtCQ/zsUTQONpaeOAfnOGZDds5rT\nJojJpuxf4t69e3H27Fns2rUL7e3t2LZtG3bt2pV3zM6dO+FwOMZ0zkzAztrQk4lM9jIIYlpht7DI\ncALCsQwcVlbb0SoX1q4slaQ6Z5uFRSrDIxzLoLn+/Nu9CKIalA1r79mzBxs3bgQAtLa2IhKJIB6P\nV/2c6YidtSEr8sgK2fIHEwRREXaluCw0nNKKwYCxh7XVSu1ZAVmQh0oUkkUTHH784gmkOQp9ExND\nWXEeGBhAXV2d9rXP50MoFMo7Zvv27bjtttuwY8cOSJJU0TkzAZu2MxXlnQmiWqi9zoIoaTtbAYBX\ncc5DOnEWRQmZrFDwOmqPs7rxRqmisDeP9eGFtzrwTtvMe44Rk8OYEyySlN9DeO+992LdunXweDy4\n5557sHv37rLnFKKuzg52DDN2KyEQcFX1emPF53QD/YDNbUDAfX5rmex7qSZ0L1OTC+Ve/HW6na38\njrx1u+wmDEZT2vf+7cm3cOT0IB75wrXwKOKtIilzvpfO9+P373YhlRWLvgeMQT5WADPh79OF8nup\nBLqXyikrzsFgEAMDA9rX/f39CAQC2tebN2/W/r1+/Xq0tbWVPacQ4XB1C6cCARdCoVhVrzlWDLz8\n9nb2D8CcGX8uayrcS7Wge5maXFD3otve0WJk8tbtdpgxFM1o3zvREcZgJI3Hnj2AOz94Ud5lBgbl\nZ47XJv9/2hOKF30PBpTnU3d/bELfpwvq91IGupfC1ylG2bD22rVrNTd85MgRBINBOJ1yGCgWi+Gu\nu+4Cx8lVk/v27cOiRYtKnjOTsNGezgRRdey6imqP05z3mstmQiKVBS/IAh5LyqHr1w714PjZcN6x\niYz8mt9thcVsLBnWVovHogmqHyEmhrLOedWqVVi+fDluvfVWMAyD7du347nnnoPL5cINN9yA9evX\nY8uWLbBYLFi2bBk2bdoEhmFGnTMToRGeBFF91M0vAOTlnIHctLBYMguPw4xEKguv04xInMOTz7+H\nr37yCjAMAyAnuHYrC5/LUlKc1R7oaHJsW1ISxHipKOd8//335329dOlS7d9bt27F1q1by54zE7HT\n5hcEUXX0zllfrQ3kep5jSQ5GAwMJwMJZHqQ4AUdODyHNCVovcyKdhYFhYDUbUeeyoGcwCS4rwGwa\nXfuSVMU5QeJMTAw0IayG0Hxtgqg+tlJhbbvcZhVLZjWX67Kb4bLJ39dPAUumeditLBiGKbsfdIrE\nmZhgSJxrCIW1CaL6lAxrK19Hk5yWb3bZTbDp9oFWSaR5OJTv+1xyfchQtLQ4x5JZiBV0nxDE+ULi\nXEPIORNE9aksrJ1FTOec9ftAA3J7ZzKd1Qaa+D2qOBeeSaCeJ0oS4ikqCiNqD4lzDXGbXTAwBrRH\nzkCUxPInEARRFtU5O6wsWGP+IywX1h7hnHX7QAMAx4vgBSnnnN3KdLEi4pzM5AaZUGibmAhInGuI\njbViTcNl6EuGcHjg2GQvhyCmBVazLKgjXbP+e9EEV9I56yu1gdJh7Swvaq1Z6rUJotaQONeYjXOu\nAQC8eO6VSV4JQUwPDAYG117ajKtXNo16TV8QVsg558RZfs2hhLVV51worD1yK0kSZ2IioP3Rakyz\nsxHL/UtxZPA4TkXOYoFn7mQviSAueP5q09KC37dbWBgNDGJJDqxR7md2280YssiOWC0IS4xwzlYz\nC4eVLRjWVsXZaTMhnsqSOBMTAjnnCeAGxT3/oXPPJK+EIKY3DMPA47QgmuQQTWbBQBZV+4hqbTWs\nrTpnQJ4UNhTNjNoLQD2noU4u8IzQIBJiAiBxngAWehfAYjSjK9492UshiGmPx2nWqrUdNhMMBkYX\n1pYLuxJKWFvfluVzW5HJCpqrVtHE2SdvuEHOmZgISJwnAIZhELTVI5QapKptgqgxHqcFaU5AOJbR\nctB2rVo7q/xXdc45cfa7C7dTpdIjxZlaqYjaQ+I8QQTtAWTFLCKZ6GQvhSCmNeq+zmlO0CaD2Ys6\n51xY2+cp3E6l5pzrnBaYWQM5Z2JCIHGeIAL2egBAf3KgzJEEQZwPbt1IT5fSWmU2GWBgmFGtVIWd\nc347lXqOzcLC7TBrY0FpEwyilpA4TxBBmyLOqdAkr4QgpjeqcwZyE8MYhoHdyhat1gbknDMw2jmr\n59gtRlmcExzefi+Ez3/rNew53Fu7GyFmNCTOE0SQnDNBTAhuh06cbbmwtc1iHLX1o342d9GcsxIK\nt1lZuO1mCKKEXS+fAAD8du+5UdXdBFENSJwniKA9AIDEmSBqjVcf1rbrxZnVwtnD8QzsFjZve0iP\nwwyjgSmac7YrYW0AGIikwQDo6I+jvYvqSIjqQ+I8QThMdjhYO0IpEmeCqCUeV84560d82i0sMlkB\ngihiOJaBV3ccIE8e8zotZXPOAMAAuOPGxQCAl9/prMVtEDMcEucJJGiX26kEUSh/MEEQ48JTNKwt\n55ejiSwSaT7PYav43RYMxzN5s7STOnH2KOK8ekkA1102C01+O/Yd76cKbqLqkDhPIAF7PURJxFB6\neLKXQhDTFk9eWFvnnJXir57BBID8wjEVn8cKSQKGYzn3nMzwMLMGsEYDLlnox6rFAfz5ta1gGAbX\nXTYLgijhzaN9tbodYoZC4jyBBG1K3pkqtgmiZtgsLEys/GgbmXMGgJ7BJIDC4twScAIAnnqhDVle\nds+pDK+dW++x4bO3XIyGOnkgySUL5ULPE12RWtwKMYMhcZ5AgnY/ACoKI4hawjCMJsoOXVhbHUTS\nrTnn0WHtGy5vwYoFPhxsH8Sj/3MIgijmifNI6j1WuO0mnOomcSaqC4nzBEIV2wQxMcxtcKEl4ABr\nzD3iVHHuGSge1jaxRnzulouxfF4dDrYP4lD7UElxZhgGC5o9GIpmEI6N3guaIMYLifMEErDJzpkq\ntgmitnz6w8vx5Y9dnvc9m+aclbC2a7Q4A7JAf/CqeQCAw6cHwQsS7BZjwWMBoHWWGwCq7p4TaXnz\nDmJmQuI8gVhZK0wGExLZxGQvhSCmNSbWCIs5X1Bz1dqy4BUKa6ssaHbDaGCw/+RA3rmFaG32AADa\nu6vb7/zvu/bjs//2OxLoGQqJ8wRjZ21I/v/svXe4JOddJvpWd1V1zqdPnhylkTXSWJKNJQdsCeTV\nLtisbQ3IwrDLfZbrC89eL+JhmYuRYXcGra+FgRV3DQuWQRgr2GNEkC3bwlawR5pRHE1OJ6fOuaor\n3j+qvq+r+vTp7jPTJ2im3ufRo+k+Fb7qqvre7/1FWVjrYThwcM3BWqoTsKdcNYPn3NgyHKY5z837\nWrF5KASGAS71MChM13VMp6vIlUT8zXfPOlXIrkE45LzK8HM+1BSHnB04WG1Y1W/Qx9GI7qWwa0O0\n5b7N8PIsRpNBjM+XbfnRV4KqqNBo8dfPpfHS23M9Oa6Ddw66IudDhw7h3nvvxf79+3H8+PGW2zz8\n8MO4//77AQCapuHzn/889u/fj/vvvx8XL17s3Yjf4fCzPgiK6PR1duBglWFVv62CwZrRLTkDwLbh\nMCRFw3S6cvkDtIAEl91y3QC8vBuHn7/Uk+M6eOegIzkfPXoUExMTeOKJJ3Dw4EEcPHhw0TYXLlzA\nsWPH6OfnnnsO5XIZjz/+OA4ePIgvfvGLvR31Oxh+zgcdOkTFiex04GA1YSXYaGhpfzPBtpEIGGbx\nvq2w1fQ7n5/qjWmbkPN1m+PYuSGKYlWiPagdXBvoSM5HjhzBnXfeCQDYtm0bisUiKhX76vChhx7C\n5z73Ofp5fHwcN954IwBg48aNmJ2dhao6JSsBwM8axQsc07YDB6sLv5Wc2/ibCXweFpsGQov2bYU9\nW+JwMQxePD7bE/9woWKQcyLipQVPFnLOnHEtoSM5ZzIZxGIx+jkejyOdblS4Onz4MG677TaMjIzQ\n73bu3ImXXnoJqqri0qVLmJqaQj6f7/HQ35nwsz4AQE2prfFIHDi4tsC6XeBNP3M3yhkAdpqm7XYB\nYQAQC3lwy+4kptNVnJm48rmOKOdExIv+mDFnLOSdOeNaQvsnrgWsq8JCoYDDhw/j0UcfxcJCo7bs\nBz/4Qbz++uu47777sGvXLmzdurXjajIW84Nll84lvBwkk6GeHq8X6FuIANOAJ8Asa3zr8VouF861\nrE9cC9cS8HGQynWMDka6ut5f+uh1CAU9+NCtm2ztJVvhU3ftwtHTKTx/fB4fuHXTZY2bQJCNmJRE\nxIedW4z6CJW6+o6/R+/08Vux0tfSkZz7+/uRyTSKZqRSKSSTRqWrl19+GblcDvfddx8kScLk5CQO\nHTqEAwcO2Mzcd955JxKJRNvz5Hu8KkwmQ0inyz09Zi+gS8YLPpvJYcDV3fjW67VcDpxrWZ+4Vq7F\nYxIsC73r673nPRtRLHSen+J+DtuGwzh2ah4nzi5gIO7vftBNmDMDyxIRL8olw5w9NlN4R9+ja+UZ\nW+5xlkJHs/btt9+OZ599FgBw8uRJ9Pf3Ixg0isPffffdeOaZZ/Dkk0/ikUcewZ49e3DgwAGcOXMG\nv/u7vwsAeOGFF3D99dfD5XKytgDHrO3AwVqCmKe7ida+HNx16wboAA6/cGXR1flyHV7eDb+XQyLs\nBetmHJ/zNYaOynnfvn3Ys2cP9u/fD4Zh8OCDD+Lw4cMIhUK46667Wu6zc+dO6LqOT3ziE/B4PPjS\nl77U84G/U+HnTHJ2CpE4cLDqIFHX7aqDXQlu2dWPbSNTOHYmhXefXsBt1w1c1nHyZRExs7yoy8Ug\nGfUh5ficryl05XN+4IEHbJ937969aJvR0VE89thjAACXy4WHHnqoB8O7+uCjytkhZwcOVhu3v2sQ\nsZBnybraVwqXi8Gv3XM9Hnz0KB579ix2boi2VOm6ruOx753D1qEw7rhxyPY3SVZRFRVsGmyYPPuj\nPsxla6gIMoKWTlsOrl44tuZVht8hZwcO1gzvvX4Q/+HfXAcXSWBeAQzE/fjkh7ajKir4/qtTLbfJ\nFEX86I2Zln8naVQxC6kT/7UTsX3twCHnVQYxawuOWduBg6sWN+/oA4Al20hOzBvBRHPZGjTNnslC\n9rGq+wEznSrl+J2vGTjkvMpwipA4cHD1I2Caniu11lW9JlMGOSuqhlTBPhcQco5byLnfUc7XHBxy\nXmXwbg6si3UCwhw4uIrh4dzgORfKgkHOsqLiD752DN8/ZpixJ+YbVRZn0vYWsvnK0sp5Ie/MG9cK\nHHJeA/hZn5NK5cDBVY6Qj6PKOVUQMTFfxvdfnYKu65hYaOTIzmbs5ZDzZpvKmIWc4yEjnWomXcHY\nXAm5krgKV+BgLeGQ8xrAIGdnBezAwdWMoI9HxVTO5aoEwAgEOzmeQ6kqYWO/US9iJtNaOcdCXvqd\ny8WgP+bHdLqK//Y3r+LBrx6FJNv7FVRFGePzpRW7HgerC4ec1wB+zmgb6TRQd+Dg6kXQz6Euq5Bk\nFaWaRL9/+qUxAMDNO5PwcO5F5Fwo1+F2MQj57SlTn/zQNnxg7xC2j0RQFRWcn7Z3wPrmjy7i4N++\nhlJVsn2vqBouzBQxk6lCqCu9vEQHK4hl19Z2cOXwsz5ougZRrcPHejvv4MCBg3ccQiQoTJBRtBDm\nxRlD3W4aDGG4z4/JhQoUVQPrNrRStiQiGuQXpXvt3d6Hvdv7cGIsiz9+4i2cHMthz5Y4/fvkQgWq\npqNYlRAONIqsPPfaNJ741wsAjOYfD/7qrRjpC6zMRTvoGRzlvAbwkYhtJyjMgYOrFkELORM1a+0L\nvWkghOG+AFRNR8oM9MoUBBQqEjYOLF1zeedoFKzbhRNjOfqdrutYyBlxLLWmvs/z5vdbh8NQVA3n\npwstj5sv1/HQ11/HdKrS8u8OVhcOOa8BaAlPx+/swMFVi6Bpli5byPm9e4xynmE/h2iQx0if4Xee\nNU3bpyeNdpO7N8WaD0fBc27s3BDBdLqCoumfrggyaqbJWqjbfdHk3D93+xbbuZpxajyHc1MFvHEh\n0/LvDlYXDjmvAUiVMMGJ2Hbg4KpFyJLrTAjyg3uHwQDYMhQGwzAYNs3LxO98ZsJQtddtXJqcAVBz\n9qlxg8ytKVa1ul05l2sy3C4G20ciAIC5JciZjLHZZ00g1BX88RNv4o1z6bZjc9AbOOS8BnCaXzhw\ncPUj6Df8vhVBRqkmgXUz2NAfxAP7b8J9P7MTADCaNMk5XYGu6zgzmUfIz2E42d4nvGezQc7EtE1M\n2gBQE+1BX6WqhKCfg9/LIhbyYDbbWhQQv3i51pqcXz+XxomxHF5zyHlV4JDzGqBRX9vJVXTg4GoF\n8TmXaxJKVRnhAA+GYXDd5jj6IsYcEAt5EA978NbFLE5P5JEv17FrY6xj7e/R/iDCAR6nxnOGv9mi\nnJsjsks1CRFzoTDcF0C+XG8Ztd1JOR87kwKwmPwdrAwccl4DrGRP5/nqAp6++B2omtp5YwcOHKwY\niFm7bCrnsH9xm0qGYfALH9gKWdHwladPAgCua+NvJnAxDPZsjqFYlTCTrtraSdYsxCvJKkRJRciM\n3h5KGMGos9nFpm2inIstyLkmKjhpqvSK2Lok6XIgyeq6TuvSNB1f+OpRPPnDC2s2Boec1wB+znhB\nVqL5xU/mjuF7Ez/EVGWm58d24MBB9yABYZmCCFnRbOlNVrx3zyC2DIVowZLdG6NdHf96i2l7wdIQ\nw6psSX512BwL8XHPZRYLg3bK+a0LGahmg45eKOevPH0Sf/g3r67bWg9lQcZkqrJkZPtqwCHnNcBK\nto2sq8aLVVdam6YcOHCwOiBmbVKes5VyBgwV/IsfMXzQkSCPQbPJRSeQoLCTY1ks5GuUgK2KtFQ1\nCJ8sDIYTBjm3U85VUYGiara/EZM2z7lQFa5cOU+lyljI1ZAurk/XHlmgiPW1s0A65LwGWMlUKlk1\nXhxJc8jZgYO1BOt2wedxI2vWyl5KOQPA9tEIfuWju/GZu3eD6bLXdDTowWgygNMTBYiSis1DYQB2\ns3ZDOTd8zsDiiG1F1ahyB4wIbwKhruDEWA4jyQBG+oKoivIVKV5d11Eyjz8+tz7LjRarxj2rraHp\n3SHnNQBVzitg1pY0k5xVh5wdOFhrEPUMtCdnAPjA3mHctL1vWcffsyUOzSTKkb4AeNZlMzuTmt7k\n3EEfh7CfW6Scy02tLa2m7fPTRSiqhpu29yHgZaGoOiTFrqyXg7qsQjb3H+sxOY/NlfDZP34el2av\n7LhUOUsOOV9T4FxG28iq3PuAMKqc1Ss3PTlw4ODKEPQ1CDkc4NpseXmwlu8ciPvh87AtlXPIYlIf\n7gsgUxBtjTOIUmzeDwAuzRo1vLePRGif6isxbZcsC4GxuXKbLZeP8fkyREnFuakr8xUXLWZtbY38\n4g45rwEYhkGIC6Is975MHlXOmkPODhysNazNKyJL+JyvBKSUJ2D0fPZ72SV8zo1xDPUFoKNR1tPY\nziCj/qjP9hloEOiW4TACXqP86JUEhZUtx56YL0PTekd+ZMGRL9c7bNke5Pp1AHVpbfzODjmvEcJ8\nCCWp3PNoxYZydszaDhysNQLe7s3alwOec2P3xigYBhhMBOD3sKiJCp1Xmn3OAGjAmZWcixVjuw1m\nG0uyn67rGJsroS/iRdjPw29eT3WZ6VRCXYFoLhrIsRkYJu65FsFpl4s6IefKlZGzNZ1srVK+HHJe\nI4Q9QSiaAqHHhUhkx+fswMG6gVU5rwQ5A8Av370Ln/vUXkQCPHweFqrW8AkTBWg1a8eCHgANQgYa\nZDRKyNn8nC4IqAgytg4bwWZBUzlXhO4JS6gr+MKjR/H7f3kEQMO/vcU8Zi9N24ScCz1SzsDaBYU5\n5LxGCPNG15my1FufC4nSdszaDhysPUhAmIthqL+21+iL+HDDlgQAwG+Sp2BRqX4PC45tTPVRk5wL\nFj8zIaPRpJ2cSWDVVjMSnFxDc+erdvj2i5eQLoi4OF2Aruu0POiN24wxj833LihMkoxFyZWata3K\nea3SqRxyXiMQci71mJxl1XgpHeXswMHagxQiCfm5jiU5ewG/x+4TLlclWh2MIBI0PhfKi5Xzhn4j\n1YoEbV0yo6m3DhtNMwj5V7v0OV+aLeG5V6cBAJKioVSTqR/8+k1xuF1MT9Op6oqpnCv1KwrkcpTz\nNYzQCpEzVc5OtLYDB2sOUsJzpUzazSD9omt1BZqmoyzItDgJQdQkZ2uEdrEqgQEQD3vh5d2UnMZm\nS3C7GGwcMBR1YJk+57/73lnoMLpwAUCuJKIsGMeOhz3YOBDC5ELFlmN9JSABYaqmo1Lr7piXZku2\n86uaZtt3rdKpHHJeIzSUc28jtp0iJA4crB8EV5mcrWbtiiBD1xefm2PdCHhZm8+ZdK5i3S6E/TxK\nVQmKqmFioYLRZBA85wYAGq3djXKey1YxPl/G3m0JvOd6o491tijSaO2Qn8Nt1/VD1XT85MT8lV88\n7JHVhSWCwlIFgVZAyxZFHHzsVTz90hj9e7kmQ4cRsAasc+V86NAh3Hvvvdi/fz+OHz/ecpuHH34Y\n999/PwCgWq3iN37jN3D//fdj//79ePHFF3s34qsEK2HW1nUdsuaYtR04WC8gxBhdLXK2mLVbRWoT\nRIIeG3kVqxIdazjAo1yTMTZXgqJqNBgMwLLynF83W0vesrsfibDh584URZRqMry8GxzrxvtuGATr\nZvDCW7M9yVypW3K3W/md57JV/O5fHMH3X50CYESs67oR+EZAFi2JiBfA2kVrs502OHr0KCYmJvDE\nE0/g4sWLOHDgAJ544gnbNhcuXMCxY8fAccaN+/a3v40tW7bgt37rt7CwsIDPfOYz+O53v7syV/AO\nRYg3Ay96SM6KpkCH8YA7Zm0HDtYeg3E/fvEjO3C9pVjISsJq1i41VQezIhLgMZupQjZ9tEJdQSQQ\nottruo7vHTMIjARuAbDkOXdHzi6Gwd7tfcgUDfLLlUSULR26Qn4e+3YmcfR0ChdnShjuC+D8dAEL\nuRp8Hhbv3zu8rOu3Vi5rlU41lapA1438agDIloxsGZsVwVzUDMb9yBRFCGsUENaRnI8cOYI777wT\nALBt2zYUi0VUKhUEg0G6zUMPPYTPfe5zeOSRRwAAsVgMZ8+eBQCUSiXEYp1boF1rWAnlLFsitJ1o\nbQcO1h4Mw+CuWzes2vmsZu3mjlRWUL9zRaL22whRzub2r59NIxby4F1bG+TMsW7wrAsV06ydKQgI\nB3hq9ibIlUSMzZVx3aYYgj6OquJMUUS5JmPzoJdu+4G9wzh6OoW/ffYsciXRZkbevSmGpFkYpRtY\nlXOrdCpCxhmz4UbW/L+1Ihoh6oG4HyfGcutXOWcyGezZs4d+jsfjSKfTlJwPHz6M2267DSMjI3Sb\ne+65B4cPH8Zdd92FUqmEv/iLv+g4kFjMD5Z1d9xuOUgmQz09Xm8Rgof1QFBrXY2zm21ytcaDqTLK\nur3+9Tquy0Gna3l74QxenDiKX7/l03C51neIx7V0X95JWM61DFfMRbnLBY0xnreRociiYwz3h4CT\nC2A4FiSIfKAviGQyhEFzWx3Az753MwYGwrZ9QwEedVmFL+DB7z38PHZujOLgr98Ol6sRjX70rGHS\n/uC+USSTIfTpOnjOjdlcDaqmoy/mp2N6fyKIv/v+OUynKwj5eXzqzp2Yz1TxwpszqCk6kskQzozn\n8L1XJvDZT+ylFdFaQVEbpnFB1hZdd9VUwblSHclkCFXTR12qSkgkgnC5GKiM4f/evjGG516bhs4w\nLe/BSj9jHcm5GVa/QKFQwOHDh/Hoo49iYWGBfv/0009jeHgYf/3Xf40zZ87gwIEDOHz4cNvj5vO9\nrTOdTIaQTvc2ErrXCLEB5GrFjuPs9lpStTz9t1AX1+X1vxPuS7fo5lqePfMijs6/jg8M3IEBf3KV\nRrZ8XGv35Z2C5V5LXTRUXyZfQ42oQUVddAzOJNLx6TwlVd7FIJ0ug2WMOZ4BsG9bfNG+Pt6NXKmO\n107OQZJVnLiYxePfPW2zEDz/upE+tX2oMf7+mA/TKSMA1sMytuP+2j3XY2yuhPfdMAifh8WrZ1J4\n4c0ZnL2UwaY+P576wVkcPZ3CbbuS2DYSWfL6BVFGNMijUJEwl6ksGvv0gvG5UKljeqaA2ZTxWdV0\nTEznEfRxmDW3CfDGIiBfFBYdp1fPWDuC70jO/f39yGQy9HMqlUIyaUwyL7/8MnK5HO677z5IkoTJ\nyUkcOnQI9Xod6t+MGwAAIABJREFUd9xxBwBg9+7dSKVSUFUVbndvlfE7HWFPCOOlKWi6Bhdz5arK\nMWuvP5D+2k6AnoPVAAkIE+oK9fO26g9NzNqFigTJ9Dv3mQFQxB98/ZY4+lqYlP1eDjPpKiYWGuT0\nrecv4sZtCQzE/ZAVo/HEpoEQ4uGG+ToZbZBzqClIbetw2BZ4NpgwxjxnlhidThslPpu7ZzWjrmgY\niPlQl9XWZm1L/+hMSaTmbcAIigv6OGriHogZYxDXa23t22+/Hc8++ywA4OTJk+jv76cm7bvvvhvP\nPPMMnnzySTzyyCPYs2cPDhw4gE2bNuGtt94CAMzMzCAQCDjE3AJhPgRN13rWncoaBOaQwfpAXTEm\nCCdAz8FqgJBzVZBxcaaEgZhvyYAwwFCQpHzm5kFDxe3aGMV1m2L42B1bWp4j4GWhAzg7aXR++vj7\nt0BSNHzr+YsAgMlUBaqmY3uTwu23LBKaybkZAzEfGADz2RpkRcN81pgjrb7hZui6DklS4eHciAY9\ni6K1dV1HptQg43ResG1TMgPIiub/YyEPeM61ZqlUHZXzvn37sGfPHuzfvx8Mw+DBBx/E4cOHEQqF\ncNddd7Xc595778WBAwfw6U9/Goqi4Atf+EKvx31VoFHCs0Kjt68EzcpZ1/WuG7c7WBk4ytnBaoLn\nXHC7GFyaLUGoK9i3o3V/6GioUV97fK6EoI+jqUMhP4/f/sWblzwHSac6N1UAz7lwz09txgtvzeHM\npFGek0RCbx6ym2yTsYYKbxWkZgXHupGIeDGXq2EuW6XVvsptyFlWNOgAPJwbHOvCXLYGSVZpsFpV\nVFCXVDAMoOvAxdkiVEtHrKJ57FJNRsDLgnW74OPZ9RsQBgAPPPCA7fPu3bsXbTM6OorHHnsMABAI\nBPCnf/qnPRje1Q1rOtUwBq/4eFYC0HQNqq6CZZYdVnBVoyxV4Gd9cLtWx5LTqHXukLODy8O3zv8T\nylIVv7Jnf8dtGYaBz8PSilfbR1v7Z6MBg5yn0hVkiiLetTXR9UKepFOJkoqtw2G4XAx2bIjg5ZML\nmM/VMN6kxAmSUYty7iLveygRwNuXsrbezKT0ZyuQSG2ec8PHG+93oVJHv2meJibtzYMhjM2VqfLv\nj/qQKggomVHaxUodEbP+uM/DLquOeC+xvsNHr3L0Op2KFCAhcNSaHcV6GZ//ySF8b+JHq3ZOYtau\nO/fCwWXijdTbeDPduvhTKxDTNgBsH4223MbDu+Hl3Q2VO9h95LG1DeZGs4vVDvM856eLGJ8vgedc\nGEoEbPv1W5RzqIsmIMRX/uqZFP2unVlbko0cZw/nopYBq9ma+Jd3bzRSe8fMmt6kO1axZlRFq4oK\nVfY+jxs1p/HFtYfek7N9hecEhdmRE3OQNQWXiuOrdk5CyrLjc3ZwmagqNcia0vVi22cqW7+HxVBi\ncTAYAVGHQKP2dTewdtci/Z93mP7lk2M5zGSq2DQQsqVWAUAy1hhLN+VMSVDY+eki/c7akKIZRDkT\nnzNgL0SSNQPktgyFwbMuatIm116qSLRSGDHx+zwsFFWDbCluslpwyHkN0evmF+Tl5Vyc7bMDA6Kp\nYlO19Kqds26as+uOWdvBZUBSZfoedxs4SpTz9tFI205YsWCDILcMLUc5N5T5hgFjv+FkAD4Pi9fP\npaHrwObBxWSfiHhpTnWwC+U8ZCpnHUAi7IHPw7b1OVvN2nGinEsW5WwGgyUiXkq+QKMdZrEmYcaM\nCietM328Gf2+Bs0vHHJeQ1DlXO9N8wti1g5wxkNdd9SaDYJqVgUS81C0lX/ZNF2jE+tqRmuPlyZR\nrPeuDZ+DtUOlXqX/rilCmy0bIOTcLh8YaCjnWMhjU9GdQMzaDIDRpGG6djEMto9EqBptDgYDANbt\nQjzkQdDHtS0kQjBoUf0jySDCfo62smwFyULOhHyt0dnE55yIeG1Vx4b7/PBwbpQqEqbTFfN8xnX5\nPPb+2KsJh5zXEGEzIKzcY+Uc5IwHS3bUmg2iYrycOnRkhNyKn28tUtsERcSXX/tf+PaFZ1blfA5W\nFtaFe9fK2VS2OzqRs2laXo5J23r8ZMwHL99Q0TsswWdL+bA//TO78Jm7d3V1nkiAh89jBHaNJoMI\nBXiUa9KSfZrrFp8zydm25jVniyJ4zoWQj6N/93nc8Hs5hAOcTTmP9JnK2SRncQ38zk4o7xqCc3Pw\nsz7k64XOG3cB4nMm5Ozk1tpByBkA0kIGg4H+FT2fNUJ7taK1S1IZiq6iJDnK+WpAxdJSttYlOX94\n3yjCAR47NrQnZ+KXXY5JG2iQejMBE3L28m4MtCh8AgB7t7dO7WoFhmEwGPdjbK6M0f4A7SBVFeSW\nedKSxefs93LweVg7OZdEJMJeMAyDvoihnEmRlEjAg0uzJUylKgh4WVqkhSwO1iLX2VHOa4yhwABS\ntUxPAoZIAFjDrO0oZysECzkvrILfua5YyHmVFkpV2Vj5i2rrXrYO3lkoWczaVaU7ct40GMK//+A2\nuDvUcr9pRx92jkZw63UDyxpTPOzF//XxG/CJD22zfb9lKAy/h8XODdG2vu7lYMT0/W4aCNEI6lJN\nhqJqmFywWxytAWGAUfEsUxSh6zqEuoKqqFBzdzJq/D9hkjPpxJUqCBhJBmlaWUM5rz45O8p5jTES\nHMLF4jjmqgvYGB69omMRgg/yplnbIWcbrISVrmXabNkb1C3nWy2zdkWqmud27v3VAKty7lUlQYLB\nuB//9dPvvqx9371rsdWJ59z4wq/eCq+nd7Ty8fdvxb6dSQwlAlQtl6sS3r6YxZM/vIDf+aWbsctM\njbIGhAEG8U6lKqiKCi3lSRTzoJnmRdK1IpboceJHB+wtOFcbDjmvMUaCQwCAmcrcFZMzVc6s3/bZ\ngQGrck6tAjnbzdqrcy8q5gRO8qu7QUkqg6nJADpH0DpYXViVc03uLiBsLdGqFveVIBbyIGZGXpP0\nq1JNwoUZI73qrYtZSs6NPOeGcgaATFFApmC8+/3m+Eb6Avgv9+6lUeXW1C6i1gGLcl6D+toOOa8x\nKDlX5674WEQ5B0zl7KgnO4jPmXdxSAmroJxtZu3VuReXY9b+328/hqpaxe/f9tsrNSwHl4nKZQSE\nXa2g5FyVMGV2kzo13gjsbJi1DXN+whIURpp0bBxoEO8NWxp9qm3k3GdRzrzjc75mMRQYBAMGM+Ur\nJ2epKSDMKXxhB1HOG0IjKNSLK754WROztkzM2nVbe9d2KNVLSFezXW/vYPVQkqypVOuPnH8w+TzG\nipOrci7ic04VBKRNJTy5UKFVw6hZm29WzlZybh381sms7aRSXYPwsh70+eKYqc5d8eRIfc4kWttJ\npbJBVOtgGTeGTWvFSvudreTfa3LWdR0/nnkFedEe6U/IWdO1rnO5ZU2GpmsQVbHzxg5WFetZOZek\nMr594V/w3OTzq3I+4nM+PWH0redNhXzG/EyVM0vI2TBhZ4oiJufLSIS9SxY/Ico5FvLAbylPupYB\nYQ45rwOMBIdQlWsoXmH6i6TJYMDAxxorRsesbYegiPCyXvT7jXSOlTZt28i5xz7nmcoc/v7st/CD\nponROoF3e/9J8ZrqGvk0a3INh45+GW+lT6zJ+dczyvUqWBcLH+vtugjJaoH4wFdrniEESnKR3/+u\nYQBGyVDAUoTEVM7ErH1xpohSTcamNvXDo+axR5P27oBrGRDmkPM6gDUo7EogazI4NwfebTxoTkCY\nHSIhZ59JziuunOuWf/d2AivLhqJqrgRGfM5A935nkh9v3Xc1MVOZw0xlDiezZ9fk/OsZZamCIBdA\ngPWvO+VMLC2rRc5+L2tL0brjxiEEvCxOjeeg67qlCIlBzgEvCw/vxrjZ3GPTwNJteRMRLz7109vx\nsffbe1iTPOe1CAhzyHkdoFfkLKkyeBcH3mWQs+NztkNQRfjcHiR9RiBIRsiu6Pmsk1av7wUpSFGS\n7KVfKxaCrXdBzrquU+W8VtHAVVMRrjfyWQ8oS1UEOD/83PojZxLDsVqVCF0Mg5Dpd3a7GAz3BbB7\nUwzZUh2pvGApQmLQGsMw6As3amgv5W8m2979no22ammaroFjGbhdDG3BuZpwyHkdYCRomGd6opxd\nHHi38QBfS2bthWqqrc+e1Ln2sl7aR3ulJzviZ3YxrstqfNGuIhQxcVZkOzlXJatZuzM5W/3SK62c\nc2IeDx37U0yWpm3fk+tcK+W+XqFoCgRZRIALIMD5IWvyFRWzOZk9gwMv/beeWYxII5mVKLCj6zpU\nbbFaJX7nwYQfHOvCTrNV5dh8CXVT3fJso1e7tcFFO7N2M1RNxReOfBFPnvsHJKM+zGdrqx4w6ZDz\nOkDcG4XX7cXbmVP4weTzl62yZFUG7+YtZu1rg5zfzpzCH77yJZzInl5yGzKReFkvvKwXDJgVj34l\n5BjkAlA0BZq+dNs5WVPweuo43eZScQK//eIXcCLT+pqIyi1blLOqqTa/pNhFrrO1zWh1hX2aFwpj\nmCrP4NjCG7bva45ybgkSAxDg/PCzRnDTlTyzJzJnUJTKbd+T5YAo55Vwnz0z9n38Pz8+uEhgRAKG\n8CCtKpNmj+hsUYSkqOBYl61VJSHnSICn5Uq7wUItjayYw6ncOQz3BVCrKyi2aVe5EnDIeR3Axbhw\n766PwcW48e0L/4I/f+uvL+s4kiaBd7HgXEYQw2qZtU9kTmOqPNNxu5osoCj2vubzufxFAO1LcpKJ\nxOv2wsW44GW9K27GJRMLUertIra/N/FD/PWJv8PJ7BkADSvKRGmq5faE0GqKQNVvTRGgo7G678Zy\nYp1YV1q5kt+7+VkhpFy5gvN/7eQ38NjpJy9/cOsQ5H4ETeUMXJnrIW0GQI73KPVJNJ/BlUgTnK7M\noSxXkBfztu9DZuAWIWdrg4u6rFF/MwH5+3JUMwDMVucBGNae/oQxn85mVtey45DzOsFtg/vwBz/1\nO4h7Y5i+DPM28R1ybg4uxgXOxa2KWVvVVPzl23+Lp8493XHbR0/+PT7/3Jd6PgYy2bebuEjwColk\n97O+FY9+JZNWiDPJeQmFoes6js69BgAo1I3KR8TUm68XW+5jNXkTUiOTOcuYQSzLNGuv9GKFqL6p\n8ozNitAwa1++6fBE9nRLK4OiKbbKcATH0yfxd6efamvNWGuQ+0l8zsZ3l6+ciTl7vNQbchbM52sl\nlDN5dypN1xsNGOq3mZwzRRF1SaX+ZoKkmU7Vzt/cCrOVefpvb8QYw4xDztcuApwfUU8EoiJ2PUnV\nVQmpWgaqrkLTNRoMxru5VYnWrsg1qLqKnNi5s9ZEaQqpHhe70HWdLmbakS1Vzqzxcvu53pHzUtdD\nFkek1vlSCuNScQIZ0UgHIZMvaXLQnMdMYB172SxUQSayuNesNdyFWdvqL+y2scLlgoxZVOtIW4Lx\niDld1dXLatih6RoERURVrtnItlgv4Y+O/gkOvvLHi/b5ydwxHJk7hrzYevGzHkDupxGt3b1ZuyrX\ncD5/yfadoinImSo0K+ZRrF95m1pScU9W5Z77Y8m707wY+fC7R/Cpn96O6zfHAQBenkXQxyFdFFGX\nVVpXm2Dv9j587I4tuPOW5ZVGnrVUbNQ8hrVvziHnaxte1gMdelfBPADwjxe/g4OvPIys2Z+Ycxsm\nGN7FX1HjC0ER8T+O/SneTL3ddjsSkFSSym1VSE0WUFWMyVPVe5eWkBXzEIiJt42qIBOJVTlLqkSD\nTlRNvSwV9fXTT+GPjv1Jy33rah1uxk39hUsFzhydf43+m0xGRMUWllTOVnIum/sak0fCF6fn7wR7\nQNgKk7NlzFbTtmD5/nJM64KlTze5hkK9iD954yuYr6WQrxcW3Z+qpZLaekVr5dx5Qfnd8efwJ298\nxRb4lRVy0KGDgeGP7YV6tvZHl7sseNMtSLxM8zPZF/Hh7vdstKVU9UW8pll7MTlzrAs/d8cWhFu0\nmGyH2co8/a2qyIJhHLP2NQ+f2yCPbhXEWGkSiq5i3PRNci4jYIJ3823N2mWpgq+e+PqSymy2Mo/J\n8gzeSLcnZxKQpOpq28ndmrYk91DRT1sm+XZKWLT4nAFYAmwMn+3vH3kIh8//87LPfyZ/ATOVuZbV\nxuqqBI8lQK/V/SCBYCROoEHOxKxdaKlK7MrZuAfEvE2UczfPkNW60m2/4MuFdcyT5UbEtlWxE/X7\nd6efWjIYbtFx5cW/xddPfxOpWqYRf9FEHuR3Xs/kTJRzwOJz7mbxQhbqC7UU/Y4U3NkRM9o89oKc\nre6C5QafPj/9Ezx98TtL/r2hnDtfb1/EC0XVICuLfc6XA1ERkRXz2BrZBBfjwnxtAf1RH2Yy1VWN\n2HbIeZ2BKLtWfrJm6LqO+eoCgEbjjG7N2sczJ/Fa6q1FkbMERI2ma+1zga1BPM0FMaywmjEl9fJW\n2ZquYc68XoKpyiz9d7vFAfGPUeXM+eg+xXoZhXoRF4vjHcdQkspUzSqqQhc3Yy0mO4OcPeDNBVOr\nfNA3UsdRUwTcOrAPQMNsSQirrkoty2par5UUJCFpVH3eON23E2zR2itMzuSZYsBgqty4b83+87nq\nAo7MHcPjZ7/dMp2mGVZTLyHnmcoc4t4Y9iR2A1gcHNmoQb5+MxoaAWH+RkBYF64YkvuetQRTkcXj\nrQM3gQHTk6Aw6+JvucGnP5h8Ht+b+OGS4kBawqzdCqRMJ4DLIueskLNZtWbNOWZjeBT9/iRmK/MY\n6vOjKioo1VYv39kh53UGr0keYhfkbG3eQAIYODPH2TBrL+0LKpk1e8kquxlkcdCpUEfFUpi/KNn9\nWKdz53A6e27RcTopZ0ERbZM3wQszR/DfX3nYFsFMzKO8i+tOOROfM9uY7MqyMe6lJgorvnL8a/iz\nN/7SuKZajkZHjxUnFm1bV+u21LZmIpivLuCJs/8AzsXiIxs/AAbMIrO2Ma7Fpm3Bcq0V6nM2lbOv\ne5+zzay90j5nWYCP9SLpT2CqPEOfzarNrF2j9yFfL3S03ABNVgS5Ak3TUJLKiHoi1JJkfeY0XbOU\nnlz6N+q1qXa5qFqUM3leuyErslizvtsp8/3bEBrFQKAf4+WpKw6GsynnZSxyZK2xqCXZCc1YDjlb\nc5mtAWHFeqmj0i3Ui/jDV76Efxn7Hv1u1oxhGQ4MYSQwCFGtI54wfqvZdKXlcVYCDjmvMyxHOVtV\nJEm94alZm2vrCyqZdbyzTakKBGTyryq1tuZOaxEMq3LWdA1fPfF1fPXk16Hp2rLI+bvjz+GLr/7Z\nooXBhcIYALu/cro8g6gngrgvbvNdNqPhczZW2UQ51+QaVVtludKxoMJcdQELtTQERcBCtWHKbqWc\npSazti34Sq7hK8e/BlEVcd/uT2Iw0I8A56dEZSWc5ohtVTMCp2IeowBDs1k7YSrnrszaqtWsLaxo\n9HJNEeBnfdgYGoWgCMgIOfNaGs96Ra7aggt/MPl8xwm22axdrJehQ0fEE7aYte3XSRZVS+WCH51/\nHb/9wu8vKpiymrD6nBupVJ3JqmI+D7kWyjnpS2BLeCMkVbJFJFshKCL+5tTjmK+mWv7duh3BcoJP\nc0JjUXuiBTnrut4wa3exYOyzkDPxOV8sjOPAj/97x5zuqfIMFE3BpGVOIWlUI8FBWr2RDxn3Yja7\nern4XZHzoUOHcO+992L//v04fvx4y20efvhh3H///QCAp556Cvfffz/97+abb+7diK9yeC+TnEum\naiXK2dOhEAlRuUsp55rl/Ok26rm8hFl7vppCTRFQUwSkamlkLOfptMrOCjloumZ7YQBgxjRhk2OV\npDKKUhkbQsMImKlRS5GLoBKfM1HODZ+ztZBHob60ehYVkY59vprCQqVBzrOVeZu1Q9M1SJrcRM6N\n6/7XqReRFrL4mU0/jVsHjffDz/nohGxVDIUmRU+ejQF/EoDFrE0Dwkzl3AU5W0lLhw5REVGRq7b7\n1SsQct4QGgEATFVm6LWQTmpVuYa8eQ/i3himyjM4X7jU+oCW4xJU5CrygrF/lA/T98G6SLWXOG39\nLE6VZyBrCp6bemFZ19hLVOQa3C43vG4PfV47FYqRVJkuyuzKOYMIH4KX9WBb1KgffTp3ruUxTmZO\n4+j863h1CZcXgWhTzt2Ts3U+OZs7v8gkrmgKJe+ufM7RxWZtMld0WmCQv2ctYyLBYEOBAQwHBwEA\nKm8skFczKKwjOR89ehQTExN44okncPDgQRw8eHDRNhcuXMCxY8fo509+8pN47LHH8Nhjj+E3f/M3\n8bGPfay3o76K0QgI60zOxN9M1CDQCAjjmupri0odL868TH14JTOVIifmWxKa1WzazrRdsRCbtauW\n1cw7Vpy0vZCdzIVksp2zrOzrqkT932Q8xPQ9GhyBn/OZ5NKakMj3DZ+zxaxtuYZ2qTXNi49U1RjH\n1shm6NBtQU6EiD1unlozrOriTO48XIwLP7vpw/S7ABtATREgqbJZitVQffmmBQNRE3FvFJyLpdHa\nFbkGN+NGiAuCAdMlORv3gpBYVRbwt6eewBeP/VlX/t5uoWoqJFWCj/Nj1CxXO1uZp9eSNJuRGMrZ\nUHw/v+2jAICX515te2yhSTnnBOMeGsrZ/O3V1r71pX4jss0bqbd7knZ0OajIVYT4ABiGoSTdSTlb\nn2ViFSNm5KTZje3GvuvhYlx4LfVWy2PMm8V8yh2IUVAvLyCMzAVxbwySJi9afFlL3Xblcw5bzdoG\nOZPfoZP1iJBzTixA1VTouo7Zyjz6fHHwbp4q57KeA8MAc9l1RM5HjhzBnXfeCQDYtm0bisUiKhW7\n3f2hhx7C5z73uZb7//mf/zk++9nP9mCo1waWUs6iUl8UcDVXTcHFuLA7voN+R+pqN9fXfmX+NTx+\n9jDeNNvyEbO2oqstA7ms5Jxuo6LKUmvlfMlCzucLl2wpQZ2CRwg5z1osA7OVebqaJuRMVsejoWGL\nD7n1y9zwOdujtQVZoMoTAHJtlLN18TFfSyFlKudbBm4CAFwqTuJM7jx+NP1j1M1r9Lg9DSuGeS8E\nRcBEaQqbwxupDxwAApwPmq5RchoKDABYbNYmZlwf50OQC9J7UJWrCHJ+MAwDL+tZVkBY3GeYyKtK\nFZeKE6gqtSVdHpcDck/9rA99ZqpXVsxRsiFtPKtSFXmxAAYMro/vpNfVzbEBY7GYb0HOVgtBN81B\nyHOk6ip+PPtyl1fZO2i6hoJYQMIfo9910/zC6maqKQIERUBWyEKHTruxBTg/dsd3YKo807LONony\nrkhL+1dJrXqCy1HOHxx9H4DFpm3rcbsx43t4N22IQXo8k4VFp9iduZoxx+jQkRMLKEolVJUa7fke\n8xgL4Fw9h3/7U5txy+7+juPpFTqScyaTQSzWeEDi8TjS6UaZxMOHD+O2227DyMjIon2PHz+OoaEh\nJJPJHg336ofPnKytD5Wu6/j/3voqDh39MlUzuq4bIf6+Pgz5Gw8MmYyazdqEHNNCBrqu24K3Wk3C\nVrN2W+UsG11zWMZtJ+fSOLxuLzgXh7fSJ21lJTv5pwj5zFUbynnGEpWdFowWccTPPhocsviQW5v9\nGuU725u1SbnAE5nTi0pnWhXUfHUBC9U0OBeLvck9AIAXZ47gkTf/Ck+dexopc4IzzNpEvRn34kJh\nDDp07DLTWggCpmmXlFkkE0ShSc0TMgqwfoT4IMpyBbquoyLX6DE8bk93tbVVQs4RAMBMeY4uzFJt\nyqEuF2SS9bM+xDxRMGCQFfKUbIhyrso15MQCop4IXUh1mvht0dpyhQYbRfhwy1Sqahdm7aosgAED\nr9uLl2ZetgXOrQZKUhmKriIZSNDvAqzh9mjngyfPMsnRzQp5SsBEOQPAu/v3AgBeb6GeSRlc68K7\nGc2kt5yAMPJ8v3fwFnjdHpxsSpmzHqvbqnHE79xQzmVznEu/A9ZsFwDIiFnMVYzPwwHDnM0wDAJc\nADVZwMc/sBUf3re8YiZXAna5O1h/qEKhgMOHD+PRRx/FwsLCom2/+c1v4uMf/3hXx43F/GDZK89R\nsyKZXF7JtvWAstt4GRlOo+N/c+4kLhaNYCg+DMR9IeSEAgRFxI2D12HbwAZg3Ni/LxpGMhlCdN6Y\npP0hFslkCOq42bNXL8MfddsmG4mtLfqtNFdjQiwoBfp3XdfxtTeewg0Du3DryF7UlCqivjAkVUZZ\nKSOZDKFUryBVy2Dv4PWQVBmn0+cBAAl/DNlaHt6Au+29EVWTHIQMInEveDeH7KTxQid8MWSFPHwR\nFxbEFDysB7s2bMTJ8ilgCuACre+7whj+38EBQyFqJhlobgV1NAhdQBXhmAd/+aO/xdboBhy863ca\nx8g2XvS0mEFFqmIgmMSO0VH0+ePI1BoWhgXFWFhEgkEMJIzFrdtjjG1q2iD927bcaBtrIhwB5oGa\ny5hgh2NJBLMB+rsSsDXDDdEfi6FPiGKyPA1fxAVBEbA1vgHJZAgB3oeSVOn4DvALxvo8ZirnSaER\n2Ca4O+/fLfJM2rzGMAYHokj4YyhIBRCPzHAiAf+0DzWthqJUwo7EFgz0R8C5OWgute041HPGs+x2\nuSGoAjVrbxkaQhHGPfEH2cYznLEQreU9s6Kuiwjyftyx6TZ85/wPMafOYN/ADVf8O3SLXMZY3PUH\nEnR8A5Ekpiqz8EZcCHuW6E1cNq5tJDyI6dIcZF5AzSTsbQOj9FgfibwX3zh7GG9lT+D+WxtuR03X\naE60oC2eFwhS1UbHNU3X4PG7unpWkskQ8lIeIU8QW0YG8a7B3Tg28xbgl+hCpORuvEeKriIc4+Hl\nvEsdEgAw3B/C2FwZiZgfyWQIom680zq79LOTqeVQVyW4XW6omoq6u0bJfPfwZrpf2BtEppZbdJyV\n5peO5Nzf349MpmH6SKVSVAm//PLLyOVyuO+++yBJEiYnJ3Ho0CEcOHAAAPDKK6/g937v97oaSD7f\n2yi4ZDKEdHptfEVXAtGceHPlMtLpMnRdx1MnGsUxLs3OQg27cSZn+GnibBweJUD/LlQV47ol49bO\npDPowyCDGHNrAAAgAElEQVQyJUMRThdSGJs1FGfCG0dWzGEsPYvrg/bfqlCrgHWxCHFBzJVS9LfM\nCnl85/wPcT49jg3sJpSlKgb8/fAwHoyXprCQKtL0iFHfiEHOMMh5wNePbC2PbKGMtL/1vTHSXMzK\nQ7qOExOXsCE0jAvpSbgYF66L78JLMy/j5OQYZkrz2BQaRTZThS4ZJDObyWLIvfjYZbEKj9tDr0M0\nG7NnK0VkhSI4FwtZUzBXzOCNsbNQNRWZasH2DM3mDILxur00UntLeDPS6TJ+avBWnMqexU3JG/Ct\nC/+ME3NGsI0mMaiWjIVOsVJFOl3Gm7OnwLlYxLQ+2/FdsnHPxtJGIBwjuRHhw0jXskilSmDMqkjz\nOeNeaiIDHsak9ea4cT4eXqTTZbjBQpTFju9AoUzSrwxyfnv+LP3bpdQ00rHevEOzWdP6IrNIp8uI\n8VFcKIxhOmuQkCq64Hf7MFtegKZrCLmN95dnOFTrQtvryFcNi02fN46CWKJmbbXqRl0w7nMmX0La\nYxxjodCY/IuVSstjl8QKfKwPm32bAQDHp89hA7cJmq7hZPYMro/vgtvVWzFhxcV54xlI+hN0fCGX\nQQZnpyewObyx5X6zWeO5HA2MYLo0h7HULM7mjPcvoids13pdfCfezpzC2+MXMGi6ULJCjlpTikJ5\nyd99tmLczxAXRFEqIVtceluCZDKE+YUCUpUsNoZGkE6XsdG/EcfwFl65+DbeM/RuAMBC3m7JG59L\n0SDHpRD2Gu+OVDfmv1zNeAaK1db3FwBOZY05dEdkK87kz2M8PUsDa4NqlO7nYTyoyQLmFwr0nveK\nX9oRfEez9u23345nn30WAHDy5En09/cjGDRWbXfffTeeeeYZPPnkk3jkkUewZ88eSswLCwsIBALg\n+eWVTbvWQXyQJCDsVO4szufG4TabGRDTMYnUHgwMUH8d0EilIp2QSk1pNjkhR82zWyObALSO2BYU\nMyfVl0ChXqSmRWIeX6imaTBPiAsiwoeh6RrKUpX6m7dGNmNLpDGJjAQMM227VCpREW0m8LnqPDRd\nw2xlDv3+JIb8xiRyMnsGmq7RaMoA275Ig6jUaTAYgEbbSNkwa8c8UQQ4P3JigQZ2EXMxAfntd8S2\n0O9Iqcy7N38E/+Xdn8UNfdcBMILgACyqEFaRqpipzGFrZDMNwiIg6TIk7cXP+RHzRCCpki0GgZju\n/ZyfNtX40fSPAQDbzUhcr9sDWVM6BnWRe5EwydnqV+9V31/bmE2pHPfGoEPHdNlYKAZYPwJ8gAYn\nkjQx3s13YdYWwLs4RD0RCIqIVDUD3s3DaysAs1RA2GJzrK4bZUADrB8bw4YZk6RUHZk7hq8c/xrN\nv66rEp489w+2tKVegLiarGbtPq/x76UyLIBG5D4h79nKPM7kL2AkOER9/QR7+wx3DOnqBtg7u1WV\nGg2SenHmCE7nztH3gTyPEY9BLt0WIcnXi1B1FX0+41rI80rSJIHF96SqdA7CIo0tiHmbmOTbBUXO\nm64n8s5mhCxmq/NgXSySvsbvToJHu8mg6SU6Kud9+/Zhz5492L9/PxiGwYMPPojDhw8jFArhrrvu\nWnK/dDqNeDy+5N8dtEZznvMPp14CAPz0hjvwg8nnqa+Y+EqGAgO032tNEah/M8wbD2sjktck53qB\nEuzm8Ea8uvBmy7QZQRbh47zo8yVwrnARGSGL4eAgTTUqyxUsVI0XOcgH4WaMdV5JKuFScRwMGGwK\nb7BNrCTAqR05E3KNe2PIiXnMVuaRDechqnWMWiaYtzInATT8sta85VYQVJGWtQRA20ZWlRoqchX9\n/j543DwWamnqa1Y0BaLaIPWiVAIDBjui2/C26SdrnvD6fAnwLo7+3h63p+Fz1mScKxgT4c7Y9kVj\nJJMA8ckFOD+iXoOkXpl/Dc9NvoBP7fx56mP1sz66CHs7Y6jx28xKYx5zkVdXJfhdPiwF2azWFvdH\n6Xdetwe8m6fmzV6A+LHJfSKLmqnKjPl9I5cXAGLeBjl3CgqqyYKxUDF/i9nyApLeBBiGAdvR57x4\n8q6rdWi6Bj/nR5gPIeqJYMpcsJ0xVSghsbczp/D89E/AuTh8fPs9Lcc3XZ5Fny9OfejdIGc2QukP\nJACTq4h6bJfmRnzOhJxfW3gTiqbgRpOIrRgJGe/OnCXdiFwXy7ih6Coqcg05MY/Hz34bgOGL/dU9\nv0R9zmSe6TZam8SvEPIbDQ7D6/bgQrERsU1iUkJ8EGWp0lXE9m3X9WMkGcBIXwCyptDnTWhHzuYc\nuiu2HZyLRUrIIFVLY8jfb7OKBGgaW402sVkNdJXn/MADD+Dxxx/HN77xDezevRu/8Au/sIiYR0dH\n8dhjj9HPN9xwA/7qr/6qt6O9BsCa/ZgJOadrGcS8EVwf3wWgod5S9CE3VDMJ9uCalHNzgQpN1zBh\nTjQxbxRRTwRZ0f6y67puKmcfkn7jJSIvlTVymKx2g1wAET4MwOjDOlacxGhoGD7Wi4gnhH5fH8J8\nCBGPsU27VTYh520RY0U9V52nwWAjwSH6UpMXa8QM3PC3Uc6ypkDRFJtyBoyXjkSyhvggot4oJE3G\nmfx5uo01WKxYLyHIB2h6BdAolUngYlwYMscEGORCSqpKqoSL5m+2M7Z10TgJOZEiHH7WTxXkN8//\nI62Y1VChDUICgH39eyn5eahabx8URhZKhAwBYDg4iAF/EnmxYLtXFwvj+KeL3+26vjAJ2tN13Rat\nDQAJc6G0YBKDn/XRXGfASBMDSBna9hM/yZ8mVgRd1+mzxrtbRGtLNbgYF3gXR3+fo/Ov4w9e/iJq\nco0WgiHP1MbQKIpm2VaS9kNyz0nw2VJ9t380/WP80bE/wT9eerbtNTQjKxjKuS/QeL5IcZnm99UK\n8rwO+pPwuHlKdCRo0YpBfz8YMLbAS6ImN5nkXpErNAe439+H2eo8vjfxQzo/hc33vttobbLwJPOV\ni3Fha3QzUrUMtegR5Rz3GM9IN+R8sTiO/3nmYcxW521R5u2q5JFsl35/HxK+BGYr85A1hS74Cfw9\n6KV9OXAqhK1DeN1eiKpIo6qjvjCdbAg554QcwnyITj4kTYKYUK3kbC1XCABjZg3pMB9Cny+OYr1k\nUxaypkDRVTPtxU7OBRs5GxNVkA/Q8f1g4kdQdRXvGXw33e7/eNcv4//c+6s2BbkUyDiT/gTCfAhT\n5Vk8P/0TAMYqO+6L00hUoDvl3Fy6k8DP+eh1h/gQJQQrIZPUFF3XUayXEOXDGAw0ouP7LOavxpga\n5NwcrT1TmQMDBiNmrq8VxDRPzPoBzoeYx4iiNnp0sxgvTTaIjvPZyPn24ffQf5Oo9G7JOeG3kvMQ\nkr4+6NBpK0sA+MdL38F3J/61bVEaK95Iv41DR7+Mk9kzFlM8Ief4omu1kjM1a7sMs/ZSCwJN1yAq\nInysD0HLb0GeR5pKZctzNjIMPKyH5sGeyZ1HqpbBTGWOWiaCHCFnIxPl1YU36bNBFqnEnN3coxoA\nfjz7Cu1zbs026AY5MY8gF7A9szQFTbCb0PNigZqmy3LFyJJwc/Q3jnmiNLfcCt7NI+GN2Qp1LFRT\nYMBQl1dZqtD0wo9tuwcs40aqlqFut7CHKOfW7/SXX/9f+Ku3G6KN1Cqwvjc7IsZClQS9kmhtsmDs\nhpxPZs+gIldxNn/Blhq5VL0IEqmd9CXAulj0eeP0WbS+v4DVZbZ61cEAh5zXJXysF4IiQlRFyJqM\nmDdCfTslqWTkwtYL9OUDgA9tuB0fHL2dmo6DXAAMGJRMs5DVj0v6H0c8ISTMhzJv8ZkRkxDxOQMN\npV6wpUsZvuUQF6ST4XwtBZZx06pXgPGwbwyNtpwom2FVWMOBQRSlEs4VLmJPYjc1P0VNwop6IlRt\ntvM5kwjMZrMiUUbGNQQoIQAN3721mIGkyQh7wojwYUp+VlM5gVVZe9w8JVZJlTFbNQocEGVrhdWs\nCxgr9p2xbdgS3ohfu+HT2BbZglQtg6yYAwMGPtZL1eJgYIBOqMZ5SexCJ3Imi5MAWDOuYSQwROMY\nSDqVoIg0lqDbSWrGLBIzZl1QEOXss7sYPG6P7fqtyrldGVpRqUOHbi5UGuROLDlciwIwFbmKIBeA\nx+1BXbHXcC7US/TfZCFB/M4/mvoxPQZRzMTKIap1m782I+TwjTOHaenNVl3LlgLJdU80WWV4N48Q\nH1yU2vit8/+EP3vjL5ET8yhLFYTNRQr5jfcm99BgwmYMBgZQliu0PnuqlkbcG6XPdUWq0Gvs88XR\n50sgJWQaPmdi1m7hu9d0DZeKEzZfdbNZGwC2x4jf+ZLtWHFKzp19ziQ+Il3L2FLARKXecmFXkiqo\nKQINhLO6p4YDdnK2NslZTTjkvA7hZT1m0RHDzBP1ReB1e8G7OBTrJRTqRWi6ZpvgNoc34lM7fx4u\n0/frYlwIcH6U5TI1aZMXgqzwQ3yIHsO6Gicvno/1UbM5maQLYpEqV/IShSzKGQBu6n+XTQURtJoo\nmyFYAofIpPjTG+7Ar9/4K9QPRF4k60tETNatXiBCJqT6Gt2Ha/hiQ3yQTgYAsMss7ELImVgsInwY\nDMNgd3wHtsY2LlLjzeMiJMy7eWSELKpybZHZjMDfTM6sDzFvFA/c8hvYm7wBm8IbABh11L2s1zSh\nD+A9g+/GJ7b/O9sE7KH58h3I2Vwo8W6enn84OGghZ2PSO5+/SJ+bdpOUNQAta07qc9UFCzkb54h6\nIjTI0c/6aD4pYPxmpOpdpzK0VtIPci2Us9teW1vVVNQUwVCl7kahFkIAhXrRQs4NszbQqNTmdXtp\nK09r9TaraXuiNAkdOn5204exITiColTuKu8caOQ4x1tEKPd548jVC7bfeaYyBx06zucvGQsPk5wH\nzeDJm5JLp4CRxfxcdR6CIqAolTHg76e+1bKlYlvcG0W/PwlBEagCDnuIWXvx/SHtP0W1TgMNU0IG\nXrfHNj8YC3eWusnql6GcyfyUFrI2s7YOveV8s2AWHxk0a0RYlXyzcnbM2g4ovKwPsibT1XnMGwHD\nMIh4wihIJRqt2byybkaYD6EsVenDSiZ3wFCanIulx8iIjdW4YJnwvKwHUU+EqoJCvYioJ2IzpwbN\naG2C9w3d1nI8rao1NcNqsv3o5jvxu7f+3/jEjp+jiw6g8SJZFapR3tDbUjkTP+FIyG7a87NWcg7Z\n/K43mK0GySqckrM5Gf3HGz6NQ5YcaCvsZm2DJHkXT6PbR5pW5gRet4dep8fN02AmAuv9I2N3u9z4\n5evvxXWJnbZtPe6le0hbIWsy3IwbLpeLKteR4KBlUWaQ8+lcww/farJUNRVPnP0H/NYLn6fxAPm6\nManPVxeou4EsolyMi/7e1Pph/j/midKFRqu65IARDb9QTTWC45pM/IvM2qbypgVcuAA8bh51tU6j\nswEj6I8ck1hjQnyQWmtCXBDboptRNyPoc2KeLjKs5Eyq240Eh6h/tVt3ACHDRAurTMIXN6qHmWZ1\nWVPocd/KnDTS0Mzf4a5NH8Jn9/5H2sO5FRrknKLv+EAgSS0yFcko6uJ1e+FjfXTRRjIaGgFhi9/p\nkqXQ0UI1DVmVsVBLYygwYFtIsi4WG0MbqM+XKmfP0uSsair9vTVdo77sVC1Dz0usX62qhNFrNevT\nkwV/gPXb5jLyHbDyXdua4ZDzOgSZwMgDFPMZD0uYD6MiVWkUbafcvxAfhKAI1BS9ITRCVS/xFRH/\n6WylUUSmRpWzMY5Bfz8K9aK5si4h5o3QFSdg+Jx9rBcBzo+kL4EdLYKdAFh8r92Ytf3g3RxGQ4t9\nZcS/PrIocMPXcnV7InMaDBhKuHR71q6ciVk7yAUoERL/FVn5E/eCi3HZFgxWhPggnbSsyplgKeXM\nMAydCKwmd4LN1sUVt3QENrAcn7NCCez9Iz+FD294v2kxSYABQye907lG/nPzZCkqIr5y/Gt4YeYn\nkDWFmr+JOTQtZFGSKvC4eVsULCEfcq1ETVkXSbQuuYWcNV3DI2/+b3zt1DdsKVohq3LmW/ucrT2S\nPW6PaTKXm8zajVQ1gk2met4e20pNvnPVBQiKiG3RLXAzbkxYOliRuvBDgUH0m4vJdJfR78SK1Yqc\n+5qCwtK1DHVZnTLrC4TM3zHA+bEnsavtuRrkvIAzuQsAgJHgMHURlOUKcmKeWpUIOZPOTeE2Zu2S\npaLeQi2NmZKRw9783gJA1BOGDh1VuUpra7dTzv9w8Rl88dX/iQuFMeTFIl185cQ8XbiQRXwr1451\nIQI0hM5wcHCRC6ChnFeXnJddIczByoNMrKTGbdRLfKzGAzxeNFaMnZQzWUETJRM21WFOzNPJazgw\nCBfjwnSl0QGq4XM2CGAgkMSZ/HlcKIxB0zVETF/v+cIlMDAIhWEY/Oeb/xM8FvXXjK6Uc5O/rxXu\nGHkPXC4Xbu5/l+37AOtblP5Tk2u4WBzH5vAGm7JqPkeIN/zmMU8UO2Pb6LaVFmbtbjAcGERJKlPz\nMm/JaR4JtlbOxpj8KMuVRf5nAHR8+XqhJXlbsRQ567qOx88eRtgTxj1b7rI12CC1jgGjEUbMG8VC\nLY2MkEVayJotLRe3EH1u6kWcyp3FYGAA89UFpGoZqJpKJ0lN15Cqpan6JCDPL1lokL/b8vaXaLcp\nqnVMV+bofTEi1y0+50XK2di/YumRTDo8CYpIF4XFepEqtqDlHmwMj+KtzEnsiG6lSuySGVjZ70tA\nVERMV2bNxQ6Lmeo8glwAYT5IlXO3eeNZakZurZwBICPksTPWiK42rrER3NgtBgONiO2T2TPgXRxu\nSt4AVVfpmEW13iBnn0FmxMXhZ31GPEUn5VxLYaJg3JNW5EzM6FW5RuMAglwQrItdRM4lqYwXZ44A\nAM7lL2CLJdZCh45xU1GT6PKWyrlqV84D/iRuGbgJe1u4AALU5+yYta95LFbOxqRFJhwyKXRLzqRg\nSYgL0tU4eYE5N4dBfz+mK3P0hWuYtY1xDJgq+WzeWFnHPA3l7Od8VA21KnRgBU1r6UI5N6c9WeHn\n/Lhz4wcXmX19nB91VbL5407lzkHTNVpowHYcq3LmgnAxLvz+e38bv7T731MVR33Okt2s3Qkf3XIn\n/u2Wn6VkTtKpOBfXMsKbgJBys/+ZgCh6XwflvJTP+dWFN/HS7Cu005OkyouKoRCMBI0Fxpdf/woA\n4OaksRhqNu9NmpPhr93waQCGX7FoBi4SkKAtK4jlx2cuNPr9ffjs3v+Aj26+k27DtzDPW0mfuCz8\nnA8et4cuNBqpVHafc8WmnI1j5+sFqj6L1oAwywLojpH34t9svhPvHbqFKrqL5nsY98awKbwBqq5i\ntjKHuiohK+So+ZZYeroNCiM5zokW71IjYtswZROS2RrZTLdpXoS2A+/mEffGcKEwhqyYw77+vfCx\nXvhZH1yMi3Z+IwsF68IJMOJjeBff8p22k3MaEwXDstAuU6EiVWl8gcfNI8D6FwWEPTf5Al2IXCpO\n0EUPsQKQfu9UObfw9S/U0gjzISpA3C43fnXPL2Ff/42Ltu3UVGel4JDzOgQhJpLiQMiZmJDma0a6\nQ8wbaX0AE2HO2J6Qc4D30xeemGcBo6uTpErUdyXI9g5OZHV51jR7RS3kbA3C6QQX44Lb5aYTZV4s\nLMoPrTXlmC4HpFiA1e98wiwWckOiBTmbBMgybvqb824OrIsF62LhZ30Ns3Z9eeS8PboFH93yEYvv\n1CBAYqlY8hpMAguwrcmXmLb9S/ydwGNRzqSQg6AI+NaFfwLQsFAoFrN2Mz6542O4sW8PJUPSfatZ\nycxU5hHhwxj098Pr9iJVS1OTNklDajXmZuUMAHsSu23k4iE54hZlZk3nI0VBSFBZkAuaRG3sx1Kz\ntjGZV80YgoAZEAY0zO+ASc4K2aYxriAXwD1bfwYeN0/T28giOeaNUrP3WHES89UF6NCp+yJhpv8t\nVdQlJ+Ztvykxa7dUzjRGxCBwYl27w5JGtxxyBgxSI4uT9w0b8SIkoJSkIpEFSZgP0d+WxEhwbq61\nWbuZnIsGaTYHXAFAgChnpUaPxbt5w1pjIcWKVMULM0cQ4UPo88bN39us9GW+46qugndxNJK82awt\nqTJyYp7Oa53gcfNwM27HrO2gQYpErUU9YeSrgo0Yop7IIuXYDBK1SUg3xAWpzypiMX2NBodxFK9j\nujyDAX+S9mklSoc8xMTPFPVEqK+6VVR2O1gLIzx1/h9xPH0S//XW/0x9y4IignWxNjNwt7DmOof4\nIDRdw6nsWUQ9kZamNEIWQT7YMtWEVCgCjI5UDBibX3M5IBNaq4nJigBrNixZQjlvM8sdWtO+2p0v\nVcvgD17+f1FX6+jzxVGWKmDAQFTrUDXVZtZuRsIXw3+68TMYK06iKleparcSSU2uIV8v4Lr4TkMl\nmqZEErS4J7Ebk6aSaSbnjeFRuBl3SyVF0CogzJrOR6Klyb3/+W0fhTfQ8GuzjBsMGLogJGMP8gG6\ngLGW3lR0FalaxuhItYT1ptkXGvfG6P14I328ca9NJce6WCS8sZbKuVAv4uArf4wtkU34jZt+jY4n\naAasNSPqicDFuOjvO19LgXOxuLn/Xfj7s9+CoinU59wthgIDOJE9jQF/vy0dL8Q1nn+yUDDucRJT\n5Rn6+/BurqU6JeSc9CWQFrKoqyIS3nhLqxiZRypSFXVVAgMGnItFgPNjtjoPVVNxsTiOwxf+GZIq\n4d9t/VnMVObw8tyrOG5WC9yT2IXvT/7IGDsftFiP7GbttGD46bslZ4Zh4Od8TkCYA3vKT5ALgDVN\nc1ELOXcKBgNA8x0bhR4C2B7dAgYMNltqXm8wiZHkP9eafM5RT8QW0BTzRhD1RPDB0ffh/SPvXda1\n8W6OmsAK9SJ06Pj2hX+huYg1pfb/t3fuwVFVeR7/3tt9+91Jd5N0HkAAeYYQhfAwPEREFB1XRlAw\nSmTYxUILBXWxAF1W2JoSRcc/dLBKx51xZnxU6VCUS5Xs4o6OWzMuREt2EdhRh4GRp5D3O+mkc/eP\n2+fcezs33X3TST/i7/NXHp3OOX3uOb/ze8fVCgfCxaMqlfF/13Ie7b0dmDFqmqHwZQf6QJqGR/Lw\ndJDm7mZ4bO5BNzuQxMSEMxuTkc8ZUOqhPzZrI5aMXRjzfZhW+OXV4+js7YTf7sPVjjoUuPJ5kFBn\nbxdCfT0Das6MCbklmJFXCpvFBinKB6iNSgYUs2dvXy8vKDE+p4Q/R9Gm+AJXPvbesAuVRbMxENHt\nNgHFLxwNe2bmFs7C0mvUz0YQhEhTk2iztir8mKbKgiWvdtTBJTkHtHDkRvnOAw4fRjn9mOy7Bqeb\nzuJE3f8B0Af+5bvy0NrT1q8+86Gz/4mucDf+1nIOsiyjJ9yDuq6GfuZjhkW0IGD3oa6rAX1yH660\nX0XQlQ+bxYZxXuXyZFZzZimLi0Zfr9sn2qIu2jRDZqbnwllU658frz2p1qaPCPZJPiVAtL2n0/CS\nDKjPe3tPO0LhEGwWSZdet+vIXrz8P6/jfOtFzC2YhRuKK/lFoqm7GV6bB2M1VhqvzQtH5ByNjruI\njtROBLfVRalUhL6SVY5Gw83RBCPF8zcD+k1qi1SqmuyfiJeX7NH5qJjmwnw1LNeY3XAFQUCh5kH2\n2ZXUrjVT7tIVG0kEpRyjspGZb/vrxr/wTlasFONgiK4SxjQ27Vx1r2epMgNow16bR6mS1VmP+q5G\nXYS6WewRIcOafwwEO4xifQZT/JNi+uQB1ecMKAJ99/xt2F25Hf84exN/ptp6lOpxA2nORrisLp15\nj/XU5sI5cnCzyN+Aw4+iiJXFaE7OSL72QNgM8pyZ5qzdG84YbhDJIiEU8VEy4cwqhAGqj5fVAZAh\ncx+o4fuJVr63BAg8rqCyaA4Ate4784ECqq9Wqz1fbr+C/770BQDlotTY3YTL7SyieWBrQpGnEK2h\nNvxv7UmE+nr4c7l8/E1YUDQXQRNCB1DyoB+f9RCWjNFf+Ly6cqqqMsDe32lRgx1DfSH0hHvwryff\nxvvffABA0ZxdVqdOIA8knJnm3N7TERHO+stsqC+E6/JnYNuczVhfdh8ki6Tb1wWufDisDr4uXpta\nXS1aq+fBYO7E97MrEgwZXQVuOCHhnIE4NIeY1pSt/doozSIarXDWbrRo7c8tuRBw+HEhUmKQ3e61\nhykLCtMeRoPBZrFxLaazpwvOSHeoA6c/RJ/ch87erpiR2rGIrhLGSiYapWMBSmBb2ahphkEggPr5\nnaj7M2TIuqhQs8wrnI35RXNxjW98zNd5onJ+Bwsz2YqCiKqpqyAKIvJdo+CR3NxkzoTcQAFhRkT7\nAPsJ58jBzVJ9Ag4fL4YxmDgCo3xt5nPWxhHEemYkUdKkUrFobRf/jFh0tNaqMZBbgcHM2D57Lt9P\nM/PLuVAJOPy6CxTPG9f4nf/tr4cgQ+Ya4MW2y9x6NWYAIQYAN4+9AQDwXqQZBRMyZaOmYW3p6piX\nHSNEQcRk/8R+f8c0Z1EQdRchdtHQas59ch/qI9r8pfbvIcsyWkKtyLF5dRpqPOHc1qOYtVmswe3j\nb8Zzi/4Zexftwsbydbpc/wJXPj+jWBQ5uxx6JQ/XnDujSngyP70pzVlyQoYcNzVxKCHhnIFoN7V2\nU7BuQYBxJGc0Wo3QHccPNdZTjNZQG5q7W9DZ2wmLYNGZO9mD7LV5kupja7cokZ2suUbQlY/ZBdfh\nSsdVnG0+p3QDGqTmzIJK2OF9ofUyrIJlwE1oES3YdN0/YH7xXMPfswsN82ldk4Rwnugbj+rS1XG1\n1IrgdVhWciNmDXBhSBSHxY4FRXOxevKK/vngkc+XBbnFM2trcUsudPZ28Yj4S22XIQoi/4y15lil\niI0DRZ6IcB7EpUsyyHNu6m6G0+rg+fSSKMX8XCXRil6NtUaAALvFzgV/AxfO6ucUb6zM76zNyXZY\n7TyivVijNQP9Nef6zkacqPszrskdj2UlSwAo7R3VJi8Da86T/RMx1T+JWwEKTWrKicKef7/dpxPc\nbPnIvFUAABeHSURBVC7snGKXuyuRuXWHQ6jvakB7JPajQGNxim/W7kCoT9Wc2cXAyC0lCiK/MLMx\nsbQ1j83DNWcjs7ZVtBoG3A0Ed5ml0LRNwjkDYf5CQK8tC4LAA7kSMWtLFolvoHitzkZzv/MldPRG\nNFrNhmC38+hcVbPYrIpZu6evhzfXmBzxSbEiCrFMlLFgkcx/bTqLcF8Yl9ovo8hdEDdwbiCY5vzX\npr9F3t+4wf1Q4pKcWDnpjrhm63gIgoC1pauxWJO7rP0fgBqwY8aszQ7Rjt5OriUVuoL8M9YKZ3b4\nzSmYGckh7d8ZKR52gzznpu4W5NpzeVnNeJc5SVTN2p29Xbz0qZoLrgh+reCIZdYGwCO2tb5YQI12\nHp+jv8gxzY5ZGk7UK37puQWz+P+91P49b4wSLzbhjgm38q8Loy4CQwU7M6LnONpdhNLAFN4PWg0+\nVGuLf9uopLgptRVyIYkS7Fb7gKmWSjyDpGrOBsFwRkyJVD9jnyGzUHhtHr6+WrO2LMu40nEVQWee\nKQuDOw2FSChaOwMZSHMGFGFd21mfUEAYwKqEdcWNqmYpL2ea/oZOA78v04ySFs4WfUMJp9XBb7+n\nIlWoBmvW9tlzEXTm4XTT3/B9x1X09PXG1EDi4dEE1OU5R5kOtMlUmBbALAxmNGee89nTge5wN7rD\nIZ0gcUZ6TLeG2rhwzrF58fdl9w9qrNE+51A4hM7eTozzjkHQlQe31cWL9AyEZJFUV0rk4gmopn+G\nNh4gnluBaczR2tck3wQ8Pe8Jbl5l5DkDCLrycLL+z+js7cSJWkU4l+eVIteeA7vFFimq0ox816i4\nwmmibzzK80pxtvlcv/81VDDLW/QcJYvEI8sBtYqbXjgrMQc5di9EQcRt42+G1xM7vsAtudASakWf\n3KcLQI3FkjELUewuRGlAKV87fdQU/NeFzzDZdw03u2tTqZpDLegOh0yZtAFtsCkJ5x80WuEcnVe7\nsPh65DvzEhaSXsmDq6iLK5wn+ybCYbGj5vtjPLpXS5G7AMtKbuxXAtMsUmTTNUe0NqfViSJ3AWwW\nGw9IG6xZGwAm+6/BZ5c+50U2BvI3J4LWLTAhZ/Am7UyDXX6aB+lzBpRDil2wok2VQWceWkNtOpPv\nYFF7YasR/oCaUrRl1sa441fM2r08poFdbKMFYMDhi9TbDsX1ObPmJtEdjABj060gCKgsnIODZ/4D\nf7pYg780ncFY72j+GRW7i3A20uVtWmBKv783YsOMB9ATDplaPzOw+IGBTNEMtqevaILduHCO1Fq4\nbfxS5Od7UVvb2v8NIngkN7csJCqcraIV0zUlSku8Y/Dcon8GoDZh0aZSXY6UKS40EQwGqDnvqdSc\nyaydgdgsNrUGdpTmPK+wAtUmgj5YJbB4wtlhtWN2wUw0djehp6+3n1lVFESsnHRHzCL6icCillt4\n2UUlZWW8V9PUYZCaM6BcMgDwKNh4B0sstJryNbnDb9JOFcxky/LoB2PWbu/p6BcMxmCHerQ5dDBE\n5zmzIDaWVjjGWxxXC1L91j3oCquaszYrwml1wCJa+KU3nlm7NDAF2+duweyC6xKey/VFsyFAwIdn\nP0JYDqM8bzr/ndb6YNR72QhJtMa9RCRDsacQO6/fqivragTXnDtVzZldvr32xEuJeiQ3T/tM1Kwd\nC4togSRadWbt7yJpXmMjLpFEcUnkcyagCEJ2cCQTGQ2ouc7xfM4AsLBY7SaVrM9zINhh2xQRDOz/\naPOunUlqzoDaZD1W1Gs8tMI5mUjtTKOf5mwyIAxQhPOlSHOHaO2RpRANhbmVuUG6uXBWNOfoXOOY\n7xGZHwugUttRqsKZCWO23+JdEAVBQIl3jCm/pc+ei9JRU3jpyXJNSVntBSeZZ3aoSSRmQ+uqkkSr\nLu4gWrmIhdaVwCwmyWK32HVmbVZqdlyOOeHsTkMJTxLOGQpLA8g1cfM0gm2ORCpblXjH8EMiGQEZ\nCxvXnFWzNgBM0ARbJWPW9tlz+eEQcPiT0ixYDq7NYjM0X2Yr0dHaNjM+Z01gzMX2y3BZnf1cLDeM\nrsRPplcZ1jM3i9rPuUc35nila7Uws29L1IVQJ5wjliUm9ONlNwyW+UVKZoDPnouxHrVoRiK5wJmK\nrkCR3YcizV4xJ5zVz3woNGdASffSmrW/a72AXJvXdOyMize/IOH8g8dnz0WuzZuw72UgFhTPw/Jx\nS1Eap3UcoGgDrBezUxouzVk5KJlJ1WWgOScjnAHVtJ3sIScKImYHZ2JR8fVJpY9lGo5IbnlvpPMQ\nq0CXCEyDaOxuRm1HPUZ7ivqludgsNswrrDCdb2uEKIiwilaNWZtpzolblJhlQL0QRlKARCt3H2l7\nWQsQYjZwSYbyvOmY6p+Em0sW6z43lnrltrqSDrpMNdrLnc+eiyJN6lSOiSBKbRewZM89htNi56lU\nzd2taOpu5hXRzKBGa6fOrE0BYRnKT6ZXxWytmCi59hysmHhbwq+vLJqDKx21qCyck/T/NsJuVTYd\nPygjN9IcmxejHAHUdzUk5XMGgKn+SfjsUo2u6cJgWV9WlfR7ZBqiIMJpdfBiLYMxa59uOqtr7jCc\n2ETJwOdsQnOOmGW1QYiAchlVzJ5dfF5Lxi5Ced5009G8ZsayZdbGfj93SS5UBK+F3+EzzOnNZCSN\nIPU5cnlqlwDBVO19reY8VMLZblXM2n1yH861KibtEpP+ZoCitQkN+a6B2woOJw6rHfdOvWvY3t/G\no7WZiVEVxFP9E/H5982mtCIjZgXLUR1ejZnB/r1ZCQWX5OLC2YxZmwkxFlk/OgXmfpvFxoVzc3cz\nLILF1KHPNeeQXnMGlOe9K6ymGkqi1XQk71CxIdJyM9vQmqCVpjiKcPZI5mrRazXnITNr8/raIZxr\nUYLBtFXGEn4fqx0CBMpzJkYu3KzdrTdrA8CqyXdiacli052uohEFccCqX4SC1nVgpkgL8zmzqNqU\naM4WiZeUbepuQY7Na8pkLvWLc1CfOSYEki2X+kNGe7nz23NR4MqHAAE5JuNl9Jrz0KSHaauEsUjt\nwWjOoiBirHd0Sl0OJJyJlMI2XXTkrPK1Y9iixAk9WuFs5iCURCvXZAUIuuYOw4VdtKElrBSnaA61\n8O5LidJfc1bnTsI5ebTPj9/hg80ioWrqStMWMLdNozkPUbS2WiWsC+daLiDg8A+6mNCTsx9JqcuB\nhDORUqJ9ScMVFU7ERiuMzPicASVoKRQOIc8Z0OUKDxe2SGEQVj0qXkWwaJjPOTpaG1Ajtkk4Dx5J\n1Ju1AWCRyVaygL4Ww1D5nFmVsMvtV9Ha04aZvvJBv1eqg0ITsg3t2bMH9957L6qqqvDVV18Zvual\nl17CAw88wL8/ePAgVqxYgVWrVuHTTz8dksES2Y/2lm0VLKYKYBBDh7a3stna40yQpcKkDSgHtQwZ\ntR31ANS61omias5qyViGKpyHJ3Xqh4B2Tydj9h2OgDCmOX964U8AgIlZVK8grnD+/PPP8d133+G9\n997Ds88+i2effbbfa06fPo0vvviCf9/Y2IhXX30V7777Ll577TV8/PHHQztqImvRbjqn1Zl1kakj\nBW0FLDMBYYAm7ShFud82Xh5SafVntiyomudMZu3hgBUMsZoM1Ov/Pmp3saHMcwaU7AKP5OaNSbKB\nuML5yJEjWLZsGQBg4sSJaG5uRltbm+41zz//PJ544gnd38yfPx8ejwfBYBA//elPh3jYRLaivWUP\nVy41ER9XEpqzi+cEp0hzFllLQqU8ZHTd9/h/r8yvT+4DoNecJ/kmIM8RQL4zPdkRIwG2p3323KQu\n24IgcO15yFKpNIVmbp+wjAvrbCCucK6rq4Pfr3YlCQQCqK1Va6geOHAA8+bNw+jRak7phQsX0NXV\nhYcffhj3338/jhw5MsTDJrIVnXAmf3PacGk1Z5ORsUWuICTROqiUlMGgas4R4WzS52yNsgw4LeoB\nvXjMAvzLgh1ZdWhnGmx9zMYCGMEsGEOnOSvCOc8RwKLi64fkPVOFaYefLMv866amJhw4cABvvvkm\nrly5ontdU1MT9u3bh0uXLmHdunX4wx/+EPNW5fe7YLUOrcM9Pz+50peZxEiZS0dTM//a5/Jk/byy\ndfxF3WoFrMJ85fKd6FzWBlZgxbU3I+BMvrFFIvguKNG1dV1K16NJxWPgd8Yeq3Yu+WE1algSrSgu\nHJ7qX8NF5j9jXtxTdgcmBcbFHWu83wfcObjYdhlFwQB8juTnPdtVin//rgA/mXUPigoSa7ObKMO9\nLnGFczAYRF2d2grs6tWryM9XquccPXoUDQ0NWLt2LUKhEM6dO4c9e/Zg6tSpmDVrFqxWK0pKSuB2\nu9HQ0IBRowY2HTU2Dm1yd7z2ZNnESJqLzaHeiC19UlbPK5vXRVtLoaWxC35nrsm5WFDblpq5h7sV\nheBqez0sggWhVsT839Hr0tHay792WBxZtWbZ8ozdVHAjAMQcayJzmZ5binAP0N0io7Z1KOYt4Z/m\nbo07NrMM1brEEvBxzdoLFy7E4cOHAQCnTp1CMBiEx6PcZG+77TYcOnQI77//Pvbt24eysjI8/fTT\nWLRoEY4ePYq+vj40Njaio6NDZxonfriQWTszSCaVKtUws6kMGT57juma3dqMAIpzyGwWj1mAR2Zu\nGJK67NlOXM25oqICZWVlqKqqgiAI2LVrFw4cOACv14tbbrnF8G8KCgqwfPlyrFmzBgCwc+dOiCJ9\n2IReOCfb4IIYPIOtEJYObLrykOZN6drLh9NCzxyRHSS0K5988knd99OmTev3mjFjxuCtt97i31dV\nVaGqauQ1DSCSw65LpSItJl2wiGtJtGZ8Opu+ApX5oCOdcKZnjsgSSJ0lUopEZu2MwCZKsAiWjDdp\nA/pSjmbTqABA0rTEJOFMZAsknImUIggC9wG66KBMG4IgwGV1ZoVw1rYkNFuABNAXWSHhTGQLme1s\nIkYkkiihp69XV0KSSD3zCit4d6lMRusKMVu6E4g2a9MzR2QHJJyJlGOz2NDR20kHZZpZNfnv0j2E\nhLAlqTlrA95IcyayBTJrEymHzNqEGfT9gs0LZ60rhS6ERLZAwplIOczMSCUTiURgmrMkWgfdoII9\nc6Q5E9kCCWci5bCIbcpzJhKB+Zz9dt+g075IOBPZBvmciZQzxlOM7t7urIgUJtIPF86D8DczVLM2\nCWciOyDhTKSc+6auQp/cl/HFL4jMwGF14P5pd6PYPfgWlcxaQz5nIlsg4UykHEEQYBGGtgMZMbJZ\nmGS7PzJrE9kG+ZwJghjxkHAmsg3SnAmCGPGM9hSiubuZMgSIrIGEM0EQI541U+5Cn9xHrQiJrIGE\nM0EQIx6KcyCyDbpGEgRBEESGQcKZIAiCIDIMEs4EQRAEkWGQcCYIgiCIDIOEM0EQBEFkGCScCYIg\nCCLDIOFMEARBEBkGCWeCIAiCyDBIOBMEQRBEhkHCmSAIgiAyDBLOBEEQBJFhCLIsy+keBEEQBEEQ\nKqQ5EwRBEESGQcKZIAiCIDIMEs4EQRAEkWGQcCYIgiCIDIOEM0EQBEFkGCScCYIgCCLDsKZ7AMPB\nnj17cPz4cQiCgKeffhrXXnttuodkihdeeAFffvklent78dBDD+GTTz7BqVOn4PP5AAAbNmzAkiVL\n0jvIBKipqcFjjz2GyZMnAwCmTJmCBx98ENu2bUM4HEZ+fj5efPFF2Gy2NI80Pr/73e9w8OBB/v3J\nkycxY8YMdHR0wOVyAQC2b9+OGTNmpGuIcfn222+xadMmrF+/HtXV1bh8+bLhWhw8eBC/+c1vIIoi\n1qxZg9WrV6d76P0wmstTTz2F3t5eWK1WvPjii8jPz0dZWRkqKir43/3617+GxWJJ48j7Ez2XHTt2\nGO73bFyXLVu2oLGxEQDQ1NSEmTNn4qGHHsKdd97J94rf78crr7ySzmEbEn0Ol5eXp3a/yCOMmpoa\neePGjbIsy/Lp06flNWvWpHlE5jhy5Ij84IMPyrIsyw0NDfKNN94ob9++Xf7kk0/SPDLzHD16VN68\nebPuZzt27JAPHToky7Isv/TSS/I777yTjqElRU1Njbx79265urpa/uabb9I9nIRob2+Xq6ur5Z07\nd8pvvfWWLMvGa9He3i7feuutcktLi9zZ2SnfcccdcmNjYzqH3g+juWzbtk3+8MMPZVmW5bffflve\nu3evLMuyPG/evLSNMxGM5mK037N1XbTs2LFDPn78uHz+/Hl55cqVaRhh4hidw6neLyPOrH3kyBEs\nW7YMADBx4kQ0Nzejra0tzaNKnLlz5+Lll18GAOTk5KCzsxPhcDjNoxo6ampqcPPNNwMAbrrpJhw5\nciTNIzLPq6++ik2bNqV7GKaw2Wx44403EAwG+c+M1uL48eMoLy+H1+uFw+FARUUFjh07lq5hG2I0\nl127dmH58uUAFE2sqakpXcMzhdFcjMjWdWGcOXMGra2tWWPFNDqHU71fRpxwrqurg9/v598HAgHU\n1tamcUTmsFgs3Ey6f/9+LF68GBaLBW+//TbWrVuHJ554Ag0NDWkeZeKcPn0aDz/8MO677z589tln\n6Ozs5GbsUaNGZdXaAMBXX32FoqIi5OfnAwBeeeUVrF27Fs888wy6urrSPLqBsVqtcDgcup8ZrUVd\nXR0CgQB/TSbuH6O5uFwuWCwWhMNhvPvuu7jzzjsBAKFQCFu3bkVVVRXefPPNdAw3JkZzAdBvv2fr\nujB++9vforq6mn9fV1eHLVu2oKqqSucuyhSMzuFU75cR6XPWImdpddLf//732L9/P371q1/h5MmT\n8Pl8KC0txS9+8Qvs27cPzzzzTLqHGJfx48fj0Ucfxe23347z589j3bp1OitANq7N/v37sXLlSgDA\nunXrMHXqVJSUlGDXrl145513sGHDhjSPcHAMtBbZtEbhcBjbtm1DZWUl5s+fDwDYtm0bVqxYAUEQ\nUF1djTlz5qC8vDzNI43Nj3/84377fdasWbrXZNO6hEIhfPnll9i9ezcAwOfz4bHHHsOKFSvQ2tqK\n1atXo7KyMq71IB1oz+Fbb72V/zwV+2XEac7BYBB1dXX8+6tXr3ItJ1v44x//iNdeew1vvPEGvF4v\n5s+fj9LSUgDA0qVL8e2336Z5hIlRUFCAH/3oRxAEASUlJcjLy0NzczPXMK9cuZKRGzIWNTU1/KC8\n5ZZbUFJSAiC71oXhcrn6rYXR/smWNXrqqacwbtw4PProo/xn9913H9xuN1wuFyorK7NijYz2ezav\nyxdffKEzZ3s8Htx9992QJAmBQAAzZszAmTNn0jhCY6LP4VTvlxEnnBcuXIjDhw8DAE6dOoVgMAiP\nx5PmUSVOa2srXnjhBbz++us8WnPz5s04f/48AEU4sOjnTOfgwYP45S9/CQCora1FfX09Vq1axdfn\no48+wg033JDOIZriypUrcLvdsNlskGUZ69evR0tLC4DsWhfGggUL+q3FddddhxMnTqClpQXt7e04\nduwY5syZk+aRxufgwYOQJAlbtmzhPztz5gy2bt0KWZbR29uLY8eOZcUaGe33bF0XADhx4gSmTZvG\nvz969Ciee+45AEBHRwe+/vprTJgwIV3DM8ToHE71fhlxZu2KigqUlZWhqqoKgiBg165d6R6SKQ4d\nOoTGxkY8/vjj/GerVq3C448/DqfTCZfLxR/sTGfp0qV48skn8fHHH6Onpwe7d+9GaWkptm/fjvfe\new/FxcW466670j3MhKmtreX+JUEQsGbNGqxfvx5OpxMFBQXYvHlzmkc4MCdPnsTevXtx8eJFWK1W\nHD58GD/72c+wY8cO3VpIkoStW7diw4YNEAQBjzzyCLxeb7qHr8NoLvX19bDb7XjggQcAKMGgu3fv\nRmFhIe655x6IooilS5dmXECS0Vyqq6v77XeHw5GV6/Lzn/8ctbW13MIEAHPmzMEHH3yAe++9F+Fw\nGBs3bkRBQUEaR94fo3P4+eefx86dO1O2X6hlJEEQBEFkGCPOrE0QBEEQ2Q4JZ4IgCILIMEg4EwRB\nEESGQcKZIAiCIDIMEs4EQRAEkWGQcCYIgiCIDIOEM0EQBEFkGCScCYIgCCLD+H/viJJufk7+cAAA\nAABJRU5ErkJggg==\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7fc213335320>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    }
  ]
}