R9_External_Lab_Questions.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"name":"R9_External_Lab_Questions.ipynb","version":"0.3.2","provenance":[],"collapsed_sections":[]},"language_info":{"codemirror_mode":{"name":"ipython","version":3},"file_extension":".py","mimetype":"text/x-python","name":"python","nbconvert_exporter":"python","pygments_lexer":"ipython3","version":"3.7.0"},"kernelspec":{"name":"python3","display_name":"Python 3"},"accelerator":"GPU"},"cells":[{"cell_type":"markdown","metadata":{"colab_type":"text","id":"WPJHV_CIn6Bg"},"source":["# U - Net\n","## Problem: \n","\n","Seismic data is collected using reflection seismology, or seismic reflection. The method requires a controlled seismic source of energy, such as compressed air or a seismic vibrator, and sensors record the reflection from rock interfaces within the subsurface. The recorded data is then processed to create a 3D view of earth’s interior. Reflection seismology is similar to X-ray, sonar and echolocation.\n","\n","A seismic image is produced from imaging the reflection coming from rock boundaries. The seismic image shows the boundaries between different rock types. In theory, the strength of reflection is directly proportional to the difference in the physical properties on either sides of the interface. While seismic images show rock boundaries, they don't say much about the rock themselves; some rocks are easy to identify while some are difficult.\n","\n","There are several areas of the world where there are vast quantities of salt in the subsurface. One of the challenges of seismic imaging is to identify the part of subsurface which is salt. Salt has characteristics that makes it both simple and hard to identify. Salt density is usually 2.14 g/cc which is lower than most surrounding rocks. The seismic velocity of salt is 4.5 km/sec, which is usually faster than its surrounding rocks. This difference creates a sharp reflection at the salt-sediment interface. Usually salt is an amorphous rock without much internal structure. This means that there is typically not much reflectivity inside the salt, unless there are sediments trapped inside it. The unusually high seismic velocity of salt can create problems with seismic imaging.\n","\n","### Data\n","The data is a set of images chosen at various locations chosen at random in the subsurface. The images are 101 x 101 pixels and each pixel is classified as either salt or sediment. In addition to the seismic images, the depth of the imaged location is provided for each image. The goal of the competition is to segment regions that contain salt.\n","\n","#### Source: \n","https://www.kaggle.com/c/tgs-salt-identification-challenge\n","\n","\n","### Note: \n","Accept the terms and download data from the above link"]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"3OKrTNUtn-od"},"source":["### Aim: \n","\n","Implement U-Net neural model architecture in keras to solve this problem.\n","\n","\n","In this, you are asked to segment salt deposits beneath the Earth’s surface. Given a set of seismic images that are 101 x 101 pixels each and each pixel we need to classify as either salt or sediment. Our goal is to segment regions that contain salt. A seismic image is produced from imaging the reflection coming from rock boundaries. The seismic image shows the boundaries between different rock types. "]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"ZUOAF8WWoA1x"},"source":["### Broad Steps:\n","\n","1. Download the dataset\n","2. Upload to Drive\n","3. Import from drive to colab\n","4. Load the images and create training data.\n","5. Build U-net Model\n","6. Train your model.\n","7. Check the validation accuracy and plot sample."]},{"cell_type":"code","metadata":{"colab_type":"code","id":"b7TLNzB6oDep","outputId":"6d75bec8-cd58-4d11-e57c-d6161a152429","executionInfo":{"status":"ok","timestamp":1563714788847,"user_tz":-330,"elapsed":3832,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}},"colab":{"base_uri":"https://localhost:8080/","height":34}},"source":["#Importing necessary libraries\n","import os\n","import random\n","import pandas as pd\n","import numpy as np\n","import matplotlib.pyplot as plt\n","plt.style.use(\"ggplot\")\n","%matplotlib inline\n","\n","from tqdm import tqdm_notebook, tnrange\n","from itertools import chain\n","from skimage.io import imread, imshow, concatenate_images\n","from skimage.transform import resize\n","from skimage.morphology import label\n","from sklearn.model_selection import train_test_split\n","\n","import tensorflow as tf\n","from keras.models import Model, load_model\n","from keras.layers import Input, BatchNormalization, Activation, Dense, Dropout\n","from keras.layers.core import Lambda, RepeatVector, Reshape\n","from keras.layers.convolutional import Conv2D, Conv2DTranspose\n","from keras.layers.pooling import MaxPooling2D, GlobalMaxPool2D\n","from keras.layers.merge import concatenate, add\n","from keras.callbacks import EarlyStopping, ModelCheckpoint, ReduceLROnPlateau\n","from keras.optimizers import Adam\n","from keras.preprocessing.image import ImageDataGenerator, array_to_img, img_to_array, load_img"],"execution_count":1,"outputs":[{"output_type":"stream","text":["Using TensorFlow backend.\n"],"name":"stderr"}]},{"cell_type":"code","metadata":{"colab_type":"code","id":"6qiCvW9coD_E","outputId":"c00d5312-5875-4edc-b13c-22079e5ce689","executionInfo":{"status":"ok","timestamp":1563714826411,"user_tz":-330,"elapsed":25112,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}},"colab":{"base_uri":"https://localhost:8080/","height":121}},"source":["#Mount drive\n","from google.colab import drive\n","drive.mount('/content/drive/')"],"execution_count":2,"outputs":[{"output_type":"stream","text":["Go to this URL in a browser: https://accounts.google.com/o/oauth2/auth?client_id=947318989803-6bn6qk8qdgf4n4g3pfee6491hc0brc4i.apps.googleusercontent.com&redirect_uri=urn%3Aietf%3Awg%3Aoauth%3A2.0%3Aoob&scope=email%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdocs.test%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive.photos.readonly%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fpeopleapi.readonly&response_type=code\n","\n","Enter your authorization code:\n","··········\n","Mounted at /content/drive/\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"B4IaUgploION"},"source":["### 1.Set your project path where you have your data and related files for this lab."]},{"cell_type":"code","metadata":{"colab_type":"code","id":"0tPkbXWyoK-z","colab":{}},"source":["os.chdir('/content/drive/My Drive/Resedency9')\n","project_path = '/content/drive/My Drive/Resedency9/'"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"Kxy5LexhoLMT","colab":{}},"source":[""],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"g15qSXZSoSuL"},"source":["### 2. Set the necessary parameters"]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"Dzrz2Or_HdxR"},"source":["## Setup"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"3IEaEnuzoW2a","colab":{}},"source":["im_width = 128      #width of your train image\n","im_height = 128     #hight of your train image\n","#border =        \n","path_train = project_path + 'train/'   #Path for your train data"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"uZQyyn55oKW-","colab":{"base_uri":"https://localhost:8080/","height":34},"outputId":"aa89ad0a-0567-47da-f238-5e4359258069","executionInfo":{"status":"ok","timestamp":1563715603389,"user_tz":-330,"elapsed":731,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}}},"source":["print(path_train)"],"execution_count":28,"outputs":[{"output_type":"stream","text":["/content/drive/My Drive/Resedency9/train/\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"Q3IoS6RCoZPh"},"source":["# 3. Make directory for train data at in your project/lab folder.\n","\n","Hint - use !mkdir function."]},{"cell_type":"code","metadata":{"colab_type":"code","id":"MRPpQ2QrocBQ","colab":{}},"source":["try:\n","    os.mkdir(path_train)\n","except OSError as exc:\n","    if exc.errno != errno.EEXIST:\n","        raise\n","    pass"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"RWATodbkocN9","colab":{}},"source":[""],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"fRbC22DGoeb7"},"source":["# 4. Extract your train images to the train directory you have just created above. \n","train.zip and test.zip files available at your google drive/local system.\n","\n","As a good practice - Upload or copy the data at your project path folder.\n","\n","Make sure you are providing the right project_path."]},{"cell_type":"code","metadata":{"id":"HGxtYcqNgE-n","colab_type":"code","outputId":"858503f9-6a14-4111-c8c9-3f0107103d64","executionInfo":{"status":"ok","timestamp":1563715611644,"user_tz":-330,"elapsed":959,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}},"colab":{"base_uri":"https://localhost:8080/","height":34}},"source":["print(project_path)"],"execution_count":29,"outputs":[{"output_type":"stream","text":["/content/drive/My Drive/Resedency9/\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"colab_type":"code","id":"5BtPznFbogjG","colab":{}},"source":["#For simplicity we have added the required code here.\n","from zipfile import ZipFile\n","\n","with ZipFile(project_path + 'train.zip', 'r') as zf:\n","  zf.extractall('train/')\n"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"7YZ2prb9oiX4","colab":{}},"source":["#The train file have both images and masks with the same names_ids."],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"zxQ1u8Bvoj1M","colab":{}},"source":[""],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"iwCH81ZWomLA"},"source":["### 5. Get the list of names of images and masks and name the list imagelist and masklist.\n","\n","Hint - Use os.listdir() funtions."]},{"cell_type":"code","metadata":{"id":"ElKEu_aTtVZj","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":101},"outputId":"f349e393-aed0-4670-89f6-e302cd2ec49e","executionInfo":{"status":"ok","timestamp":1563715616244,"user_tz":-330,"elapsed":976,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}}},"source":["os.listdir(project_path)"],"execution_count":30,"outputs":[{"output_type":"execute_result","data":{"text/plain":["['data',\n"," 'data.zip',\n"," 'train.zip',\n"," 'train (1)',\n"," 'R9_External_Lab_Questions.ipynb']"]},"metadata":{"tags":[]},"execution_count":30}]},{"cell_type":"code","metadata":{"colab_type":"code","id":"Clt5ya_JoplC","colab":{}},"source":["imagelist  = os.listdir(project_path+'train (1)/images')\n","masklist = os.listdir(project_path+'train (1)/masks')"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"-vHPQoHVoppK","outputId":"217550a9-6bfc-41b3-a98f-9de74582242a","executionInfo":{"status":"ok","timestamp":1563715647905,"user_tz":-330,"elapsed":1166,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}},"colab":{"base_uri":"https://localhost:8080/","height":84}},"source":["#Test your list names by printing some of the names as given below.\n","print(imagelist[-1])\n","print(masklist[-1])\n","print(imagelist[10])\n","print(masklist[10])"],"execution_count":32,"outputs":[{"output_type":"stream","text":["429bf7c665.png\n","423ae1a09c.png\n","bfc47bbe13.png\n","c092a99d23.png\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"colab_type":"code","id":"WUZw-kSKoqIr","colab":{}},"source":[""],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"1t8raFq5ous8"},"source":["# 6. Read and test your images and respective masks.\n","\n","Hint -\n","\n","import  matplotlib.pyplot as plt\n","\n","import cv2\n","\n","plt.imshow(cv2.imread('path of image'))\n","\n","plt.imshow(cv2.imread('path of mask'))\n"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"1gGpCSPpovvH","colab":{}},"source":["import cv2\n","import matplotlib.pyplot as plt\n","from PIL import Image"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"9E-McW2toyaR","outputId":"565d4f4e-2cf5-412c-c879-0d8b7a679657","executionInfo":{"status":"ok","timestamp":1563715679045,"user_tz":-330,"elapsed":1575,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}},"colab":{"base_uri":"https://localhost:8080/","height":269}},"source":["\n","for file in imagelist:\n","   n = imagelist.index(file)\n","   path1 = os.path.join(project_path+'train (1)/images',file)\n","   plt.imshow(cv2.imread(path1))\n","   break "],"execution_count":37,"outputs":[{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAQoAAAD8CAYAAACPd+p5AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"cell_type":"code","metadata":{"id":"_JKlJ1B8zaZt","colab_type":"code","outputId":"3b576876-3cbd-4587-93b7-7448f0885d7b","executionInfo":{"status":"ok","timestamp":1563715690755,"user_tz":-330,"elapsed":1635,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}},"colab":{"base_uri":"https://localhost:8080/","height":269}},"source":["for file in masklist:\n","   n = masklist.index(file)\n","   path2 = os.path.join(project_path+'train (1)/masks',file)\n","   plt.imshow(cv2.imread(path2))\n","   break"],"execution_count":38,"outputs":[{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAQoAAAD8CAYAAACPd+p5AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nkqGQVPpfO37bIpiEIUEAAAAASUVORK5CYII=\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"_qFeroMNoylc"},"source":["# 7. Create your training data.\n","\n","Hints - \n","\n","image_path = os.path.join(project_path +'path of your image directory' +n )\n","\n","mask_path = os.path.join(project_path +'path of your mask directory'+n )"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"gJbRVbL4o1qT","outputId":"8cbcf3b9-d0ba-4243-a1a2-413e17a9980b","executionInfo":{"status":"ok","timestamp":1563386867243,"user_tz":-330,"elapsed":16468,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}},"colab":{"base_uri":"https://localhost:8080/","height":34}},"source":["# Get and resize train images and masks\n","def get_data():\n","    X = np.zeros(((len(imagelist), im_height, im_width, 1)), dtype=np.float32) # Create an array for image\n","    y = np.zeros(((len(masklist), im_height, im_width, 1)), dtype=np.float32) #Create an array for mask\n","    \n","    for n in imagelist: \n","        k = imagelist.index(n)\n","        ##Add image_path\n","        image_path = os.path.join(project_path+'train (1)/images', n)\n","        ##Add mask_path\n","        mask_path = os.path.join(project_path+'train (1)/masks', n)                                                           \n","        \n","        # Load images and resize to (128,128,1)\n","        resized_img = Image.open(image_path)\n","        resized_img = resized_img.resize((im_width, im_height))\n","        resized_img = np.reshape(resized_img.convert('L'), (128,128,1)) ## Converting the image to greyscale # Minimizing the parameters\n","        # Load masks and resize to (128,128,1)\n","        resized_mask = Image.open(mask_path)\n","        resized_mask = resized_mask.resize((im_width, im_height))\n","        resized_mask = np.reshape(resized_mask.convert('L'), (128,128,1)) ## Converting the image to greyscale # Minimizing the parameters\n","        # Save images\n","        X[k] = resized_img /255\n","        y[k] = resized_mask/255\n","    print('Done!')\n","    return X, y\n","    \n","X, y = get_data()"],"execution_count":0,"outputs":[{"output_type":"stream","text":["Done!\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"colab_type":"code","id":"seheYxTBo32V","outputId":"9248bef0-ec80-4230-af55-e23d3696574d","executionInfo":{"status":"ok","timestamp":1563387056419,"user_tz":-330,"elapsed":907,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}},"colab":{"base_uri":"https://localhost:8080/","height":34}},"source":["X.shape"],"execution_count":0,"outputs":[{"output_type":"execute_result","data":{"text/plain":["(4000, 128, 128, 1)"]},"metadata":{"tags":[]},"execution_count":76}]},{"cell_type":"code","metadata":{"id":"btizSpjEIX5b","colab_type":"code","outputId":"92ffcf75-913a-4da3-a242-3ba932f96ce4","executionInfo":{"status":"ok","timestamp":1563387073072,"user_tz":-330,"elapsed":934,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}},"colab":{"base_uri":"https://localhost:8080/","height":34}},"source":["y.shape"],"execution_count":0,"outputs":[{"output_type":"execute_result","data":{"text/plain":["(4000, 128, 128, 1)"]},"metadata":{"tags":[]},"execution_count":77}]},{"cell_type":"code","metadata":{"id":"p5BTyrkZIqp_","colab_type":"code","outputId":"fdc4a48f-b979-4be1-f3b8-c76e81be7df1","executionInfo":{"status":"ok","timestamp":1563387911637,"user_tz":-330,"elapsed":1204,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}},"colab":{"base_uri":"https://localhost:8080/","height":286}},"source":["plt.imshow(X[1].squeeze())"],"execution_count":0,"outputs":[{"output_type":"execute_result","data":{"text/plain":["<matplotlib.image.AxesImage at 0x7f3918f87400>"]},"metadata":{"tags":[]},"execution_count":90},{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAQUAAAD8CAYAAAB+fLH0AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nHg48U/Dw8HDgmYKHh4cDzxQ8PDwc/B+VuHFs3T7dkwAAAABJRU5ErkJggg==\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"cell_type":"markdown","metadata":{"id":"qLgr93EFuqMC","colab_type":"text"},"source":["# Not Done the Below CNN execution as the Unet Segmentation topics are still not covered"]},{"cell_type":"markdown","metadata":{"id":"c9ZO8qUvuocx","colab_type":"text"},"source":[""]},{"cell_type":"code","metadata":{"colab_type":"code","id":"kt6zXJPHo35y","colab":{}},"source":["# Split train and valid\n","from sklearn.model_selection import train_test_split\n","X_train, X_valid, y_train, y_valid = train_test_split(X, y, test_size=0.15, random_state=2019)"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"mh_SRFGro6wr","colab":{}},"source":["## Test your data whether it looks fine - Random check\n","import random \n","# Check if training data looks all right\n","ix = random.randint(0, len(X_train))\n","has_mask = y_train[ix].max() > 0\n","\n","fig, ax = plt.subplots(1, 2, figsize=(10, 10))\n","\n","ax[0].imshow(X_train[ix, ..., 0], cmap='seismic', interpolation='bilinear')\n","if has_mask:\n","    ax[0].contour(y_train[ix].squeeze(), colors='k', levels=[0.5])\n","ax[0].set_title('Seismic')\n","\n","ax[1].imshow(y_train[ix].squeeze(), interpolation='bilinear', cmap='gray')\n","ax[1].set_title('Salt');\n","\n","\n"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"_-JXuxrRo8a6","colab":{}},"source":[""],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"em5jGV93o8qR"},"source":["### 8. Define loss and dice_coeff function."]},{"cell_type":"code","metadata":{"colab_type":"code","id":"Uyxf8uhQpA78","colab":{}},"source":[""],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"EmCjyEUUpA_V","colab":{}},"source":[""],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"-ofKZA-9pDVJ"},"source":["### 9. Build and compile UNet Model for your data.\n","\n","Hint - \n","You can install and use segmentation models from this github repository.\n","\n","#Install segmentation models\n","\n","!pip install git+https://github.com/qubvel/segmentation_models"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"456utznCpGco","colab":{}},"source":["#model = Unet(backbone_name='resnet34', encoder_weights=None, input_shape=(None, None, 1))\n","#compile your model by adding the parameters like optimizer, loss and metrics.\n","\n"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"eL1lRlXDpIHl","colab":{}},"source":["#Get the summary of your model\n"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"cA4f4oYYpJA6","colab":{}},"source":[""],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"6UYvb8OdpLV5"},"source":["### 10. Fit your model using model.fit function.\n","Hint - As it might take long time to run. Run it for only 1 or 2 epochs."]},{"cell_type":"code","metadata":{"colab_type":"code","id":"bFw1Yq06pQbp","colab":{}},"source":[""],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"iH34vMq9pSnh"},"source":["### 11.Predict on val set using model.predict funtion and store in preds_val variable.\n","\n"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"FymN5QpopXW-","colab":{}},"source":[""],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"uKkpE3vEpXi8","colab":{}},"source":["#Get the threshold predictions to look at refined results.\n","preds_val_t = (preds_val > 0.5).astype(np.uint8)"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"Pyn3RyCRpaRH","colab":{}},"source":["#Plot a sample\n","def plot_sample(X, y, preds, binary_preds, ix=None):\n","    if ix is None:\n","        ix = random.randint(0, len(X))\n","\n","    has_mask = y[ix].max() > 0\n","\n","    fig, ax = plt.subplots(1, 4, figsize=(20, 10))\n","    ax[0].imshow(X[ix, ..., 0], cmap='seismic')\n","    if has_mask:\n","        ax[0].contour(y[ix].squeeze(), colors='k', levels=[0.5])\n","    ax[0].set_title('Seismic')\n","\n","    ax[1].imshow(y[ix].squeeze())\n","    ax[1].set_title('Salt')\n","\n","    ax[2].imshow(preds[ix].squeeze(), vmin=0, vmax=1)\n","    if has_mask:\n","        ax[2].contour(y[ix].squeeze(), colors='k', levels=[0.5])\n","    ax[2].set_title('Salt Predicted')\n","    \n","    ax[3].imshow(binary_preds[ix].squeeze(), vmin=0, vmax=1)\n","    if has_mask:\n","        ax[3].contour(y[ix].squeeze(), colors='k', levels=[0.5])\n","    ax[3].set_title('Salt Predicted binary');"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"s_h5fcgYpbyR","colab":{}},"source":["# Check if valid data looks all right\n","plot_sample(X_valid, y_valid, preds_val, preds_val_t, ix=15)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"oasnhlaDpeU0"},"source":["If you are getting good results- Congratulations.\n","If you are not, try to explore what might be the reason."]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"yllJR0PQpm4M"},"source":["# Text generation using a RNN"]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"2W6g03QsptwG"},"source":["Given a sequence of words from this data, train a model to predict the next word in the sequence. Longer sequences of text can be generated by calling the model repeatedly."]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"hbd3E0IuHwjz"},"source":["**Mount your Google Drive**"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"Dikrw1ylHtAL","colab":{"base_uri":"https://localhost:8080/","height":121},"outputId":"7981386d-a04a-4024-e711-c3cc29568159","executionInfo":{"status":"ok","timestamp":1563717838528,"user_tz":-330,"elapsed":25696,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}}},"source":["#Mount drive\n","from google.colab import drive\n","drive.mount('/content/drive/')"],"execution_count":2,"outputs":[{"output_type":"stream","text":["Go to this URL in a browser: https://accounts.google.com/o/oauth2/auth?client_id=947318989803-6bn6qk8qdgf4n4g3pfee6491hc0brc4i.apps.googleusercontent.com&redirect_uri=urn%3Aietf%3Awg%3Aoauth%3A2.0%3Aoob&scope=email%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdocs.test%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive.photos.readonly%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fpeopleapi.readonly&response_type=code\n","\n","Enter your authorization code:\n","··········\n","Mounted at /content/drive/\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"colab_type":"code","id":"8_uTQu4nIGzm","colab":{}},"source":["import os\n","os.chdir('/content/drive/My Drive/Resedency9/data')"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"fnIX_mLXHdxS"},"source":["### Import Keras and other libraries"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"0TmrQuvpHdxU","colab":{"base_uri":"https://localhost:8080/","height":34},"outputId":"3923d571-7d27-4f30-8ec2-4bb328ad5cbb","executionInfo":{"status":"ok","timestamp":1563717846210,"user_tz":-330,"elapsed":3177,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}}},"source":["import glob\n","\n","from sklearn.utils import shuffle\n","import numpy as np\n","\n","from keras.preprocessing.text import Tokenizer\n","from keras.models import Sequential, load_model\n","from keras.layers import LSTM, Dense, Dropout, Embedding, Masking, Bidirectional\n","from keras.optimizers import Adam\n","from keras import backend"],"execution_count":4,"outputs":[{"output_type":"stream","text":["Using TensorFlow backend.\n"],"name":"stderr"}]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"zKBXQflGlPjG"},"source":["## Download data\n","Reference: Data is collected from http://www.gutenberg.org\n","\n","For the lab purpose, you can load the dataset provided by Great Learning"]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"s321mV4DHdxZ"},"source":["### Load the Oscar Wilde dataset\n","\n","Store all the \".txt\" file names in a list"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"VUanlzNJHdxa","colab":{"base_uri":"https://localhost:8080/","height":54},"outputId":"ed99d233-502d-4a4a-94d5-4fadfc75a689","executionInfo":{"status":"ok","timestamp":1563717849615,"user_tz":-330,"elapsed":1200,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}}},"source":["oscarWildelist = glob.glob(\"*.txt\")\n","print(oscarWildelist)"],"execution_count":5,"outputs":[{"output_type":"stream","text":["['A Critic in Pall Mall.txt', 'A House of Pomegranates.txt', 'A Woman of No Importance a play.txt', 'Children in Prison and Other Cruelties of Prison Life.txt', 'An Ideal Husband.txt', 'De Profundis.txt', 'Charmides and Other Poems.txt', 'For Love of the King.txt', 'Essays and Lectures.txt', 'Impressions of America.txt', 'Intentions.txt', 'Lady Windermere_s Fan.txt', 'Lord Arthur Savile_s Crime.txt', 'Selected poems of oscar wilde including The Ballad of Reading Gaol.txt', 'Rose Leaf and Apple Leaf.txt', 'Poems with the Ballad of Reading Gaol.txt', 'Oscar Wilde Miscellaneous.txt', 'Salomé A tragedy in one act.txt', 'Miscellaneous Aphorisms_ The Soul of Man.txt', 'Selected prose of oscar wilde with a Preface by Robert Ross.txt', 'Reviews.txt', 'Miscellanies.txt', 'Shorter Prose Pieces.txt', 'The Ballad of Reading Gaol.txt', 'The Duchess of Padua.txt', 'The Canterville Ghost.txt', 'The Happy Prince and other tales.txt', 'The Picture of Dorian Gray.txt', 'The Importance of Being Earnest.txt', 'Vera or, The Nihilists.txt', 'The Soul of Man.txt']\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"glr4hv6uZkL-"},"source":["### Read the data\n","\n","Read contents of every file from the list and append the text in a new list"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"zLrMMjrkRt9x","colab":{}},"source":["codetext = []\n","bookranges = []\n","for oscarWildefile in oscarWildelist:\n","  oscartext = open(oscarWildefile, \"r\")\n","  start = len(codetext)\n","  codetext.append(oscartext.read())\n","  end = len(codetext)\n","  bookranges.append({\"start\": start, \"end\": end, \"name\": oscarWildefile.rsplit(\"/\", 1)[-1]})\n","  oscartext.close()"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"jMX-Fu-GHdxj"},"source":["## Process the text\n","Initialize and fit the tokenizer"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"zQf1AV8wHdxl","colab":{}},"source":["tokenizer = Tokenizer(lower=True, filters='!\"#$%&()*+,-./:;<=>?@[\\\\]^_`{|}~\\t\\n')\n","tokenizer.fit_on_texts(codetext)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"vpZ0A2-xHdxp"},"source":["### Vectorize the text\n","\n","Before training, we need to map strings to a numerical representation. Create two lookup tables: one mapping words to numbers, and another for numbers to words."]},{"cell_type":"code","metadata":{"colab_type":"code","id":"_Nsq-rSPHdxq","colab":{}},"source":["word_idx = tokenizer.word_index\n","idx_word = tokenizer.index_word"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"YMYdjx4aHdxu"},"source":["Get the word count for every word and also get the total number of words."]},{"cell_type":"code","metadata":{"colab_type":"code","id":"ioEZ2c21Hdxw","colab":{}},"source":["word_counts = tokenizer.word_counts\n","num_words = len(word_idx) + 1"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"dWUBr9rHHdx0"},"source":["Convert text to sequence of numbers"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"dwLl0BWKHdx2","colab":{}},"source":["sequences = tokenizer.texts_to_sequences(codetext)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"GkpK8McUHdx6"},"source":["### Generate Features and Labels"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"zxhQamjwHdx7","colab":{}},"source":["features = []\n","labels = []\n","\n","training_length = 50\n","# Iterate through the sequences of tokens\n","for seq in sequences:\n","    # Create multiple training examples from each sequence\n","    for i in range(training_length, training_length+300):\n","        # Extract the features and label\n","        extract = seq[i - training_length: i - training_length + 20]\n","\n","        # Set the features and label\n","        features.append(extract[:-1])\n","        labels.append(extract[-1])"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"bbmsf23Bymwe"},"source":["### The prediction task"]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"wssHQ1oGymwe"},"source":["Given a word, or a sequence of words, what is the most probable next word? This is the task we're training the model to perform. The input to the model will be a sequence of words, and we train the model to predict the output—the following word at each time step.\n","\n","Since RNNs maintain an internal state that depends on the previously seen elements, given all the words computed until this moment, what is the next word?"]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"T2bsVOl7HdyA"},"source":["### Generate training and testing data"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"j7-IsvynHdyB","colab":{}},"source":["from sklearn.utils import shuffle\n","import numpy as np\n","\n","features, labels = shuffle(features, labels, random_state=1)\n","\n","# Decide on number of samples for training\n","train_end = int(0.7 * len(labels))\n","\n","train_features = np.array(features[:train_end])\n","valid_features = np.array(features[train_end:])\n","\n","train_labels = labels[:train_end]\n","valid_labels = labels[train_end:]\n","\n","# Convert to arrays\n","X_train, X_valid = np.array(train_features), np.array(valid_features)\n","\n","# Using int8 for memory savings\n","y_train = np.zeros((len(train_labels), num_words), dtype=np.int8)\n","y_valid = np.zeros((len(valid_labels), num_words), dtype=np.int8)\n","\n","# One hot encoding of labels\n","for example_index, word_index in enumerate(train_labels):\n","    y_train[example_index, word_index] = 1\n","\n","for example_index, word_index in enumerate(valid_labels):\n","    y_valid[example_index, word_index] = 1"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"juT1mZrUHdyE"},"source":["This is just to check the features and labels"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"wkdmNbgjHdyF","colab":{"base_uri":"https://localhost:8080/","height":151},"outputId":"18e385e6-0812-40a1-e2fd-3d3a3a8ad6fe","executionInfo":{"status":"ok","timestamp":1563717920551,"user_tz":-330,"elapsed":1314,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}}},"source":["for i, sequence in enumerate(X_train[:2]):\n","    text = []\n","#     print(i, sequence)\n","    for idx in sequence:\n","        text.append(idx_word[idx])\n","        \n","    print('Features: ' + ' '.join(text)+'\\n')\n","    print('Label: ' + idx_word[np.argmax(y_train[i])] + '\\n')"],"execution_count":13,"outputs":[{"output_type":"stream","text":["Features: down on him making satirical remarks on the photographs suddenly there leaped out two figures 'poor poor ghost '\n","\n","Label: she\n","\n","Features: and other poems author oscar wilde release date september 19 2014 ebook 1031 this file was first posted on\n","\n","Label: 17\n","\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"r6oUuElIMgVx"},"source":["## Build The Model"]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"m8gPwEjRzf-Z"},"source":["Use `keras.Sequential` to define the model. For this simple example three layers are used to define our model:\n","\n","* `keras.layers.Embedding`: The input layer. A trainable lookup table that will map the numbers of each character to a vector with `embedding_dim` dimensions;\n","* `keras.layers.LSTM`: A type of RNN with size `units=rnn_units` (You can also use a GRU layer here.)\n","* `keras.layers.Dense`: The output layer, with `num_words` outputs."]},{"cell_type":"code","metadata":{"colab_type":"code","id":"GKpCQFZLHdyN","colab":{"base_uri":"https://localhost:8080/","height":591},"outputId":"b3c30076-d168-429b-9446-7d4649159eae","executionInfo":{"status":"ok","timestamp":1563717929012,"user_tz":-330,"elapsed":1979,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}}},"source":["model = Sequential()\n","\n","# Embedding layer\n","model.add(\n","    Embedding(\n","        input_dim=len(word_idx) + 1,\n","        output_dim=100,\n","        weights=None,\n","        trainable=True))\n","\n","# Recurrent layer\n","model.add(\n","    LSTM(\n","        64, return_sequences=False, dropout=0.1,\n","        recurrent_dropout=0.1))\n","\n","# Fully connected layer\n","model.add(Dense(64, activation='relu'))\n","\n","# Dropout for regularization\n","model.add(Dropout(0.5))\n","\n","# Output layer\n","model.add(Dense(num_words, activation='softmax'))\n","\n","# Compile the model\n","model.compile(\n","    optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])\n","\n","model.summary()"],"execution_count":14,"outputs":[{"output_type":"stream","text":["WARNING: Logging before flag parsing goes to stderr.\n","W0721 14:05:30.629365 139995219507072 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:74: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.\n","\n","W0721 14:05:30.668369 139995219507072 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:517: The name tf.placeholder is deprecated. Please use tf.compat.v1.placeholder instead.\n","\n","W0721 14:05:30.674100 139995219507072 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:4138: The name tf.random_uniform is deprecated. Please use tf.random.uniform instead.\n","\n","W0721 14:05:30.791287 139995219507072 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:133: The name tf.placeholder_with_default is deprecated. Please use tf.compat.v1.placeholder_with_default instead.\n","\n","W0721 14:05:30.804279 139995219507072 deprecation.py:506] From /usr/local/lib/python3.6/dist-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","W0721 14:05:31.217445 139995219507072 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/optimizers.py:790: The name tf.train.Optimizer is deprecated. Please use tf.compat.v1.train.Optimizer instead.\n","\n","W0721 14:05:31.240889 139995219507072 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:3295: The name tf.log is deprecated. Please use tf.math.log instead.\n","\n"],"name":"stderr"},{"output_type":"stream","text":["_________________________________________________________________\n","Layer (type)                 Output Shape              Param #   \n","=================================================================\n","embedding_1 (Embedding)      (None, None, 100)         3283900   \n","_________________________________________________________________\n","lstm_1 (LSTM)                (None, 64)                42240     \n","_________________________________________________________________\n","dense_1 (Dense)              (None, 64)                4160      \n","_________________________________________________________________\n","dropout_1 (Dropout)          (None, 64)                0         \n","_________________________________________________________________\n","dense_2 (Dense)              (None, 32839)             2134535   \n","=================================================================\n","Total params: 5,464,835\n","Trainable params: 5,464,835\n","Non-trainable params: 0\n","_________________________________________________________________\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"vL3tUp1UHdyS"},"source":["For each word the model looks up the embedding, runs the LSTM one timestep with the embedding as input, and applies the dense layer to generate logits predicting the log-liklihood of the next word."]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"LJL0Q0YPY6Ee"},"source":["## Train the model"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"6o84puBcHdyV","colab":{"base_uri":"https://localhost:8080/","height":1000},"outputId":"d37d01b1-bfdd-422b-ee73-19396e9ddf94","executionInfo":{"status":"ok","timestamp":1563720490859,"user_tz":-330,"elapsed":2554481,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}}},"source":["h = model.fit(X_train, y_train, epochs = 500, batch_size = 50, \n","          verbose = 1)## Train the model"],"execution_count":15,"outputs":[{"output_type":"stream","text":["W0721 14:05:40.092331 139995219507072 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_grad.py:1250: add_dispatch_support.<locals>.wrapper (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version.\n","Instructions for updating:\n","Use tf.where in 2.0, which has the same broadcast rule as np.where\n"],"name":"stderr"},{"output_type":"stream","text":["Epoch 1/500\n","6510/6510 [==============================] - 10s 2ms/step - loss: 8.2992 - acc: 0.0581\n","Epoch 2/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 6.7376 - acc: 0.0614\n","Epoch 3/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 6.4581 - acc: 0.0584\n","Epoch 4/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 6.2733 - acc: 0.0594\n","Epoch 5/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 6.1360 - acc: 0.0625\n","Epoch 6/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 5.9883 - acc: 0.0630\n","Epoch 7/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 5.8559 - acc: 0.0668\n","Epoch 8/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 5.7357 - acc: 0.0705\n","Epoch 9/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 5.6345 - acc: 0.0757\n","Epoch 10/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 5.5578 - acc: 0.0782\n","Epoch 11/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 5.4422 - acc: 0.0848\n","Epoch 12/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 5.3001 - acc: 0.0909\n","Epoch 13/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 5.1521 - acc: 0.1078\n","Epoch 14/500\n","6510/6510 [==============================] - 5s 802us/step - loss: 4.9761 - acc: 0.1269\n","Epoch 15/500\n","6510/6510 [==============================] - 5s 822us/step - loss: 4.8152 - acc: 0.1504\n","Epoch 16/500\n","6510/6510 [==============================] - 5s 810us/step - loss: 4.6729 - acc: 0.1740\n","Epoch 17/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 4.5115 - acc: 0.1948\n","Epoch 18/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 4.3792 - acc: 0.2178\n","Epoch 19/500\n","6510/6510 [==============================] - 5s 816us/step - loss: 4.2756 - acc: 0.2338\n","Epoch 20/500\n","6510/6510 [==============================] - 5s 802us/step - loss: 4.1668 - acc: 0.2499\n","Epoch 21/500\n","6510/6510 [==============================] - 5s 800us/step - loss: 4.0427 - acc: 0.2731\n","Epoch 22/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 3.9292 - acc: 0.2820\n","Epoch 23/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 3.8202 - acc: 0.3031\n","Epoch 24/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 3.7112 - acc: 0.3178\n","Epoch 25/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 3.6081 - acc: 0.3352\n","Epoch 26/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 3.5163 - acc: 0.3441\n","Epoch 27/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 3.4384 - acc: 0.3499\n","Epoch 28/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 3.3673 - acc: 0.3608\n","Epoch 29/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 3.2993 - acc: 0.3685\n","Epoch 30/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 3.1905 - acc: 0.3863\n","Epoch 31/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 3.1365 - acc: 0.3889\n","Epoch 32/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 3.0756 - acc: 0.3916\n","Epoch 33/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 2.9917 - acc: 0.4017\n","Epoch 34/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 2.9196 - acc: 0.4131\n","Epoch 35/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 2.8465 - acc: 0.4184\n","Epoch 36/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 2.8098 - acc: 0.4195\n","Epoch 37/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 2.7337 - acc: 0.4315\n","Epoch 38/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 2.6803 - acc: 0.4324\n","Epoch 39/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 2.6203 - acc: 0.4430\n","Epoch 40/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 2.5462 - acc: 0.4587\n","Epoch 41/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 2.4817 - acc: 0.4636\n","Epoch 42/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 2.4302 - acc: 0.4736\n","Epoch 43/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 2.3916 - acc: 0.4736\n","Epoch 44/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 2.3453 - acc: 0.4751\n","Epoch 45/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 2.2699 - acc: 0.4908\n","Epoch 46/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 2.2453 - acc: 0.4916\n","Epoch 47/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 2.2009 - acc: 0.4888\n","Epoch 48/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 2.1370 - acc: 0.5020\n","Epoch 49/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 2.0956 - acc: 0.5083\n","Epoch 50/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 2.0428 - acc: 0.5195\n","Epoch 51/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 1.9765 - acc: 0.5336\n","Epoch 52/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 1.9454 - acc: 0.5330\n","Epoch 53/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 1.9092 - acc: 0.5435\n","Epoch 54/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 1.8908 - acc: 0.5459\n","Epoch 55/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 1.8195 - acc: 0.5576\n","Epoch 56/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 1.7997 - acc: 0.5614\n","Epoch 57/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 1.7782 - acc: 0.5644\n","Epoch 58/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 1.7401 - acc: 0.5747\n","Epoch 59/500\n","6510/6510 [==============================] - 5s 806us/step - loss: 1.6960 - acc: 0.5802\n","Epoch 60/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 1.6672 - acc: 0.5813\n","Epoch 61/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 1.6268 - acc: 0.5902\n","Epoch 62/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 1.5941 - acc: 0.5972\n","Epoch 63/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 1.5833 - acc: 0.6015\n","Epoch 64/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 1.5338 - acc: 0.6029\n","Epoch 65/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 1.4979 - acc: 0.6164\n","Epoch 66/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 1.4958 - acc: 0.6111\n","Epoch 67/500\n","6510/6510 [==============================] - 5s 800us/step - loss: 1.5080 - acc: 0.6092\n","Epoch 68/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 1.4511 - acc: 0.6218\n","Epoch 69/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 1.4362 - acc: 0.6220\n","Epoch 70/500\n","6510/6510 [==============================] - 5s 800us/step - loss: 1.3991 - acc: 0.6359\n","Epoch 71/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 1.3663 - acc: 0.6384\n","Epoch 72/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 1.3607 - acc: 0.6364\n","Epoch 73/500\n","6510/6510 [==============================] - 5s 805us/step - loss: 1.3599 - acc: 0.6398\n","Epoch 74/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 1.3301 - acc: 0.6467\n","Epoch 75/500\n","6510/6510 [==============================] - 5s 815us/step - loss: 1.2935 - acc: 0.6555\n","Epoch 76/500\n","6510/6510 [==============================] - 5s 807us/step - loss: 1.2757 - acc: 0.6535\n","Epoch 77/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 1.2632 - acc: 0.6585\n","Epoch 78/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 1.2721 - acc: 0.6558\n","Epoch 79/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 1.2367 - acc: 0.6624\n","Epoch 80/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 1.2034 - acc: 0.6760\n","Epoch 81/500\n","6510/6510 [==============================] - 5s 800us/step - loss: 1.2129 - acc: 0.6733\n","Epoch 82/500\n","6510/6510 [==============================] - 5s 807us/step - loss: 1.1805 - acc: 0.6777\n","Epoch 83/500\n","6510/6510 [==============================] - 5s 807us/step - loss: 1.1731 - acc: 0.6790\n","Epoch 84/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 1.1602 - acc: 0.6868\n","Epoch 85/500\n","6510/6510 [==============================] - 5s 808us/step - loss: 1.1599 - acc: 0.6796\n","Epoch 86/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 1.1567 - acc: 0.6854\n","Epoch 87/500\n","6510/6510 [==============================] - 5s 823us/step - loss: 1.1266 - acc: 0.6903\n","Epoch 88/500\n","6510/6510 [==============================] - 5s 819us/step - loss: 1.1353 - acc: 0.6833\n","Epoch 89/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 1.1134 - acc: 0.6920\n","Epoch 90/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 1.0843 - acc: 0.6998\n","Epoch 91/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 1.0798 - acc: 0.6972\n","Epoch 92/500\n","6510/6510 [==============================] - 5s 819us/step - loss: 1.0864 - acc: 0.7025\n","Epoch 93/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 1.0840 - acc: 0.7000\n","Epoch 94/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 1.0496 - acc: 0.7097\n","Epoch 95/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 1.0449 - acc: 0.7081\n","Epoch 96/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 1.0131 - acc: 0.7158\n","Epoch 97/500\n","6510/6510 [==============================] - 5s 826us/step - loss: 1.0463 - acc: 0.7126\n","Epoch 98/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.9911 - acc: 0.7195\n","Epoch 99/500\n","6510/6510 [==============================] - 5s 815us/step - loss: 1.0005 - acc: 0.7187\n","Epoch 100/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.9908 - acc: 0.7230\n","Epoch 101/500\n","6510/6510 [==============================] - 5s 822us/step - loss: 0.9746 - acc: 0.7197\n","Epoch 102/500\n","6510/6510 [==============================] - 5s 809us/step - loss: 0.9615 - acc: 0.7289\n","Epoch 103/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.9819 - acc: 0.7232\n","Epoch 104/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.9770 - acc: 0.7197\n","Epoch 105/500\n","6510/6510 [==============================] - 5s 824us/step - loss: 0.9419 - acc: 0.7250\n","Epoch 106/500\n","6510/6510 [==============================] - 5s 809us/step - loss: 0.9282 - acc: 0.7327\n","Epoch 107/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.9396 - acc: 0.7316\n","Epoch 108/500\n","6510/6510 [==============================] - 5s 828us/step - loss: 0.9216 - acc: 0.7367\n","Epoch 109/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.9078 - acc: 0.7441\n","Epoch 110/500\n","6510/6510 [==============================] - 5s 826us/step - loss: 0.8929 - acc: 0.7433\n","Epoch 111/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.9257 - acc: 0.7352\n","Epoch 112/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.9024 - acc: 0.7435\n","Epoch 113/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.8889 - acc: 0.7469\n","Epoch 114/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.8830 - acc: 0.7436\n","Epoch 115/500\n","6510/6510 [==============================] - 5s 767us/step - loss: 0.8587 - acc: 0.7484\n","Epoch 116/500\n","6510/6510 [==============================] - 5s 766us/step - loss: 0.8721 - acc: 0.7479\n","Epoch 117/500\n","6510/6510 [==============================] - 5s 766us/step - loss: 0.8600 - acc: 0.7478\n","Epoch 118/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.8743 - acc: 0.7487\n","Epoch 119/500\n","6510/6510 [==============================] - 5s 765us/step - loss: 0.8679 - acc: 0.7478\n","Epoch 120/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.8308 - acc: 0.7538\n","Epoch 121/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.8335 - acc: 0.7570\n","Epoch 122/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.8332 - acc: 0.7622\n","Epoch 123/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.8153 - acc: 0.7607\n","Epoch 124/500\n","6510/6510 [==============================] - 5s 771us/step - loss: 0.8418 - acc: 0.7550\n","Epoch 125/500\n","6510/6510 [==============================] - 5s 765us/step - loss: 0.7991 - acc: 0.7668\n","Epoch 126/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.7991 - acc: 0.7667\n","Epoch 127/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.8057 - acc: 0.7515\n","Epoch 128/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.8012 - acc: 0.7642\n","Epoch 129/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.8069 - acc: 0.7654\n","Epoch 130/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.7772 - acc: 0.7690\n","Epoch 131/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.7640 - acc: 0.7711\n","Epoch 132/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.7588 - acc: 0.7739\n","Epoch 133/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.7729 - acc: 0.7690\n","Epoch 134/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.7657 - acc: 0.7753\n","Epoch 135/500\n","6510/6510 [==============================] - 5s 807us/step - loss: 0.7602 - acc: 0.7716\n","Epoch 136/500\n","6510/6510 [==============================] - 5s 803us/step - loss: 0.7624 - acc: 0.7742\n","Epoch 137/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.7544 - acc: 0.7779\n","Epoch 138/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.7576 - acc: 0.7767\n","Epoch 139/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.7345 - acc: 0.7874\n","Epoch 140/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.7359 - acc: 0.7773\n","Epoch 141/500\n","6510/6510 [==============================] - 5s 808us/step - loss: 0.7177 - acc: 0.7871\n","Epoch 142/500\n","6510/6510 [==============================] - 5s 768us/step - loss: 0.7384 - acc: 0.7833\n","Epoch 143/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.7233 - acc: 0.7866\n","Epoch 144/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.7033 - acc: 0.7865\n","Epoch 145/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.7138 - acc: 0.7912\n","Epoch 146/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.7087 - acc: 0.7899\n","Epoch 147/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.7213 - acc: 0.7856\n","Epoch 148/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.6930 - acc: 0.7903\n","Epoch 149/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.6903 - acc: 0.7917\n","Epoch 150/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.6794 - acc: 0.8017\n","Epoch 151/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.7020 - acc: 0.7912\n","Epoch 152/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.6795 - acc: 0.7991\n","Epoch 153/500\n","6510/6510 [==============================] - 5s 764us/step - loss: 0.6906 - acc: 0.7919\n","Epoch 154/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.7005 - acc: 0.7946\n","Epoch 155/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.6719 - acc: 0.7983\n","Epoch 156/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.6989 - acc: 0.7896\n","Epoch 157/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.6615 - acc: 0.7974\n","Epoch 158/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.6758 - acc: 0.7998\n","Epoch 159/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.6744 - acc: 0.7982\n","Epoch 160/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.6602 - acc: 0.8072\n","Epoch 161/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.6634 - acc: 0.7998\n","Epoch 162/500\n","6510/6510 [==============================] - 5s 768us/step - loss: 0.6561 - acc: 0.8084\n","Epoch 163/500\n","6510/6510 [==============================] - 5s 760us/step - loss: 0.6544 - acc: 0.8032\n","Epoch 164/500\n","6510/6510 [==============================] - 5s 752us/step - loss: 0.6571 - acc: 0.8055\n","Epoch 165/500\n","6510/6510 [==============================] - 5s 760us/step - loss: 0.6445 - acc: 0.8032\n","Epoch 166/500\n","6510/6510 [==============================] - 5s 758us/step - loss: 0.6595 - acc: 0.8052\n","Epoch 167/500\n","6510/6510 [==============================] - 5s 762us/step - loss: 0.6299 - acc: 0.8094\n","Epoch 168/500\n","6510/6510 [==============================] - 5s 759us/step - loss: 0.6498 - acc: 0.8043\n","Epoch 169/500\n","6510/6510 [==============================] - 5s 770us/step - loss: 0.6417 - acc: 0.8081\n","Epoch 170/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.6339 - acc: 0.8103\n","Epoch 171/500\n","6510/6510 [==============================] - 5s 761us/step - loss: 0.5851 - acc: 0.8192\n","Epoch 172/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.6233 - acc: 0.8140\n","Epoch 173/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.6332 - acc: 0.8104\n","Epoch 174/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.6166 - acc: 0.8132\n","Epoch 175/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.6089 - acc: 0.8206\n","Epoch 176/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.6213 - acc: 0.8138\n","Epoch 177/500\n","6510/6510 [==============================] - 5s 803us/step - loss: 0.6154 - acc: 0.8138\n","Epoch 178/500\n","6510/6510 [==============================] - 5s 815us/step - loss: 0.6223 - acc: 0.8155\n","Epoch 179/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.5856 - acc: 0.8204\n","Epoch 180/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 0.6167 - acc: 0.8112\n","Epoch 181/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.6063 - acc: 0.8198\n","Epoch 182/500\n","6510/6510 [==============================] - 5s 765us/step - loss: 0.6081 - acc: 0.8164\n","Epoch 183/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.6101 - acc: 0.8161\n","Epoch 184/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.5925 - acc: 0.8157\n","Epoch 185/500\n","6510/6510 [==============================] - 5s 758us/step - loss: 0.6036 - acc: 0.8189\n","Epoch 186/500\n","6510/6510 [==============================] - 5s 768us/step - loss: 0.5835 - acc: 0.8212\n","Epoch 187/500\n","6510/6510 [==============================] - 5s 767us/step - loss: 0.5975 - acc: 0.8174\n","Epoch 188/500\n","6510/6510 [==============================] - 5s 767us/step - loss: 0.5771 - acc: 0.8243\n","Epoch 189/500\n","6510/6510 [==============================] - 5s 764us/step - loss: 0.5905 - acc: 0.8218\n","Epoch 190/500\n","6510/6510 [==============================] - 5s 766us/step - loss: 0.5918 - acc: 0.8221\n","Epoch 191/500\n","6510/6510 [==============================] - 5s 763us/step - loss: 0.5885 - acc: 0.8227\n","Epoch 192/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.5857 - acc: 0.8246\n","Epoch 193/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.5673 - acc: 0.8258\n","Epoch 194/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.5695 - acc: 0.8266\n","Epoch 195/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.5676 - acc: 0.8252\n","Epoch 196/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.5872 - acc: 0.8195\n","Epoch 197/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.5727 - acc: 0.8284\n","Epoch 198/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 0.5654 - acc: 0.8237\n","Epoch 199/500\n","6510/6510 [==============================] - 5s 768us/step - loss: 0.5571 - acc: 0.8303\n","Epoch 200/500\n","6510/6510 [==============================] - 5s 762us/step - loss: 0.5548 - acc: 0.8304\n","Epoch 201/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.5612 - acc: 0.8309\n","Epoch 202/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.5523 - acc: 0.8332\n","Epoch 203/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.5409 - acc: 0.8350\n","Epoch 204/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.5600 - acc: 0.8264\n","Epoch 205/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.5582 - acc: 0.8310\n","Epoch 206/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.5408 - acc: 0.8329\n","Epoch 207/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.5310 - acc: 0.8353\n","Epoch 208/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.5486 - acc: 0.8339\n","Epoch 209/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.5633 - acc: 0.8321\n","Epoch 210/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.5482 - acc: 0.8323\n","Epoch 211/500\n","6510/6510 [==============================] - 5s 768us/step - loss: 0.5482 - acc: 0.8283\n","Epoch 212/500\n","6510/6510 [==============================] - 5s 770us/step - loss: 0.5282 - acc: 0.8366\n","Epoch 213/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.5345 - acc: 0.8333\n","Epoch 214/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.5435 - acc: 0.8359\n","Epoch 215/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.5480 - acc: 0.8367\n","Epoch 216/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.5235 - acc: 0.8349\n","Epoch 217/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.5487 - acc: 0.8296\n","Epoch 218/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.5298 - acc: 0.8389\n","Epoch 219/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.5379 - acc: 0.8376\n","Epoch 220/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.5309 - acc: 0.8359\n","Epoch 221/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.5217 - acc: 0.8402\n","Epoch 222/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.5080 - acc: 0.8455\n","Epoch 223/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.5234 - acc: 0.8361\n","Epoch 224/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 0.5133 - acc: 0.8401\n","Epoch 225/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.4984 - acc: 0.8492\n","Epoch 226/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.5194 - acc: 0.8415\n","Epoch 227/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.5011 - acc: 0.8433\n","Epoch 228/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.5141 - acc: 0.8404\n","Epoch 229/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.5046 - acc: 0.8461\n","Epoch 230/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.5012 - acc: 0.8459\n","Epoch 231/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.5121 - acc: 0.8406\n","Epoch 232/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 0.4938 - acc: 0.8445\n","Epoch 233/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.5134 - acc: 0.8425\n","Epoch 234/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.5143 - acc: 0.8467\n","Epoch 235/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.5057 - acc: 0.8456\n","Epoch 236/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.4964 - acc: 0.8464\n","Epoch 237/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.4916 - acc: 0.8456\n","Epoch 238/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 0.4859 - acc: 0.8470\n","Epoch 239/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.5068 - acc: 0.8401\n","Epoch 240/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.4874 - acc: 0.8484\n","Epoch 241/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.4973 - acc: 0.8502\n","Epoch 242/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.4956 - acc: 0.8473\n","Epoch 243/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.5080 - acc: 0.8402\n","Epoch 244/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.4798 - acc: 0.8513\n","Epoch 245/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.5089 - acc: 0.8402\n","Epoch 246/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.5009 - acc: 0.8508\n","Epoch 247/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.5012 - acc: 0.8445\n","Epoch 248/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.4727 - acc: 0.8522\n","Epoch 249/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.4854 - acc: 0.8450\n","Epoch 250/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.4840 - acc: 0.8512\n","Epoch 251/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.4778 - acc: 0.8496\n","Epoch 252/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.4740 - acc: 0.8582\n","Epoch 253/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.4811 - acc: 0.8467\n","Epoch 254/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.4740 - acc: 0.8564\n","Epoch 255/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.4703 - acc: 0.8541\n","Epoch 256/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.4886 - acc: 0.8507\n","Epoch 257/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 0.4689 - acc: 0.8524\n","Epoch 258/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.4884 - acc: 0.8488\n","Epoch 259/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.4593 - acc: 0.8547\n","Epoch 260/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.4601 - acc: 0.8584\n","Epoch 261/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.4670 - acc: 0.8564\n","Epoch 262/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.4697 - acc: 0.8531\n","Epoch 263/500\n","6510/6510 [==============================] - 5s 771us/step - loss: 0.4542 - acc: 0.8588\n","Epoch 264/500\n","6510/6510 [==============================] - 5s 759us/step - loss: 0.4638 - acc: 0.8576\n","Epoch 265/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.4386 - acc: 0.8625\n","Epoch 266/500\n","6510/6510 [==============================] - 5s 763us/step - loss: 0.4686 - acc: 0.8524\n","Epoch 267/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.4874 - acc: 0.8487\n","Epoch 268/500\n","6510/6510 [==============================] - 5s 768us/step - loss: 0.4655 - acc: 0.8499\n","Epoch 269/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.4582 - acc: 0.8594\n","Epoch 270/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.4519 - acc: 0.8594\n","Epoch 271/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.4535 - acc: 0.8590\n","Epoch 272/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.4557 - acc: 0.8625\n","Epoch 273/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.4523 - acc: 0.8591\n","Epoch 274/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.4518 - acc: 0.8591\n","Epoch 275/500\n","6510/6510 [==============================] - 5s 766us/step - loss: 0.4471 - acc: 0.8619\n","Epoch 276/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.4713 - acc: 0.8515\n","Epoch 277/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.4420 - acc: 0.8562\n","Epoch 278/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.4391 - acc: 0.8607\n","Epoch 279/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.4500 - acc: 0.8602\n","Epoch 280/500\n","6510/6510 [==============================] - 5s 771us/step - loss: 0.4448 - acc: 0.8598\n","Epoch 281/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.4335 - acc: 0.8650\n","Epoch 282/500\n","6510/6510 [==============================] - 5s 770us/step - loss: 0.4423 - acc: 0.8662\n","Epoch 283/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.4444 - acc: 0.8647\n","Epoch 284/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.4455 - acc: 0.8610\n","Epoch 285/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.4333 - acc: 0.8667\n","Epoch 286/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.4554 - acc: 0.8573\n","Epoch 287/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.4462 - acc: 0.8639\n","Epoch 288/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.4364 - acc: 0.8673\n","Epoch 289/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.4327 - acc: 0.8656\n","Epoch 290/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.4318 - acc: 0.8644\n","Epoch 291/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.4352 - acc: 0.8654\n","Epoch 292/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.4326 - acc: 0.8684\n","Epoch 293/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.4285 - acc: 0.8644\n","Epoch 294/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.4377 - acc: 0.8639\n","Epoch 295/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.4449 - acc: 0.8639\n","Epoch 296/500\n","6510/6510 [==============================] - 5s 770us/step - loss: 0.4436 - acc: 0.8601\n","Epoch 297/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.4284 - acc: 0.8685\n","Epoch 298/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.4316 - acc: 0.8671\n","Epoch 299/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.4322 - acc: 0.8668\n","Epoch 300/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.4191 - acc: 0.8697\n","Epoch 301/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.4110 - acc: 0.8751\n","Epoch 302/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.4463 - acc: 0.8604\n","Epoch 303/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.4163 - acc: 0.8696\n","Epoch 304/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.4110 - acc: 0.8713\n","Epoch 305/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.4202 - acc: 0.8674\n","Epoch 306/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.4090 - acc: 0.8734\n","Epoch 307/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.4052 - acc: 0.8740\n","Epoch 308/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.4352 - acc: 0.8708\n","Epoch 309/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.4138 - acc: 0.8754\n","Epoch 310/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.4259 - acc: 0.8684\n","Epoch 311/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.4086 - acc: 0.8731\n","Epoch 312/500\n","6510/6510 [==============================] - 5s 770us/step - loss: 0.4126 - acc: 0.8728\n","Epoch 313/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.4235 - acc: 0.8690\n","Epoch 314/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.4155 - acc: 0.8677\n","Epoch 315/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.4121 - acc: 0.8685\n","Epoch 316/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.4178 - acc: 0.8697\n","Epoch 317/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.4098 - acc: 0.8708\n","Epoch 318/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.4076 - acc: 0.8768\n","Epoch 319/500\n","6510/6510 [==============================] - 5s 800us/step - loss: 0.4044 - acc: 0.8768\n","Epoch 320/500\n","6510/6510 [==============================] - 5s 838us/step - loss: 0.4240 - acc: 0.8647\n","Epoch 321/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.4029 - acc: 0.8753\n","Epoch 322/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.4240 - acc: 0.8673\n","Epoch 323/500\n","6510/6510 [==============================] - 5s 768us/step - loss: 0.4063 - acc: 0.8740\n","Epoch 324/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.3981 - acc: 0.8748\n","Epoch 325/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.3917 - acc: 0.8779\n","Epoch 326/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.4142 - acc: 0.8690\n","Epoch 327/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.4131 - acc: 0.8676\n","Epoch 328/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.4066 - acc: 0.8708\n","Epoch 329/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.3946 - acc: 0.8786\n","Epoch 330/500\n","6510/6510 [==============================] - 5s 764us/step - loss: 0.3918 - acc: 0.8780\n","Epoch 331/500\n","6510/6510 [==============================] - 5s 766us/step - loss: 0.3996 - acc: 0.8743\n","Epoch 332/500\n","6510/6510 [==============================] - 5s 771us/step - loss: 0.3920 - acc: 0.8803\n","Epoch 333/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.3855 - acc: 0.8788\n","Epoch 334/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.3885 - acc: 0.8779\n","Epoch 335/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.4112 - acc: 0.8671\n","Epoch 336/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 0.4015 - acc: 0.8745\n","Epoch 337/500\n","6510/6510 [==============================] - 5s 770us/step - loss: 0.4016 - acc: 0.8765\n","Epoch 338/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.3944 - acc: 0.8782\n","Epoch 339/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.3992 - acc: 0.8720\n","Epoch 340/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.3978 - acc: 0.8720\n","Epoch 341/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.3919 - acc: 0.8783\n","Epoch 342/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.3824 - acc: 0.8805\n","Epoch 343/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.3931 - acc: 0.8768\n","Epoch 344/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.3936 - acc: 0.8759\n","Epoch 345/500\n","6510/6510 [==============================] - 5s 771us/step - loss: 0.3970 - acc: 0.8734\n","Epoch 346/500\n","6510/6510 [==============================] - 5s 770us/step - loss: 0.3686 - acc: 0.8799\n","Epoch 347/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.3819 - acc: 0.8806\n","Epoch 348/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.3883 - acc: 0.8806\n","Epoch 349/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.3995 - acc: 0.8762\n","Epoch 350/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.3924 - acc: 0.8745\n","Epoch 351/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.4045 - acc: 0.8760\n","Epoch 352/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.3900 - acc: 0.8724\n","Epoch 353/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.4008 - acc: 0.8771\n","Epoch 354/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.3856 - acc: 0.8846\n","Epoch 355/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.4006 - acc: 0.8756\n","Epoch 356/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.3975 - acc: 0.8802\n","Epoch 357/500\n","6510/6510 [==============================] - 5s 771us/step - loss: 0.3709 - acc: 0.8860\n","Epoch 358/500\n","6510/6510 [==============================] - 5s 768us/step - loss: 0.3666 - acc: 0.8810\n","Epoch 359/500\n","6510/6510 [==============================] - 5s 800us/step - loss: 0.4020 - acc: 0.8693\n","Epoch 360/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.3829 - acc: 0.8786\n","Epoch 361/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.3793 - acc: 0.8799\n","Epoch 362/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.3712 - acc: 0.8826\n","Epoch 363/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.3832 - acc: 0.8813\n","Epoch 364/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.3829 - acc: 0.8799\n","Epoch 365/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.3758 - acc: 0.8808\n","Epoch 366/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.3720 - acc: 0.8860\n","Epoch 367/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.3854 - acc: 0.8777\n","Epoch 368/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 0.3922 - acc: 0.8771\n","Epoch 369/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.3930 - acc: 0.8814\n","Epoch 370/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.3916 - acc: 0.8771\n","Epoch 371/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.3818 - acc: 0.8796\n","Epoch 372/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.3793 - acc: 0.8805\n","Epoch 373/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.3766 - acc: 0.8820\n","Epoch 374/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.3728 - acc: 0.8840\n","Epoch 375/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.3734 - acc: 0.8837\n","Epoch 376/500\n","6510/6510 [==============================] - 5s 770us/step - loss: 0.3853 - acc: 0.8776\n","Epoch 377/500\n","6510/6510 [==============================] - 5s 762us/step - loss: 0.3770 - acc: 0.8831\n","Epoch 378/500\n","6510/6510 [==============================] - 5s 767us/step - loss: 0.3827 - acc: 0.8805\n","Epoch 379/500\n","6510/6510 [==============================] - 5s 768us/step - loss: 0.3902 - acc: 0.8760\n","Epoch 380/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.3689 - acc: 0.8868\n","Epoch 381/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 0.3772 - acc: 0.8863\n","Epoch 382/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.3528 - acc: 0.8905\n","Epoch 383/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.3604 - acc: 0.8899\n","Epoch 384/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.3677 - acc: 0.8849\n","Epoch 385/500\n","6510/6510 [==============================] - 5s 759us/step - loss: 0.3785 - acc: 0.8780\n","Epoch 386/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.3618 - acc: 0.8874\n","Epoch 387/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.3516 - acc: 0.8869\n","Epoch 388/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.3715 - acc: 0.8820\n","Epoch 389/500\n","6510/6510 [==============================] - 5s 771us/step - loss: 0.3383 - acc: 0.8922\n","Epoch 390/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.3735 - acc: 0.8823\n","Epoch 391/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.3599 - acc: 0.8856\n","Epoch 392/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 0.3516 - acc: 0.8903\n","Epoch 393/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.3619 - acc: 0.8854\n","Epoch 394/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.3569 - acc: 0.8877\n","Epoch 395/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 0.3935 - acc: 0.8767\n","Epoch 396/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.3696 - acc: 0.8868\n","Epoch 397/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.3557 - acc: 0.8905\n","Epoch 398/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.3678 - acc: 0.8863\n","Epoch 399/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.3481 - acc: 0.8920\n","Epoch 400/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.3665 - acc: 0.8849\n","Epoch 401/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.3686 - acc: 0.8879\n","Epoch 402/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.3782 - acc: 0.8865\n","Epoch 403/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.3656 - acc: 0.8889\n","Epoch 404/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.3454 - acc: 0.8965\n","Epoch 405/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.3660 - acc: 0.8819\n","Epoch 406/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.3514 - acc: 0.8896\n","Epoch 407/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.3228 - acc: 0.8974\n","Epoch 408/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.3403 - acc: 0.8929\n","Epoch 409/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.3614 - acc: 0.8891\n","Epoch 410/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.3480 - acc: 0.8949\n","Epoch 411/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.3440 - acc: 0.8922\n","Epoch 412/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.3539 - acc: 0.8888\n","Epoch 413/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.3419 - acc: 0.8925\n","Epoch 414/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.3598 - acc: 0.8834\n","Epoch 415/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 0.3428 - acc: 0.8902\n","Epoch 416/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.3451 - acc: 0.8932\n","Epoch 417/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.3475 - acc: 0.8909\n","Epoch 418/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.3504 - acc: 0.8888\n","Epoch 419/500\n","6510/6510 [==============================] - 5s 801us/step - loss: 0.3498 - acc: 0.8889\n","Epoch 420/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 0.3547 - acc: 0.8889\n","Epoch 421/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.3491 - acc: 0.8902\n","Epoch 422/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 0.3373 - acc: 0.8940\n","Epoch 423/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.3626 - acc: 0.8848\n","Epoch 424/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.3331 - acc: 0.8940\n","Epoch 425/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.3429 - acc: 0.8928\n","Epoch 426/500\n","6510/6510 [==============================] - 5s 771us/step - loss: 0.3563 - acc: 0.8842\n","Epoch 427/500\n","6510/6510 [==============================] - 5s 770us/step - loss: 0.3515 - acc: 0.8923\n","Epoch 428/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.3552 - acc: 0.8912\n","Epoch 429/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.3459 - acc: 0.8932\n","Epoch 430/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 0.3550 - acc: 0.8882\n","Epoch 431/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 0.3538 - acc: 0.8897\n","Epoch 432/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.3538 - acc: 0.8876\n","Epoch 433/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.3486 - acc: 0.8905\n","Epoch 434/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.3463 - acc: 0.8912\n","Epoch 435/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.3501 - acc: 0.8926\n","Epoch 436/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.3308 - acc: 0.8963\n","Epoch 437/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.3518 - acc: 0.8903\n","Epoch 438/500\n","6510/6510 [==============================] - 5s 770us/step - loss: 0.3415 - acc: 0.8906\n","Epoch 439/500\n","6510/6510 [==============================] - 5s 771us/step - loss: 0.3386 - acc: 0.8957\n","Epoch 440/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.3329 - acc: 0.9005\n","Epoch 441/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.3366 - acc: 0.8925\n","Epoch 442/500\n","6510/6510 [==============================] - 5s 806us/step - loss: 0.3356 - acc: 0.8942\n","Epoch 443/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.3469 - acc: 0.8914\n","Epoch 444/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.3459 - acc: 0.8923\n","Epoch 445/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.3455 - acc: 0.8931\n","Epoch 446/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.3377 - acc: 0.8960\n","Epoch 447/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.3427 - acc: 0.8937\n","Epoch 448/500\n","6510/6510 [==============================] - 5s 765us/step - loss: 0.3381 - acc: 0.8923\n","Epoch 449/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.3757 - acc: 0.8813\n","Epoch 450/500\n","6510/6510 [==============================] - 5s 768us/step - loss: 0.3454 - acc: 0.8916\n","Epoch 451/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.3355 - acc: 0.8912\n","Epoch 452/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.3394 - acc: 0.8959\n","Epoch 453/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.3411 - acc: 0.8940\n","Epoch 454/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.3446 - acc: 0.8900\n","Epoch 455/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.3383 - acc: 0.8948\n","Epoch 456/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.3224 - acc: 0.9012\n","Epoch 457/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.3312 - acc: 0.8959\n","Epoch 458/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.3103 - acc: 0.9028\n","Epoch 459/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.3409 - acc: 0.8948\n","Epoch 460/500\n","6510/6510 [==============================] - 5s 771us/step - loss: 0.3392 - acc: 0.8959\n","Epoch 461/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.3361 - acc: 0.8955\n","Epoch 462/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.3276 - acc: 0.8975\n","Epoch 463/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.3414 - acc: 0.8931\n","Epoch 464/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.3278 - acc: 0.8969\n","Epoch 465/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.3378 - acc: 0.8980\n","Epoch 466/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.3327 - acc: 0.8980\n","Epoch 467/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.3359 - acc: 0.8920\n","Epoch 468/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.3272 - acc: 0.8978\n","Epoch 469/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.3411 - acc: 0.8949\n","Epoch 470/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.3267 - acc: 0.8959\n","Epoch 471/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.3534 - acc: 0.8922\n","Epoch 472/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.3299 - acc: 0.8975\n","Epoch 473/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.3251 - acc: 0.8965\n","Epoch 474/500\n","6510/6510 [==============================] - 5s 770us/step - loss: 0.3386 - acc: 0.8945\n","Epoch 475/500\n","6510/6510 [==============================] - 5s 767us/step - loss: 0.3357 - acc: 0.8949\n","Epoch 476/500\n","6510/6510 [==============================] - 5s 766us/step - loss: 0.3316 - acc: 0.8954\n","Epoch 477/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.3392 - acc: 0.8902\n","Epoch 478/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.3279 - acc: 0.8957\n","Epoch 479/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.3324 - acc: 0.8945\n","Epoch 480/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.3266 - acc: 0.8975\n","Epoch 481/500\n","6510/6510 [==============================] - 5s 767us/step - loss: 0.3241 - acc: 0.8952\n","Epoch 482/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.3217 - acc: 0.8962\n","Epoch 483/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.3175 - acc: 0.9031\n","Epoch 484/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.3360 - acc: 0.8992\n","Epoch 485/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.3261 - acc: 0.8965\n","Epoch 486/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.3382 - acc: 0.8960\n","Epoch 487/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.3330 - acc: 0.8965\n","Epoch 488/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.3259 - acc: 0.9015\n","Epoch 489/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.3137 - acc: 0.9000\n","Epoch 490/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.3258 - acc: 0.8972\n","Epoch 491/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.3138 - acc: 0.9028\n","Epoch 492/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.3220 - acc: 0.9015\n","Epoch 493/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.3255 - acc: 0.8969\n","Epoch 494/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.2996 - acc: 0.9052\n","Epoch 495/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.2825 - acc: 0.9109\n","Epoch 496/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.3203 - acc: 0.8977\n","Epoch 497/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.3259 - acc: 0.8985\n","Epoch 498/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.3222 - acc: 0.8974\n","Epoch 499/500\n","6510/6510 [==============================] - 5s 817us/step - loss: 0.3363 - acc: 0.8991\n","Epoch 500/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.3225 - acc: 0.9022\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"82716QWAJrXG"},"source":["### Save Model"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"_8MFkQgYJm-D","colab":{}},"source":["# save the model to file\n","model.save('.SemiTrainedModel_500epochs.h5')"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"4AYVKydVJv5C"},"source":["## If you have already trained the model and saved it, you can load a pretrained model"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"IqsQUz04J0GP","colab":{}},"source":["# load the model\n","model = load_model('.SemiTrainedModel_500epochs.h5')"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"ZFe2Y0SJJ3Hb"},"source":["### Note: After loading the model run  model.fit()  to continue training form there, if required."]},{"cell_type":"code","metadata":{"colab_type":"code","id":"e9yLm_xnJ5JV","colab":{"base_uri":"https://localhost:8080/","height":1000},"outputId":"e73b16ab-640f-465d-f8ec-c25e5a4b160c","executionInfo":{"status":"ok","timestamp":1563723160626,"user_tz":-330,"elapsed":2589423,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}}},"source":["model.fit(X_train, y_train, batch_size=50, epochs=500)"],"execution_count":18,"outputs":[{"output_type":"stream","text":["Epoch 1/500\n","6510/6510 [==============================] - 6s 919us/step - loss: 0.2926 - acc: 0.9092\n","Epoch 2/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.3189 - acc: 0.8962\n","Epoch 3/500\n","6510/6510 [==============================] - 5s 806us/step - loss: 0.3186 - acc: 0.9000\n","Epoch 4/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 0.3340 - acc: 0.8985\n","Epoch 5/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 0.3042 - acc: 0.9048\n","Epoch 6/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.3156 - acc: 0.9025\n","Epoch 7/500\n","6510/6510 [==============================] - 5s 800us/step - loss: 0.3173 - acc: 0.9031\n","Epoch 8/500\n","6510/6510 [==============================] - 5s 801us/step - loss: 0.3294 - acc: 0.8971\n","Epoch 9/500\n","6510/6510 [==============================] - 5s 812us/step - loss: 0.3341 - acc: 0.8994\n","Epoch 10/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.3233 - acc: 0.8998\n","Epoch 11/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.3192 - acc: 0.8980\n","Epoch 12/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.3260 - acc: 0.9015\n","Epoch 13/500\n","6510/6510 [==============================] - 5s 803us/step - loss: 0.3022 - acc: 0.9097\n","Epoch 14/500\n","6510/6510 [==============================] - 5s 802us/step - loss: 0.3294 - acc: 0.8978\n","Epoch 15/500\n","6510/6510 [==============================] - 5s 803us/step - loss: 0.3158 - acc: 0.8992\n","Epoch 16/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.3110 - acc: 0.9018\n","Epoch 17/500\n","6510/6510 [==============================] - 5s 803us/step - loss: 0.3174 - acc: 0.9000\n","Epoch 18/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.3271 - acc: 0.8972\n","Epoch 19/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.3279 - acc: 0.8966\n","Epoch 20/500\n","6510/6510 [==============================] - 5s 804us/step - loss: 0.3028 - acc: 0.9063\n","Epoch 21/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 0.3083 - acc: 0.9018\n","Epoch 22/500\n","6510/6510 [==============================] - 5s 812us/step - loss: 0.3082 - acc: 0.9011\n","Epoch 23/500\n","6510/6510 [==============================] - 5s 813us/step - loss: 0.3173 - acc: 0.9015\n","Epoch 24/500\n","6510/6510 [==============================] - 5s 808us/step - loss: 0.3115 - acc: 0.9011\n","Epoch 25/500\n","6510/6510 [==============================] - 5s 810us/step - loss: 0.3048 - acc: 0.9063\n","Epoch 26/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.3089 - acc: 0.9035\n","Epoch 27/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.2869 - acc: 0.9140\n","Epoch 28/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.2834 - acc: 0.9101\n","Epoch 29/500\n","6510/6510 [==============================] - 5s 816us/step - loss: 0.3358 - acc: 0.8922\n","Epoch 30/500\n","6510/6510 [==============================] - 6s 889us/step - loss: 0.3139 - acc: 0.8991\n","Epoch 31/500\n","6510/6510 [==============================] - 5s 811us/step - loss: 0.3056 - acc: 0.9043\n","Epoch 32/500\n","6510/6510 [==============================] - 5s 804us/step - loss: 0.3093 - acc: 0.9058\n","Epoch 33/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.2895 - acc: 0.9118\n","Epoch 34/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.3059 - acc: 0.9041\n","Epoch 35/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.3164 - acc: 0.9038\n","Epoch 36/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.3137 - acc: 0.9028\n","Epoch 37/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 0.3195 - acc: 0.8992\n","Epoch 38/500\n","6510/6510 [==============================] - 5s 800us/step - loss: 0.3283 - acc: 0.8982\n","Epoch 39/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.3113 - acc: 0.9023\n","Epoch 40/500\n","6510/6510 [==============================] - 5s 812us/step - loss: 0.3060 - acc: 0.9041\n","Epoch 41/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 0.3057 - acc: 0.9072\n","Epoch 42/500\n","6510/6510 [==============================] - 5s 800us/step - loss: 0.3275 - acc: 0.8960\n","Epoch 43/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.2910 - acc: 0.9078\n","Epoch 44/500\n","6510/6510 [==============================] - 5s 804us/step - loss: 0.3004 - acc: 0.9071\n","Epoch 45/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.3086 - acc: 0.9040\n","Epoch 46/500\n","6510/6510 [==============================] - 5s 815us/step - loss: 0.3166 - acc: 0.8998\n","Epoch 47/500\n","6510/6510 [==============================] - 5s 817us/step - loss: 0.3113 - acc: 0.8997\n","Epoch 48/500\n","6510/6510 [==============================] - 5s 803us/step - loss: 0.3237 - acc: 0.8980\n","Epoch 49/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.3047 - acc: 0.9022\n","Epoch 50/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.3021 - acc: 0.9029\n","Epoch 51/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.3176 - acc: 0.9017\n","Epoch 52/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.3164 - acc: 0.9015\n","Epoch 53/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.3184 - acc: 0.9009\n","Epoch 54/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.2967 - acc: 0.9061\n","Epoch 55/500\n","6510/6510 [==============================] - 5s 804us/step - loss: 0.3302 - acc: 0.8992\n","Epoch 56/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.3127 - acc: 0.9000\n","Epoch 57/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.3080 - acc: 0.9051\n","Epoch 58/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.3182 - acc: 0.9022\n","Epoch 59/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.3069 - acc: 0.9058\n","Epoch 60/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.3144 - acc: 0.9018\n","Epoch 61/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.3084 - acc: 0.9014\n","Epoch 62/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.2812 - acc: 0.9094\n","Epoch 63/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.3040 - acc: 0.9011\n","Epoch 64/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 0.2901 - acc: 0.9104\n","Epoch 65/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.3059 - acc: 0.9068\n","Epoch 66/500\n","6510/6510 [==============================] - 5s 804us/step - loss: 0.2824 - acc: 0.9103\n","Epoch 67/500\n","6510/6510 [==============================] - 5s 803us/step - loss: 0.2957 - acc: 0.9074\n","Epoch 68/500\n","6510/6510 [==============================] - 5s 804us/step - loss: 0.2931 - acc: 0.9086\n","Epoch 69/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.2991 - acc: 0.9058\n","Epoch 70/500\n","6510/6510 [==============================] - 5s 803us/step - loss: 0.2829 - acc: 0.9158\n","Epoch 71/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.2840 - acc: 0.9109\n","Epoch 72/500\n","6510/6510 [==============================] - 5s 805us/step - loss: 0.2904 - acc: 0.9117\n","Epoch 73/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.2902 - acc: 0.9100\n","Epoch 74/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.2799 - acc: 0.9140\n","Epoch 75/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.2931 - acc: 0.9086\n","Epoch 76/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.2967 - acc: 0.9074\n","Epoch 77/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 0.2947 - acc: 0.9091\n","Epoch 78/500\n","6510/6510 [==============================] - 5s 801us/step - loss: 0.2782 - acc: 0.9124\n","Epoch 79/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 0.2917 - acc: 0.9071\n","Epoch 80/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 0.2952 - acc: 0.9071\n","Epoch 81/500\n","6510/6510 [==============================] - 5s 810us/step - loss: 0.2912 - acc: 0.9072\n","Epoch 82/500\n","6510/6510 [==============================] - 5s 813us/step - loss: 0.3041 - acc: 0.9023\n","Epoch 83/500\n","6510/6510 [==============================] - 5s 800us/step - loss: 0.3083 - acc: 0.9060\n","Epoch 84/500\n","6510/6510 [==============================] - 5s 809us/step - loss: 0.2839 - acc: 0.9124\n","Epoch 85/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.2939 - acc: 0.9074\n","Epoch 86/500\n","6510/6510 [==============================] - 5s 811us/step - loss: 0.2921 - acc: 0.9100\n","Epoch 87/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 0.3058 - acc: 0.9052\n","Epoch 88/500\n","6510/6510 [==============================] - 5s 806us/step - loss: 0.2951 - acc: 0.9101\n","Epoch 89/500\n","6510/6510 [==============================] - 5s 807us/step - loss: 0.2990 - acc: 0.9049\n","Epoch 90/500\n","6510/6510 [==============================] - 5s 804us/step - loss: 0.2878 - acc: 0.9104\n","Epoch 91/500\n","6510/6510 [==============================] - 5s 805us/step - loss: 0.3095 - acc: 0.9054\n","Epoch 92/500\n","6510/6510 [==============================] - 5s 804us/step - loss: 0.2982 - acc: 0.9069\n","Epoch 93/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.3032 - acc: 0.9069\n","Epoch 94/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 0.3165 - acc: 0.8971\n","Epoch 95/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.2983 - acc: 0.9094\n","Epoch 96/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.2961 - acc: 0.9084\n","Epoch 97/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 0.3021 - acc: 0.9068\n","Epoch 98/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.2917 - acc: 0.9071\n","Epoch 99/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.3000 - acc: 0.9089\n","Epoch 100/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.2872 - acc: 0.9108\n","Epoch 101/500\n","6510/6510 [==============================] - 5s 802us/step - loss: 0.2952 - acc: 0.9049\n","Epoch 102/500\n","6510/6510 [==============================] - 5s 825us/step - loss: 0.2951 - acc: 0.9080\n","Epoch 103/500\n","6510/6510 [==============================] - 6s 849us/step - loss: 0.2855 - acc: 0.9106\n","Epoch 104/500\n","6510/6510 [==============================] - 5s 831us/step - loss: 0.2881 - acc: 0.9140\n","Epoch 105/500\n","6510/6510 [==============================] - 6s 845us/step - loss: 0.2799 - acc: 0.9126\n","Epoch 106/500\n","6510/6510 [==============================] - 6s 863us/step - loss: 0.2791 - acc: 0.9123\n","Epoch 107/500\n","6510/6510 [==============================] - 5s 833us/step - loss: 0.3025 - acc: 0.9071\n","Epoch 108/500\n","6510/6510 [==============================] - 5s 829us/step - loss: 0.2920 - acc: 0.9108\n","Epoch 109/500\n","6510/6510 [==============================] - 5s 825us/step - loss: 0.3056 - acc: 0.9066\n","Epoch 110/500\n","6510/6510 [==============================] - 5s 825us/step - loss: 0.3041 - acc: 0.9065\n","Epoch 111/500\n","6510/6510 [==============================] - 5s 827us/step - loss: 0.2854 - acc: 0.9138\n","Epoch 112/500\n","6510/6510 [==============================] - 5s 830us/step - loss: 0.2939 - acc: 0.9080\n","Epoch 113/500\n","6510/6510 [==============================] - 5s 807us/step - loss: 0.3015 - acc: 0.9065\n","Epoch 114/500\n","6510/6510 [==============================] - 5s 800us/step - loss: 0.2935 - acc: 0.9075\n","Epoch 115/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.2900 - acc: 0.9080\n","Epoch 116/500\n","6510/6510 [==============================] - 5s 814us/step - loss: 0.2872 - acc: 0.9081\n","Epoch 117/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.2872 - acc: 0.9111\n","Epoch 118/500\n","6510/6510 [==============================] - 5s 808us/step - loss: 0.2794 - acc: 0.9111\n","Epoch 119/500\n","6510/6510 [==============================] - 5s 814us/step - loss: 0.2832 - acc: 0.9111\n","Epoch 120/500\n","6510/6510 [==============================] - 5s 843us/step - loss: 0.2918 - acc: 0.9095\n","Epoch 121/500\n","6510/6510 [==============================] - 5s 838us/step - loss: 0.3074 - acc: 0.8980\n","Epoch 122/500\n","6510/6510 [==============================] - 5s 836us/step - loss: 0.2787 - acc: 0.9135\n","Epoch 123/500\n","6510/6510 [==============================] - 5s 831us/step - loss: 0.2904 - acc: 0.9081\n","Epoch 124/500\n","6510/6510 [==============================] - 5s 839us/step - loss: 0.2917 - acc: 0.9080\n","Epoch 125/500\n","6510/6510 [==============================] - 5s 844us/step - loss: 0.2901 - acc: 0.9104\n","Epoch 126/500\n","6510/6510 [==============================] - 6s 845us/step - loss: 0.2832 - acc: 0.9109\n","Epoch 127/500\n","6510/6510 [==============================] - 5s 826us/step - loss: 0.2855 - acc: 0.9077\n","Epoch 128/500\n","6510/6510 [==============================] - 5s 841us/step - loss: 0.2832 - acc: 0.9063\n","Epoch 129/500\n","6510/6510 [==============================] - 5s 840us/step - loss: 0.2854 - acc: 0.9097\n","Epoch 130/500\n","6510/6510 [==============================] - 6s 857us/step - loss: 0.2879 - acc: 0.9097\n","Epoch 131/500\n","6510/6510 [==============================] - 5s 814us/step - loss: 0.2878 - acc: 0.9077\n","Epoch 132/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.2891 - acc: 0.9115\n","Epoch 133/500\n","6510/6510 [==============================] - 5s 803us/step - loss: 0.2952 - acc: 0.9068\n","Epoch 134/500\n","6510/6510 [==============================] - 5s 819us/step - loss: 0.3047 - acc: 0.9020\n","Epoch 135/500\n","6510/6510 [==============================] - 5s 808us/step - loss: 0.2864 - acc: 0.9086\n","Epoch 136/500\n","6510/6510 [==============================] - 5s 809us/step - loss: 0.3014 - acc: 0.9066\n","Epoch 137/500\n","6510/6510 [==============================] - 5s 811us/step - loss: 0.2893 - acc: 0.9095\n","Epoch 138/500\n","6510/6510 [==============================] - 5s 825us/step - loss: 0.2765 - acc: 0.9121\n","Epoch 139/500\n","6510/6510 [==============================] - 5s 841us/step - loss: 0.2873 - acc: 0.9124\n","Epoch 140/500\n","6510/6510 [==============================] - 5s 821us/step - loss: 0.2938 - acc: 0.9063\n","Epoch 141/500\n","6510/6510 [==============================] - 5s 823us/step - loss: 0.2783 - acc: 0.9132\n","Epoch 142/500\n","6510/6510 [==============================] - 5s 820us/step - loss: 0.2784 - acc: 0.9127\n","Epoch 143/500\n","6510/6510 [==============================] - 5s 814us/step - loss: 0.2885 - acc: 0.9084\n","Epoch 144/500\n","6510/6510 [==============================] - 5s 826us/step - loss: 0.2720 - acc: 0.9144\n","Epoch 145/500\n","6510/6510 [==============================] - 5s 820us/step - loss: 0.2882 - acc: 0.9072\n","Epoch 146/500\n","6510/6510 [==============================] - 5s 826us/step - loss: 0.2932 - acc: 0.9061\n","Epoch 147/500\n","6510/6510 [==============================] - 5s 827us/step - loss: 0.2971 - acc: 0.9054\n","Epoch 148/500\n","6510/6510 [==============================] - 5s 827us/step - loss: 0.2935 - acc: 0.9100\n","Epoch 149/500\n","6510/6510 [==============================] - 5s 822us/step - loss: 0.2813 - acc: 0.9146\n","Epoch 150/500\n","6510/6510 [==============================] - 5s 832us/step - loss: 0.2741 - acc: 0.9112\n","Epoch 151/500\n","6510/6510 [==============================] - 5s 819us/step - loss: 0.2982 - acc: 0.9114\n","Epoch 152/500\n","6510/6510 [==============================] - 5s 822us/step - loss: 0.2767 - acc: 0.9167\n","Epoch 153/500\n","6510/6510 [==============================] - 5s 818us/step - loss: 0.2861 - acc: 0.9135\n","Epoch 154/500\n","6510/6510 [==============================] - 5s 826us/step - loss: 0.2685 - acc: 0.9160\n","Epoch 155/500\n","6510/6510 [==============================] - 5s 822us/step - loss: 0.2797 - acc: 0.9101\n","Epoch 156/500\n","6510/6510 [==============================] - 5s 830us/step - loss: 0.2997 - acc: 0.9046\n","Epoch 157/500\n","6510/6510 [==============================] - 5s 815us/step - loss: 0.2890 - acc: 0.9121\n","Epoch 158/500\n","6510/6510 [==============================] - 5s 802us/step - loss: 0.2903 - acc: 0.9072\n","Epoch 159/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.2861 - acc: 0.9175\n","Epoch 160/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.2728 - acc: 0.9144\n","Epoch 161/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.2913 - acc: 0.9095\n","Epoch 162/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.2789 - acc: 0.9124\n","Epoch 163/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.2896 - acc: 0.9123\n","Epoch 164/500\n","6510/6510 [==============================] - 5s 813us/step - loss: 0.2792 - acc: 0.9114\n","Epoch 165/500\n","6510/6510 [==============================] - 5s 810us/step - loss: 0.2828 - acc: 0.9117\n","Epoch 166/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.2812 - acc: 0.9158\n","Epoch 167/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.2779 - acc: 0.9151\n","Epoch 168/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2856 - acc: 0.9146\n","Epoch 169/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2761 - acc: 0.9134\n","Epoch 170/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.2765 - acc: 0.9131\n","Epoch 171/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 0.2948 - acc: 0.9054\n","Epoch 172/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.2769 - acc: 0.9141\n","Epoch 173/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.2909 - acc: 0.9106\n","Epoch 174/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.2921 - acc: 0.9120\n","Epoch 175/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.2855 - acc: 0.9098\n","Epoch 176/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.2828 - acc: 0.9146\n","Epoch 177/500\n","6510/6510 [==============================] - 5s 802us/step - loss: 0.2830 - acc: 0.9126\n","Epoch 178/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 0.2743 - acc: 0.9147\n","Epoch 179/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 0.2821 - acc: 0.9141\n","Epoch 180/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.2799 - acc: 0.9095\n","Epoch 181/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.2740 - acc: 0.9140\n","Epoch 182/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2836 - acc: 0.9117\n","Epoch 183/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.2781 - acc: 0.9134\n","Epoch 184/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.2818 - acc: 0.9147\n","Epoch 185/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2747 - acc: 0.9154\n","Epoch 186/500\n","6510/6510 [==============================] - 5s 812us/step - loss: 0.2866 - acc: 0.9084\n","Epoch 187/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.2895 - acc: 0.9117\n","Epoch 188/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 0.2803 - acc: 0.9151\n","Epoch 189/500\n","6510/6510 [==============================] - 5s 819us/step - loss: 0.2979 - acc: 0.9098\n","Epoch 190/500\n","6510/6510 [==============================] - 5s 836us/step - loss: 0.2795 - acc: 0.9146\n","Epoch 191/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.2835 - acc: 0.9098\n","Epoch 192/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.2632 - acc: 0.9237\n","Epoch 193/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.2833 - acc: 0.9129\n","Epoch 194/500\n","6510/6510 [==============================] - 5s 801us/step - loss: 0.2770 - acc: 0.9138\n","Epoch 195/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.2763 - acc: 0.9149\n","Epoch 196/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.2584 - acc: 0.9174\n","Epoch 197/500\n","6510/6510 [==============================] - 5s 825us/step - loss: 0.2708 - acc: 0.9174\n","Epoch 198/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 0.2479 - acc: 0.9215\n","Epoch 199/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.2601 - acc: 0.9161\n","Epoch 200/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.2854 - acc: 0.9147\n","Epoch 201/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.2619 - acc: 0.9183\n","Epoch 202/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.2713 - acc: 0.9141\n","Epoch 203/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.2748 - acc: 0.9152\n","Epoch 204/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.2579 - acc: 0.9172\n","Epoch 205/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.2600 - acc: 0.9207\n","Epoch 206/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.2900 - acc: 0.9078\n","Epoch 207/500\n","6510/6510 [==============================] - 5s 801us/step - loss: 0.2782 - acc: 0.9127\n","Epoch 208/500\n","6510/6510 [==============================] - 5s 804us/step - loss: 0.2699 - acc: 0.9137\n","Epoch 209/500\n","6510/6510 [==============================] - 5s 802us/step - loss: 0.2862 - acc: 0.9124\n","Epoch 210/500\n","6510/6510 [==============================] - 5s 805us/step - loss: 0.2786 - acc: 0.9115\n","Epoch 211/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.2699 - acc: 0.9143\n","Epoch 212/500\n","6510/6510 [==============================] - 5s 800us/step - loss: 0.2612 - acc: 0.9189\n","Epoch 213/500\n","6510/6510 [==============================] - 5s 804us/step - loss: 0.2697 - acc: 0.9129\n","Epoch 214/500\n","6510/6510 [==============================] - 5s 803us/step - loss: 0.2628 - acc: 0.9147\n","Epoch 215/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.2902 - acc: 0.9075\n","Epoch 216/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 0.2565 - acc: 0.9177\n","Epoch 217/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.2827 - acc: 0.9135\n","Epoch 218/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.2752 - acc: 0.9167\n","Epoch 219/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 0.2619 - acc: 0.9161\n","Epoch 220/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 0.2801 - acc: 0.9126\n","Epoch 221/500\n","6510/6510 [==============================] - 5s 802us/step - loss: 0.2651 - acc: 0.9178\n","Epoch 222/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 0.2672 - acc: 0.9184\n","Epoch 223/500\n","6510/6510 [==============================] - 5s 811us/step - loss: 0.2615 - acc: 0.9206\n","Epoch 224/500\n","6510/6510 [==============================] - 5s 822us/step - loss: 0.2640 - acc: 0.9160\n","Epoch 225/500\n","6510/6510 [==============================] - 5s 821us/step - loss: 0.2682 - acc: 0.9157\n","Epoch 226/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.2573 - acc: 0.9180\n","Epoch 227/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.2756 - acc: 0.9137\n","Epoch 228/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.2725 - acc: 0.9180\n","Epoch 229/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.2730 - acc: 0.9189\n","Epoch 230/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.2919 - acc: 0.9147\n","Epoch 231/500\n","6510/6510 [==============================] - 5s 808us/step - loss: 0.2596 - acc: 0.9198\n","Epoch 232/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.2584 - acc: 0.9177\n","Epoch 233/500\n","6510/6510 [==============================] - 5s 810us/step - loss: 0.2911 - acc: 0.9083\n","Epoch 234/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.2886 - acc: 0.9097\n","Epoch 235/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.2688 - acc: 0.9186\n","Epoch 236/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.2560 - acc: 0.9183\n","Epoch 237/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.2658 - acc: 0.9207\n","Epoch 238/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.2693 - acc: 0.9186\n","Epoch 239/500\n","6510/6510 [==============================] - 5s 804us/step - loss: 0.2609 - acc: 0.9189\n","Epoch 240/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 0.2768 - acc: 0.9109\n","Epoch 241/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.2697 - acc: 0.9127\n","Epoch 242/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.2740 - acc: 0.9169\n","Epoch 243/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 0.2670 - acc: 0.9172\n","Epoch 244/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 0.2599 - acc: 0.9210\n","Epoch 245/500\n","6510/6510 [==============================] - 5s 800us/step - loss: 0.2607 - acc: 0.9175\n","Epoch 246/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.2541 - acc: 0.9167\n","Epoch 247/500\n","6510/6510 [==============================] - 5s 800us/step - loss: 0.2578 - acc: 0.9194\n","Epoch 248/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.2568 - acc: 0.9160\n","Epoch 249/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 0.2742 - acc: 0.9174\n","Epoch 250/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.2660 - acc: 0.9180\n","Epoch 251/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.2550 - acc: 0.9214\n","Epoch 252/500\n","6510/6510 [==============================] - 5s 803us/step - loss: 0.2582 - acc: 0.9204\n","Epoch 253/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 0.2585 - acc: 0.9186\n","Epoch 254/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 0.2612 - acc: 0.9207\n","Epoch 255/500\n","6510/6510 [==============================] - 5s 803us/step - loss: 0.2751 - acc: 0.9166\n","Epoch 256/500\n","6510/6510 [==============================] - 5s 810us/step - loss: 0.2517 - acc: 0.9230\n","Epoch 257/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.2603 - acc: 0.9181\n","Epoch 258/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 0.2710 - acc: 0.9140\n","Epoch 259/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.2747 - acc: 0.9121\n","Epoch 260/500\n","6510/6510 [==============================] - 5s 800us/step - loss: 0.2663 - acc: 0.9200\n","Epoch 261/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.2680 - acc: 0.9160\n","Epoch 262/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2694 - acc: 0.9147\n","Epoch 263/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2675 - acc: 0.9180\n","Epoch 264/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.2577 - acc: 0.9186\n","Epoch 265/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 0.2607 - acc: 0.9180\n","Epoch 266/500\n","6510/6510 [==============================] - 5s 800us/step - loss: 0.2599 - acc: 0.9178\n","Epoch 267/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2673 - acc: 0.9180\n","Epoch 268/500\n","6510/6510 [==============================] - 5s 800us/step - loss: 0.2703 - acc: 0.9178\n","Epoch 269/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.2469 - acc: 0.9258\n","Epoch 270/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.2554 - acc: 0.9201\n","Epoch 271/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.2652 - acc: 0.9172\n","Epoch 272/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 0.2618 - acc: 0.9129\n","Epoch 273/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.2605 - acc: 0.9183\n","Epoch 274/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.2666 - acc: 0.9163\n","Epoch 275/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.2733 - acc: 0.9149\n","Epoch 276/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.2654 - acc: 0.9166\n","Epoch 277/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.2621 - acc: 0.9180\n","Epoch 278/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2773 - acc: 0.9171\n","Epoch 279/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.2647 - acc: 0.9157\n","Epoch 280/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.2497 - acc: 0.9233\n","Epoch 281/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.2697 - acc: 0.9126\n","Epoch 282/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.2687 - acc: 0.9161\n","Epoch 283/500\n","6510/6510 [==============================] - 5s 804us/step - loss: 0.2695 - acc: 0.9140\n","Epoch 284/500\n","6510/6510 [==============================] - 5s 827us/step - loss: 0.2531 - acc: 0.9195\n","Epoch 285/500\n","6510/6510 [==============================] - 5s 812us/step - loss: 0.2652 - acc: 0.9163\n","Epoch 286/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 0.2482 - acc: 0.9233\n","Epoch 287/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.2781 - acc: 0.9131\n","Epoch 288/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.2595 - acc: 0.9181\n","Epoch 289/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.2689 - acc: 0.9192\n","Epoch 290/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2558 - acc: 0.9177\n","Epoch 291/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2704 - acc: 0.9171\n","Epoch 292/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.2567 - acc: 0.9200\n","Epoch 293/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.2602 - acc: 0.9178\n","Epoch 294/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.2615 - acc: 0.9190\n","Epoch 295/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.2561 - acc: 0.9198\n","Epoch 296/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2732 - acc: 0.9160\n","Epoch 297/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.2557 - acc: 0.9235\n","Epoch 298/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.2351 - acc: 0.9290\n","Epoch 299/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.2604 - acc: 0.9192\n","Epoch 300/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 0.2621 - acc: 0.9161\n","Epoch 301/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.2696 - acc: 0.9140\n","Epoch 302/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.2581 - acc: 0.9184\n","Epoch 303/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.2711 - acc: 0.9169\n","Epoch 304/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.2746 - acc: 0.9164\n","Epoch 305/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.2540 - acc: 0.9195\n","Epoch 306/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.2558 - acc: 0.9206\n","Epoch 307/500\n","6510/6510 [==============================] - 5s 768us/step - loss: 0.2565 - acc: 0.9198\n","Epoch 308/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.2555 - acc: 0.9190\n","Epoch 309/500\n","6510/6510 [==============================] - 5s 767us/step - loss: 0.2629 - acc: 0.9201\n","Epoch 310/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.2503 - acc: 0.9218\n","Epoch 311/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.2609 - acc: 0.9189\n","Epoch 312/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.2411 - acc: 0.9269\n","Epoch 313/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.2644 - acc: 0.9181\n","Epoch 314/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.2613 - acc: 0.9214\n","Epoch 315/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2585 - acc: 0.9212\n","Epoch 316/500\n","6510/6510 [==============================] - 5s 801us/step - loss: 0.2587 - acc: 0.9169\n","Epoch 317/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2542 - acc: 0.9203\n","Epoch 318/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.2645 - acc: 0.9149\n","Epoch 319/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.2577 - acc: 0.9198\n","Epoch 320/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.2499 - acc: 0.9178\n","Epoch 321/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.2411 - acc: 0.9233\n","Epoch 322/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.2512 - acc: 0.9223\n","Epoch 323/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.2586 - acc: 0.9181\n","Epoch 324/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.2618 - acc: 0.9194\n","Epoch 325/500\n","6510/6510 [==============================] - 5s 768us/step - loss: 0.2552 - acc: 0.9195\n","Epoch 326/500\n","6510/6510 [==============================] - 5s 771us/step - loss: 0.2356 - acc: 0.9258\n","Epoch 327/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.2600 - acc: 0.9171\n","Epoch 328/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.2569 - acc: 0.9209\n","Epoch 329/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.2513 - acc: 0.9184\n","Epoch 330/500\n","6510/6510 [==============================] - 5s 770us/step - loss: 0.2498 - acc: 0.9223\n","Epoch 331/500\n","6510/6510 [==============================] - 5s 803us/step - loss: 0.2348 - acc: 0.9257\n","Epoch 332/500\n","6510/6510 [==============================] - 5s 803us/step - loss: 0.2533 - acc: 0.9209\n","Epoch 333/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.2610 - acc: 0.9197\n","Epoch 334/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.2539 - acc: 0.9194\n","Epoch 335/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.2499 - acc: 0.9221\n","Epoch 336/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.2371 - acc: 0.9267\n","Epoch 337/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.2432 - acc: 0.9247\n","Epoch 338/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2549 - acc: 0.9203\n","Epoch 339/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.2513 - acc: 0.9209\n","Epoch 340/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.2484 - acc: 0.9230\n","Epoch 341/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.2600 - acc: 0.9200\n","Epoch 342/500\n","6510/6510 [==============================] - 5s 766us/step - loss: 0.2437 - acc: 0.9240\n","Epoch 343/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.2516 - acc: 0.9198\n","Epoch 344/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.2558 - acc: 0.9184\n","Epoch 345/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.2522 - acc: 0.9206\n","Epoch 346/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.2493 - acc: 0.9220\n","Epoch 347/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.2514 - acc: 0.9230\n","Epoch 348/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.2623 - acc: 0.9203\n","Epoch 349/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 0.2541 - acc: 0.9172\n","Epoch 350/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.2718 - acc: 0.9157\n","Epoch 351/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.2547 - acc: 0.9206\n","Epoch 352/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.2578 - acc: 0.9180\n","Epoch 353/500\n","6510/6510 [==============================] - 5s 771us/step - loss: 0.2505 - acc: 0.9235\n","Epoch 354/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.2559 - acc: 0.9172\n","Epoch 355/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.2614 - acc: 0.9152\n","Epoch 356/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.2638 - acc: 0.9189\n","Epoch 357/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.2432 - acc: 0.9244\n","Epoch 358/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.2465 - acc: 0.9227\n","Epoch 359/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.2613 - acc: 0.9183\n","Epoch 360/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.2537 - acc: 0.9237\n","Epoch 361/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.2456 - acc: 0.9204\n","Epoch 362/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.2498 - acc: 0.9221\n","Epoch 363/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.2508 - acc: 0.9215\n","Epoch 364/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.2561 - acc: 0.9217\n","Epoch 365/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.2348 - acc: 0.9257\n","Epoch 366/500\n","6510/6510 [==============================] - 5s 771us/step - loss: 0.2538 - acc: 0.9249\n","Epoch 367/500\n","6510/6510 [==============================] - 5s 759us/step - loss: 0.2583 - acc: 0.9229\n","Epoch 368/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.2547 - acc: 0.9178\n","Epoch 369/500\n","6510/6510 [==============================] - 5s 760us/step - loss: 0.2385 - acc: 0.9252\n","Epoch 370/500\n","6510/6510 [==============================] - 5s 769us/step - loss: 0.2425 - acc: 0.9266\n","Epoch 371/500\n","6510/6510 [==============================] - 5s 768us/step - loss: 0.2531 - acc: 0.9215\n","Epoch 372/500\n","6510/6510 [==============================] - 5s 763us/step - loss: 0.2507 - acc: 0.9224\n","Epoch 373/500\n","6510/6510 [==============================] - 5s 763us/step - loss: 0.2367 - acc: 0.9283\n","Epoch 374/500\n","6510/6510 [==============================] - 5s 766us/step - loss: 0.2614 - acc: 0.9172\n","Epoch 375/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.2502 - acc: 0.9227\n","Epoch 376/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 0.2416 - acc: 0.9233\n","Epoch 377/500\n","6510/6510 [==============================] - 5s 772us/step - loss: 0.2530 - acc: 0.9240\n","Epoch 378/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.2598 - acc: 0.9183\n","Epoch 379/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.2555 - acc: 0.9197\n","Epoch 380/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.2642 - acc: 0.9183\n","Epoch 381/500\n","6510/6510 [==============================] - 5s 779us/step - loss: 0.2624 - acc: 0.9181\n","Epoch 382/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.2478 - acc: 0.9235\n","Epoch 383/500\n","6510/6510 [==============================] - 5s 771us/step - loss: 0.2521 - acc: 0.9244\n","Epoch 384/500\n","6510/6510 [==============================] - 5s 775us/step - loss: 0.2346 - acc: 0.9230\n","Epoch 385/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.2384 - acc: 0.9229\n","Epoch 386/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.2652 - acc: 0.9226\n","Epoch 387/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.2541 - acc: 0.9206\n","Epoch 388/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.2614 - acc: 0.9166\n","Epoch 389/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2392 - acc: 0.9250\n","Epoch 390/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.2730 - acc: 0.9147\n","Epoch 391/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.2518 - acc: 0.9201\n","Epoch 392/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.2449 - acc: 0.9244\n","Epoch 393/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 0.2523 - acc: 0.9253\n","Epoch 394/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 0.2455 - acc: 0.9243\n","Epoch 395/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.2381 - acc: 0.9263\n","Epoch 396/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.2422 - acc: 0.9223\n","Epoch 397/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.2372 - acc: 0.9266\n","Epoch 398/500\n","6510/6510 [==============================] - 5s 799us/step - loss: 0.2353 - acc: 0.9261\n","Epoch 399/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.2529 - acc: 0.9250\n","Epoch 400/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.2332 - acc: 0.9275\n","Epoch 401/500\n","6510/6510 [==============================] - 5s 774us/step - loss: 0.2613 - acc: 0.9210\n","Epoch 402/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.2497 - acc: 0.9214\n","Epoch 403/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.2578 - acc: 0.9177\n","Epoch 404/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.2518 - acc: 0.9247\n","Epoch 405/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.2494 - acc: 0.9250\n","Epoch 406/500\n","6510/6510 [==============================] - 5s 815us/step - loss: 0.2364 - acc: 0.9246\n","Epoch 407/500\n","6510/6510 [==============================] - 5s 804us/step - loss: 0.2420 - acc: 0.9233\n","Epoch 408/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.2408 - acc: 0.9238\n","Epoch 409/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.2463 - acc: 0.9215\n","Epoch 410/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 0.2555 - acc: 0.9206\n","Epoch 411/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.2528 - acc: 0.9247\n","Epoch 412/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 0.2380 - acc: 0.9230\n","Epoch 413/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.2400 - acc: 0.9247\n","Epoch 414/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.2363 - acc: 0.9252\n","Epoch 415/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.2469 - acc: 0.9220\n","Epoch 416/500\n","6510/6510 [==============================] - 5s 836us/step - loss: 0.2556 - acc: 0.9212\n","Epoch 417/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.2668 - acc: 0.9194\n","Epoch 418/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.2496 - acc: 0.9209\n","Epoch 419/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.2478 - acc: 0.9215\n","Epoch 420/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.2424 - acc: 0.9232\n","Epoch 421/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.2387 - acc: 0.9280\n","Epoch 422/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.2474 - acc: 0.9238\n","Epoch 423/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.2370 - acc: 0.9300\n","Epoch 424/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.2442 - acc: 0.9209\n","Epoch 425/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.2488 - acc: 0.9226\n","Epoch 426/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.2551 - acc: 0.9212\n","Epoch 427/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.2532 - acc: 0.9200\n","Epoch 428/500\n","6510/6510 [==============================] - 5s 802us/step - loss: 0.2332 - acc: 0.9273\n","Epoch 429/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.2398 - acc: 0.9269\n","Epoch 430/500\n","6510/6510 [==============================] - 5s 798us/step - loss: 0.2644 - acc: 0.9175\n","Epoch 431/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.2465 - acc: 0.9261\n","Epoch 432/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 0.2520 - acc: 0.9237\n","Epoch 433/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.2538 - acc: 0.9197\n","Epoch 434/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.2453 - acc: 0.9241\n","Epoch 435/500\n","6510/6510 [==============================] - 5s 802us/step - loss: 0.2486 - acc: 0.9241\n","Epoch 436/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.2421 - acc: 0.9243\n","Epoch 437/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.2302 - acc: 0.9296\n","Epoch 438/500\n","6510/6510 [==============================] - 5s 795us/step - loss: 0.2512 - acc: 0.9195\n","Epoch 439/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.2401 - acc: 0.9230\n","Epoch 440/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2505 - acc: 0.9230\n","Epoch 441/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.2393 - acc: 0.9244\n","Epoch 442/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.2499 - acc: 0.9190\n","Epoch 443/500\n","6510/6510 [==============================] - 5s 780us/step - loss: 0.2412 - acc: 0.9276\n","Epoch 444/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.2451 - acc: 0.9233\n","Epoch 445/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.2422 - acc: 0.9275\n","Epoch 446/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.2478 - acc: 0.9267\n","Epoch 447/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.2515 - acc: 0.9180\n","Epoch 448/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.2443 - acc: 0.9249\n","Epoch 449/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.2496 - acc: 0.9249\n","Epoch 450/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.2569 - acc: 0.9192\n","Epoch 451/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.2423 - acc: 0.9260\n","Epoch 452/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.2425 - acc: 0.9244\n","Epoch 453/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.2442 - acc: 0.9283\n","Epoch 454/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.2401 - acc: 0.9287\n","Epoch 455/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.2337 - acc: 0.9272\n","Epoch 456/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.2483 - acc: 0.9217\n","Epoch 457/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 0.2539 - acc: 0.9200\n","Epoch 458/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.2531 - acc: 0.9240\n","Epoch 459/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2362 - acc: 0.9255\n","Epoch 460/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2599 - acc: 0.9172\n","Epoch 461/500\n","6510/6510 [==============================] - 5s 793us/step - loss: 0.2425 - acc: 0.9255\n","Epoch 462/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.2340 - acc: 0.9289\n","Epoch 463/500\n","6510/6510 [==============================] - 5s 789us/step - loss: 0.2493 - acc: 0.9220\n","Epoch 464/500\n","6510/6510 [==============================] - 5s 791us/step - loss: 0.2415 - acc: 0.9249\n","Epoch 465/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 0.2381 - acc: 0.9269\n","Epoch 466/500\n","6510/6510 [==============================] - 5s 810us/step - loss: 0.2442 - acc: 0.9244\n","Epoch 467/500\n","6510/6510 [==============================] - 5s 812us/step - loss: 0.2364 - acc: 0.9246\n","Epoch 468/500\n","6510/6510 [==============================] - 5s 786us/step - loss: 0.2417 - acc: 0.9275\n","Epoch 469/500\n","6510/6510 [==============================] - 5s 781us/step - loss: 0.2425 - acc: 0.9233\n","Epoch 470/500\n","6510/6510 [==============================] - 5s 773us/step - loss: 0.2501 - acc: 0.9210\n","Epoch 471/500\n","6510/6510 [==============================] - 5s 777us/step - loss: 0.2259 - acc: 0.9257\n","Epoch 472/500\n","6510/6510 [==============================] - 5s 770us/step - loss: 0.2315 - acc: 0.9252\n","Epoch 473/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.2364 - acc: 0.9263\n","Epoch 474/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.2176 - acc: 0.9339\n","Epoch 475/500\n","6510/6510 [==============================] - 5s 783us/step - loss: 0.2560 - acc: 0.9200\n","Epoch 476/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.2290 - acc: 0.9303\n","Epoch 477/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.2411 - acc: 0.9253\n","Epoch 478/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.2385 - acc: 0.9238\n","Epoch 479/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.2530 - acc: 0.9263\n","Epoch 480/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.2460 - acc: 0.9223\n","Epoch 481/500\n","6510/6510 [==============================] - 5s 785us/step - loss: 0.2496 - acc: 0.9243\n","Epoch 482/500\n","6510/6510 [==============================] - 5s 794us/step - loss: 0.2443 - acc: 0.9253\n","Epoch 483/500\n","6510/6510 [==============================] - 5s 790us/step - loss: 0.2317 - acc: 0.9298\n","Epoch 484/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.2453 - acc: 0.9240\n","Epoch 485/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2385 - acc: 0.9260\n","Epoch 486/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.2603 - acc: 0.9184\n","Epoch 487/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.2561 - acc: 0.9207\n","Epoch 488/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.2351 - acc: 0.9241\n","Epoch 489/500\n","6510/6510 [==============================] - 5s 784us/step - loss: 0.2430 - acc: 0.9250\n","Epoch 490/500\n","6510/6510 [==============================] - 5s 788us/step - loss: 0.2278 - acc: 0.9290\n","Epoch 491/500\n","6510/6510 [==============================] - 5s 792us/step - loss: 0.2469 - acc: 0.9275\n","Epoch 492/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.2595 - acc: 0.9197\n","Epoch 493/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2461 - acc: 0.9249\n","Epoch 494/500\n","6510/6510 [==============================] - 5s 797us/step - loss: 0.2590 - acc: 0.9194\n","Epoch 495/500\n","6510/6510 [==============================] - 5s 796us/step - loss: 0.2392 - acc: 0.9244\n","Epoch 496/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.2334 - acc: 0.9269\n","Epoch 497/500\n","6510/6510 [==============================] - 5s 776us/step - loss: 0.2540 - acc: 0.9220\n","Epoch 498/500\n","6510/6510 [==============================] - 5s 782us/step - loss: 0.2477 - acc: 0.9237\n","Epoch 499/500\n","6510/6510 [==============================] - 5s 778us/step - loss: 0.2220 - acc: 0.9281\n","Epoch 500/500\n","6510/6510 [==============================] - 5s 787us/step - loss: 0.2270 - acc: 0.9303\n"],"name":"stdout"},{"output_type":"execute_result","data":{"text/plain":["<keras.callbacks.History at 0x7f52e0106278>"]},"metadata":{"tags":[]},"execution_count":18}]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"EmkaxXdjHdyd"},"source":["## Evaluation"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"7RraFX9YHdye","colab":{"base_uri":"https://localhost:8080/","height":118},"outputId":"bcce1469-96e2-42b5-9d4c-749d97466696","executionInfo":{"status":"ok","timestamp":1563723734700,"user_tz":-330,"elapsed":6592,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}}},"source":["print(model.evaluate(X_train, y_train, batch_size = 20))\n","print('\\nModel Performance: Log Loss and Accuracy on validation data')\n","print(model.evaluate(X_valid, y_valid, batch_size = 20))"],"execution_count":19,"outputs":[{"output_type":"stream","text":["6510/6510 [==============================] - 4s 615us/step\n","[0.02733311530364629, 0.993855605293895]\n","\n","Model Performance: Log Loss and Accuracy on validation data\n","2790/2790 [==============================] - 2s 550us/step\n","[9.317935273638764, 0.3720430124511001]\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"u5CKxykLHdyj"},"source":["## Generate text"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"4JSW5EwKHdyk","colab":{"base_uri":"https://localhost:8080/","height":154},"outputId":"35d30c3a-764b-4ddf-a36c-f0e56fccfb62","executionInfo":{"status":"ok","timestamp":1563724522507,"user_tz":-330,"elapsed":3259,"user":{"displayName":"Ezhirko Arulmozhi","photoUrl":"https://lh4.googleusercontent.com/-477BLY1dxIY/AAAAAAAAAAI/AAAAAAAAB1o/N0VNqCcpb28/s64/photo.jpg","userId":"16774972310128518178"}}},"source":["seed_length=50\n","new_words=50\n","diversity=1\n","n_gen=1\n","\n","import random\n","\n","# Choose a random sequence\n","seq = random.choice(sequences)\n","\n","# print seq\n","\n","# Choose a random starting point\n","seed_idx = random.randint(0, len(seq) - seed_length - 10)\n","# Ending index for seed\n","end_idx = seed_idx + seed_length\n","\n","gen_list = []\n","\n","for n in range(n_gen):\n","    # Extract the seed sequence\n","    seed = seq[seed_idx:end_idx]\n","    original_sequence = [idx_word[i] for i in seed]\n","    generated = seed[:] + ['#']\n","\n","    # Find the actual entire sequence\n","    actual = generated[:] + seq[end_idx:end_idx + new_words]\n","        \n","    # Keep adding new words\n","    for i in range(new_words):\n","\n","        # Make a prediction from the seed\n","        preds = model.predict(np.array(seed).reshape(1, -1))[0].astype(np.float64)\n","\n","        # Diversify\n","        preds = np.log(preds) / diversity\n","        exp_preds = np.exp(preds)\n","\n","        # Softmax\n","        preds = exp_preds / sum(exp_preds)\n","\n","        # Choose the next word\n","        probas = np.random.multinomial(1, preds, 1)[0]\n","\n","        next_idx = np.argmax(probas)\n","\n","        # New seed adds on old word\n","        #             seed = seed[1:] + [next_idx]\n","        seed += [next_idx]\n","        generated.append(next_idx)\n","    # Showing generated and actual abstract\n","    n = []\n","\n","    for i in generated:\n","        n.append(idx_word.get(i, '< --- >'))\n","\n","    gen_list.append(n)\n","\n","a = []\n","\n","for i in actual:\n","    a.append(idx_word.get(i, '< --- >'))\n","\n","a = a[seed_length:]\n","\n","gen_list = [gen[seed_length:seed_length + len(a)] for gen in gen_list]\n","\n","print('Original Sequence: \\n'+' '.join(original_sequence))\n","print(\"\\n\")\n","# print(gen_list)\n","print('Generated Sequence: \\n'+' '.join(gen_list[0][1:]))\n","# print(a)"],"execution_count":32,"outputs":[{"output_type":"stream","text":["/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:36: RuntimeWarning: divide by zero encountered in log\n"],"name":"stderr"},{"output_type":"stream","text":["Original Sequence: \n","left me as she came unwarned and fell from sighing into sleep before the summer morning dawned what wonder now i find no maiden fair who dwells between these mountains and the seas and go unloving and unloved or ere i turn to such as these what wonder if the\n","\n","\n","Generated Sequence: \n","day is in the sting of beauty in archaeology herod suppliant the tetrarch's remorse the tetrarch's treasure salome rose 25 lady and a cappadocian by seeing crowned again with paltry mrs an was has new background without frontiers the poetry of archaeology herod suppliant the tetrarch's remorse the tetrarch's treasure\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"colab_type":"code","id":"9aXeiz7ZHdyp","colab":{}},"source":[""],"execution_count":0,"outputs":[]}]}