{ "cells": [ { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "DK_9xTbrzykg" }, "source": [ "# Convolutional Neural Network Example\n", "\n", "Build a convolutional neural network with TensorFlow.\n", "\n", "This example is using TensorFlow layers API, see 'convolutional_network_raw' example\n", "for a raw TensorFlow implementation with variables.\n", "\n", "- Author: V.M. adapted from Aymeric Damien\n", "- Project: https://github.com/aymericdamien/TensorFlow-Examples/" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "RKyOnXmczykl" }, "source": [ "## CNN Overview\n", "\n", "![CNN](http://personal.ie.cuhk.edu.hk/~ccloy/project_target_code/images/fig3.png)\n", "\n", "## MNIST Dataset Overview\n", "\n", "This example is using MNIST handwritten digits. The dataset contains 60,000 examples for training and 10,000 examples for testing. The digits have been size-normalized and centered in a fixed-size image (28x28 pixels) with values from 0 to 1. For simplicity, each image has been flattened and converted to a 1-D numpy array of 784 features (28*28).\n", "\n", "![MNIST Dataset](http://neuralnetworksanddeeplearning.com/images/mnist_100_digits.png)\n", "\n", "More info: http://yann.lecun.com/exdb/mnist/" ] }, { "cell_type": "code", "execution_count": 20, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 35 }, "colab_type": "code", "id": "vluMT3pXzykn", "outputId": "d0499245-bce5-4199-862a-33026f2b1137" }, "outputs": [], "source": [ "from __future__ import division, print_function, absolute_import\n", "\n", "import tensorflow as tf\n", "import matplotlib.pyplot as plt\n", "import numpy as np\n", "from keras import backend as k \n", "from keras import optimizers\n", "from keras.engine import Input, Model\n", "from keras.models import Sequential\n", "from keras.layers import Dense, Conv2D, Dropout, Flatten, MaxPooling2D\n", "from keras.layers import Activation\n", "from keras.preprocessing.image import ImageDataGenerator\n", "from keras.callbacks import ModelCheckpoint, EarlyStopping\n", "\n" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "qDQa-kHhCtST" }, "source": [ "Import the data" ] }, { "cell_type": "code", "execution_count": 4, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 72 }, "colab_type": "code", "id": "4s0qX6MKClqp", "outputId": "00799fc0-6c70-4372-dfbb-989e3eae9755" }, "outputs": [], "source": [ "(x_train, y_train), (x_test, y_test) = tf.keras.datasets.mnist.load_data()" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "ooUX2a8qDz8X" }, "source": [ "Plot some images" ] }, { "cell_type": "code", "execution_count": 5, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 219 }, "colab_type": "code", "id": "HoPljxXwD3Z5", "outputId": "f1a385c0-7aee-4899-d29a-9d6a1f660ca3" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "8\n", "7\n", "5\n", "8\n", "7\n" ] }, { "data": { "text/plain": [ "" ] }, "execution_count": 5, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "%matplotlib inline\n", "plt.subplot(1,5,1)\n", "image_index = 7777 # You may select anything up to 60,000\n", "print(y_train[image_index]) # The label is 8\n", "plt.imshow(x_train[image_index], cmap='Greys')\n", "plt.subplot(1,5,2)\n", "image_index = 1302 # You may select anything up to 60,000\n", "print(y_train[image_index]) \n", "plt.imshow(x_train[image_index], cmap='Greys')\n", "plt.subplot(1,5,3)\n", "image_index = 31302 # You may select anything up to 60,000\n", "print(y_train[image_index]) \n", "plt.imshow(x_train[image_index], cmap='Greys')\n", "plt.subplot(1,5,4)\n", "image_index = 21432 # You may select anything up to 60,000\n", "print(y_train[image_index]) \n", "plt.imshow(x_train[image_index], cmap='Greys')\n", "plt.subplot(1,5,5)\n", "image_index = 54378 # You may select anything up to 60,000\n", "print(y_train[image_index]) \n", "plt.imshow(x_train[image_index], cmap='Greys')" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "_6P3i2JyDT4k" }, "source": [ "Reshaping the images" ] }, { "cell_type": "code", "execution_count": 6, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 72 }, "colab_type": "code", "id": "SKzxXjOCDOwk", "outputId": "ce494dff-9eee-49e2-f392-f591ffb38a16" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "x_train shape: (60000, 28, 28, 1)\n", "Number of images in x_train 60000\n", "Number of images in x_test 10000\n" ] } ], "source": [ "# Reshaping the array to 4-dims so that it can work with the Keras API\n", "x_train = x_train.reshape(x_train.shape[0], 28, 28, 1)\n", "x_test = x_test.reshape(x_test.shape[0], 28, 28, 1)\n", "input_shape = (28, 28, 1)\n", "# Making sure that the values are float so that we can get decimal points after division\n", "x_train = x_train.astype('float32')\n", "x_test = x_test.astype('float32')\n", "\n", "n_samples = x_train.shape[0]\n", "print('x_train shape:', x_train.shape)\n", "print('Number of images in x_train', x_train.shape[0])\n", "print('Number of images in x_test', x_test.shape[0])" ] }, { "cell_type": "code", "execution_count": 8, "metadata": { "colab": {}, "colab_type": "code", "id": "ppHoJpWxi45l" }, "outputs": [], "source": [ "batch_size = 128\n", "\n", "\n", "train_datagen = ImageDataGenerator(rescale=1./255)\n", "test_datagen = ImageDataGenerator(rescale=1./255)\n", "\n", "train_generator = train_datagen.flow(x_train, y_train,batch_size=batch_size)\n", "validation_generator = train_datagen.flow(x_train, y_train,batch_size=batch_size)\n", "test_generator = test_datagen.flow(x_test)" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "bYb5nn6jC3_T" }, "source": [ "Initializations" ] }, { "cell_type": "code", "execution_count": 16, "metadata": { "colab": {}, "colab_type": "code", "id": "N8sE73g9zyk0" }, "outputs": [], "source": [ "# Training Parameters\n", "learning_rate = 0.001\n", "epochs = 40 # small number of epoch to save time (it should be increased)\n", "\n", "# Network Parameters\n", "num_input = 784 # MNIST data input (img shape: 28*28)\n", "num_classes = 10 # MNIST total classes (0-9 digits)\n", "dropout = 0.25 # Dropout, probability to drop a unit\n", "\n", "img_width, img_height = 28, 28" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "L_BmiePyN_RZ" }, "source": [ "Model definition" ] }, { "cell_type": "code", "execution_count": 10, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 753 }, "colab_type": "code", "id": "gaW7XDmjzyk8", "outputId": "0f241e59-3f2b-42ec-c791-43fd796e5c5e", "scrolled": true }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "WARNING:tensorflow:From /home/valerie/.local/lib/python3.6/site-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n", "Instructions for updating:\n", "Colocations handled automatically by placer.\n", "WARNING:tensorflow:From /home/valerie/anaconda3/envs/conda_tf_env/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py:3445: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n", "Instructions for updating:\n", "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n", "_________________________________________________________________\n", "Layer (type) Output Shape Param # \n", "=================================================================\n", "input_1 (InputLayer) (None, 28, 28, 1) 0 \n", "_________________________________________________________________\n", "conv2d_1 (Conv2D) (None, 24, 24, 32) 832 \n", "_________________________________________________________________\n", "activation_1 (Activation) (None, 24, 24, 32) 0 \n", "_________________________________________________________________\n", "max_pooling2d_1 (MaxPooling2 (None, 12, 12, 32) 0 \n", "_________________________________________________________________\n", "conv2d_2 (Conv2D) (None, 12, 12, 64) 18496 \n", "_________________________________________________________________\n", "activation_2 (Activation) (None, 12, 12, 64) 0 \n", "_________________________________________________________________\n", "max_pooling2d_2 (MaxPooling2 (None, 6, 6, 64) 0 \n", "_________________________________________________________________\n", "flatten_1 (Flatten) (None, 2304) 0 \n", "_________________________________________________________________\n", "dense_1 (Dense) (None, 1024) 2360320 \n", "_________________________________________________________________\n", "activation_3 (Activation) (None, 1024) 0 \n", "_________________________________________________________________\n", "dropout_1 (Dropout) (None, 1024) 0 \n", "_________________________________________________________________\n", "dense_2 (Dense) (None, 10) 10250 \n", "_________________________________________________________________\n", "activation_4 (Activation) (None, 10) 0 \n", "=================================================================\n", "Total params: 2,389,898\n", "Trainable params: 2,389,898\n", "Non-trainable params: 0\n", "_________________________________________________________________\n" ] } ], "source": [ "# Creating a Sequential Model and adding the layers\n", "\n", "conv_net_in = Input(shape=(img_width, img_height, 1))\n", "# First 2D convolution Layer\n", "# Convolution Layer with 32 filters and a kernel size of 5\n", "conv_net = Conv2D(32, (5, 5))(conv_net_in)\n", "conv_net = Activation(\"relu\")(conv_net)\n", "# Max Pooling (down-sampling) with strides of 2 and kernel size of 2\n", "conv_net = MaxPooling2D()(conv_net)\n", "# Second 2D convolution Layer\n", "conv_net = Conv2D(64, (3, 3), padding=\"same\")(conv_net)\n", "conv_net = Activation(\"relu\")(conv_net)\n", "conv_net = MaxPooling2D()(conv_net)\n", "# Flatten the data to a 1-D vector for the fully connected layer\n", "conv_net = Flatten()(conv_net)\n", "\n", "# first fully connected\n", "conv_net = Dense(1024)(conv_net)\n", "conv_net = Activation('relu')(conv_net)\n", "conv_net = Dropout(dropout)(conv_net)\n", "\n", "# Output\n", "conv_net = Dense(10)(conv_net)\n", "conv_net = Activation('softmax')(conv_net)\n", "\n", "\n", "conv_model = Model(conv_net_in, conv_net)\n", "\n", "conv_model.summary()" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "2nUji-ZGODlM" }, "source": [ "Hyper parameter for the learnng task" ] }, { "cell_type": "code", "execution_count": 21, "metadata": { "colab": {}, "colab_type": "code", "id": "rA8rqtBYzylD" }, "outputs": [], "source": [ "adam = optimizers.Adam(lr=learning_rate)\n", "\n", "conv_model.compile(optimizer=adam, \n", " loss='sparse_categorical_crossentropy', \n", " metrics=['accuracy'])\n", "\n", "patience = 4\n", "early = EarlyStopping(monitor = 'val_loss', patience = patience,\n", " verbose = 1, mode = 'auto')\n", "\n", "checkpointer = ModelCheckpoint(filepath=\"CNN_MNIST_best_weights.hdf5\", \n", " monitor = 'val_acc',\n", " verbose=1, \n", " save_best_only=True)\n" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "hyzedBVJO3V5" }, "source": [ "Fit/learn the model" ] }, { "cell_type": "code", "execution_count": 22, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 478 }, "colab_type": "code", "id": "kotE3EDmN4d3", "outputId": "d3d2376b-62dc-4fa7-df20-ee2f06eda3f4", "scrolled": true }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Epoch 1/40\n", "469/468 [==============================] - 33s 71ms/step - loss: 0.0069 - acc: 0.9978 - val_loss: 0.0062 - val_acc: 0.9981\n", "\n", "Epoch 00001: val_acc improved from -inf to 0.99813, saving model to CNN_MNIST_best_weights.hdf5\n", "Epoch 2/40\n", "469/468 [==============================] - 34s 72ms/step - loss: 0.0071 - acc: 0.9977 - val_loss: 0.0037 - val_acc: 0.9988\n", "\n", "Epoch 00002: val_acc improved from 0.99813 to 0.99875, saving model to CNN_MNIST_best_weights.hdf5\n", "Epoch 3/40\n", "469/468 [==============================] - 35s 74ms/step - loss: 0.0062 - acc: 0.9978 - val_loss: 0.0023 - val_acc: 0.9991\n", "\n", "Epoch 00003: val_acc improved from 0.99875 to 0.99906, saving model to CNN_MNIST_best_weights.hdf5\n", "Epoch 4/40\n", "469/468 [==============================] - 35s 75ms/step - loss: 0.0048 - acc: 0.9986 - val_loss: 0.0017 - val_acc: 0.9992\n", "\n", "Epoch 00004: val_acc improved from 0.99906 to 0.99922, saving model to CNN_MNIST_best_weights.hdf5\n", "Epoch 5/40\n", "469/468 [==============================] - 36s 76ms/step - loss: 0.0038 - acc: 0.9988 - val_loss: 0.0012 - val_acc: 0.9995\n", "\n", "Epoch 00005: val_acc improved from 0.99922 to 0.99953, saving model to CNN_MNIST_best_weights.hdf5\n", "Epoch 6/40\n", "469/468 [==============================] - 38s 81ms/step - loss: 0.0058 - acc: 0.9982 - val_loss: 0.0030 - val_acc: 0.9991\n", "\n", "Epoch 00006: val_acc did not improve from 0.99953\n", "Epoch 7/40\n", "469/468 [==============================] - 36s 77ms/step - loss: 0.0031 - acc: 0.9991 - val_loss: 9.7505e-04 - val_acc: 0.9998\n", "\n", "Epoch 00007: val_acc improved from 0.99953 to 0.99977, saving model to CNN_MNIST_best_weights.hdf5\n", "Epoch 8/40\n", "469/468 [==============================] - 37s 79ms/step - loss: 0.0061 - acc: 0.9981 - val_loss: 0.0025 - val_acc: 0.9994\n", "\n", "Epoch 00008: val_acc did not improve from 0.99977\n", "Epoch 9/40\n", "469/468 [==============================] - 36s 77ms/step - loss: 0.0050 - acc: 0.9985 - val_loss: 8.2932e-04 - val_acc: 0.9998\n", "\n", "Epoch 00009: val_acc improved from 0.99977 to 0.99984, saving model to CNN_MNIST_best_weights.hdf5\n", "Epoch 10/40\n", "469/468 [==============================] - 36s 77ms/step - loss: 0.0046 - acc: 0.9988 - val_loss: 0.0038 - val_acc: 0.9991\n", "\n", "Epoch 00010: val_acc did not improve from 0.99984\n", "Epoch 11/40\n", "469/468 [==============================] - 37s 78ms/step - loss: 0.0024 - acc: 0.9992 - val_loss: 0.0023 - val_acc: 0.9991\n", "\n", "Epoch 00011: val_acc did not improve from 0.99984\n", "Epoch 12/40\n", "469/468 [==============================] - 38s 82ms/step - loss: 0.0027 - acc: 0.9992 - val_loss: 0.0034 - val_acc: 0.9994\n", "\n", "Epoch 00012: val_acc did not improve from 0.99984\n", "Epoch 13/40\n", "469/468 [==============================] - 39s 83ms/step - loss: 0.0043 - acc: 0.9988 - val_loss: 0.0056 - val_acc: 0.9982\n", "\n", "Epoch 00013: val_acc did not improve from 0.99984\n", "Epoch 00013: early stopping\n" ] }, { "data": { "text/plain": [ "" ] }, "execution_count": 22, "metadata": {}, "output_type": "execute_result" } ], "source": [ "conv_model.fit_generator(\n", " train_generator,\n", " steps_per_epoch=n_samples/batch_size,\n", " epochs=epochs,\n", " validation_data=validation_generator,\n", " validation_steps=100,\n", " callbacks=[checkpointer,early])" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "AiC7sNeeOle_" }, "source": [ "Save the best model" ] }, { "cell_type": "code", "execution_count": 23, "metadata": { "colab": {}, "colab_type": "code", "id": "V9Lnntx0OoP2" }, "outputs": [], "source": [ "conv_model.load_weights('CNN_MNIST_best_weights.hdf5')\n", "conv_model.save('CNN_MNIST.h5')\n" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "v7IMEt1MO8Mn" }, "source": [ "Validation on the test set" ] }, { "cell_type": "code", "execution_count": 24, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 237 }, "colab_type": "code", "id": "0mK7eh3CNiIo", "outputId": "a0abff9e-ce63-4477-8bd7-82bbcf977d4a" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Confusion matrix\n", "[[ 979 0 0 0 0 0 0 0 0 1]\n", " [ 0 1134 0 0 0 0 1 0 0 0]\n", " [ 2 0 1023 0 0 0 0 6 1 0]\n", " [ 0 0 0 1007 0 2 0 0 1 0]\n", " [ 0 0 0 0 974 0 2 0 0 6]\n", " [ 1 1 0 7 0 882 1 0 0 0]\n", " [ 5 2 0 0 1 0 948 0 2 0]\n", " [ 0 0 1 0 0 0 0 1025 1 1]\n", " [ 3 0 2 0 0 1 0 3 962 3]\n", " [ 1 0 0 0 4 1 0 3 2 998]]\n", "Classification error: 0.68 %\n" ] } ], "source": [ "from sklearn.metrics import confusion_matrix\n", "y_proba = conv_model.predict(x_test/255.)\n", "C = y_proba.argmax(axis=-1) \n", "#print(C[:10])\n", "\n", "M = confusion_matrix(y_test,C)\n", "print(\"Confusion matrix\")\n", "print(M)\n", "\n", "print(\"Classification error: \", np.round((1-np.sum(np.diag(M))/np.shape(x_test)[0])*100,2),\"%\")\n" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "nPZ1uvLp9qiQ" }, "source": [ "Plot the 32 filters of the first convolution layer\n", "\n" ] }, { "cell_type": "code", "execution_count": 25, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 470 }, "colab_type": "code", "id": "kUEg8gbn9rS5", "outputId": "b5edd3b1-6031-4e2c-f951-210e38e8493e" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "(5, 5, 1, 32)\n" ] }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "weights, biases = conv_model.layers[1].get_weights()\n", "print(weights.shape)\n", "\n", "plt.figure(figsize=(8,8))\n", "for k in range(weights.shape[3]):\n", " plt.subplot(4,8,k+1)\n", " plt.imshow(weights[:,:,0,k])\n", " plt.title(str(k+1))" ] }, { "cell_type": "markdown", "metadata": { "colab_type": "text", "id": "q0oPrM3yPsYs" }, "source": [ "Plot some intermediate layers" ] }, { "cell_type": "code", "execution_count": 26, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 816 }, "colab_type": "code", "id": "xu-HIWN3PwAb", "outputId": "cecc4bbe-5410-4c91-ec91-23ca8274617c" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[, , , , , , ]\n", "(28, 28, 1)\n", "(1, 28, 28, 1)\n", "(28, 28, 1)\n" ] }, { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAADftJREFUeJzt3X+MXXWZx/HP0zJtsVRo01K7pVKK7UJhQ9FJFdFdCIuLxFhMFtZm1x2M7rhZ2dWkiZJmEzGKIUZAN2vcVGksCT9k+VkjKrVqAHdSOmVZWqnaLjuLtZMOTUdbdLftTB//mFMytnO+9/be8+NOn/crae695zn3nCcXPnPuvd9zz9fcXQDimVJ3AwDqQfiBoAg/EBThB4Ii/EBQhB8IivADQRF+ICjCDwR1RpU7m2bTfYZmVrlLIJT/1291xA9bM+u2FX4zu07SVyRNlfQNd78jtf4MzdTb7Zp2dgkgYYtvbnrdlt/2m9lUSV+V9F5JyyWtNrPlrW4PQLXa+cy/UtJud3/Z3Y9IelDSqmLaAlC2dsK/UNIvxz3eky37A2bWa2b9ZtZ/VIfb2B2AIrUT/om+VDjp98Huvs7du929u0vT29gdgCK1E/49khaNe3yepL3ttQOgKu2Ef6ukpWZ2gZlNk/RBSRuLaQtA2Voe6nP3ETO7RdL3NTbUt97df1pYZwBK1dY4v7s/KenJgnoBUCFO7wWCIvxAUIQfCIrwA0ERfiAowg8ERfiBoAg/EBThB4Ii/EBQhB8IivADQRF+IKhKL92N1gx8/opkfXTGSRdQet28S15NPrfvskda6um4C3/44WR91nNn5tbm/8t/tLVvtIcjPxAU4QeCIvxAUIQfCIrwA0ERfiAowg8ExTh/Bxj+ztJkfceKfy1t30fzTxFoys+u/kayfl/3gtzaQ5v+LPnc0Z27WuoJzeHIDwRF+IGgCD8QFOEHgiL8QFCEHwiK8ANBtTXOb2YDkg5JGpU04u7dRTR1umk0jv+TFQ+Wtu9/+/WSZP2uvmuT9cXnp68H8NTyR5P1v541mFu7/ea5yecu+TTj/GUq4iSfq919fwHbAVAh3vYDQbUbfpf0lJltM7PeIhoCUI123/Zf6e57zexcSZvM7Gfu/vT4FbI/Cr2SNENvaHN3AIrS1pHf3fdmt0OSHpO0coJ11rl7t7t3d2l6O7sDUKCWw29mM81s1vH7kt4jaUdRjQEoVztv++dLeszMjm/nfnf/XiFdAShdy+F395clXVZgL5PWyDVvS9Z/eNlXG2yhK1n98vCyZP1Hf5U4vWLvUPK5y4b7k/UpM2Yk61/Y8ifJ+tq523NrI7NHks9FuRjqA4Ii/EBQhB8IivADQRF+ICjCDwTFpbsL8NrCacn6lAZ/YxsN5f34/enhtNGXf56st2P3Zy9P1u+fc2eDLeSf1Xne9zj21IlXHwiK8ANBEX4gKMIPBEX4gaAIPxAU4QeCYpy/AOfc25es/2X/3yTrNnwwWR8ZHDjFjorz0et/kKyfNYWrM01WHPmBoAg/EBThB4Ii/EBQhB8IivADQRF+ICjG+Ssw+tIv6m4h18DtVyTrHznnSw22kL6095rBd+TWZv1gZ/K5ow32jPZw5AeCIvxAUIQfCIrwA0ERfiAowg8ERfiBoBqO85vZeknvkzTk7pdmy+ZI+pakxZIGJN3k7sPltYlW/fpD6XH8n/xtehz/7Cnpcfy+w1OT9Rc+n3/d/zMPPpd8LsrVzJH/m5KuO2HZrZI2u/tSSZuzxwAmkYbhd/enJR04YfEqSRuy+xsk3VBwXwBK1upn/vnuPihJ2e25xbUEoAqln9tvZr2SeiVpht5Q9u4ANKnVI/8+M1sgSdntUN6K7r7O3bvdvbsrMWkjgGq1Gv6Nknqy+z2SniimHQBVaRh+M3tAUp+kPzazPWb2EUl3SLrWzHZJujZ7DGASafiZ391X55SuKbgXlGD/Wz1ZbzSO30jPjz+arC97nLH8TsUZfkBQhB8IivADQRF+ICjCDwRF+IGguHT3aeDIpvNza30X3dng2emhvsv6epL1i9f8d7LO5bc7F0d+ICjCDwRF+IGgCD8QFOEHgiL8QFCEHwiKcf5J4Iwli5P1z73l33Nrsxv8ZHfb4fS+z/9ceqR+dJgrtk9WHPmBoAg/EBThB4Ii/EBQhB8IivADQRF+ICjG+SeBCx/6VbJ++bTW/4av3vz3yfqy/9ra8rbR2TjyA0ERfiAowg8ERfiBoAg/EBThB4Ii/EBQDcf5zWy9pPdJGnL3S7Nlt0n6O0mvZqutdfcny2rydDfcc0Wy/tn5ja69Pz230jPw58lnXvyp3ck6190/fTVz5P+mpOsmWH63u6/I/hF8YJJpGH53f1rSgQp6AVChdj7z32JmL5rZejObXVhHACrRavi/JulCSSskDUrK/VBqZr1m1m9m/UfV4IJxACrTUvjdfZ+7j7r7MUlfl7Qyse46d+929+6uxBdTAKrVUvjNbMG4hx+QtKOYdgBUpZmhvgckXSVprpntkfQZSVeZ2QpJLmlA0sdK7BFACRqG391XT7D4nhJ6OW2dsfCPkvV3/9OWZP2sKa1/XOp76S3J+rJhfq8fFWf4AUERfiAowg8ERfiBoAg/EBThB4Li0t0V2Ll2UbL++Ju+3db2r95+Y26Nn+wiD0d+ICjCDwRF+IGgCD8QFOEHgiL8QFCEHwiKcf4KbHv/3Q3WaO8KR2f/w7Hc2sjwcFvbxumLIz8QFOEHgiL8QFCEHwiK8ANBEX4gKMIPBMU4/2ng6Pyzc2tdRxZW2MnJRl/dn1vzw+np22x6+vyHqfPmttSTJI3OOydZ37VmWsvbboaPWm7ton9scA2GgwcL6YEjPxAU4QeCIvxAUIQfCIrwA0ERfiAowg8E1XCc38wWSbpX0pskHZO0zt2/YmZzJH1L0mJJA5Jucnd+PF6D7zy8vu4Wcr3zPyea4X3M/n1vTD539rxDyfqWt93fUk+dbvk/35KsL/lUXyH7aebIPyJpjbtfLOkdkj5uZssl3Spps7svlbQ5ewxgkmgYfncfdPfns/uHJO2UtFDSKkkbstU2SLqhrCYBFO+UPvOb2WJJl0vaImm+uw9KY38gJJ1bdHMAytN0+M3sLEmPSPqkuzd9crGZ9ZpZv5n1H1X6XG4A1Wkq/GbWpbHg3+fuj2aL95nZgqy+QNLQRM9193Xu3u3u3V1tXqgSQHEaht/MTNI9kna6+13jShsl9WT3eyQ9UXx7AMpi7p5ewexdkp6RtF1jQ32StFZjn/sfkvRmSa9IutHdD6S29Uab42+3a9rtedL5v+9fkKxvvvThijqJ5Xd+JLd21PMvd96M61+8OVn/zQut/9x4wbMjyfr0727NrW3xzTroB/J/LzxOw3F+d39WUt7G4iUZOE1whh8QFOEHgiL8QFCEHwiK8ANBEX4gKC7dXYEz/+J/kvVLvpD+CaeX+F9p1kXJUzNK/dnsJc98OFn3V2a2tf0lD7+WX3xue1vbnq1dbdU7AUd+ICjCDwRF+IGgCD8QFOEHgiL8QFCEHwiq4e/5ixT19/xAVU7l9/wc+YGgCD8QFOEHgiL8QFCEHwiK8ANBEX4gKMIPBEX4gaAIPxAU4QeCIvxAUIQfCIrwA0ERfiCohuE3s0Vm9iMz22lmPzWzT2TLbzOzX5nZC9m/68tvF0BRmpkOYkTSGnd/3sxmSdpmZpuy2t3u/qXy2gNQlobhd/dBSYPZ/UNmtlPSwrIbA1CuU/rMb2aLJV0uaUu26BYze9HM1pvZ7Jzn9JpZv5n1H9XhtpoFUJymw29mZ0l6RNIn3f2gpK9JulDSCo29M7hzoue5+zp373b37i5NL6BlAEVoKvxm1qWx4N/n7o9Kkrvvc/dRdz8m6euSVpbXJoCiNfNtv0m6R9JOd79r3PIF41b7gKQdxbcHoCzNfNt/paQPSdpuZi9ky9ZKWm1mKyS5pAFJHyulQwClaObb/mclTXQd8CeLbwdAVTjDDwiK8ANBEX4gKMIPBEX4gaAIPxAU4QeCIvxAUIQfCIrwA0ERfiAowg8ERfiBoAg/EJS5e3U7M3tV0v+OWzRX0v7KGjg1ndpbp/Yl0VuriuztfHef18yKlYb/pJ2b9bt7d20NJHRqb53al0RvraqrN972A0ERfiCousO/rub9p3Rqb53al0Rvraqlt1o/8wOoT91HfgA1qSX8Znadmf3czHab2a119JDHzAbMbHs283B/zb2sN7MhM9sxbtkcM9tkZruy2wmnSaupt46YuTkxs3Str12nzXhd+dt+M5sq6ReSrpW0R9JWSavd/aVKG8lhZgOSut299jFhM/tTSa9JutfdL82WfVHSAXe/I/vDOdvdP90hvd0m6bW6Z27OJpRZMH5maUk3SLpZNb52ib5uUg2vWx1H/pWSdrv7y+5+RNKDklbV0EfHc/enJR04YfEqSRuy+xs09j9P5XJ66wjuPujuz2f3D0k6PrN0ra9doq9a1BH+hZJ+Oe7xHnXWlN8u6Skz22ZmvXU3M4H52bTpx6dPP7fmfk7UcObmKp0ws3THvHatzHhdtDrCP9HsP5005HClu79V0nslfTx7e4vmNDVzc1UmmFm6I7Q643XR6gj/HkmLxj0+T9LeGvqYkLvvzW6HJD2mzpt9eN/xSVKz26Ga+3ldJ83cPNHM0uqA166TZryuI/xbJS01swvMbJqkD0raWEMfJzGzmdkXMTKzmZLeo86bfXijpJ7sfo+kJ2rs5Q90yszNeTNLq+bXrtNmvK7lJJ9sKOPLkqZKWu/ut1fexATMbInGjvbS2CSm99fZm5k9IOkqjf3qa5+kz0h6XNJDkt4s6RVJN7p75V+85fR2lcbeur4+c/Pxz9gV9/YuSc9I2i7pWLZ4rcY+X9f22iX6Wq0aXjfO8AOC4gw/ICjCDwRF+IGgCD8QFOEHgiL8QFCEHwiK8ANB/R5UEeYO44sn+QAAAABJRU5ErkJggg==\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" }, { "name": "stdout", "output_type": "stream", "text": [ "(1, 24, 24, 32)\n" ] }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "from keras import models\n", "from keras.preprocessing import image\n", "\n", "layer_outputs = [layer.output for layer in conv_model.layers[:7]] \n", "# Extracts the outputs of the top layers\n", "\n", "print(layer_outputs )\n", "activation_model = models.Model(inputs=conv_model.input, outputs=layer_outputs[1]) # Creates a model that will return these outputs, given the model input\n", "#activation_model.summary()\n", "\n", "img = np.reshape(x_test[1,:,:],(28,28,1))\n", "print(img.shape)\n", "img_tensor = image.img_to_array(img)\n", "img_tensor = np.expand_dims(img_tensor, axis=0)\n", "img_tensor /= 255.\n", "\n", "print(img_tensor.shape)\n", "\n", "print(img_tensor[0].shape)\n", "\n", "plt.imshow(img_tensor[0,:,:,0])\n", "plt.show()\n", "\n", "activations = activation_model.predict(img_tensor/255.) # Returns a Numpy array corresponding to the output of layer 2\n", "print(activations.shape)\n", "\n", "# Show the result of the 1st convolution\n", "\n", "plt.figure(figsize=(8,8))\n", "for k in range(activations.shape[3]):\n", " plt.subplot(4,8,k+1)\n", " plt.imshow(activations[0,:,:,k])\n", " plt.title(str(k+1))" ] }, { "cell_type": "code", "execution_count": 27, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 470 }, "colab_type": "code", "id": "SzyUWMy3ia9Y", "outputId": "165b9fd0-b9da-49c3-ea61-e302d7690c1c" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "(1, 24, 24, 32)\n" ] }, { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeMAAAGzCAYAAAAc+X/PAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzs3XmcpGV97/3Pr6ur9+6Z7ll7dpYBZkAWHYO4IAYQNI9LQvBg9Ejyioc8MUTJMYlo4hZNjCaP5yFR44tzUEHJURQinATDoyMoGsEZIYAw7Aww+9r7Ur38nj+uq4aesbfprqqrq/r7fr36VVN3Vd31+87dXVdd93Xf123ujoiIiKRTlboAERGR+U6NsYiISGJqjEVERBJTYywiIpKYGmMREZHE1BiLiIgkpsZYREQksbJvjM3sajPbamaDZva11PUUg5nVmtkNZva8mXWb2YNm9qbUdRWamX3DzHabWZeZPWlm701dUzGY2XozGzCzb6SupRjM7J6Yryf+PJG6pmIwsyvMbJuZ9ZrZM2b2utQ1FcqYbZf/GTGzf0xdVyGZ2Tozu9PMDpvZHjP7gplVp6qn7BtjYBfwaeArqQspomrgReD1wALgo8AtZrYuYU3F8Blgnbu3AG8FPm1mr0hcUzF8EdiSuogiu9rdm+LPqamLKTQzuxj4LPB7QDNwPvBs0qIKaMy2awKWAf3AtxOXVWhfAvYB7cDZhM/X96UqpuwbY3e/zd2/CxxMXUuxuHuvu3/C3be7+6i7/yvwHFBRDZW7P+rug/m78eekhCUVnJldAXQAm1PXIrPySeCv3P2++De50913pi6qSH6b0Gjdm7qQAjsBuMXdB9x9D/DvwOmpiin7xng+MrNlwCnAo6lrKTQz+5KZ9QGPA7uBOxOXVDBm1gL8FfDB1LWUwGfM7ICZ/dTMLkhdTCGZWQbYBCwxs6fNbEfcxVmfurYiuRK4yStv7uTrgCvMrMHMVgJvIjTISagxLjNmlgVuBm5098dT11No7v4+wm6/1wG3AYOTv6KsfAq4wd1fTF1IkX0IOBFYCVwP/B8zq6Q9HMuALKHH+DrCLs5zgL9MWVQxmNkawu7bG1PXUgQ/IvSEu4AdwFbgu6mKUWNcRsysCvg6kAOuTlxO0bj7iLv/BFgF/GHqegrBzM4GLgL+R+pais3d73f3bncfdPcbgZ8Cb05dVwH1x9t/dPfd7n4A+DyVlTHvPcBP3P251IUUUvwsvYvwhb8RWAy0Eo4DSEKNcZkwMwNuIHwrv8zdhxKXVArVVM6Y8QXAOuAFM9sD/ClwmZk9kLKoEnHAUhdRKO5+mNCTqrTdtuN5D5XZK24DVgNfiF8aDwJfJeEXqrJvjM2s2szqgAyQMbO6lIenF9E/ARuAt7h7/1RPLjdmtjSeKtJkZhkzuwR4J/DD1LUVyPWELxZnx58vA/8GXJKyqEIzs4Vmdkn+79DM3kU40viu1LUV2FeBP46/t63ANcC/Jq6poMzs1YShhko7ipq4N+M54A/j7+lCwtj4Q6lqKvvGmDBO0w9cC7w7/ruixm7MbC3wB4QP8T1jzv17V+LSCskJu6R3AIeBvweucffbk1ZVIO7e5+578j9ADzDg7vtT11ZgWcKphvuBA8AfA29390o71/hThNPTngS2AQ8Cf520osK7ErjN3btTF1IkvwVcSvhdfRoYBv4kVTFWeQfIiYiIlJdK6BmLiIiUNTXGIiIiic2qMTazS83siXji+7WFKmouUcbKoIyVQRkrw3zIeLxmPGYcZ6F5EriYcNDNFuCd7v5Y4cpLSxkrgzJWBmWsDPMh40zMpmf8a8DT7v6su+eAbwJvK0xZc4YyVgZlrAzKWBnmQ8bjNpvzcVcSriSUtwM4d7IX1Fit19E4i7csrToaGWGIFmvzAXohnAA/acbG1hpvWzn1FLUW5wvI31bl5w+IUyMMe/ieNDRaffR9z4T7o1VHvS5TNQpAtYXbuqowJ4hNMdVC26p6BnqGWX3GAj+0sx+mkXE+bEdlnHuUcXyZpkavbmub9nvEjwaqcvH1g+Ezw3LDYcHoaHxC1VG3nom32fChMlITbuNHFD7N1qR6URujAwPUrlntw4cOQQVux7EG6CXng1NOejObxni8lf/KPm8zuwq4CqCOBs61C2fxlqW113dwkD1stE3c75sZIgdTZFzYXscHbnnVhOvM2ggADVVhyuX1tXuOevyB/nUA3L7rLAAO9jYA0NMRbqsOZsNt3LbDzeEPp6ot/GWtXRYuXnXuou0ALM72TJrxobv28ORPD3L5X53Ode+4j76OoSkzzoftqIxzjzK+ZGzGTOtC2v/8AxOu00bCZ0XDztCYNu8InxkNe3Lx8bD6vuU1AHStjV/4m8LrR6uP7ihU9xz90Z9vhIeawvNG6kcnzdj74EMMbHuSRb9zObs/dx253r4pM5bbdhzrfp/eBdpms5t6B2E6sbxVhGsLH8Xdr3f3Te6+KUvtLN6u9GqpZ4CjJruaMmNTW03J6iuEBcvq6Ng9MHaRtqMyzknK+JKxGTNNTSWrrxCqFy5g+HDH2EUVtx1nYjY94y3AejM7AdgJXAH8TkGqmiNaaKWfHvq9Fw9f3Gadsa069FTXZcPES2ur+wD4SX/4XvOF/3wDADWPh13d2Tj3zYK4aymTi18g45fPkbrwLXZgUXj+7toWAPY3NwNT94xXn9HCgRf6OLSjDx8tTMa5phjbca5RxspQ0IweerA1h0OfK37UEEe8OHxKaOA6NoTPlFe88ikAPrL8pwAsz3QB8C+d4bLpd2w/A4DufaHxr+oLnz1xZ9+RHvhUatasZnj/AYYOHCIeQFxx23EmZtwYu/uwmV1NmHM2A3zF3Svq+rpVVsWpfjYPci/99EG4EHVFZcxUV/H2vziN/3nVAxze1Q8VmHE+bEdlrAzzIaNlMrRd/nb2fel/MnzoMFRgxpmY1QUV3P1OKuji7+NZbO0spp37fTNdfmjGc8/WxqMmVmfDmO6p2bAr6tnh8O30o78IBxO2fyfs5q7f2wvAYFt4vLc9fAsdXBC+fdbvD99mG3aEgy6yPeHx/SvC2PLwqumPQGw4fwkbzl/Cde+4jxd/2Vlp8+sChduOc5kyVoZCZayKx2PFYz4Zjsc/dS4Knw3ZVxwG4L+sDe3guroDAPy8N1wo7f/bdRoA+x9dAkDDrvC6urDzjVxL2D030hK7xrFnbLmpe8j1p29g5ekb2P256xh84cWK3I7HSzNwiYiIJFaJlxqck/JHUa+PY8ZDcej3czveBMCCzWHMt/mJ0HPOLQ3jMrtfHb7WZk8N4zcNtaGH3fngIgDavht60NX94fUH+8ImzZ/iJCLzU75HnGsLnz22MQwaX7juSQBObtgLwJN9ywH4+4cuAiCzLXz2tG0LnyGrumPPtyrc9reFz5judaEvlz9pZ3SKo6hlcuoZi4iIJKaecYlk4uHP3fEIx/v61wLwi0dPBGDdzjDA07WxFYBd4aBqNp0VvsW+r/1uALYPLQbgs1t/O6x3bzhFoLo2bMrR+vD9amG2r0hJRKQcjNaGz5xsazh1cVVb+KzYMxDOtHi0I/SIt+8InylNj4XjUxY+E3rA2a7wmeTV4TOle0X4jBloi5N+1MUJhwbyY8T5yUF0Wd6ZUM9YREQkMfWMS+zF4XAo4iN9q49avudVYWatwRPDt9g3bwhHOL5/6Q8BWFsdjrK+uycc2dgax3NGD4Qx5uGN7cBL34Jbq9UzFpnPbDj0WIe6Qo/3mcHw2eFDoQ/W8Ez4TFl0IN/DDbfDdeF1AwvC44Ot4X7vmviZUxd6zlX9x/TlpneasUxAPWMREZHE1DMuke7ROgC25+I5e5kwL+yJJ4e5qYdOCIc+vnbpMwBcuuBhAE6oDq97NE7iftPPXw3Axvt2AjAaJ3HvWht61hvbw7U7svmTDEVkfoo9VYs94arueGZGd5yRK86Cm7/gw0h+xkk/+gIQvatij7ghHlWdHxPuqzrqfX51dmk5HuoZi4iIJKaecYmMxglhO0fCDFkLMmEGrguXPgG81JNdUh0mo14Xz0fuHA1fOz/2wmUArPh+vKLK86EHXHX2RgAOvjx8a31jy+6wvnhec/59RWSeyZ/2mz8POH7a5xaGB4YawwN1B/K38bKusYfbsyYsz64KcxlkMuF1ffvCVF7VvUcPEg83qms8G/qkFhERSUw948Tq4pzVy7KdAJxWE3q2i6vCkYzf6F4HwNM/iOcj/+jp8MLFYQauHb++EICzNoax5qU1YaYu9YhFBHhpLDd2eT12lbN98bb3qIfpXxqPpl4a572Pe+cGe8Isf9VdYe9c1XB+rDmuN1Ok+ucJfWKLiIgkpp5xYvmrOS3KhDHiE6vDt9Hn48HQn/mPNwNw6vdCj9c7w233b5wFgJ0frryyqfX50hQsImUtf6JFdX+8HyfKz48pDyzOz9w1CMBQfzhTI7s33nbHo63DzrsjM3Ed6VrLjKhnLCIikph6xolUxasq5XvEm2oPATDg4dvlp3a+BYCVd4aBGHvsWQB8Q7jW6M6Lw/N+a3U4Gjt/dPaQBm5EZBwWx35rDsfrGYcTN8gfXtITJwUcWRT21tVkwxkZwwfDXAeZ/mOOnm6IY8XV6hEXgnrGIiIiialnnEhDVZiBa3U2zC29OBPO3ftub7iW6EP/tgGAtZsfjS8IRzLufMMCAM572bbw+rrQo1aPWEQmk7+6UiYMBWOjoUc7UhuWDzfFGbZG4tHTO8JnUU1nvBpT7AEPNcSn1apHXEjqGYuIiCSmnnGJ5WfGyo8Vn1UTesgvDIdDHP9q238BYOWPwlWXRjrC+ce5SzYBMHReGOg5q2UHoPOJRWRy+as3ZTvjbXcc662K5xMviffzPeN4VLQNhcfjCR86errI9EkuIiKSmHrGJdYW55w+KbsfgHoLXze/1XV6eML32gCo+vkvAMisDYc4vnBeeN5r14Sx4vyYs8aKRWQymcE4Vhw+Mo70iIfCpdXJLQ575TK1oWc80hnOJ44nfDBSH89DzuZfrx5xMahnLCIikph6xiWWP3r65Gz4dvnkULio6JcfPB+AU+85AMDISPiW2n12OwALfm0fACfUh8cHfPJNl4kT0ubPZx6MX2vz9/NjzQeGwhGTtVXDDGv8WaRiDYUTNhiJY7/9y8NngQ2EvWveG24b9h/9OXBkrLgmjiUP5y+UfPT64+EwR+aszo9Rx+nyqYlj1fkZv3Itxr6BWQSqMPr0FRERSUw94xLLX6e4ivB18we9pwLQcn84j3j0qUcAyJy8DoC9vxa+rZ7eFI6q3pMLAz3Dx4wVV8evpU3xJMJsVbhfa2E8KD+2nL/tGakFYHtPuPpTbjRD33BNARKKyFwyWnP03NP5o6Pz5x03bw/3Fz8UzuDIPBBm9RsdCN3W6lUrARhZFq4QN9wSPjtGauPsgMNxLuuuMChdvT92hfuP6fbWhs8XbwyfdUNLGqnuH0UC9YxFREQSU8+4xJZlwrfK7tHwLfLewycD0LA39GQzq1YAcPicJQCMnhDmnM73hLd1LgdgcDhsumwmnrdcFy5KurQ29LxH44DO0Gh43eFc+DbanQvzzL5wqBWAgY5wnxFjYCBbqJgiMkfke8Qe55r2wdAHq98TblteCF3l7J6w943W0AO2obDc68NnVlV36OnWdIYeNN3xQshxJi+rCZ8fo61h793I8rCewcXh9V2rQyF9K+JMXq2jDD6r/mCe/idEREQSU8+4xAY8P4Ybvh0eHAiHOB6ZDeek0CMeaIvXDN0Teq6/PLAWABvJz4oTH49HZW9fEsaKq+O339GR8D1ruD9s4qru2JOO88zmr2Xa2hFn38kY+4+5KouIlD+vj3vdmkJPt64u3PYvDmO4z58S9p5Vxb1uowvDZ5RVx6OtD4Xn1XTEvlsc5h1ujHNbN4YFtctCj/ll7bsA+K0lDwBwbt2LR9WzfzT0lLf0n8jffbFz9gErhHrGIiIiiU3ZMzaz1cBNwHLCd6Lr3f06M2sDvgWsA7YD73D3w8UrtXgGvI9H2cIgAxjGSk5gja1nyHM8wn3000eOAcysdbYZdw2Hb5Odo2EMd2Q0fB/qXZ7/XhTGXarjsEzjjrA8Pw1spj/8I3/lldFsvMLK3rC+/Ll++dfnz+kbPnyY7T/+3wz3dYMZS9a/iuWnvY7caD/P3nMTg72HGe7rLkjGVEq5HVNRRmU8XlYTe661oUe8amEHAN0NoYd62sIwh8Ga+nAFuBNqw/2TasJtXTwjozkeht0d5yyoix82yzJh/fkrz414uP+TF3q49k862LnPqDI4/bKTOO/dJ/KT55fz4CfuonPHQwx3Dpb1diyk6fSMh4EPuvsG4FXAH5nZRuBaYLO7rwc2x/tlyTDWcyavtkt4JW9gB8/Q411s53HaWMpr7FIyoZEs34xVGVb92lt52Vv/nI2X/jH7nvgp/R172PPwZlra13Pauz9CVU0tlHPG+bAdlVEZy0QmA3/+ly184a5T+ex3TmbLN59n3zPdPHPzL1j08lWc+MX3U1Vf3p85hTRlz9jddwO747+7zWwbsBJ4G3BBfNqNwD3Ah4pSZZHVWj21hJ5ltWVp8GYG6Wc/u3gFrwcgSw05+t/OLDM+PxyOYh6JRzvXZsK3zu6T4rywdWH85qV5ZOML4zhNPH2Y+sNxbDiTH0MOTxxqOnrcd7QmHlV94gJgAYcbHain6tFl7FvTxaEtv2TZB/9vOhbnsO82QG/XrDOmUsrtmIoyKuPx8r7wmdLn4fiT7SNh/vvBntAz3vV8mGsg2xI+dOrqw21DTegJN9eG3XD5MzYWZMMBJ5m4u+6F3vCZ9vS+xWG9h2KujvC+NT+Nn0FVN3DnV09j1z23cdLb3kf2B81kaWKE7rLdjoV0XGPGZrYOOAe4H1gWG+p8g7200MWl0O+9dNPBAtrIMUithV+sqvBfVREZhw4fIrdzJ7Vr1zDS1UNmQTgVwaozUCEZ58N2VEZlLBe5jkP0799Jw/K1DPV3k22MnzmZyvnMma1pH01tZk3ArcA17t5lNr0jb83sKuAqgDoaZlJjyQz7MA/zM07lbKotC9O8OMnYjAvb6yZ97jO58HuXv57xGQvDkYcHTgj/N91L6uNK49hwJs7jOhA2VU9nOLKxqysz9mkMNYeus9fFIyDjOFF9czg3cGFj+DZbM9zLg9d8m1f+6StZ+doX2ZsZ4Y0vf4SMObc29HNgGhnnw3ZUxvSUcWJjM2biecFTviaeUTHUEcZ8a/ri3NGH41WdBsNyrwpjvz3xo6w7Tva3K3/cStxrl8lPsBWbgoaR8ITmeDxL/qpPQz7A8//6NVa85m1UZ2rBYKTOGK6HyS46V07bsRCm1TM2syyhIb7Z3W+Li/eaWXt8vB3YN95r3f16d9/k7puy1Bai5qIY9VEe5mcsZw1LLUz/VkMtgx4asdGwn3jKjE1tc3dKydHhEX758X9l2UWnsfKCEwGob6un70A42mt0aHoZ58N2VMa0lHH6GTNNTaUq+bj56AjPff9rtJ30chaedCYA1Q3NDPWEKTM9XBCnrLdjoUznaGoDbgC2ufvnxzx0B3Al8Lfx9vaiVFgC7s5jbKWRZtbaKUeWL2EFu3medZzGEDkoQMb81ZP2e9hNs7ouHMH46yufAqBjKHwDrK0KY8n5qyz1Dodfxv2D4Q+vdyg0+tXx8da60KDmx3PyFsb7S6s7+eZHfsnpG6p5+x/XMOrPAXDgwlYO3/Uw5/y3Exju6i9IxlRKuR1TUUZlPF5HrrJUlX/vcDtSG/e6LQj3Mzk75nVxeewBZ3viTFuxx5u/vvFIbXjdcHOc+yC2m8O1o+y75VvYycvI/tb5dFSFF9a9fCN799zPgjN/neFcb0EyVoLp7KZ+DfBfgUfM7D/jso8QGuFbzOz3gReAy4tTYvF1cpA9vEATC7jPvw/AyZzBWk7lEe5jp29nhCEImcvS9gc6eOCO3bSf0sT/85v34RhvuuZk3vDeE/jGf3+YLbftZLB3GMo443zYjsqojOViYPtz9DzwC7Lt7ez83OcBp/Utb6Ll4jdw4CvfoOe+LYwODEIZZywkc5/mIEUBtFibn2sXluz9Cul+30yXH5pyoHz1GQv8A7e8qhQlFdx177iPF3/ZOWXG+bAdlXFuU8aX1K5Z7e1//oFSlFRwuz93HYMvvKjtiGbgEhERSa6kPWMz2w/0woQH7aa2mIlrW+vuS6ZagTLOCcqojIAyzhHKOJ2MpWyMAcxsq7tvKumbTlOhalPGtJSx9OspBmUs/XqKQRmnR7upRUREElNjLCIikliKxvj6BO85XYWqTRnTUsbSr6cYlLH06ykGZZyGko8Zi4iIyNG0m1pERCQxNcYiIiKJlawxNrNLzewJM3vazJJfTNrMVpvZ3Wa2zcweNbMPxOVtZvZ9M3sq3rYexzqVscSUURknWacylpgyziwjECYsL/YPkAGeAU4EaoCHgI2leO9JamoHXh7/3Qw8CWwEPgdcG5dfC3xWGZVRGZVRGZWxGBmPrLdExZ8H3DXm/oeBD6f8Dx2nxtuBi4EngPYx/+lPKKMyps6ljMqojJWZMf9Tqt3UK4EXx9zfEZfNCWa2DjgHuB9Y5u67AeLt0mmuRhkTU0ZlHEMZE1PGaWcESjdmPN4VK+bEOVVm1gTcClzj7l2zWdU4y5SxRJTx+FY1zjJlLBFlPL5VjbOs0jICpWuMdwCrx9xfBewq0XtPyMyyhP/Mm939trh4r5m1x8fbgX3TXJ0yJqKMyjgOZUxEGY87I1C6xngLsN7MTjCzGuAK4I4Svfe4zMyAG4Bt7v75MQ/dAVwZ/30lYTxgOpQxAWVUxgkoYwLKOKOMQQkHud9MOOrsGeAv5sCg+2sJuzseBv4z/rwZWARsBp6Kt23KqIzKqIzKqIzFyujumg5TREQkNc3AJSIikpgaYxERkcTUGIuIiCSmxlhERCQxNcYiIiKJqTEWERFJTI2xiIhIYmqMRUREElNjLCIikpgaYxERkcTUGIuIiCSmxlhERCQxNcYiIiKJqTEWERFJTI2xiIhIYmqMRUREElNjLCIikpgaYxERkcTUGIuIiCSmxlhERCQxNcYiIiKJqTEWERFJTI2xiIhIYmqMRUREElNjLCIikpgaYxERkcTUGIuIiCSmxlhERCQxNcYiIiKJqTEWERFJTI2xiIhIYmqMRUREElNjLCIikpgaYxERkcTUGIuIiCRWEY2xmW0wsx+aWaeZPW1mv5m6ptkws6vNbKuZDZrZ14557EIze9zM+szsbjNbm6jMWZkoo5nVmNl3zGy7mbmZXZCuytmZJOOrzOz7ZnbIzPab2bfNrD1hqTM2ScaNcfnh+PMDM9uYsNQZm+zvccxzPh5/Xy8qcXkFMcl2XBdz9Yz5+WjCUmdsis/VBjP7kpkdiO3Ij0tdX9k3xmZWDdwO/CvQBlwFfMPMTkla2OzsAj4NfGXsQjNbDNwGfJSQdSvwrZJXVxjjZox+Arwb2FPSigpvooytwPXAOmAt0A18taSVFc5EGXcBv034PV0M3AF8s7SlFcxkv6uY2UmErLtLWVSBTZoRWOjuTfHnUyWsq5Amy3g94Xd1Q7z9kxLWBUB1qd+wCE4DVgD/w90d+KGZ/RT4r4RGq+y4+20AZrYJWDXmod8CHnX3b8fHPwEcMLPT3P3xkhc6CxNldPcc8P/Gx0bSVFcYk2T83tjnmdkXgB+VtrrCmCRjB9ARHzNgBDg5RY2zNcnfY94XgA8BXyplXYU0jYxlb6KMZnYq8FZglbt3xcW/KHV9Zd8zBmyCZWeUupASOB14KH/H3XuBZ+JyKV/nA4+mLqIYzKwDGAD+EfibxOUUnJldDuTc/c7UtRTZ82a2w8y+GvfQVZJzgeeBT8bd1I+Y2WWlLqISGuPHgX3An5lZ1szeCLweaEhbVlE0AZ3HLOsEmhPUIgVgZmcCHwP+LHUtxeDuC4EFwNXAg4nLKSgzayJ8wbgmdS1FdAB4JWE45RWEz5qbk1ZUeKsInbdOwl7Wq4EbzWxDKYso+8bY3YeAtwO/QRhj/CBwC7AjZV1F0gO0HLOshTDmKGXGzE4Gvgd8wN3vTV1PscQ9OF8GbjKzpanrKaBPAl939+dSF1Is7t7j7lvdfdjd9xIaqjea2bGfQ+WsHxgCPu3uOXf/EXA38MZSFlH2jTGAuz/s7q9390XufglwIvDz1HUVwaPAWfk7ZtYInESF7uKsZPEo+B8An3L3r6eupwSqCHurVqYupIAuBN5vZnvMbA+wGrjFzD6UuK5i8ng73vBguXo4dQFQIY2xmZ1pZnXx8PQ/BdqBryUua8bMrNrM6oAMkInZqoF/Ac4ws8vi4x8DHi63g7dg0oyYWW18DKAmPlZ2f/wTZTSzlcAPgS+6+5fTVjk7k2S82MzOMbNM7EV9HjgMbEta8AxM8rt6IWH35tnxZxfwB8AXkxU7Q5Nsx3PN7FQzqzKzRcA/APe4+7HDZXPeJNvxx8ALwIfjc14DXADcVdIC3b3sf4C/I/yh9xB2+52cuqZZ5vkE4Rvo2J9PxMcuIoyT9wP3AOtS11uEjNvHeazsck6UEfh4/HfP2J/U9RY44+Xx97QH2A/cCZyZut5CZhzneduBi1LXW+Dt+E7gOaCXcOrWTcDy1PUWejsSDoL9Wcz5GPCbpa7PYiEiIiKSSEXsphYRESlnaoxFREQSm1VjbGaXmtkTFuaDvrZQRc0lylgZlLEyKGNlmA8Zj9eMx4zNLAM8CVxMOKd3C/BOd3+scOWlpYyVQRkrgzJWhvmQcSZm0xifRzgS7ZJ4/8MA7v6ZiV5TY7VeR+OM3i+FEYYZpJ8GmhmglyFyHwFlVMa5RxnHNx8yVi9o8NqlC6Zcd6ZqNDw/3mbj1O/1VUMA1FXlAKgh//jRO04HYlvRN1oDQH+8HRgNlzgYGs2EJ07RpIz05xg61EPdyjYG93Uy0tVfcdtxrAF6yfnglKdmzuZCESuBF8fc30GY4/MoZnYV4UpK1NHAuXbhLN6ytPb6Dg6yh422ift9M0PklBFlnIuU8SXzLWPN0hY2/MPvTbnu1rp+ABbV9QLQXhdOFd7YsAuA02rD7erqvvB4pv6o1z89NAjAloE1ADxj/OdmAAAgAElEQVTStxqAp7rDpGp7esOsvKM+ebtz+CeP07n1WdZd82a2vf+r9HX1V9x2HOt+3zyt582mMR7vf/xXvhO5+/WEy1PRYm2VcB6VMipjuVDGeZCxcX37pBnzjfCC2nBbnwk94UXZ0Cg3Z8LyZ3KhUX2gP/RAa2OPeWl1uJDRkky4XZk9DEB3XWise0dqAegYDPf7hrJTpBm33PmwHSc1mwO4dhCmf8tbRZiBpmLUUs8A/WMXKWMZUsbKoIyVIbu4maEDXWMXVVzGmZhNz3gLsN7MTgB2AlcAv1OQquaIFlrpp4d+78XDFzdlLEPKWBmU8fhkM2FMOD9WXGWhc7l3IOxOfrwj9IQPdIY9xMO7w4XuWp4KfbRsT3h+riXsBO1ZF9bTdtpBADYtDaOUI3G3dG1mGJi6Z9x4ygoGdh1mcE9HfmasituOMzHjxtjdh83sasL8nRngK+5eURcsqLIqTvWzeZB76acP4BZlLD/KWBmUsTJYpoo1f3gxT/3lN8nt64IKzDgTJZ0Os8XavJwH4bv80JRHxCnj3KaML1HGuW26GRvXt/t4B3Dle8Y18fZQb+j59vSEa7CMdoUebN3e0Cdb+GTo+dbvD2PF2Z5wa0Ph9QNLw+v3nx1eN/iycKDXktZwBdfa+D5TjhmPse39X6X3qd3ajmgGLhERkeRmM2YsIiJzVM9AOMq5a18TAJmucB5w9UAc442dtdoOj8vD7cDi0LPtWVlz1Pri6cRkwhlOVD0fjp7eMxjW29Iaesq12eHCBpkn1DMWERFJTD1jEZEKdGTyjXiTiT3iugPhNp5GzGgc4j1wVujhDiwPPVurC2PARw4rGgiPZ/pCHy7TH9Zjcey5tzb0xGsXqGc8E+oZi4iIJKaesYhIBcofTV23cACAAY9HUdeEj/36PaFnO1wfbgeWhR5t3aIw6UhTfRgc7ukPPd7BqtAD9sEwllw1HHvGo6HrnD+PWWZGPWMREZHE1DMWEalADTVhUHh5czgPmMXhZmdnuMJT1+J4FaSh0MNtXBKOhl7cFOasHhiOV2PKhVs7HHrEDTtjHy52hIcaw+ursyNFSDF/qGcsIiKSmHrGIiIVKDcSjn4+1N9w1PL8ecAnr90LQH310FHPy/eIO3vDecQjfeF+TW/oAR87Z/VobbxOcrV6xrOhnrGIiEhi6hmLiFSwI+cbR/mjnquOuYRwdby6U2d/OOp6oDscRV19ODQTdfvCekbixFy5BfH1C0PPulY941lRz1hERCQx9YxFROaRhXXhPOJ8T7hrMPSEj73uMfE22x1n7IoTa+XCwdgMtYWecHNLWJ/pPONZUc9YREQkMfWMRUTmgfx1jfM9346BcLR0V7y6U293nKFrIDQLVd3haOz8HNb5HvHA0tCDrmkNM3vV6SpNBaGesYiISGLqGYuIVLB8T7gxmwNeOro6vzw/U1dfJvSQbSD00Rp2h1sPHeQjR09XLQ094pbGgWKXPq+oZywiIpKYesYiIhWsPht6vvmrOB2MM23lhkOXt7MrzFF9ZKatztBHOzJWHCfwGl4YxoZbm3T0dDGoZywiIpKYesYiIhWspir0iAdH4lHSsUc7kAvXJx4dCn2yTFfoKdcdzF/nOLw+1xqPnl4Yrm9cnRktQdXzj3rGIiIiialnLCJSwTrjDFuDQ+HjPn809eBg6Bl7LvTJsl3hdjS2CsMNoQftreEo7LaW3knfJxN73PkZvlqy4Wjr+kwYfF6YDddLHhwN73so18Cz1TpHOU89YxERkcTUMxYRqWBdfaFnnIs94ZF8T7guHh29ohOAjqZw2HRbWxcAG1vD9Y5f2fIcAIsyPQD0jtaM+z41FsamG6rC2PLSTDcAC+L9Ojt6rPmR3FK2Vk/e255P1DMWERFJTD1jEZEKlsuFj/mRgXC0tPWE+8Pd4fZAX+gxL19xGICL2p8A4DdbHgDgZTXh8YyFvlvPaBgL7hgNPet8f7fOwlj0gIex413xcOxHcu3h+SONR9U15Bly/tzsA1YI9YxFREQSU89YRKSC+Wici7ozfNzXHQh9sMadoQdbPRAe7zh5GQA3b2wBYMdJrQCsqT8EwLJsGFvOxrHhzpEwxvxIz0oAnupYAsC+w80ADHWFseXswfi+8fzl/NHag4ucvZ1PFSpm2ZuyMTaz1cBNwHLCHonr3f06M2sDvgWsA7YD73D3w8UrtXgGvI9H2cIgAxjGSk5gja1nyHM8wn3000eOAcysVRnnLmVUxnIxHzIOH+rg4Fe+xejhHjBj0ennsegV5zPS38vO27/OYM8hRrq7yzpjIU2nZzwMfNDdHzCzZuAXZvZ94HeBze7+t2Z2LXAt8KHilVo8hrGeM2mxVoZ9iJ+zmTZfxm6208ZS1tlp/Nj/jRGGlXEOU0ZlLBelzFhTG8Z2hzxclan2UOgRtz0ajnb2rb8EoLkqjClnmsLY7u7Vocf74sKTARhsDT3dXHPoWR87NXVNZ+gxr+0O5xUPd+VYXX0RzSevYnhkkJ//xz9xwv4V7N73ICsyq1h+7u9z/w/+mhyDZbsdC2nKMWN33+3uD8R/dwPbgJXA24Ab49NuBN5erCKLrdbqabGwS6basjTQzCD97GcX7awFIEsNKOOcpozKWC7mRcaaZloaVwBQnamlsX4Jg7lu9nU+yYpFZ4XlNQ1QxhkL6bjGjM1sHXAOcD+wzN13Q2iwzWxpwatLoN976aaDBbSRY5BaC0cEVoXvLcpYJpRRGctFsTM2N4Sjnw8tDx/3vfG8487OJgBau0PP1wbDTFvEW9+5J9TxRDgXOE5VTWNzGBNmcfgyMdIaetKj9bE5ycS5rZtCT3q4McNA3yE6c3tZc9apDG7/Nl2vXsVQozG6pRq6K2M7zta0j6Y2sybgVuAad+86jtddZWZbzWzrEIMzqbFkhn2Yh/kZp3I21Zad9uuUcW5Rxokp49xSiIzDXX1FrHD2RoYH2faLb7DmFW8jk62b9uvKaTsWwrR6xmaWJTTEN7v7bXHxXjNrj73idmDfeK919+uB6wFarG3OXgBz1Ed5mJ+xnDUstTBWUkMtg95PrdUzGs6mU0aUMTVlVMaxGRvXt0+acUFd6Bk3LwsNWteC0CDuWRV6uIc3hKOgaw+HHm11X1hdtifc1nWEseCqoXB/tDrObb0gjDH3Lw59usG28H5DLeHM45GGUXxkhP1f+jp1l5xN7k0b2MMw9sMm9m08QHZJEyPfGZpWxrm8HQtlyp6xmRlwA7DN3T8/5qE7gCvjv68Ebi98eaXh7jzGVhppZq2dcmT5Elawm+cBGCIHyjinKaMylov5kvHgTd8m276Ulkted2R5/dkb6PlJmFBktKcPyjhjIZn75F84zOy1wL3AI7w02cpHCOPGtwBrgBeAy9390GTrarE2P9cunG3NBdfhB9jKPTSx4MiykzmDFtp4hPsYoJ8cA4wwvEgZlTElZVTGYzWub/cN//B7U75nQzYc5ZypOnqO6L6hsHu8bzCM8dZmw9HX+esej4yGPltnT+hRD/UdMzd1fF5VNqzXqsL9oaeeZcfHvkLd2qVQZZg5K698PY2nruDZz3yX3P4uhg73MtqfK9vtOB33+2a6/JBN9bwpG+NCmg//oco4tynjS5RxbptuxrnaGFdXh93b9bXhfbPx/ljb3v9Vep/are2IZuASEZkX8o3uRPKNcF7+useWv05xc7hOMflbKSjNTS0iIpJYSXdTm9l+oBc4ULI3PT6Lmbi2te6+ZKoVKOOcoIzKCCjjHKGM08lYysYYwMy2uvumkr7pNBWqNmVMSxlLv55iUMbSr6cYlHF6tJtaREQkMTXGIiIiiaVojK9P8J7TVajalDEtZSz9eopBGUu/nmJQxmko+ZixiIiIHE27qUVERBIrWWNsZpea2RNm9rSZXVuq952kntVmdreZbTOzR83sA3F5m5l938yeiretx7FOZSwxZVTGSdapjCWmjDPLCITJvIv9A2SAZ4ATgRrgIWBjKd57kpragZfHfzcDTwIbgc8B18bl1wKfVUZlVEZlVEZlLEbGI+stUfHnAXeNuf9h4MMp/0PHqfF24GLgCaB9zH/6E8qojKlzKaMyKmNlZsz/lGo39UrgxTH3d8Rlc4KZrQPOIVyJapm77waIt0unuRplTEwZlXEMZUxMGaedESjdmPF4V6yYE4dxm1kTcCtwjbt3zWZV4yxTxhJRxuNb1TjLlLFElPH4VjXOskrLCJSuMd4BrB5zfxWwq0TvPSEzyxL+M29299vi4r1m1h4fbwf2TXN1ypiIMirjOJQxEWU87oxA6RrjLcB6MzvBzGqAK4A7SvTe4zIzA24Atrn758c8dAdwZfz3lYTxgOlQxgSUURknoIwJKOOMMgYlHOR+M+Gos2eAv5gDg+6vJezueBj4z/jzZmARsBl4Kt62KaMyKqMyKqMyFiuju2sGLhERkdQ0A5eIiEhiaoxFREQSU2MsIiKSmBpjERGRxNQYi4iIJKbGWEREJDE1xiIiIompMRYREUlMjbGIiEhiaoxFREQSU2MsIiKSmBpjERGRxNQYi4iIJKbGWEREJDE1xiIiIompMRYREUlMjbGIiEhiaoxFREQSU2MsIiKSmBpjERGRxNQYi4iIJKbGWEREJDE1xiIiIompMRYREUlMjbGIiEhiaoxFREQSU2MsIiKSmBpjERGRxNQYi4iIJKbGWEREJDE1xiIiIompMRYREUlMjbGIiEhiaoxFREQSU2MsIiKSWNk1xmZ2tZltNbNBM/vamOXvMrOeMT99ZuZm9oqE5c7IRBnjY+8ws21m1m1mj5nZ2xOVOStTZHyvmT0dt+O/m9mKRGXOipnVmtkNZvZ83F4Pmtmbxjx+oZk9Hn9X7zaztSnrnYnJMppZjZl9x8y2x7/FCxKXOyNTZHyVmX3fzA6Z2X4z+7aZtaeu+XhNkXFj/Fs9HH9+YGYbU9d8vKb6exzzvI/H39eLSllf2TXGwC7g08BXxi5095vdvSn/A7wPeBZ4IEGNszVuRjNbCXwD+O9AC/BnwD+b2dKSVzh7E2V8PfA3wNuANuA54H+XvLrCqAZeBF4PLAA+CtxiZuvMbDFwW1zWBmwFvpWq0FmYMGN8/CfAu4E9KYorkMkytgLXA+uAtUA38NUURc7SZBl3Ab9N+D1dDNwBfDNJlbMz1e8qZnYSIevuklfn7mX5Q/gg/9okj98NfDx1nYXMCJwL7DvmOfuB81LXWsCMfw98ccz9FYADJ6WutUB5HwYuA64C/mPM8kagHzgtdY2FynjMsh3ABalrK2bGuPzlQHfq+oq4HauBPwL6UtdXjIzA94A3A9uBi0pZSzn2jKcUd/edD9yUupYC2wpsM7O3mlkm7qIeJPxCVQqLP2PvA5yRoJaCMrNlwCnAo8DpwEP5x9y9F3gmLi9bx2SsSFNkPH+C5WVlvIxm1gEMAP9I2HtV1o7NaGaXAzl3vzNFPdUp3rQE3gPc6+7PpS6kkNx9xMxuAv4ZqANywOXxg7xS3Al8y8y+DDwFfIzQM25IWtUsmVkWuBm40d0fN7Mmwl6NsTqB5pIXVyDHZkxdTzFMltHMziT8vr4tRW2FMlFGd19oZo3AlcDzqeorhAn+Hv8GeGOqmiqyZ0xojG9MXUShxQMKPgdcANQQxj7+l5mdnbKuQnL3zcDHgVsJf/DbCeNwOxKWNStmVgV8nfDl6eq4uIcw7j9WCyFr2ZkgY0WZLKOZnUzYxfkBd783QXkFMdV2jF/8vwzcVKbHqkyU8ZPA11N24CquMTaz1xDGGb+TupYiOBv4sbtvdfdRd98C3A+U9Ki/YnP3L7r7endfSmiUq4FfJi5rRszMgBuAZYSxqaH40KPAWWOe1wicRBnu4pwkY8WYLGMcFvsB8Cl3/3qiEmftOLZjFWFP1cpS1VYok2S8EHi/me0xsz3AasLBXR8qVW1l1xibWbWZ1QEZIGNmdWY2dnf7lcCt7l6WPQyYNOMW4HX5nrCZnQO8jjIcM54oY7w9w4I1hCNVr3P3w2krnrF/AjYAb3H3/jHL/wU4w8wui/8PHwMeLtPduxNlzJ9OUhfv1sTta7+yhrlv3IzxDIcfEg46/HKq4gpkoowXm9k58TiVFuDzwGFgW6I6Z2Oi39ULCcelnB1/dgF/AHyxZJWlPpptBke/fYIwhjj25xPxsTqgA7gwdZ1FzHg18DRhd+azwAdT11vIjMBCwpeLXsLpMJ8BMqnrnWHGtTHXAGG3dP7nXfHxi4DHCUdR3wOsS11zETJuH2c7l1XOyTIShlT8mOU9qWsucMbL4+9pD+E4hzuBM1PXXMiM4zx3OyU+mtriG4uIiEgiZbebWkREpNKoMRYREUlsVo2xmV1qZk9YmEf42kIVNZcoY2VQxsqgjJVhPmQ8XjMeMzazDPAkcDHhHNAtwDvd/bHClZeWMlYGZawMylgZ5kPGmZhNz/jXgKfd/Vl3zxEmDi/rmWfGoYyVQRkrgzJWhvmQ8bjNZjrMlYQrYOTtIFzIYEI1Vut1NM7iLUurjkZGGKLF2nyAXggzeymjMs45yjg+ZZx75kPGsQboJeeDU55bP5vGeLyV/8o+bzO7inCFGupo4Fy7cBZvWVp7fQcH2cNG28T9vpkhcqCMyjgHKeNLlHFumw8Zx7rfN0/rebPZTb2DMGVY3irCrCVHcffr3X2Tu2/KUjuLtyu9WuoZ4KgJhZRRGeckZXyJMs5t8yHjTMymMd4CrDezE8ysBriCcNHpitFCK/300O+9ePjipoxlSBkrgzJWhvmQcSZmvJva3YfN7GrgLsL8wl9x97Kb5H4yVVbFqX42D3Iv/fQB3KKM5UcZK4MyVob5kHEmSjodZou1eTnv9+/yQ1MOwivj3KaML1HGuU0ZXzIfMmoGLhERkcTUGIuIiCSmxlhERCQxNcYiIiKJqTEWERFJTI2xiIhIYmqMRUREElNjLCIikpgaYxERkcTUGIuIiCSmxlhERCQxNcYiIiKJqTEWERFJTI2xiIhIYmqMRUREElNjLCIikpgaYxERkcTUGIuIiCSmxlhERCQxNcYiIiKJqTEWERFJTI2xiIhIYmqMRUREElNjLCIikpgaYxERkcTUGIuIiCSmxlhERCQxNcYiIiKJqTEWERFJTI2xiIhIYmqMRUREEque6glmthq4CVgOjALXu/t1ZtYGfAtYB2wH3uHuh4tXavEMeB+PsoVBBjCMlZzAGlvPkOd4hPvop48cA5hZqzLOXcqojOVCGSsjYyFNp2c8DHzQ3TcArwL+yMw2AtcCm919PbA53i9LhrGeM3m1XcIreQM7eIYe72I7j9PGUl5jl5IhC8o4pymjMpYLZayMjIU0ZWPs7rvd/YH4725gG7ASeBtwY3zajcDbi1VksdVaPS3WCkC1ZWmgmUH62c8u2lkLQJYaUMY5TRmVsVwoY2VkLKQpd1OPZWbrgHOA+4Fl7r4bQoNtZksneM1VwFUAdTTMptaS6PdeuulgAW3kGKTW6gGoCt9blBFlnCuUURlBGSvFtA/gMrMm4FbgGnfvmu7r3P16d9/k7puy1M6kxpIZ9mEe5mecytlUW3bar1PGuUUZJ6aMc4syTqycMhbCtBpjM8sSGuKb3f22uHivmbXHx9uBfcUpsTRGfZSH+RnLWcNSWwlADbUMen94nFFQxjlPGZWxXChjZWQslCkbYzMz4AZgm7t/fsxDdwBXxn9fCdxe+PJKw915jK000sxaO+XI8iWsYDfPAzBEDpRxTlNGZSwXylgZGQvJ3H3yJ5i9FrgXeIRwahPARwjjxrcAa4AXgMvd/dBk62qxNj/XLpxtzQXX4QfYyj00seDIspM5gxbaeIT7GKCfHAOMMLxIGZUxJWVUxmMp49x2v2+myw/ZVM+bsjEupPnwH6qMc5syvkQZ5zZlfMl8yKgZuERERBIrac/YzPYDvcCBkr3p8VnMxLWtdfclU61AGecEZVRGQBnnCGWcTsZSNsYAZrbV3TeV9E2nqVC1KWNaylj69RSDMpZ+PcWgjNOj3dQiIiKJqTEWERFJLEVjfH2C95yuQtWmjGkpY+nXUwzKWPr1FIMyTkPJx4xFRETkaNpNLSIikpgaYxERkcRK1hib2aVm9oSZPW1myS8mbWarzexuM9tmZo+a2Qfi8jYz+76ZPRVvW49jncpYYsqojJOsUxlLTBlnlhEIk3kX+wfIAM8AJwI1wEPAxlK89yQ1tQMvj/9uBp4ENgKfA66Ny68FPquMyqiMyqiMyliMjEfWW6LizwPuGnP/w8CHU/6HjlPj7cDFwBNA+5j/9CeUURlT51JGZVTGysyY/ynVbuqVwItj7u+Iy+YEM1sHnEO4EtUyd98NEG+XTnM1ypiYMirjGMqYmDJOOyNQujHj8a5YMSfOqTKzJuBW4Bp375rNqsZZpowloozHt6pxliljiSjj8a1qnGWVlhEoXWO8A1g95v4qYFeJ3ntCZpYl/Gfe7O63xcV7zaw9Pt4O7Jvm6pQxEWVUxnEoYyLKeNwZgdI1xluA9WZ2gpnVAFcAd5TovcdlZgbcAGxz98+PeegO4Mr47ysJ4wHToYwJKKMyTkAZE1DGGWUMSjjI/WbCUWfPAH8xBwbdX0vY3fEw8J/x583AImAz8FS8bVNGZVRGZVRGZSxWRnfXdJgiIiKpaQYuERGRxNQYi4iIJKbGWEREJDE1xiIiIompMRYREUlMjbGIiEhiaoxFREQSU2MsIiKSmBpjERGRxNQYi4iIJKbGWEREJDE1xiIiIompMRYREUlMjbGIiEhiaoxFREQSU2MsIiKSmBpjERGRxNQYi4iIJKbGWEREJDE1xiIiIompMRYREUlMjbGIiEhiaoxFREQSU2MsIiKSmBpjERGRxNQYi4iIJKbGWEREJDE1xiIiIompMRYREUlMjbGIiEhiaoxFREQSU2MsIiKSmBpjERGRxNQYi4iIJKbGWEREJLGya4zNrNbMbjCz582s28weNLM3xcfWmZmbWc+Yn4+mrvl4TZYxPt5gZl8yswNm1mlmP05Z70xMsR3fdcw27Ivb9RWp6z4e09iO7zCzbfGxx8zs7SnrnYlpZHyvmT0dt+O/m9mKlPXOlJl9w8x2m1mXmT1pZu8d89iFZvZ4/D2928zWpqx1pibKaGY1ZvYdM9se/w4vSFzqjE2S8VVm9n0zO2Rm+83s22bWXtLi3L2sfoBG4BPAOsKXif8L6I731wEOVKeus1gZ4+PfAL4JLAEywCtS11zojMc893eBZwBLXXehMgIrgRzwJsCA3wD6gKWp6y5gxtcD+4DTgRrgn4Afpa55hjlPB2rjv08D9gCvABYDncDlQB3wd8B9qestcMYa4BrgtcBu4ILUtRYh45viNmwBGoCvAP9eytosFlXWzOxh4JPAL4DngKy7D6etqrDGZPwlsAVY5e5daasqrHxGd7/1mOV3A/e4+yfTVFY4Y7bjDuD/uPvSMY/tB97q7j9LVV8hjMl4HlDv7n8Ul68AdgInu/szCUucFTM7FbgH+ACwEPhdd391fKwROACc4+6PJytylsZmdPdbxizfAbzb3e9JVFrBTJQxPvZywhfH5lLVU3a7qY9lZsuAU4BHxyx+3sx2mNlXzWxxotIK5piM5wLPA5+Mu6kfMbPLkhZYABNsR+Iuv/OBm1LUVUjHZNwKbDOzt5pZJu6iHgQeTlnjbB2T0eLPkYfj7RmlrqsQ4tBQH/A4oYd4J6Gn9VD+Oe7eS9iLc3qSImdpgowVZZoZz+eYz6JiK+vG2MyywM3AjfFb6AHglcBawq6H5vh42Ron4yrCh1knsAK4GrjRzDakq3J2xsk41nuAe939udJXVjjHZnT3EcIXjH8mNML/DPxB/DAvS+NsxzuBd5jZmWZWD3yMMIzUkLDMGXP39xE+U14H3EbYbk2Ev8WxOuPzys4EGSvKVBnN7EzC7+qflbKusm2MzawK+Dph3O1qAHfvcfet7j7s7nvj8jeaWUvCUmdsvIxAPzAEfNrdc+7+I+Bu4I1pqpydCTKO9R7gxpIWVWDjZTSzi4DPARcQxuReD/wvMzs7UZmzMsHf42bg48CthL052wnjyTvSVDl77j7i7j8hfCn+Q6CHMM44VgshZ1kaJ2PFmSijmZ0MfI+w6/reUtZUlo2xmRlwA7AMuMzdhyZ4an5A3CZ4fM6aJGNZ78Yca6rtaGavIfT+v5OgvIKYJOPZwI/jl8dRd98C3A9clKjUGZtsO7r7F919fRwbvxWoJhz3UO6qgZMIuzLPyi+MY8b55eUun7GSHckYh8R+AHzK3b9e6kLKsjEmHJW5AXiLu/fnF5rZuWZ2qplVmdki4B8IB/4cuxupHIybEfgx8ALwYTOrjg3WBcBdpS9x1ibKmHclcKu7l20vg4kzbgFel+8Jm9k5hN1m5fhla6K/xzozO8OCNcD1wHXufjhVoTNhZkvN7Aoza4rj+5cA7wR+CPwLcIaZXWZmdYTdmw+X28FbU2TMn8JWF59eE7dtWXVyJstoZisJWb/o7l9OUmDqQ82P94cwHuzAAGEXUf7nXfE/9jmglzAwfxOwPHXNhczoLx2e/7OY8zHgN1PXXISMdUAHcGHqWouY8WrgacIuzWeBD6auuZAZCUcaPxx/T/cAnwEyqWueQcYlwI/i72MX8Ajw38Y8fhHhYKB+wtG561LXXISM2+N2HvtTVjkny0gYTvFjfod7SllfRZzaJCIiUs7KdTe1iIhIxVBjLCIiktisGmMzu9TMnrAw9+y1hSpqLlHGyqCMlUEZK8N8yHi8ZjxmbGYZ4EngYsJ5g1uAd7r7Y4UrLy1lrAzKWBmUsTLMh4wzMZvG+DzgE+5+Sbz/YQB3/8xEr6mxWq+jcUbvl8IIwwzSTwPNDNDLELmPgDIq49yjjONTxrlnJhnrW2t9Qfs0MsaTrRqqwqRajZYbu5iO0XoARj0sGRjNAtA3HG5HRsLO4iMnbVloH2w1ba0AAA1hSURBVOuqw6UOaquGj1rfdHTu7qX/8OCUL6k+jnUeayXw4pj7OwjzJh/FzK4CrgKoo4Fz7cJZvGVp7fUdHGQPG20T9/tmhsgpI8o4FynjS5RxbptJxub2Bt71z1PPh1MVG89zGp4H4JV1oYmqjU3h7T1h1uCekXDK9OO9ywF4aF+4smd3T2isM9UjYX1VYX2nLNkPwLqmgwBkbWTaeW/+nR9M63mzaYzHa+l/pZvt7tcTTvanxdoq4TwqZVTGcqGMylguJs24fOPkGfON8HlNTwPwloZwQbuMNQHw+y+8FoB7nl4PwOhAJjzeEZvA2JqNNIyG21E7qqruhWF93UOhEW+rKfwU8rM5gGsHsHrM/VXArtmVM7fUUs8AR00MpYxlSBkrgzJWhvmQcSZm0zPeAqw3sxMI1yi9AvidglQ1R7TQSj899HsvHr4iKWMZUsbKoIyVoZAZ8z3iU+r2APD6un0AHB4Ny3/9F+8FoPGWBQCsfyLMrOvZ0DM+tCGMFR/cFHrENW0DAOQOhh5wdW943uG+sPt6TVPxZgCdcWPs7sNmdjVhTuQM8BV3r4TJ0Y+osipO9bN5kHvppw/gFmUsP8pYGZSxMsyHjDMxm54x7n4nFXjx6bEWWzuLaed+30yXH/rr1PUUgzJWBmWsDMo4fXVV4QJh62rCAVZPxaOi/+jRdwHQ/rfhPvfdB0DVyScAsPMt7QB0vSwcbf2yk8JVPRfW9AHwkz0bw/PjlY4z8UCulXUdAAyOzqrpHJdm4BIREUms8M27iIhICdRaOO/3Z73hKOnbXzgTgOF/XwxA5pmnAMi94eUAPPmeMOb7lpdtAeDE+tCjXpgJPeK/f+xiABY8Gfqp/UvD+5zaFsail2XDUdUvDLYVPIt6xiIiIompZywiImVp1+BCAP5j9zoAOp9uBaA5HvT84pX/f3v3EhtXdcdx/Pu3PX6/87CdxIlDCKQplAQoiDa0oIJKWbRVpVZ9LNJV1UWlUnXRUPZVW1VU3VWRaEUlkIoACVCpIE2BogKBQIEUkpAEJRDsOI6N49d4ZjxzuvjfCXbq2GPwzJ0xv49kzcydmXPvz4p08vc59xyvmK/5lq+0+buepwC4POFjwENZr6x/P3QrADXP+KzrpkFf1GOy1+vV7c0DAHTWTACqjEVERFYkVcYiIlKRxmbqAJiargUg1+wVbXKXjwHfstnHjL+36gAAq6t99nUWv3/44fFrAHjyuesAuOygr6w1tc7vM860e3tfaPJ2TmZWFyuKKmMREZG4qTIWEZGK1J7wZTWv3xBtCLHRx4BX1/nYbmOV30c8lvNKt958Ba7/pHwHqAdPfB6A3qf9ezXD/r3Ja/39m6/5r79f47OoX0luLlYUVcYiIiJxU2UsIiIVKb+/cFedV641Vbk570/laue8Ph+tWf3QyA0AhP0+K7rh5cMApHZeBsDkzV4h/6R7PwBDWV+beipbt7wBZlFlLCIiEjNVxiIiUtHyuzflgs15vSXaxakvMQLAv5N9ADz97E4ArtjnK3BZWwsA733VK+l7djwBQIv57OsDKf/exZX2smYoWssiIiJSEFXGIiKyoiTM7w/eUX8KgKZoDes/vbcLgM2P+SxsRs4DMHTHFgC+fpvfj3xj/UkAjkX3FZ9K+WO+8i4GVcYiIiIxU2UsIiIrQn6s+KoG3584F7ze/OPwzQCMPrEOgHXHjgOQ3LHRj9/hK3btXvUCAP1ZH0N+N+3bNk0UcRb1hWsv+hlERERkQaqMRUSkouUr4s4aX1t6W+0ZAE5k1gDw8Bu+n/G2Z3xWdVjj9xf3fykBwI+v/icACfw+5SMpr6AH0u1Fv/Y8VcYiIiIxU2UsIiIVrTqqaK9q8DWqJ4NXvH/u/yIAa571+4Orhj8AYOSWPgA6dvp9xl9r9jWo87On8xVxMWdPX0yVsYiISMxUGYuISEVbW+trU3dXz91d6cirmwC48vkBALLrvfIdvMnHmH+1xdeeTgXf3/h4qttf50rfNaoyFhERiZkqYxERqWid1b7LUn729JNDVwPQ/YJXwAR/HPms3z/85Rt8jPjqun4AjqS7ABjMtAL/P1ac3x0qP2v7wnmj2dtt1X6f8ni0b/LZtLcznGkiS2HjzqqMRUREYqbKWEREKtrp9CoATk37/cPvDvvr1UmfZX32Vr9v+PztXsH+rHsfAMM536f4/Yx/b3zGK9uxjD+2J3wN6956vz+5PaqA82tfZ6MKeirnK3S9Md4LwGDSK/Az4y1MZgrb6UmVsYiISMxUGYuISEU7Muljvh9M+v3B6ZR3baOX+/3G4zd4hXtdr69Z/dq0r0n93OiVALwz6mPNybR/vqbaK+rr1/p9yx/ONAFwfMrPc3TM16zOZH0Wdv+IjxHPDHulbRmvmKuTRi5ZWDerylhERCRmqoxFRKSinZtuBmBkshGAzKiP4Wa8oMWqfBb06+9vAODQgI8hT4/URx+IGsrNnfn894E2f5L2ujUx5pVw02n/XO2Yt7tqxh+jTaIY3+hPkl25C8cWs2hnbGa9wF+AbiAH7A0h/MHMOoG/An3ASeA7IYQPCztteZkOU7zFK6SYxjDWs5mNtpVMSHOIl0gyRZppzKxDGcuXMipjpVDGlZFxORVSGc8APw8hvGZmLcCrZrYP+CGwP4TwazPbA+wBflG8Sy0ew9jK52i1DmZChpfZT2foYoCTdLKWPtvGv8LfyDKjjGVMGZWxUijj8mYcmvTK+Px5r4xrogq2OuXvN77qx7PRxOZsVBA3JaMGotuHE5PRtUeVbu24l7VVPnma6rSPJVvW35/o8fNkG/xz+Uo8tc0b3tozxEhDuqAMixbQIYSBEMJr0fNx4DCwHvgGcH/0sfuBbxZ0xjJUZw20WgcANZagkRZSJBminx58ObUEtaCMZU0ZlbFSKOPKyLicljRmbGZ9wE7gANAVQhgA77DNbO2yX10MkmGScUZpo5M0KerMZ8dV+f9blLFCKKMyVgpl/OQZGxIZ4KOx4WyDV7Dpdq8385XwTHN2zvcy0VhwzYSPAYcqf8wl/HFiU/573l5+TDlE7Xdt8F2fNrSMAtCWmAZgc+M5P147wrF8ub2IgmdTm1kz8AhwVwhhbAnf+5GZHTSzgxlShX4tFjNhhjd5kSvZQY0lCv6eMpYXZbw0ZSwvynhpszNOjZZ3xuVQUGVsZgm8I34ghPBodHjQzHqiqrgHODvfd0MIe4G9AK3WGeb7TDnIhRxv8iLdbGStrQegljpSIUmdNZDz/TKVEWWMmzIq46ctY/f2hTN+puMMAO31PlZ7psNXwEplvIvraPTjzbXeqQ9O+PvnJ7xkzua8Lq1v9Pe3dHpl21Hr3+uq8/ozvztUfi3s1mqvhMei0ju/Eld9lVfqk7k6cgXWvIt+yswMuA84HEK4d9ZbjwO7o+e7gccKOmMZCiHwNgdpooVNdsWF42tYxwCnAMiQBmUsa8qojJVCGVdGxuVkISz8nyoz2wU8DxzCb20C+CU+bvwQsBF4D/h2CGFkobZarTPcaF/5pNe87EbDOQ7yLM20XTh2OVfRSieHeIlpkqSZJsvMKmVUxjgpozJe7NOQsXt7Z/jBg7ctes5kNF26LqpMsxfVm201Xulmcj4LOlE1dwy5I9qFqbFq7gzoevPXI1mftZ3Kzf1z+3hUGc+3D/ID3/8HZ94eWXTrpkU74+VUrv9oCnEg7GcsLP4LVcbypowfUcbypowf+TR0xlqBS0REVoSG6rmdaBVzO9up7NwdlPJrS1/q/VLS2tQiIiIxK+mfqc1sCJgEzpXspEuzmktf26YQwprFGlDGsqCMyggoY5lQxkIylrIzBjCzgyGE60t60gIt17UpY7yUsfTtFIMylr6dYlDGwujP1CIiIjFTZywiIhKzODrjvTGcs1DLdW3KGC9lLH07xaCMpW+nGJSxACUfMxYREZG59GdqERGRmJWsMzazO8zsqJkdN7M9pTrvAtfTa2bPmNlhM3vLzH4aHe80s31mdix67FhCm8pYYsqojAu0qYwlpowfLyPgi3kX+weoBk4AlwG1wBvA9lKce4Fr6gGujZ63AO8A24HfAnui43uA3yijMiqjMiqjMhYj44V2S3TxNwFPzXp9N3B3nL/Qea7xMeB24CjQM+uXflQZlTHuXMqojMq4MjPmf0r1Z+r1wPuzXp+OjpUFM+sDduI7UXWFEAYAose1BTajjDFTRmWcRRljpowFZwRKN2Y8344VZTGN28yagUeAu0IIY5+kqXmOKWOJKOPSmprnmDKWiDIural5jq20jEDpOuPTQO+s1xuA/hKd+5LMLIH/Mh8IITwaHR40s57o/R7gbIHNKWNMlFEZ56GMMVHGJWcEStcZvwJsNbPNZlYLfBd4vETnnpeZGXAfcDiEcO+stx4HdkfPd+PjAYVQxhgoozJegjLGQBk/VkZXwkHuO/FZZyeAe8pg0H0X/ueON4HXo587gVXAfuBY9NipjMqojMqojMpYrIwhBK3AJSIiEjetwCUiIhIzdcYiIiIxU2csIiISM3XGIiIiMVNnLCIiEjN1xiIiIjFTZywiIhIzdcYiIiIx+x+vY9xrkv2HsgAAAABJRU5ErkJggg==\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "activation_model = models.Model(inputs=conv_model.input, outputs=layer_outputs[2]) # Creates a model that will return these outputs, given the model input\n", "\n", "\n", "activations = activation_model.predict(img_tensor/255.) # Returns a Numpy array corresponding to the output of layer 2\n", "print(activations.shape)\n", "\n", "# Show the result of the 1st Relu activation\n", "\n", "plt.figure(figsize=(8,8))\n", "for k in range(activations.shape[3]):\n", " plt.subplot(4,8,k+1)\n", " plt.imshow(activations[0,:,:,k])\n", " plt.title(str(k+1))" ] }, { "cell_type": "code", "execution_count": 16, "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 470 }, "colab_type": "code", "id": "eVO9QoV4UD83", "outputId": "e1e82841-d440-4c00-f0f3-c6fbff1778b0" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "(1, 12, 12, 32)\n" ] }, { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeMAAAGzCAYAAAAc+X/PAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi40LCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcv7US4rQAAIABJREFUeJzt3XmcXHWZ7/HP03s6e2ejk0AaQgAB\nNUJGBGVxgoArziAMjI6ZuXgZdRidRcegM4LbuMxcR2dcmNwLsqgo7igoA1E2lZjIFhISIJBAQlay\ndzqd7s5z/zgnoX5ld3V3dVX9avm+X696df1Obc+3q7qfOud36pS5OyIiIhJPXewCREREap2asYiI\nSGRqxiIiIpGpGYuIiESmZiwiIhKZmrGIiEhkasYiIiKRVXwzNrMrzWyZmXWb2Q2x6ykGM2s2s+vM\nbJ2Z7TGzR8zsjbHrKjQz+6aZbTSz3Wb2pJm9J3ZNxWBmc8xsv5l9M3YtxWBm96T59qan1bFrKgYz\nu9TMnjCzTjNbY2Znxq6pUDKeu0OnPjP7r9h1FZKZdZjZHWa2w8w2mdlXzKwhVj0V34yBF4BPA9fH\nLqSIGoDngbOB8cA/A7eaWUfEmorhs0CHu48D3gZ82sxOjVxTMXwVWBq7iCK70t3HpKfjYxdTaGb2\nBuDzwF8BY4GzgGeiFlVAGc/dGOAIoAv4XuSyCu1rwBagHZhL8v/1/bGKqfhm7O4/dPcfAy/GrqVY\n3L3T3a9x97XuftDdfwY8C1RVo3L3Fe7efWiYnmZHLKngzOxSYCewOHYtMiKfAD7p7g+mf5Mb3H1D\n7KKK5CKSpnV/7EIK7GjgVnff7+6bgF8AJ8UqpuKbcS0ys2nAccCK2LUUmpl9zcz2AauAjcAdkUsq\nGDMbB3wS+IfYtZTAZ81sm5n92szOiV1MIZlZPTAPmGJmT5vZ+nQT56jYtRXJAuAmr75jJ38JuNTM\nWs1sBvBGkoYchZpxhTGzRuBbwI3uvip2PYXm7u8n2ex3JvBDoDv3LSrKp4Dr3H197EKK7CPAMcAM\nYBHwUzOrpi0c04BG4B0kr9O5wKtIpo+qipnNItl8e2PsWorgPpI14d3AemAZ8ONYxagZVxAzqwNu\nBg4AV0Yup2jcvc/dHwBmAu+LXU8hmNlc4FzgP2LXUmzuvsTd97h7t7vfCPwaeFPsugqoK/35X+6+\n0d23AV+kujIe8hfAA+7+bOxCCin9X/oLkjf8o4HJwESS/QCiUDOuEGZmwHUk78ovcveeyCWVQgPV\nM2d8DtABPGdmm4APAReZ2UMxiyoRByx2EYXi7jtI1qQyN9tW2ybcQ95Nda4VtwFHAV9J3zS+CHyD\niG+oKr4Zm1mDmbUA9UC9mbXE3D29iL4OvAx4q7t3DXblSmNmU9OPiowxs3ozOx+4jOrZ0WkRyRuL\nuenpWuB24PyYRRWamU0ws/MP/R2a2TtJ9jSONhdXJN8A/jZ93U4E/h74WeSaCsrMziCZaqi2vahJ\nt2Y8C7wvfZ1OIJkbfyxWTRXfjEnmabqAhcC70vNVNXeTztv8Nck/8U0Zn/17Z+TSCslJNkmvB3YA\n/w78nbvfFrWqAnH3fe6+6dAJ2Avsd/etsWsrsEaSjxpuBbYBfwu83d2fjFpV4X2K5ONpTwJPAA8D\nn4laUeEtAH7o7ntiF1IkfwpcQPJafRroIXlTFYVV3w5yIiIilaUa1oxFREQqmpqxiIhIZGrGIiIi\nkY2oGZvZBWa2Oj0KzcJCFVVOlLE6KGN1UMbqUAsZhyvvHbjSQ8I9CbyBZA/YpcBl7r6ycOXFpYzV\nQRmrgzJWh1rImI+RNOPTgWvc/fx0fBWAu392oNs0WbO3MDqvx4uhj1666aKVseynkx4OfBSUsRYy\n1o8b7Y1TJgx4n5NaOoPx5Ppw3OmN4fhgczDe2xuOD+wLr29NBwd87P4c3H+A3h17aGqfRM/WnRzc\ns0/PI8pYjmohY6b9dHLAuwc96M1IDo4xg+Rr/Q5ZD5yW6wYtjOY0mz+Chyytzb6eF9nEiTaPJb6Y\nHg4oI7WRsXHKBI783HsHvHzBy5YE47+a8PtgvLR7ajB+cO+xwXjJix3BeN0j04Nx3YzwuC5Wl/tN\n894HV7DvkaeY+t638/zCa+nes0/PI8pYjmohY6YlPrTjFhX9SFVmdgVwBUALrcV+uCiUsTpkZmyY\nPD5yNcVRa8+jMlauWsiYaSTNeANwZMZ4Zros4O6LSA4FyDhrq6gjjDQziv0EayjKSG1kbJk9I8g4\nbvT+4LofaHs0GI+pGxOM//E7C4LxhCfCx9p3RLjVqn1VXzBe/5Zws3XTmAPZ5Qbq28bS8+KuzEV6\nHlHGclSOGZ+95ZXB+OjLHh3gmsUzkr2plwJzzOxoM2sCLgWq4tCFh4xjIl3spcs78eQ48MpYgWoh\nY8vsGfRs3E7Plh2k+4FUXcZaeB6VsXblvWbs7r1mdiVwJ8mXNFzv7lX1Zfd1VsfxPpeHuZ8u9gHc\nqoyVpxYyWn09U/7Xm3nhMzfRs3UnVGHGWngelbF2jehzxu5+h7sf5+6z3b3aDpIOwGRr5wy7gDGM\nRxkrVy1kHH3Kccz68gdpPmpa1WashedRGWtTNX7VoEjB9faF71uv3nxGMP7Zz8OdQacsD6e4Lr/m\nx8F4zf5wb+vfnxrOEXPeqfmUKSLAtp8eF4wnvzX3l4b95nVfC8bv5LUFr2kwOhymiIhIZGrGIiIi\nkakZi4iIRKY5Y5Eh2Le/KRj/aMXcYGwd4eeQ3/H2e4Pxu8Y9H4xf8dh5wbhjXjjHbKN78ylTRIC2\nLwzv0JnvPLL0c8TZtGYsIiISmZqxiIhIZGrGIiIikWnOWCQPjc3hnO5ZHWuC8T+0PROMH+kOvxKx\n488eC8bPfP70rPsPv5JRRIau7v6HY5cwbFozFhERiUzNWEREJDI1YxERkcg0ZyySh9aW8PuF/2PG\n4mDc5+Hnkt/zmb8Lxr0fCL/P+OD04Ptd9S5ZpMbob15ERCQyNWMREZHI1IxFREQi05yxSB7eM+fX\nwbjZwu8jfu/6M4Px1B+H36f6xKdmB+Ompr4CVicilUZrxiIiIpGpGYuIiESmZiwiIhKZ5oxF8vCu\nceEccB3NwfiFP5sUjNddMTMYN07cG96+Lvw+48bG8NjXx7ZtC8ZN9eEc8+Ob2w+fNwvvS6TWPPm1\nVwfj497/u0iVDJ3WjEVERCJTMxYREYlMzVhERCQyzRmL5GF7Xzhn+8lt84Lx+gvDOeKZ858Lxs0N\n4ZzwlOZwDnly1rjRwsc76OGxrdc0TT58vk5zxlLjxk7fE7uEYdOasYiISGRqxiIiIpGpGYuIiESm\nOWORPHxw7TuC8fInjgrG48IpXTp7wu833tXdEoxXb5oajPv6wvfJvim8/uj14eVtT/QcPl+3pX6A\nqkVqw57NY4Lx1NefEozrf/VQQR/vqa+eFoxP+PJLxwWwtQ8M6T60ZiwiIhKZmrGIiEhkg26mNrPr\ngbcAW9z95HRZG/BdoANYC1zi7juKV2ZxrfBlbGMjTTRzup0HQI8fYDkP0sU+RtGKczBylSOjjNWR\ncdst32XfypU0No/l2L/8JwB6uzpZvuRG9nftoGXURNwrO2MtPI/KWB0ZC2koc8Y3AF8BbspYthBY\n7O6fM7OF6fgjhS+vNKYziyOZzQqWHl62llW0MZUOO4G1vorneCpihSOnjIXNuHJ9ezBuejGcp23a\nHX7W16+bEownPNsZjNv27Q/Ge06YGIx3HJdsxJo059VMPulMNtz5LfbMST6rvOMHd1P3+tlMfcs5\n7PrZPXT9T/hdy5VGr1VlHKln3/p/g/GxY/8yGM/7dFswfmTDjGBsK8YG4zPe+Fgw/t3GcB+RyT8O\nNzL3Pbnm8Hn37sELZgibqd39PmB71uILgRvT8zcCbx/So5WpiTaFRsIdbLbyAu3MAqCdWfTQ099N\nK4YyVkfG1qNmU9/SGizremwFY16X7KAy5nWncDCrsVeaWngelbE6MhZSvnPG09x9Y3p+EzCtQPWU\njQN002yjAGiiBaf6jmqkjNWhb/de6ieMA6Bu/FjIOjpYNaiF51EZa9uIP9rk7m45vrPNzK4ArgBo\noXWgq5U1MyPXa0YZK8NwMjZMHl+iqgrLzMByXl5Tz6Mylq9ayDgc+TbjzWbW7u4bzawd2DLQFd19\nEbAIYJy1VczboCaa6fYumm0U3d6F5fgPp4zlK9+MLbNn5MxY35C19nlsOAe8+9jcdW37g41SLVnj\nrgFv27ulE/uZ09SWbI5umDCauu7tNEwcS++OPVj9wBu8au15VMbyUqqM50+fG4xn80gwzt7beNYf\nzMSG1n8iHE9n5XDKGZJ8N1PfBixIzy8AflKYcsrHFKazkXUAbGQdDTRGrqjwlLE6jJ53AnvufRiA\nPfc+TF1rdmOvfLXwPCpjbRvKR5tuAc4BJpvZeuBq4HPArWZ2ObAOuKSYRRbbcl/CDrbSQzf3++0c\nw4nM4niW8yAbfC2jaKX5D9ZcKosyVkfGTV/6Hl0rn6Vvzz6efe+/M+mS1zPx7Wey6T++y+5fPkTD\nlAnUTxg7+B2VsVp4HpWxOjIWkrmXbgvHOGvz02x+yR6vkJb4Ynb79hyzcQllLG9Dzdgye4Yf+bn3\nlqKkgnt+4bXsX7NBzyPKWO6U8SUlbcZmtpVkTXoysG2Qq8cyUG2z3H1KP8sDylg2lHEQylg2lHEQ\nNZGxlM348IOaLXP3eYNfs/QKVZsyxqWMpb+fYlDG0t9PMSjj4HRsahERkcjUjEVERCKL1YwXRXrc\noShUbcoYlzKW/n6KQRlLfz/FoIyDiDJnLCIiIi/RZmoREZHI1IxFREQiK2kzNrMLzGy1mT2dfg9y\nVGZ2vZltMbPHM5a1mdldZvZU+nNirvvo5z6VscSUURlz3Kcylpgy5pcRdy/JCagH1gDHAE3Ao8CJ\npXr8AWo6CzgFeDxj2ReAhen5hcDnlVEZlVEZlVEZi5XR3UvajE8H7swYXwVcFfMXmtbRkfULXQ20\np+fbgdXKqIzKqIzKqIzFyujuJd1MPQN4PmO8Pl1Wbqa5+8b0/CZg2jBuq4zlQxlzU8byoYy51UJG\n7cCViydvcar6s1/KWB2UsTooY3XIJ2Mpm/EG4MiM8cx0WbnZbGbtAOnPLcO4rTKWD2XMTRnLhzLm\nVgsZS9qMlwJzzOxoM2sCLgVuK+HjD9VtwIL0/ALgJ8O4rTKWD2XMTRnLhzLmVgsZS7cDVzqp/Sbg\nSZI94z5WBhPwtwAbgR6SeYjLgUnAYuAp4G6gTRmVURmVURmVsZgZdThMERGRyLQDl4iISGRqxiIi\nIpGpGYuIiESmZiwiIhKZmrGIiEhkasYiIiKRqRmLiIhEpmYsIiISmZqxiIhIZGrGIiIikakZi4iI\nRKZmLCIiEpmasYiISGRqxiIiIpGpGYuIiESmZiwiIhKZmrGIiEhkasYiIiKRqRmLiIhEpmYsIiIS\nmZqxiIhIZGrGIiIikakZi4iIRKZmLCIiEpmasYiISGRqxiIiIpGpGYuIiESmZiwiIhKZmrGIiEhk\nasYiIiKRqRmLiIhEpmYsIiISmZqxiIhIZGrGIiIikakZi4iIRFYVzdjMXmZmvzSzXWb2tJn9Seya\nRsLMrjSzZWbWbWY3ZF0238xWmdk+M/uVmc2KVOaIDJTRzJrM7PtmttbM3MzOiVflyOTI+Bozu8vM\ntpvZVjP7npm1Ryw1bzkynpgu35Ge7jazEyOWmrdcf48Z1/l4+no9t8TlFUSO57EjzbU34/QvEUvN\n2yD/V1vN7Gtmti3tI/eVur6Kb8Zm1gD8BPgZ0AZcAXzTzI6LWtjIvAB8Grg+c6GZTQZ+CPwLSdZl\nwHdLXl1h9Jsx9QDwLmBTSSsqvIEyTgQWAR3ALGAP8I2SVlY4A2V8AXgHyet0MnAb8J3SllYwuV6r\nmNls4GJgYymLKrCcGYEJ7j4mPX2qhHUVUq6Mi0heqy9Lf/59CesCoKHUD1gEJwDTgf9wdwd+aWa/\nBv6CpGlVHHf/IYCZzQNmZlz0p8AKd/9eevk1wDYzO8HdV5W80BEYKKO7HwC+lF7WF6e6wsiR8eeZ\n1zOzrwD3lra6wsiRcSewM73MgD7g2Bg1jlSOv8dDvgp8BPhaKesqpCFkrHgDZTSzE4C3ATPdfXe6\n+Pelrq/i14wHYMDJsYsogpOARw8N3L0TWJMul8p1FrAidhHFYGY7gf3AfwH/GrmcgjOzi4Fud78j\ndi1Fts7M1pvZN9ItdNXk1cA64BPpZurlZnZRqYuohma8GtgCfNjMGs3sPOBsoDVuWUUxBtiVtWwX\nMDZCLVIAZvYK4OPAh2PXUgzuPgEYD1wJPBy5nIIys7EkbzA+GLuWItoG/BHJdMqpJP9rvhW1osKb\nSbLytotkK+uVwI1m9rJSFlHxzdjde4C3A28mmWP8R+BWYH3MuopkLzAua9k4kjlHqTBmdizwc+CD\n7n5/7HqKJd2Ccy1wk5lNjV1PAV0D3OzuayPXUTTuvtfdl7l7r7tvJmlU56VvRKpFF9ADfNrdD7j7\nvcCvgPNKWUTFN2MAd3/M3c9290nufj5wDPC72HUVwQrglYcGZjYamE2VbuKsZule8HcDn3L3m2PX\nUwJ1JFurZsQupIDmAx8ws01mtgk4ErjVzD4Sua5i8vRnVfSO1GP9LPN+lhVVVfxCzewVZtaS7p7+\nIaAduCFyWXkzswYzawHqgfo0WwPwI+BkM7sovfzjwGOVtvMW5MyImTWnlwE0pZdZtGLzNFBGM5sB\n/BL4irtfG7fKkcmR8Q1m9iozqzezccAXgR3AE1ELzkOO1+p8ks2bc9PTC8Bfk+zQVVFyPI+nmdnx\nZlZnZpOA/wTucffs6bKyl+N5vA94Drgqvc5rgdcDd5a0QHev+BPwbyR/6HtJNvsdG7umEea5huSd\nWebpmvSyc4FVJJtW7gE6YtdbhIxr+7ms4nIOlBG4Oj2/N/MUu94CZ7w4fZ3uBbYCtwOviF1vITP2\nc721wLmx6y3w83gZ8CzQSfLRrZuAI2LXW+jnkWQn2N+mOVcCf1Lq+iwtRERERCKpis3UIiIilUzN\nWEREJDI1YxERkchG1IzN7AIzW23JlzMsLFRR5UQZq4MyVgdlrA61kHG48t6By8zqgSeBN5AcYGMp\ncJm7rxzoNk3W7C2MzuvxYnCcTnbTyhi66aKXnsdQRmUsQ8rYP2UsP7WQMdN+Ojng3YN+NHMkXxTx\nauBpd38GwMy+A1xIslt4v1oYzWk2fwQPWVo7/UWeYSWn2Jks8cXsYYcyoozlSBn7p4zlpxYyZlri\ni4d0vZFspp4BPJ8xXk8/R9cxsyss+Q7JZT10j+DhSq+bLloYlblIGZWxLCnjS5SxvNVCxnwUfQcu\nd1/k7vPcfV4jzcV+uCiUsTooY3VQxupQCxkzjWQz9QaSY7EeMjNdVjWaGcV+ujIXKWMFUsbisIbw\n38e6j746GB/1yd8U9PH0PFaHWsiYj5GsGS8F5pjZ0WbWBFwK3FaYssrDOCbSxV66vBNPjhuujBVI\nGauDMlaHWsiYj7zXjN2918yuJDmYdj1wvbtX1bcH1Vkdx/tcHuZ+utgHcKsyVh5lrA7KWB1qIWM+\nRjRn7O53uPtx7j7b3T9TqKLKyWRr5wy7gDGMRxkrlzJWB2WsDrWQcbhGMmcsIjWs7uijgnGh54hF\naokOhykiIhKZmrGIiEhkasYiIiKRac5YRPLS99QzsUsQqRpaMxYREYlMzVhERCQyNWMREZHI1IxF\nREQiUzMWERGJTM1YREQkMjVjERGRyNSMRUREIlMzFhERiUzNWEREJDI1YxERkcjUjEVERCJTMxYR\nEYlMzVhERCQyNWMREZHI1IxFREQiUzMWERGJTM1YREQkMjVjERGRyNSMRUREIlMzFhERiUzNWERE\nJDI1YxERkcgaYhcgIrWpd/6pwfg/r/tKMD6paVQwPn/63KLXJBKL1oxFREQiG3TN2MyuB94CbHH3\nk9NlbcB3gQ5gLXCJu+8oXpnFtcKXsY2NNNHM6XYeAD1+gOU8SBf7GEUrzsHIVY6MMipjpfj4h3Zw\n7y/3M31yA4/dcxQA23f08ZDfVzUZa+F5rIWMhTSUNeMbgAuyli0EFrv7HGBxOq5Y05nFq3hdsGwt\nq2hjKq+1C2hjKgfYH6m6wlBGZawUb7u4la/fOClY9vmv7KiqjLXwPNZCxkIadM3Y3e8zs46sxRcC\n56TnbwTuAT5SwLpKaqJNocs7g2VbeYFTORuAdmaxhpUxSisYZVTGYqube2Iw3jpvfDDua7Zg3P79\np4PxP3Scfvh8l3eyll8fnif+jf+CU5mV3E7PY0WohYyFlO+c8TR335ie3wRMK1A9ZeMA3TRbsgNJ\nEy04HrmiwlPG6qCM1UEZa9uI96Z2dzezAX+jZnYFcAVAC60jfbgozIxcrxllrAzKqIyVQhmrI+Nw\n5LtmvNnM2gHSn1sGuqK7L3L3ee4+r5HmPB+u9Jpoptu7AOj2Lgwb8LrKWL6UMaSM5UsZQ5WaMV/5\nrhnfBiwAPpf+/EnBKioTU5jORtbRwQlsZB0NNMYuqeCUsTqUS8aDj4Tzf5MeyX39vmHcd7lkLCZl\nrG1D+WjTLSQ7a002s/XA1SRN+FYzuxxYB1xSzCKLbbkvYQdb6aGb+/12juFEZnE8y3mQDb6WUbTS\nTEvsMkdEGZWxUiijMtYicy/dBPo4a/PTbH7JHq+Qlvhidvv2gbeppJSxvCnjS5SxvCnjS2oho47A\nJSIiEllJ14zNbCvJZu3JwLaSPfDwDFTbLHefMtiNlbFsKOMglLFsKOMgaiJjKZvx4Qc1W+bu80r+\nwENQqNqUMS5lLP39FIMylv5+ikEZB6fN1CIiIpGpGYuIiEQWqxkvivS4Q1Go2pQxLmUs/f0UgzKW\n/n6KQRkHEWXOWERERF6izdQiIiKRlbQZm9kFZrbazJ42s+jfgWxm15vZFjN7PGNZm5ndZWZPpT8n\nDvM+lbHElFEZc9ynMpaYMuaXEXcvyQmoB9YAxwBNwKPAiaV6/AFqOgs4BXg8Y9kXgIXp+YXA55VR\nGZVRGZVRGYuV0d1L2oxPB+7MGF8FXBXzF5rW0ZH1C10NtKfn24HVyqiMyqiMyqiMxcro7iXdTD0D\neD5jvD5dVm6mufvG9PwmYNowbquM5UMZc1PG8qGMudVCRu3AlYsnb3GqendzZawOylgdlLE65JOx\nlM14A3BkxnhmuqzcbDazdoD055Zh3FYZy4cy5qaM5UMZc6uFjCVtxkuBOWZ2tJk1AZcCt5Xw8Yfq\nNmBBen4B8JNh3FYZy4cy5qaM5UMZc6uFjKXbgSud1H4T8CTJnnEfK4MJ+FuAjUAPyTzE5cAkYDHw\nFHA30KaMyqiMyqiMyljMjDoCl4iISGTagUtERCQyNWMREZHI1IxFREQiUzMWERGJTM1YREQkMjVj\nERGRyNSMRUREIlMzFhERiUzNWEREJDI1YxERkcjUjEVERCJTMxYREYlMzVhERCQyNWMREZHI1IxF\nREQiUzMWERGJTM1YREQkMjVjERGRyNSMRUREIlMzFhERiUzNWEREJDI1YxERkcjUjEVERCJTMxYR\nEYlMzVhERCQyNWMREZHI1IxFREQiUzMWERGJTM1YREQkMjVjERGRyNSMRUREIlMzFhERiUzNWERE\nJDI1YxERkcjUjEVERCKruGZsZlea2TIz6zazGzKWv9PM9mac9pmZm9mpEcvNy0AZ08suMbMnzGyP\nma00s7dHKnNEBsn4HjN7On0ef2Fm0yOVOSJm1mxm15nZuvT5esTM3phx+XwzW5W+Vn9lZrNi1puP\nXBnNrMnMvm9ma9O/xXMil5uXQTK+xszuMrPtZrbVzL5nZu2xax6uQTKemP6t7khPd5vZibFrHq7B\n/h4zrvfx9PV6binrq7hmDLwAfBq4PnOhu3/L3cccOgHvB54BHopQ40j1m9HMZgDfBP4BGAd8GPi2\nmU0teYUjN1DGc4B/BS4E2oBngVtKXVyBNADPA2cD44F/Bm41sw4zmwz8EPgXkpzLgO/GKnQEBsyY\nXv4A8C5gU4ziCiRXxonAIqADmAXsAb4Ro8gRypXxBeAdJK/TycBtwHeiVDkyg71WMbPZwMXAxpJX\n5+4VeSL5R35Djst/BVwdu85CZgROA7ZkXWcrcHrsWguY8d+Br2aMpwMOzI5da4HyPgZcBFwB/CZj\n+WigCzghdo2Fypi1bD1wTuzaipkxXX4KsCd2fUV8HhuAvwH2xa6vGBmBXwBvAtYC55aylkpcMx5U\nurnvLOCm2LUU2DLgCTN7m5nVp5uou0leUNXE+jl/coxCCsnMpgHHASuAk4BHD13m7p3AmnR5xcrK\nWJUGyXjWAMsrSn8ZzWwnsB/4L5KtVxUtO6OZXQx0u/sdMeppiPGgJfBu4H53fzZ2IYXk7n1mdhPw\nbaAFOABcnP4jrxa/AL5jZtcCTwEfJ1kzbo1a1QiZWSPwLeBGd19lZmNItmpk2gWMLXlxBZKdMXY9\nxZAro5m9guT1emGM2gploIzuPsHMRgMLgHWx6iuEfv4ex5K8wXhDrJqqcs2YpBnfGLuIQkt3KPgC\ncA7QRDL38f/MbG7MugrJ3e8GrgZ+QLKpaC3JPNz6eFWNjJnVATeTvHm6Ml28l2TeP9M4kqwVZ4CM\nVSVXRjM7Fvg58EF3vz9CeQUx2POYvvG/FripQvdVGSjjNcDN7r42UlnV14zN7LUk84zfj11LEcwF\n7nP3Ze5+0N2XAkuAku71V2zu/lV3n+Pu00iacgPweOSy8mJmBlwHTCOZm+pJL1oBvDLjeqOB2VTg\nJs4cGatGrozptNjdwKfc/eZIJY7YMJ7HOpItVTNKVVuh5Mg4H/iAmW0ys03AkSQ7d32kVLVVXDM2\nswYzawHqgXozazGzzM3tC4AfuHtFrmFAzoxLgTMPrQmb2auAM6nAOeOBMqY/T7bEUSR7qn7Z3XfE\nrThvXwdeBrzV3bsylv8IONnMLkp/Dx8HHqvQzbsDZTz0cZKWdNiUPr/2B/dQ/vrNmH7C4ZfAV9z9\n2ljFFchAGd9gZq9K91MZB3wR2AE8EanOkRjotTqfZL+UuenpBeCvga+WrLLYe7PlsffbNSRziJmn\na9LLWoCdwPzYdRYx45XA0yQAX2xYAAAdPklEQVSbM58B/jF2vYXMCEwgeXPRSfJxmM8C9bHrzTPj\nrDTXfpLN0odO70wvPxdYRbIX9T1AR+yai5BxbT/Pc0XlzJWRZErFs5bvjV1zgTNenL5O95Ls53A7\n8IrYNRcyYz/XXUuJ96a29IFFREQkkorbTC0iIlJt1IxFREQiUzMWERGJbETN2MwuMLPVlhzUf2Gh\niionylgdlLE6KGN1qIWMw5X3DlxmVg88SXLEkvUkH7u5zN1XFq68uJSxOihjdVDG6lALGfMxkmZ8\nOsnHbc5Px1cBuPtnB7pNkzV7C6PzerwY+uilmy5aGct+OunhwEdBGZWx/Chj/5Sx/OSTsX7MaG9o\naxv4ThvCPjatdXcwbqvfH4y7Pfyoe6+HG4m39oRHpe3qbsoKMfSPyvdu307f3s5BbzCSY1PPIPk6\nqkPWk3yr0IBaGM1pNn8ED1lam309L7KJE20eS3wxPRxQRpSxHClj/5Sx/OSTsaGtjfaPfHDgK4wP\nDxb296cuDsaXjQtXutf1NgbjTX3hkWmve+HMYPzwmqOCse0deuvc+PkvD+l6Rf+iCDO7guTr4mip\n7GP9D0gZq4MyVgdlrA6ZGesnTohcTfGNpBlvIDl+5yEz02UBd19EckhDxllbRR1hpJlR7Cc4up8y\noozlSBlfoozlLZ+MzUcdGWbM2vX44pc/FIz/dmL4pVIP7m8Jxpc//O5g3LciXDPuntwXjM2LfwTX\nkexNvRSYY2ZHm1kTcClwW2HKKg/jmEgXe+nyThwHZaxIylgdlLE61ELGfOS9ZuzuvWZ2JXAnycH+\nr3f3ivvGmVzqrI7jfS4Pcz9d7AO4VRkrjzJWB2WsDrWQMR8j+pyxu9/h7se5+2x3/0yhiionk62d\nM+wCxjAeZaxcylgdlLE61ELG4Sr6DlwiIiKF5PXhFHI9B4Px2566IBh3n70pGB/4XDhHPPP0F4Jx\n79eOCMZbXxWutx6YGD5eIehwmCIiIpGpGYuIiESmZiwiIhKZ5oxFRKSi3bry1GA86c7wc8VbvhYe\nQes3b/m3YLy5Lzwi1z9tuiIYH2ws/oFVtGYsIiISmZqxiIhIZGrGIiIikWnOWEREKor1hMeKPrg9\n/IrDLa/tDcb/Pf+GYNxaVx+ML7rvvcG4ozVcT+0dXfjPFWfTmrGIiEhkasYiIiKRqRmLiIhEpjlj\nERGpbFlfN3zayWuC8Rkte4Lx9/fOCsbHve/pYLzmYydn3b/mjEVERKqemrGIiEhkasYiIiKRac5Y\nREQqmk04EIz/e9btwXjnwb5gfN1H/yQY7748/Nxx7+jw+qWgNWMREZHI1IxFREQiUzMWERGJTHPG\nIiJS0d58wuPB+IVeD8Z//uh7gvHkLd3BeMP85mDsFt7ePOuDzGN6guHocfuDcVPDS8fG3tIcHid7\nIFozFhERiUzNWEREJDI1YxERkcg0ZywiIhWto+XFYPyzvS8Pr/A/bcHwzf99RzCe07wpGD/TPS0Y\n/9GoZ4Lx+Lpwznlz35hg/FBXx+HzX27e13/RWbRmLCIiEpmasYiISGRqxiIiIpFpzlhERCrarc+d\nEoy3bh8bjMdlfUx4V29rMF5rU4LxtMadwXhNz9Rg/FDW9yHf9fzx4QPcN/Hw2W3bVvZfdBatGYuI\niESmZiwiIhLZoJupzex64C3AFnc/OV3WBnwX6ADWApe4+47ilVlcK3wZ29hIE82cbucB0OMHWM6D\ndLGPUbTiHIxc5cgoozJWCmVUxlo0lDnjG4CvADdlLFsILHb3z5nZwnT8kcKXVxrTmcWRzGYFSw8v\nW8sq2phKh53AWl/FczwVscKRU0ZlrBTKqIzDtWXruGA8anVLMG6/a0sw/t2Pjw7GfZu3Zt1jezDy\nvvD7jRumHxGMJ84N56h3Z0wp2xDfbwy6mdrd7wO2Zy2+ELgxPX8j8PahPVx5mmhTaKQpWLaVF2gn\n+Y22M4seevq7acVQRmWsFMqojLUo372pp7n7xvT8JmDaQFc0syuAKwBaaB3oamXnAN002ygAmmjB\n8QGvq4zlSxlDyli+lDGUmbF+4oSS1BfTiHfgcneHgX+j7r7I3ee5+7xGmge6Wlkzs5yXK2NlUEZl\nrBTKGGasHzMm53WrQb5rxpvNrN3dN5pZO7Bl0FtUmCaa6fYumm0U3d6FkfuFU4mUsTooY3VQxhHY\n0xgMu9rDOd5VV04a5A5mFqaOw156/L6WHFfLkO+a8W3AgvT8AuAned5P2ZrCdDayDoCNrKOBxkFu\nUXmUsTooY3VQxto2lI823QKcA0w2s/XA1cDngFvN7HJgHXBJMYsstuW+hB1spYdu7vfbOYYTmcXx\nLOdBNvhaRtFKM0N8e1OmlFEZK4UyKmMtGrQZu/tlA1w0v8C1RPNyO63f5ady9uHzS3xxqcopCmVM\nKGP5U8aEMtYWS/a/KtGDmW0lWZOeDGwr2QMPz0C1zXL3Kf0sDyhj2VDGQShj2VDGQdRExlI248MP\narbM3eeV/IGHoFC1KWNcylj6+ykGZSz9/RSDMg5Ox6YWERGJTM1YREQksljNeFGkxx2KQtWmjHEp\nY+nvpxiUsfT3UwzKOIgoc8YiIiLyEm2mFhERiUzNWEREJLKSNmMzu8DMVpvZ0+n3IEdlZteb2RYz\nezxjWZuZ3WVmT6U/Jw7zPpWxxJRRGXPcpzKWmDLmlxF3L8kJqAfWAMcATcCjwImlevwBajoLOAV4\nPGPZF4CF6fmFwOeVURmVURmVURmLldHdS9qMTwfuzBhfBVwV8xea1tGR9QtdDbSn59uB1cqojMqo\njMqojMXK6O4l3Uw9A3g+Y7w+XVZuprn7xvT8JmDaMG6rjOVDGXNTxvKhjLnVQkbtwJWLJ29xqvqz\nX8pYHZSxOihjdcgnYymb8QbgyIzxzHRZudlsZu0A6c8tw7itMpYPZcxNGcuHMuZWCxlL2oyXAnPM\n7GgzawIuBW4r4eMP1W3AgvT8AuAnw7itMpYPZcxNGcuHMuZWCxlLtwNXOqn9JuBJkj3jPlYGE/C3\nABuBHpJ5iMuBScBi4CngbqBNGZVRGZVRGZWxmBl1OEwREZHItAOXiIhIZGrGIiIikakZi4iIRKZm\nLCIiEpmasYiISGRqxiIiIpGpGYuIiESmZiwiIhKZmrGIiEhkasYiIiKRqRmLiIhEpmYsIiISmZqx\niIhIZGrGIiIikakZi4iIRKZmLCIiEpmasYiISGRqxiIiIpGpGYuIiESmZiwiIhKZmrGIiEhkasYi\nIiKRqRmLiIhEpmYsIiISmZqxiIhIZGrGIiIikakZi4iIRKZmLCIiEpmasYiISGRqxiIiIpGpGYuI\niESmZiwiIhKZmrGIiEhkasYiIiKRqRmLiIhEVnHN2Myazew6M1tnZnvM7BEze2N6WYeZuZntzTj9\nS+yahytXxvTyVjP7mpltM7NdZnZfzHrzMcjz+M6s53Bf+ryeGrvu4RjC83iJmT2RXrbSzN4es958\nDCHje8zs6fR5/IWZTY9Zb77M7JtmttHMdpvZk2b2nozL5pvZqvR1+iszmxWz1nwNlNHMmszs+2a2\nNv07PCdyqXnLkfE1ZnaXmW03s61m9j0zay9pce5eUSdgNHAN0EHyZuItwJ503AE40BC7zmJlTC//\nJvAdYApQD5wau+ZCZ8y67l8CawCLXXehMgIzgAPAGwED3gzsA6bGrruAGc8BtgAnAU3A14F7Y9ec\nZ86TgOb0/AnAJuBUYDKwC7gYaAH+DXgwdr0FztgE/B3wOmAjcE7sWouQ8Y3pczgOaAWuB35Rytos\nLaqimdljwCeA3wPPAo3u3hu3qsLKyLgC+B0w0913x62qsA5ldPcfZC3/FXCPu38iTmWFk/E8rgd+\n6u5TMy7bCrzN3X8bq75CyMh4OjDK3f8mXT4d2AAc6+5rIpY4ImZ2PHAP8EFgAvCX7n5GetloYBvw\nKndfFa3IEcrM6O63ZixfD7zL3e+JVFrBDJQxvewUkjeOY0tVT8Vtps5mZtOA40ia1CHrzGy9mX3D\nzCZHKq1gsjK+GlgHfCLdTL3czC6KWmABDPA8km7yOwu4KUZdhZSVcRnwhJm9zczq003U3cBjMWsc\nqX6eR8u8OP15ckmLKpB0amgfsIpkDfEOkjWtRw9dx907SbbinBSlyBEaIGNVGWLGs8j6X1RsFd2M\nzawR+BZwY/oudBvwR8Askk0PY9PLK1Y/GWeS/DPbBUwHrgRuNLOXxatyZPrJmOndwP3u/mzpKyuc\n7Izu3kfyBuPbJE3428Bfp//MK1I/z+MvgEvM7BVmNgr4OMk0UmvEMvPm7u8n+Z9yJvBDkudtDMnf\nYqZd6fUqzgAZq8pgGc3sFSSv1Q+Xsq6KbcZmVgfcTDLvdiWAu+9192Xu3uvum9Pl55lZRf5h9JcR\n6AJ6gE+7+wF3vxf4FXBenCpHZoCMmd4N3FjSogqsv4xmdi7wBZJ51SbgbOD/mdncSGWOyAB/j3cD\nVwM/ANampz0km+grkrv3ufsDJG+K3wfsJZlnzDSOJGdF6idj1Rkoo5kdC/ycZNP1/aWsqSKbsZkZ\ncB0wDbjI3XsGuOqhCfGKy5kjY3+bMSty4n+w59HMXkuy9v/9COUVRI6Mc4H70jePB919KbAEODdS\nqXnL9Ty6+1fdfY67TyNpyg3A43EqLagGYDbJpsxXHlqYzhkfWl7pDmWsZoczplNidwOfcvebS11I\nxTWp1NeBlwFvdfeuQwvN7DQzO97M6sxsEvCfJDv+ZG9GqgT9ZgTuA54DrjKzhrRhvR64M0KNIzVQ\nxkMWAD9w94pdy2DgjEuBMw+tCZvZq0g2m1XinPFAf48tZnayJY4CFgFfdvcdsQrNh5lNNbNLzWxM\nOr9/PnAZsBj4EXCymV1kZi0kmzcfq7SdtwbJeOgjbC3p1ZvS59YGvMMylCujmc0Afgl8xd2vjVJg\n7F3Nh3simQ92YD/JJqJDp3emv9hngU6SifmbgCNi11zIjP7S7vm/TXOuBP4kds1FyNgC7ATmx661\niBmvBJ4m2aT5DPCPsWsuZEaSPY0fS1+nm4DPAvWxa84j4xTg3vT1uBtYDvzvjMvPJdkZqItk79yO\n2DUXIePa9HnOPFVUzlwZSaZTPOs1vLeU9VXFR5tEREQqWaVuphYREakaasYiIiKRqRmLiIhENqJm\nbGYXmNlqSw4Ev7BQRZUTZawOylgdlLE61ELG4cp7By4zqweeBN5A8iH+pcBl7r5yoNvUjx3tDZMm\nDv1BesM955v2hrXarn3huLk5GB+Y2BCMDzYPL6u70/PCZhqnTqZ31268s+sxBsnYZM3ewuhhPU5M\njtPJbloZQzdd9NJT9Iw9U8Pb+piD4fhg+Ly3bDgQXt4zvMOO55OxfvRob2hrG/hO68PXUtvo8MBZ\nk+v3BuPsd719hBm7vDEYb+kOj1NzoCu8PJu707NpE41TptC7axfepdcqKGM5qoWMmfbTyQHvHvRj\nYA2DXSGHVwNPu/szAGb2HeBCko/a9P9gkyZyxL98YMgP0LA9LG/GfeE/4ebblwbj+o7w8+nr3jEt\nGO87eqBjg/Sve806dt12F1P//j1s+tR/cqBzw6AZWxjNaTZ/WI8T005/kWdYySl2Jkt8MXvYUfSM\nm/78jGDc9ZqwkfVkNZ4TPxYesKl346ZhPV4+GRva2pjxD3834H32je0Lxpf90ZJgfHnbb4Jxa9af\n4h4PFyzvDr+t7b/W/nEwXr/8iAFrAdi/di077/wfjvjrK9jwxS9xYP16vVZRxnJUCxkzLfHFQ7re\nSDZTzwCezxivT5cFzOwKM1tmZsv69lTWYXf7duyifuKEzEWDZuypsEO5dtNFC6MyFymj2bK+zgp7\nre7aRf0EvVZBGctdLWTMR9F34HL3Re4+z93n1Y+tzM0Mg8nM2Ejz4DeoQLWWsX60XquVShmrQy1k\nzDSSzdQbgCMzxjPTZXlrfDEs54jfhpsCNy3YH4zPvSZ8gm5fNikYz/ppuFn6uWn1wdhbw/vPVj9x\nPH07dmYuGnHGctPMKPYTHImy4BkbZh0ZjCe/NdzsfPz4LcH4d9e+Khj3dITTDTbMzdSFyOjhS4ex\nU8M54ac7pwTj/9MbHmL63h+dEox7xoRzzkef9jwjUT9+PH079VqtdMpYu0ayZrwUmGNmR5tZE3Ap\ncFthyioPTR0z6dn8Ir1btx86nFrVZRzHRLrYS5d34ihjpWo+8kh6tm6j58UX9VqtYMpYu/JeM3b3\nXjO7kuQLCuqB6929Gr6p5DCrr6ftzy9ky5euo/fFHQC3VlvGOqvjeJ/Lw9xPF/tAGSuS1dcz6U//\nhE2L/i+92/VarVTKWLtGNGfs7ne4+3HuPtvdP1OoosrJqFecwPTPfJimGUdQrRknWztn2AWMYbwy\nVrDWE1/GkVctpKm9vWoz1sLzqIy1aSRzxiNn4bzZa84O3xxd+KePBOMn9k8Pxu2NwRwZt7ecHIz3\nTm8Kxt6ce45YiqN3XTgf2pD1jb1rsq4/id8Wt6A8WNZLp2t1sOcyq+4LPz8//f7wWx+7/yz8LHX2\nZ96fWRrOq/e1htfXofJEqpv+xkVERCJTMxYREYlMzVhERCSyqHPG9aPCibiTxmzMef1fbTkuGL/4\nk5nBeHzWIQd3HZt1LOqmcB6OvkEPFyoCgGe/VLKOTW1Zx9NefUVLMJ44bUcw7u4J//TqfzM+GHdN\nCd8ne2N+x5AXkcqgNWMREZHI1IxFREQiUzMWERGJLOqccd++8OGvXXZWeIWsibrGTeFX6zUHX/wB\ne4/J+p7bxnCO2Oqzvje3L+uAwyIDsOwp26w55N3Hha+9idN2B+PWpvA46XueCL8redye8AH2Tdcc\nsUgt0ZqxiIhIZGrGIiIikakZi4iIRBb32NTZerPeG2R9DnjKI+E82tbwa2+xUeG8ndWF1z/YrTli\nKYzs7zdumhR+13b2HPH2va3BeMrD4Wtzy6n6zLtILdOasYiISGRqxiIiIpGpGYuIiERWXnPGWSYs\nD8vbdUx4eV9L1ueIs+eIs+eg/+AAwyL5OTg6PK766FHdwXj3/uZgXP+7ccF455ysO6zLOm66iNQU\nrRmLiIhEpmYsIiISmZqxiIhIZGU1Z1y/K/zwZn13OAe8f0o455s9b/cHn1POHosUSF1L+Jl2z9of\nYc/WMcF40ovha3n7K7PmiAfZn8HHhI/XNjU89nV9xv4Sm7NqE5Hyp24lIiISmZqxiIhIZGrGIiIi\nkZXVnPHB5nBebc+scB6t7Ylwnm3frvD7jbOn3XrCaTvqsqbSxj4X3t/4p/cF4/2TWw6f376z35Kl\nEOqy9hUYMzoY9+0O50ez7b34tGC844Tw/voyXlcHvvpgPhX+Ad/VFIx37g1fi5OPDF8wD3zyW8G4\n2cLrb+nrzPl4zRa+b95zMNxfYl3vS8e+/t+jtue8LxEpP1ozFhERiUzNWEREJDI1YxERkcjKas74\nxJc/F4znTlgfjL815fRgPPHRcJJ40uPhd8r2jAvjvfC6cDzhfz0fjE8avzEYb9g/4fD5FU+Fxx6W\n/K37xBnB+At/fkMwvuaJtwbjPzs6fF7qLJzr/3Dbfwfj33cfCMbvXvR3h89bgQ4B3bQ9fB877pnw\n8m0H2oLxN47pCMZ/PPrJYDzWwv0l7twXHoj95vWvCcbPrmwPxq0vvDRPvm7rF/svWkTKltaMRURE\nIht0zdjMrgfeAmxx95PTZW3Ad4EOYC1wibvvKF6ZxbXi83ex9cFneWhSI3/xvQsA2L+rm3v/6Xb2\nbdxDa/tYDvZV9rfqrPBlbGMjTTRzup0HQI8fYDkP0sU+RtGKU9kZL//7zdx+1z7GTqrjO/+TrDnu\n2tnHc9++lp5d22kc34YfrOyMq/7tTl588BkYNY7pH/0QAH2d+3julm9xYNd2mqogYy28VpWxOjIW\n0lDWjG8ALshathBY7O5zgMXpuGJNv+BETvn824NlS7+ximmnzuCNt17KtFNnsP/FrkjVFcZ0ZvEq\nXhcsW8sq2pjKa+0C2pjKAfYPcOvKsOCScdzx7XDz7Y1f383ojjnMft9HGd0xh97O3B+TKndHnH8S\nr/jsnwbLdt/1S1o75nDsez9KaxVkrIXXqjJWR8ZCGnTN2N3vM7OOrMUXAuek528E7gE+MtJiZrSG\nn818b9tvg/FH3/L7YLz3zT3BeGp9+PnUHg8/i/lcb9hQWw9NOR8Lzz/fy1817OOfpz4AwGkPbObK\nG1/JhEmPccY7e/nw9ZX9oploU+jy8LOsW3mBUzkbgHZmsYaVJall1tW/CcZfvfq4YDyF1cH4l4TP\na7a7mXv4fJd3spndfPToVwPwG/8FpwLN9/6GKQ4PsG+AexmenrHhHG/PmPB97bG3hq+1n/7b3GD8\no/VTct6/vza8ftfM5DPvzUyne+926g4YY59N5ok3PrSSKQv+hoMNMG7uH7Ht/juHHqQMldNrtViU\nsToyFlK+c8bT3P3QXjWbgGkFqqdsbN12kAlTkwM7jJ/SSF+vD3KLynOAbpptFABNtOAoYyXq3beH\nhrHjAKgfMxbPOiBINaiF51EZa9uId+Byd4eBf6NmdoWZLTOzZX17ch9lqFyZGbm+UyczYw+Vude1\nWe5vDaq1jH2dlftaHeTymnoelbF81ULG4ci3GW82s3aA9OeWga7o7ovcfZ67z6sfm3tzYzmZMrmO\nnVuSj8js3HKAuoaBXziZGRtpLlWJI9ZEM92ebE7t9i5yveWotYz1oyvntdrQOpbePck8ce+e3VjW\n4UUz1drzqIzlpRYy5ivfzxnfBiwAPpf+/Ekhiln89PHBeOmmo4Jxa1M4R3zM+G3BeExD+PnStXvD\nz3pu2DU+GO/Z+dLxfHu37mDLnhs55c4PALBjzh1c8x+tjH/T69l1x6842FI5/5yHagrT2cg6OjiB\njayjgcbBb1RhipYxa1vQnqMPZo1bwitw5CDjoevZDgeXOZ0zk8cc9coT2f7U75gw/4/Z+dDvsFHZ\nj1359FqtDrWQMV9D+WjTLSQ7a002s/XA1SRN+FYzuxxYB1xSzCKLbevXbqF71bP07e1kw4f+lfEX\nvoFxbzqbbV//Np0PLKV+0kTqx42NXeaILPcl7GArPXRzv9/OMZzILI5nOQ+ywdcyilaaqex/4rWQ\nccvN32T/02vo6+zkuU98ionnn8f4+X/MlptuZs+S39EwcSL148bFLnNEauF5VMbqyFhIQ9mb+rIB\nLppf4FqimfL+JKL3hptMpn3oisPnN33qP0taU6G93E7rd/mhPRsBlvjiUpVTFLWQcepfvKvf5e3v\ne+/h8xu++KVSlVMUtfA8KmOi0jMWko7AJSIiEpklO0OX6MHMtpJs1p4MbBvk6rEMVNssd8/94VCU\nsYwo4yCUsWwo4yBqImMpm/HhBzVb5u7zSv7AQ1Co2pQxLmUs/f0UgzKW/n6KQRkHp83UIiIikakZ\ni4iIRBarGS+K9LhDUajalDEuZSz9/RSDMpb+fopBGQcRZc5YREREXqLN1CIiIpGVtBmb2QVmttrM\nnjaz6N+BbGbXm9kWM3s8Y1mbmd1lZk+lPycO8z6VscSUURlz3Kcylpgy5pcRdy/JCagH1gDHAE3A\no8CJpXr8AWo6CzgFeDxj2ReAhen5hcDnlVEZlVEZlVEZi5XR3UvajE8H7swYXwVcFfMXmtbRkfUL\nXQ20p+fbgdXKqIzKqIzKqIzFyujuJd1MPQN4PmO8Pl1Wbqa5+8b0/CZg2jBuq4zlQxlzU8byoYy5\n1UJG7cCViydvcap6d3NlrA7KWB2UsTrkk7GUzXgD4Ze4zkyXlZvNZtYOkP7cMozbKmP5UMbclLF8\nKGNutZCxpM14KTDHzI42sybgUuC2Ej7+UN0GLEjPLwB+MozbKmP5UMbclLF8KGNutZCxdDtwpZPa\nbwKeJNkz7mNlMAF/C7AR6CGZh7gcmAQsBp4C7gbalFEZlVEZlVEZi5lRR+ASERGJTDtwiYiIRKZm\nLCIiEpmasYiISGRqxiIiIpGpGYuIiESmZiwiIhKZmrGIiEhkasYiIiKR/X9ryWWYwGfHAwAAAABJ\nRU5ErkJggg==\n", "text/plain": [ "
" ] }, "metadata": { "tags": [] }, "output_type": "display_data" } ], "source": [ "activation_model = models.Model(inputs=conv_model.input, outputs=layer_outputs[3]) # Creates a model that will return these outputs, given the model input\n", "\n", "\n", "activations = activation_model.predict(img_tensor/255.) # Returns a Numpy array corresponding to the output of layer 4\n", "print(activations.shape)\n", "\n", "# Show the result of the 1st Max Pooling\n", "\n", "plt.figure(figsize=(8,8))\n", "for k in range(activations.shape[3]):\n", " plt.subplot(4,8,k+1)\n", " plt.imshow(activations[0,:,:,k])\n", " plt.title(str(k+1))" ] } ], "metadata": { "colab": { "collapsed_sections": [], "name": "MNIST_convolutional_network_keras.ipynb", "provenance": [], "version": "0.3.2" }, "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.7.3" } }, "nbformat": 4, "nbformat_minor": 1 }