In [1]:

    

import warnings
warnings.filterwarnings('ignore')


    

In [2]:

    

%matplotlib inline
%pylab inline


    

    




Populating the interactive namespace from numpy and matplotlib

In [3]:

    

import matplotlib.pylab as plt
# https://docs.scipy.org/doc/numpy/reference/routines.math.html
import numpy as np


    

In [4]:

    

from datetime import tzinfo, timedelta, datetime


    

In [5]:

    

from distutils.version import StrictVersion


    

In [6]:

    

import sklearn

assert StrictVersion(sklearn.__version__ ) >= StrictVersion('0.18.1')

sklearn.__version__


    

    
Out[6]:





'0.18.1'

In [7]:

    

import tensorflow as tf
tf.logging.set_verbosity(tf.logging.ERROR)

assert StrictVersion(tf.__version__) >= StrictVersion('1.1.0')

tf.__version__


    

    
Out[7]:





'1.2.1'

In [8]:

    

import keras

assert StrictVersion(keras.__version__) >= StrictVersion('2.0.0')

keras.__version__


    

    




Using TensorFlow backend.






    
Out[8]:





'2.0.6'

In [9]:

    

# !curl -O https://raw.githubusercontent.com/DJCordhose/speed-limit-signs/master/data/speed-limit-signs.zip
# !curl -O https://raw.githubusercontent.com/DJCordhose/speed-limit-signs/master/data/augmented-signs.zip


    

In [10]:

    

# https://docs.python.org/3/library/zipfile.html
# from zipfile import ZipFile
# zip = ZipFile(r'speed-limit-signs.zip')
# zip.extractall('.')
# zip = ZipFile(r'augmented-signs.zip')
# zip.extractall('.')


    

In [11]:

    

# !ls -l speed-limit-signs


    

In [12]:

    

# !ls -l augmented-signs


    

In [13]:

    

import os
import skimage.data
import skimage.transform
from keras.utils.np_utils import to_categorical
import numpy as np

def load_data(data_dir, type=".ppm"):
    num_categories = 6

    # Get all subdirectories of data_dir. Each represents a label.
    directories = [d for d in os.listdir(data_dir) 
                   if os.path.isdir(os.path.join(data_dir, d))]
    # Loop through the label directories and collect the data in
    # two lists, labels and images.
    labels = []
    images = []
    for d in directories:
        label_dir = os.path.join(data_dir, d)
        file_names = [os.path.join(label_dir, f) for f in os.listdir(label_dir) if f.endswith(type)]
        # For each label, load it's images and add them to the images list.
        # And add the label number (i.e. directory name) to the labels list.
        for f in file_names:
            images.append(skimage.data.imread(f))
            labels.append(int(d))
    images64 = [skimage.transform.resize(image, (64, 64)) for image in images]
    y = np.array(labels)
    y = to_categorical(y, num_categories)
    X = np.array(images64)
    return X, y


    

In [14]:

    

# Depends on harware GPU architecture, set as high as possible (this works well on K80)
BATCH_SIZE = 500


    

In [15]:

    

# Load datasets.
ROOT_PATH = "./"


    

In [16]:

    

original_dir = os.path.join(ROOT_PATH, "speed-limit-signs")
original_images, original_labels = load_data(original_dir, type=".ppm")


    

In [17]:

    

data_dir = os.path.join(ROOT_PATH, "augmented-signs")
augmented_images, augmented_labels = load_data(data_dir, type=".png")


    

In [18]:

    

all_images = np.vstack((original_images, augmented_images))


    

In [19]:

    

all_labels = np.vstack((original_labels, augmented_labels))


    

In [20]:

    

# https://stackoverflow.com/a/4602224
p = numpy.random.permutation(len(all_labels))


    

In [21]:

    

p = numpy.random.permutation(len(all_labels))


    

In [22]:

    

shuffled_images = all_images[p]


    

In [23]:

    

shuffled_labels = all_labels[p]


    

In [24]:

    

# Turn this around if you want the large training set using augmented data or the original one

# X, y = original_images, original_labels
# X, y = augmented_images, augmented_labels
X, y = shuffled_images, shuffled_labels


    

In [25]:

    

# Same as above

# evaluation_X, evaluation_y = augmented_images, augmented_labels
evaluation_X, evaluation_y = original_images, original_labels


    

In [26]:

    

from sklearn.model_selection import train_test_split


    

In [27]:

    

checkpoint_callback = keras.callbacks.ModelCheckpoint('../tmp/model-checkpoints/weights.epoch-{epoch:02d}-val_loss-{val_loss:.2f}.hdf5');


    

In [28]:

    

early_stopping_callback = keras.callbacks.EarlyStopping(monitor='val_loss', patience=30, verbose=1)


    

In [29]:

    

# https://github.com/tensorflow/tensorflow/blob/master/tensorflow/tensorboard/README.md
# https://keras.io/callbacks/#tensorboard
# http://stackoverflow.com/questions/42112260/how-do-i-use-the-tensorboard-callback-of-keras
tb_callback = keras.callbacks.TensorBoard(log_dir='../tmp/tf_log')
#                                          histogram_freq=1, write_graph=True, write_images=True)
#                                          histogram_freq=1, write_graph=True, write_images=True)
# tbCallBack = keras.callbacks.TensorBoard(log_dir='./logs', histogram_freq=0, batch_size=32, write_graph=True, write_grads=False, write_images=False, embeddings_freq=0, embeddings_layer_names=None, embeddings_metadata=None)
# To start tensorboard
# tensorboard --logdir=/mnt/c/Users/olive/Development/ml/tf_log
# open http://localhost:6006


    

In [30]:

    

# we want to distribute our different classes equally over test and train, this works using stratify
# https://github.com/amueller/scipy-2017-sklearn/blob/master/notebooks/04.Training_and_Testing_Data.ipynb
# http://scikit-learn.org/stable/modules/generated/sklearn.model_selection.train_test_split.html

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42, stratify=y)


    

In [31]:

    

X_train.shape, y_train.shape


    

    
Out[31]:





((3335, 64, 64, 3), (3335, 6))

In [32]:

    

from keras.models import Model
from keras.layers import Dense, Dropout, Activation, Flatten, Input
from keras.layers import Convolution2D, MaxPooling2D

# drop_out = 0.9
# drop_out = 0.75
drop_out = 0.5
# drop_out = 0.25
# drop_out = 0.0

# input tensor for a 3-channel 64x64 image
inputs = Input(shape=(64, 64, 3))

# one block of convolutional layers
x = Convolution2D(64, 3, 3, activation='relu')(inputs)
# x = Dropout(drop_out)(x)
x = Convolution2D(64, 3, 3, activation='relu')(x)
# x = Dropout(drop_out)(x)
x = Convolution2D(64, 3, 3, activation='relu')(x)
x = MaxPooling2D(pool_size=(2, 2))(x)
x = Dropout(drop_out)(x)

# one more block
x = Convolution2D(128, 3, 3, activation='relu')(x)
# x = Dropout(drop_out)(x)
x = Convolution2D(128, 3, 3, activation='relu')(x)
x = MaxPooling2D(pool_size=(2, 2))(x)
x = Dropout(drop_out)(x)

# one more block
x = Convolution2D(256, 3, 3, activation='relu')(x)
x = MaxPooling2D(pool_size=(2, 2))(x)
x = Dropout(drop_out)(x)

x = Flatten()(x)
x = Dense(256, activation='relu')(x)
x = Dropout(drop_out)(x)

# softmax activation, 6 categories
predictions = Dense(6, activation='softmax')(x)
model = Model(input=inputs, output=predictions)
model.summary()
# model.compile(optimizer='rmsprop',
model.compile(optimizer='adam',
              loss='categorical_crossentropy',
              metrics=['accuracy'])


    

    




_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
input_1 (InputLayer)         (None, 64, 64, 3)         0         
_________________________________________________________________
conv2d_1 (Conv2D)            (None, 62, 62, 64)        1792      
_________________________________________________________________
conv2d_2 (Conv2D)            (None, 60, 60, 64)        36928     
_________________________________________________________________
conv2d_3 (Conv2D)            (None, 58, 58, 64)        36928     
_________________________________________________________________
max_pooling2d_1 (MaxPooling2 (None, 29, 29, 64)        0         
_________________________________________________________________
dropout_1 (Dropout)          (None, 29, 29, 64)        0         
_________________________________________________________________
conv2d_4 (Conv2D)            (None, 27, 27, 128)       73856     
_________________________________________________________________
conv2d_5 (Conv2D)            (None, 25, 25, 128)       147584    
_________________________________________________________________
max_pooling2d_2 (MaxPooling2 (None, 12, 12, 128)       0         
_________________________________________________________________
dropout_2 (Dropout)          (None, 12, 12, 128)       0         
_________________________________________________________________
conv2d_6 (Conv2D)            (None, 10, 10, 256)       295168    
_________________________________________________________________
max_pooling2d_3 (MaxPooling2 (None, 5, 5, 256)         0         
_________________________________________________________________
dropout_3 (Dropout)          (None, 5, 5, 256)         0         
_________________________________________________________________
flatten_1 (Flatten)          (None, 6400)              0         
_________________________________________________________________
dense_1 (Dense)              (None, 256)               1638656   
_________________________________________________________________
dropout_4 (Dropout)          (None, 256)               0         
_________________________________________________________________
dense_2 (Dense)              (None, 6)                 1542      
=================================================================
Total params: 2,232,454
Trainable params: 2,232,454
Non-trainable params: 0
_________________________________________________________________

In [ ]:

    

!rm -rf ../tmp/tf_log
!rm -rf ../tmp/model-checkpoints

!mkdir ../tmp/model-checkpoints
!mkdir ../tmp/tf_log


    

In [ ]:

    

# Running on a GPU bach size might be critical depdendng on the GPU memory available
# more is desirable, but we might end up using 50 only 
print(datetime.utcnow().isoformat())
# BE CAREFUL, validation data is always the last data sets and not shuffled
# https://keras.io/getting-started/faq/#how-is-the-validation-split-computed
model.fit(X_train, y_train, epochs=1000, batch_size=BATCH_SIZE, validation_split=0.3, 
#           callbacks=[tb_callback, early_stopping_callback])
          callbacks=[tb_callback])
# model.fit(X_train, y_train, epochs=50, batch_size=200, validation_split=0.3)
print(datetime.utcnow().isoformat())


    

    




2017-07-31T13:48:46.738969
Train on 2334 samples, validate on 1001 samples
Epoch 1/1000
2334/2334 [==============================] - 15s - loss: 1.8009 - acc: 0.2035 - val_loss: 1.7851 - val_acc: 0.1968
Epoch 2/1000
2334/2334 [==============================] - 5s - loss: 1.7385 - acc: 0.2134 - val_loss: 1.7857 - val_acc: 0.1968
Epoch 3/1000
2334/2334 [==============================] - 5s - loss: 1.7185 - acc: 0.2219 - val_loss: 1.7880 - val_acc: 0.1978
Epoch 4/1000
2334/2334 [==============================] - 5s - loss: 1.7196 - acc: 0.2498 - val_loss: 1.7876 - val_acc: 0.1878
Epoch 5/1000
2334/2334 [==============================] - 5s - loss: 1.7091 - acc: 0.2678 - val_loss: 1.7848 - val_acc: 0.1978
Epoch 6/1000
2334/2334 [==============================] - 5s - loss: 1.7049 - acc: 0.2768 - val_loss: 1.7782 - val_acc: 0.2288
Epoch 7/1000
2334/2334 [==============================] - 5s - loss: 1.7097 - acc: 0.2669 - val_loss: 1.7772 - val_acc: 0.2078
Epoch 8/1000
2334/2334 [==============================] - 5s - loss: 1.7046 - acc: 0.2708 - val_loss: 1.7866 - val_acc: 0.1758
Epoch 9/1000
2334/2334 [==============================] - 5s - loss: 1.6894 - acc: 0.2986 - val_loss: 1.7895 - val_acc: 0.1578
Epoch 10/1000
2334/2334 [==============================] - 5s - loss: 1.6813 - acc: 0.2969 - val_loss: 1.7880 - val_acc: 0.1748
Epoch 11/1000
2334/2334 [==============================] - 5s - loss: 1.6781 - acc: 0.3008 - val_loss: 1.7846 - val_acc: 0.1778
Epoch 12/1000
2334/2334 [==============================] - 5s - loss: 1.6827 - acc: 0.2888 - val_loss: 1.7948 - val_acc: 0.1489
Epoch 13/1000
2334/2334 [==============================] - 5s - loss: 1.6953 - acc: 0.2858 - val_loss: 1.7964 - val_acc: 0.1489
Epoch 14/1000
2334/2334 [==============================] - 5s - loss: 1.7105 - acc: 0.2656 - val_loss: 1.7741 - val_acc: 0.1918
Epoch 15/1000
2334/2334 [==============================] - 5s - loss: 1.6932 - acc: 0.2922 - val_loss: 1.7897 - val_acc: 0.1618
Epoch 16/1000
2334/2334 [==============================] - 5s - loss: 1.6834 - acc: 0.2862 - val_loss: 1.7945 - val_acc: 0.1489
Epoch 17/1000
2334/2334 [==============================] - 5s - loss: 1.6701 - acc: 0.3046 - val_loss: 1.7902 - val_acc: 0.1628
Epoch 18/1000
2334/2334 [==============================] - 5s - loss: 1.6622 - acc: 0.3059 - val_loss: 1.7817 - val_acc: 0.1808
Epoch 19/1000
2334/2334 [==============================] - 5s - loss: 1.6465 - acc: 0.3128 - val_loss: 1.7941 - val_acc: 0.1628
Epoch 20/1000
2334/2334 [==============================] - 5s - loss: 1.6449 - acc: 0.2995 - val_loss: 1.7698 - val_acc: 0.2008
Epoch 21/1000
2334/2334 [==============================] - 5s - loss: 1.6697 - acc: 0.2905 - val_loss: 1.7579 - val_acc: 0.2178
Epoch 22/1000
2334/2334 [==============================] - 5s - loss: 1.6506 - acc: 0.3012 - val_loss: 1.7857 - val_acc: 0.1818
Epoch 23/1000
2334/2334 [==============================] - 5s - loss: 1.6414 - acc: 0.3128 - val_loss: 1.8025 - val_acc: 0.1748
Epoch 24/1000
2334/2334 [==============================] - 5s - loss: 1.6245 - acc: 0.3239 - val_loss: 1.7712 - val_acc: 0.2038
Epoch 25/1000
2334/2334 [==============================] - 5s - loss: 1.6177 - acc: 0.3252 - val_loss: 1.7500 - val_acc: 0.2298
Epoch 26/1000
2334/2334 [==============================] - 5s - loss: 1.6216 - acc: 0.3196 - val_loss: 1.7818 - val_acc: 0.2018
Epoch 27/1000
2334/2334 [==============================] - 5s - loss: 1.6243 - acc: 0.3081 - val_loss: 1.7972 - val_acc: 0.1928
Epoch 28/1000
2334/2334 [==============================] - 5s - loss: 1.6051 - acc: 0.3171 - val_loss: 1.8096 - val_acc: 0.1868
Epoch 29/1000
2334/2334 [==============================] - 5s - loss: 1.6036 - acc: 0.3188 - val_loss: 1.8070 - val_acc: 0.1988
Epoch 30/1000
2334/2334 [==============================] - 5s - loss: 1.5849 - acc: 0.3188 - val_loss: 1.8612 - val_acc: 0.1818
Epoch 31/1000
2334/2334 [==============================] - 5s - loss: 1.6044 - acc: 0.3222 - val_loss: 1.8464 - val_acc: 0.1888
Epoch 32/1000
2334/2334 [==============================] - 5s - loss: 1.5933 - acc: 0.3188 - val_loss: 1.8412 - val_acc: 0.1868
Epoch 33/1000
2334/2334 [==============================] - 5s - loss: 1.5829 - acc: 0.3269 - val_loss: 1.8587 - val_acc: 0.1828
Epoch 34/1000
2334/2334 [==============================] - 5s - loss: 1.5726 - acc: 0.3312 - val_loss: 1.8418 - val_acc: 0.1878
Epoch 35/1000
2334/2334 [==============================] - 5s - loss: 1.5587 - acc: 0.3350 - val_loss: 1.7740 - val_acc: 0.2298
Epoch 36/1000
2334/2334 [==============================] - 5s - loss: 1.5457 - acc: 0.3428 - val_loss: 1.8842 - val_acc: 0.1818
Epoch 37/1000
2334/2334 [==============================] - 5s - loss: 1.5460 - acc: 0.3449 - val_loss: 1.8435 - val_acc: 0.2078
Epoch 38/1000
2334/2334 [==============================] - 5s - loss: 1.5433 - acc: 0.3428 - val_loss: 1.8587 - val_acc: 0.2108
Epoch 39/1000
2334/2334 [==============================] - 5s - loss: 1.5375 - acc: 0.3432 - val_loss: 1.9343 - val_acc: 0.1838
Epoch 40/1000
2334/2334 [==============================] - 5s - loss: 1.5279 - acc: 0.3513 - val_loss: 1.8349 - val_acc: 0.1968
Epoch 41/1000
2334/2334 [==============================] - 5s - loss: 1.5285 - acc: 0.3663 - val_loss: 1.9359 - val_acc: 0.1908
Epoch 42/1000
2334/2334 [==============================] - 5s - loss: 1.5140 - acc: 0.3582 - val_loss: 1.9073 - val_acc: 0.1868
Epoch 43/1000
2334/2334 [==============================] - 5s - loss: 1.4942 - acc: 0.3766 - val_loss: 1.6936 - val_acc: 0.2757
Epoch 44/1000
2334/2334 [==============================] - 5s - loss: 1.4943 - acc: 0.3676 - val_loss: 1.8485 - val_acc: 0.2058
Epoch 45/1000
2334/2334 [==============================] - 5s - loss: 1.5243 - acc: 0.3719 - val_loss: 1.6577 - val_acc: 0.2957
Epoch 46/1000
2334/2334 [==============================] - 5s - loss: 1.5509 - acc: 0.3410 - val_loss: 1.9223 - val_acc: 0.1998
Epoch 47/1000
2334/2334 [==============================] - 5s - loss: 1.5048 - acc: 0.3668 - val_loss: 1.9379 - val_acc: 0.1638
Epoch 48/1000
2334/2334 [==============================] - 5s - loss: 1.5298 - acc: 0.3676 - val_loss: 1.9484 - val_acc: 0.1778
Epoch 49/1000
2334/2334 [==============================] - 5s - loss: 1.5144 - acc: 0.3616 - val_loss: 1.8760 - val_acc: 0.2008
Epoch 50/1000
2334/2334 [==============================] - 5s - loss: 1.4853 - acc: 0.3728 - val_loss: 1.9106 - val_acc: 0.2078
Epoch 51/1000
2334/2334 [==============================] - 5s - loss: 1.4757 - acc: 0.3775 - val_loss: 1.8013 - val_acc: 0.2418
Epoch 52/1000
2334/2334 [==============================] - 5s - loss: 1.5307 - acc: 0.3659 - val_loss: 1.7534 - val_acc: 0.2777
Epoch 53/1000
2334/2334 [==============================] - 5s - loss: 1.5345 - acc: 0.3685 - val_loss: 1.7194 - val_acc: 0.2358
Epoch 54/1000
2334/2334 [==============================] - 5s - loss: 1.5312 - acc: 0.3522 - val_loss: 1.8328 - val_acc: 0.1848
Epoch 55/1000
2334/2334 [==============================] - 5s - loss: 1.4969 - acc: 0.3796 - val_loss: 1.7729 - val_acc: 0.2298
Epoch 56/1000
2334/2334 [==============================] - 5s - loss: 1.4782 - acc: 0.3719 - val_loss: 1.9372 - val_acc: 0.1788
Epoch 57/1000
2334/2334 [==============================] - 5s - loss: 1.4631 - acc: 0.3929 - val_loss: 1.8974 - val_acc: 0.2008
Epoch 58/1000
2334/2334 [==============================] - 5s - loss: 1.4609 - acc: 0.3766 - val_loss: 1.9179 - val_acc: 0.2048
Epoch 59/1000
2334/2334 [==============================] - 5s - loss: 1.4568 - acc: 0.3972 - val_loss: 2.0415 - val_acc: 0.1718
Epoch 60/1000
2334/2334 [==============================] - 5s - loss: 1.4437 - acc: 0.3959 - val_loss: 1.7672 - val_acc: 0.2388
Epoch 61/1000
2334/2334 [==============================] - 5s - loss: 1.4495 - acc: 0.3937 - val_loss: 1.7914 - val_acc: 0.2507
Epoch 62/1000
2334/2334 [==============================] - 5s - loss: 1.4299 - acc: 0.4083 - val_loss: 1.8929 - val_acc: 0.2318
Epoch 63/1000
2334/2334 [==============================] - 5s - loss: 1.4099 - acc: 0.4130 - val_loss: 2.0386 - val_acc: 0.1988
Epoch 64/1000
2334/2334 [==============================] - 5s - loss: 1.3891 - acc: 0.4152 - val_loss: 1.9052 - val_acc: 0.2338
Epoch 65/1000
2334/2334 [==============================] - 5s - loss: 1.4037 - acc: 0.4109 - val_loss: 2.3809 - val_acc: 0.1718
Epoch 66/1000
2334/2334 [==============================] - 5s - loss: 1.4657 - acc: 0.3912 - val_loss: 2.1282 - val_acc: 0.1768
Epoch 67/1000
2334/2334 [==============================] - 5s - loss: 1.4403 - acc: 0.3967 - val_loss: 2.2153 - val_acc: 0.1678
Epoch 68/1000
2334/2334 [==============================] - 5s - loss: 1.4406 - acc: 0.3980 - val_loss: 2.1103 - val_acc: 0.1938
Epoch 69/1000
2334/2334 [==============================] - 5s - loss: 1.4079 - acc: 0.4139 - val_loss: 2.0451 - val_acc: 0.2138
Epoch 70/1000
2334/2334 [==============================] - 5s - loss: 1.3751 - acc: 0.4173 - val_loss: 1.9309 - val_acc: 0.2288
Epoch 71/1000
2334/2334 [==============================] - 5s - loss: 1.3620 - acc: 0.4289 - val_loss: 2.0134 - val_acc: 0.2208
Epoch 72/1000
2334/2334 [==============================] - 5s - loss: 1.3493 - acc: 0.4237 - val_loss: 1.9824 - val_acc: 0.2458
Epoch 73/1000
2334/2334 [==============================] - 5s - loss: 1.3477 - acc: 0.4306 - val_loss: 1.9207 - val_acc: 0.2328
Epoch 74/1000
2334/2334 [==============================] - 5s - loss: 1.3216 - acc: 0.4469 - val_loss: 1.9449 - val_acc: 0.2707
Epoch 75/1000
2334/2334 [==============================] - 5s - loss: 1.3130 - acc: 0.4494 - val_loss: 2.5123 - val_acc: 0.1828
Epoch 76/1000
2334/2334 [==============================] - 5s - loss: 1.3446 - acc: 0.4306 - val_loss: 2.1837 - val_acc: 0.2068
Epoch 77/1000
2334/2334 [==============================] - 5s - loss: 1.3001 - acc: 0.4563 - val_loss: 1.9700 - val_acc: 0.2478
Epoch 78/1000
2334/2334 [==============================] - 5s - loss: 1.2855 - acc: 0.4614 - val_loss: 2.1198 - val_acc: 0.2527
Epoch 79/1000
2334/2334 [==============================] - 5s - loss: 1.3018 - acc: 0.4636 - val_loss: 2.3271 - val_acc: 0.1988
Epoch 80/1000
2334/2334 [==============================] - 5s - loss: 1.2792 - acc: 0.4644 - val_loss: 2.0488 - val_acc: 0.2408
Epoch 81/1000
2334/2334 [==============================] - 5s - loss: 1.2374 - acc: 0.5056 - val_loss: 2.0997 - val_acc: 0.2398
Epoch 82/1000
2334/2334 [==============================] - 5s - loss: 1.2559 - acc: 0.4790 - val_loss: 1.9409 - val_acc: 0.2587
Epoch 83/1000
2334/2334 [==============================] - 5s - loss: 1.2751 - acc: 0.4790 - val_loss: 2.4851 - val_acc: 0.1908
Epoch 84/1000
2334/2334 [==============================] - 5s - loss: 1.2590 - acc: 0.4910 - val_loss: 2.4517 - val_acc: 0.2008
Epoch 85/1000
2334/2334 [==============================] - 5s - loss: 1.2463 - acc: 0.4841 - val_loss: 2.3165 - val_acc: 0.2218
Epoch 86/1000
2334/2334 [==============================] - 5s - loss: 1.2234 - acc: 0.4970 - val_loss: 1.8542 - val_acc: 0.3037
Epoch 87/1000
2334/2334 [==============================] - 5s - loss: 1.1930 - acc: 0.4910 - val_loss: 2.0224 - val_acc: 0.2917
Epoch 88/1000
2334/2334 [==============================] - 5s - loss: 1.1775 - acc: 0.5056 - val_loss: 2.3150 - val_acc: 0.2527
Epoch 89/1000
2334/2334 [==============================] - 5s - loss: 1.2190 - acc: 0.5086 - val_loss: 2.1920 - val_acc: 0.2368
Epoch 90/1000
2334/2334 [==============================] - 5s - loss: 1.2106 - acc: 0.4979 - val_loss: 2.7417 - val_acc: 0.2018
Epoch 91/1000
2334/2334 [==============================] - 5s - loss: 1.2066 - acc: 0.5129 - val_loss: 2.1938 - val_acc: 0.2627
Epoch 92/1000
2334/2334 [==============================] - 5s - loss: 1.2162 - acc: 0.4884 - val_loss: 2.4576 - val_acc: 0.1978
Epoch 93/1000
2334/2334 [==============================] - 5s - loss: 1.1858 - acc: 0.5296 - val_loss: 2.0923 - val_acc: 0.2627
Epoch 94/1000
2334/2334 [==============================] - 5s - loss: 1.1454 - acc: 0.5317 - val_loss: 2.3953 - val_acc: 0.2198
Epoch 95/1000
2334/2334 [==============================] - 5s - loss: 1.1622 - acc: 0.5253 - val_loss: 2.5310 - val_acc: 0.2148
Epoch 96/1000
2334/2334 [==============================] - 5s - loss: 1.1479 - acc: 0.5300 - val_loss: 2.3051 - val_acc: 0.2428
Epoch 97/1000
2334/2334 [==============================] - 5s - loss: 1.1856 - acc: 0.5304 - val_loss: 2.3270 - val_acc: 0.2388
Epoch 98/1000
2334/2334 [==============================] - 5s - loss: 1.1570 - acc: 0.5308 - val_loss: 2.1741 - val_acc: 0.2537
Epoch 99/1000
2334/2334 [==============================] - 5s - loss: 1.1132 - acc: 0.5446 - val_loss: 2.2533 - val_acc: 0.2348
Epoch 100/1000
2334/2334 [==============================] - 5s - loss: 1.0798 - acc: 0.5707 - val_loss: 2.3934 - val_acc: 0.2448
Epoch 101/1000
2334/2334 [==============================] - 5s - loss: 1.0817 - acc: 0.5493 - val_loss: 2.3482 - val_acc: 0.2428
Epoch 102/1000
2334/2334 [==============================] - 5s - loss: 1.0760 - acc: 0.5643 - val_loss: 2.5586 - val_acc: 0.2318
Epoch 103/1000
2334/2334 [==============================] - 5s - loss: 1.0638 - acc: 0.5690 - val_loss: 2.8674 - val_acc: 0.2298
Epoch 104/1000
2334/2334 [==============================] - 5s - loss: 1.0994 - acc: 0.5548 - val_loss: 2.5812 - val_acc: 0.2368
Epoch 105/1000
2334/2334 [==============================] - 5s - loss: 1.1179 - acc: 0.5540 - val_loss: 2.8600 - val_acc: 0.1968
Epoch 106/1000
2334/2334 [==============================] - 5s - loss: 1.0880 - acc: 0.5673 - val_loss: 2.4889 - val_acc: 0.2278
Epoch 107/1000
2334/2334 [==============================] - 5s - loss: 1.0532 - acc: 0.5835 - val_loss: 1.8518 - val_acc: 0.3467
Epoch 108/1000
2334/2334 [==============================] - 5s - loss: 1.0184 - acc: 0.5913 - val_loss: 2.2908 - val_acc: 0.2707
Epoch 109/1000
2334/2334 [==============================] - 5s - loss: 1.0383 - acc: 0.5823 - val_loss: 2.2188 - val_acc: 0.2967
Epoch 110/1000
2334/2334 [==============================] - 5s - loss: 1.0400 - acc: 0.5831 - val_loss: 2.2860 - val_acc: 0.2807
Epoch 111/1000
2334/2334 [==============================] - 5s - loss: 0.9750 - acc: 0.6050 - val_loss: 2.5808 - val_acc: 0.2488
Epoch 112/1000
2334/2334 [==============================] - 5s - loss: 0.9675 - acc: 0.6058 - val_loss: 2.0360 - val_acc: 0.3157
Epoch 113/1000
2334/2334 [==============================] - 5s - loss: 0.9547 - acc: 0.6268 - val_loss: 2.6023 - val_acc: 0.2627
Epoch 114/1000
2334/2334 [==============================] - 5s - loss: 0.9785 - acc: 0.6123 - val_loss: 2.3848 - val_acc: 0.2797
Epoch 115/1000
2334/2334 [==============================] - 5s - loss: 0.9320 - acc: 0.6161 - val_loss: 2.2806 - val_acc: 0.2907
Epoch 116/1000
2334/2334 [==============================] - 5s - loss: 0.9203 - acc: 0.6290 - val_loss: 2.0377 - val_acc: 0.3357
Epoch 117/1000
2334/2334 [==============================] - 5s - loss: 1.0044 - acc: 0.6093 - val_loss: 2.8662 - val_acc: 0.2408
Epoch 118/1000
2334/2334 [==============================] - 5s - loss: 0.9425 - acc: 0.6350 - val_loss: 2.5516 - val_acc: 0.2637
Epoch 119/1000
2334/2334 [==============================] - 5s - loss: 0.9296 - acc: 0.6470 - val_loss: 2.1993 - val_acc: 0.3107
Epoch 120/1000
2334/2334 [==============================] - 5s - loss: 0.8810 - acc: 0.6482 - val_loss: 2.1363 - val_acc: 0.3177
Epoch 121/1000
2334/2334 [==============================] - 5s - loss: 0.9154 - acc: 0.6491 - val_loss: 2.2669 - val_acc: 0.3167
Epoch 122/1000
2334/2334 [==============================] - 5s - loss: 0.8825 - acc: 0.6607 - val_loss: 2.1928 - val_acc: 0.3207
Epoch 123/1000
2334/2334 [==============================] - 5s - loss: 0.8992 - acc: 0.6422 - val_loss: 2.2635 - val_acc: 0.2867
Epoch 124/1000
2334/2334 [==============================] - 5s - loss: 0.8733 - acc: 0.6444 - val_loss: 2.8255 - val_acc: 0.2677
Epoch 125/1000
2334/2334 [==============================] - 5s - loss: 0.8577 - acc: 0.6560 - val_loss: 2.1706 - val_acc: 0.3536
Epoch 126/1000
2334/2334 [==============================] - 5s - loss: 0.8671 - acc: 0.6641 - val_loss: 2.8523 - val_acc: 0.2657
Epoch 127/1000
2334/2334 [==============================] - 5s - loss: 0.8639 - acc: 0.6555 - val_loss: 2.3048 - val_acc: 0.3317
Epoch 128/1000
2334/2334 [==============================] - 5s - loss: 0.8530 - acc: 0.6530 - val_loss: 2.5344 - val_acc: 0.2987
Epoch 129/1000
2334/2334 [==============================] - 5s - loss: 0.8225 - acc: 0.6859 - val_loss: 2.6624 - val_acc: 0.2947
Epoch 130/1000
2334/2334 [==============================] - 5s - loss: 0.8324 - acc: 0.6718 - val_loss: 2.6191 - val_acc: 0.2897
Epoch 131/1000
2334/2334 [==============================] - 5s - loss: 0.8049 - acc: 0.6881 - val_loss: 2.1588 - val_acc: 0.3397
Epoch 132/1000
2334/2334 [==============================] - 5s - loss: 0.7697 - acc: 0.7014 - val_loss: 2.6757 - val_acc: 0.3037
Epoch 133/1000
2334/2334 [==============================] - 5s - loss: 0.7834 - acc: 0.6967 - val_loss: 3.2066 - val_acc: 0.2607
Epoch 134/1000
2334/2334 [==============================] - 5s - loss: 0.7940 - acc: 0.6962 - val_loss: 3.0269 - val_acc: 0.2957
Epoch 135/1000
2334/2334 [==============================] - 5s - loss: 0.7648 - acc: 0.7044 - val_loss: 2.5653 - val_acc: 0.2787
Epoch 136/1000
2334/2334 [==============================] - 5s - loss: 0.7674 - acc: 0.7142 - val_loss: 2.6823 - val_acc: 0.3167
Epoch 137/1000
2334/2334 [==============================] - 5s - loss: 0.7498 - acc: 0.7027 - val_loss: 2.7217 - val_acc: 0.3187
Epoch 138/1000
2334/2334 [==============================] - 5s - loss: 0.7650 - acc: 0.7035 - val_loss: 2.3137 - val_acc: 0.3227
Epoch 139/1000
2334/2334 [==============================] - 5s - loss: 0.7631 - acc: 0.7057 - val_loss: 2.5427 - val_acc: 0.3237
Epoch 140/1000
2334/2334 [==============================] - 5s - loss: 0.7581 - acc: 0.7057 - val_loss: 2.4763 - val_acc: 0.3197
Epoch 141/1000
2334/2334 [==============================] - 5s - loss: 0.7492 - acc: 0.7121 - val_loss: 2.9275 - val_acc: 0.2997
Epoch 142/1000
2334/2334 [==============================] - 5s - loss: 0.7309 - acc: 0.7249 - val_loss: 2.2279 - val_acc: 0.3646
Epoch 143/1000
2334/2334 [==============================] - 5s - loss: 0.7635 - acc: 0.7048 - val_loss: 2.1914 - val_acc: 0.3427
Epoch 144/1000
2334/2334 [==============================] - 5s - loss: 0.7498 - acc: 0.7104 - val_loss: 3.2203 - val_acc: 0.2667
Epoch 145/1000
2334/2334 [==============================] - 5s - loss: 0.7437 - acc: 0.7099 - val_loss: 2.6944 - val_acc: 0.3267
Epoch 146/1000
2334/2334 [==============================] - 5s - loss: 0.7174 - acc: 0.7271 - val_loss: 2.7480 - val_acc: 0.3217
Epoch 147/1000
2334/2334 [==============================] - 5s - loss: 0.7012 - acc: 0.7412 - val_loss: 2.4474 - val_acc: 0.3407
Epoch 148/1000
2334/2334 [==============================] - 5s - loss: 0.6920 - acc: 0.7301 - val_loss: 2.5163 - val_acc: 0.3467
Epoch 149/1000
2334/2334 [==============================] - 5s - loss: 0.6760 - acc: 0.7446 - val_loss: 2.0758 - val_acc: 0.3996
Epoch 150/1000
2334/2334 [==============================] - 5s - loss: 0.7175 - acc: 0.7232 - val_loss: 2.8141 - val_acc: 0.3157
Epoch 151/1000
2334/2334 [==============================] - 5s - loss: 0.6739 - acc: 0.7314 - val_loss: 2.5314 - val_acc: 0.3686
Epoch 152/1000
2334/2334 [==============================] - 5s - loss: 0.6940 - acc: 0.7365 - val_loss: 2.5676 - val_acc: 0.3317
Epoch 153/1000
2334/2334 [==============================] - 5s - loss: 0.6950 - acc: 0.7429 - val_loss: 2.6513 - val_acc: 0.3427
Epoch 154/1000
2334/2334 [==============================] - 5s - loss: 0.6417 - acc: 0.7622 - val_loss: 2.2526 - val_acc: 0.3696
Epoch 155/1000
2334/2334 [==============================] - 5s - loss: 0.6380 - acc: 0.7549 - val_loss: 2.6590 - val_acc: 0.3067
Epoch 156/1000
2334/2334 [==============================] - 5s - loss: 0.7044 - acc: 0.7151 - val_loss: 2.5518 - val_acc: 0.3916
Epoch 157/1000
2334/2334 [==============================] - 5s - loss: 0.6685 - acc: 0.7468 - val_loss: 2.6335 - val_acc: 0.3417
Epoch 158/1000
2334/2334 [==============================] - 5s - loss: 0.6270 - acc: 0.7601 - val_loss: 2.9593 - val_acc: 0.3347
Epoch 159/1000
2334/2334 [==============================] - 5s - loss: 0.6094 - acc: 0.7656 - val_loss: 2.5912 - val_acc: 0.3457
Epoch 160/1000
2334/2334 [==============================] - 5s - loss: 0.6190 - acc: 0.7656 - val_loss: 2.3027 - val_acc: 0.3786
Epoch 161/1000
2334/2334 [==============================] - 5s - loss: 0.6000 - acc: 0.7716 - val_loss: 3.2102 - val_acc: 0.3157
Epoch 162/1000
2334/2334 [==============================] - 5s - loss: 0.6350 - acc: 0.7609 - val_loss: 2.8088 - val_acc: 0.3117
Epoch 163/1000
2334/2334 [==============================] - 5s - loss: 0.6551 - acc: 0.7468 - val_loss: 3.8372 - val_acc: 0.3057
Epoch 164/1000
2334/2334 [==============================] - 5s - loss: 0.6581 - acc: 0.7631 - val_loss: 2.4115 - val_acc: 0.3516
Epoch 165/1000
2334/2334 [==============================] - 5s - loss: 0.7036 - acc: 0.7232 - val_loss: 3.1528 - val_acc: 0.3137
Epoch 166/1000
2334/2334 [==============================] - 5s - loss: 0.6327 - acc: 0.7455 - val_loss: 2.1200 - val_acc: 0.4236
Epoch 167/1000
2334/2334 [==============================] - 5s - loss: 0.6421 - acc: 0.7524 - val_loss: 2.4633 - val_acc: 0.3686
Epoch 168/1000
2334/2334 [==============================] - 5s - loss: 0.5979 - acc: 0.7751 - val_loss: 2.6870 - val_acc: 0.3387
Epoch 169/1000
2334/2334 [==============================] - 5s - loss: 0.5777 - acc: 0.7746 - val_loss: 2.6390 - val_acc: 0.3467
Epoch 170/1000
2334/2334 [==============================] - 5s - loss: 0.5519 - acc: 0.7892 - val_loss: 2.3434 - val_acc: 0.3856
Epoch 171/1000
2334/2334 [==============================] - 5s - loss: 0.5820 - acc: 0.7866 - val_loss: 2.2342 - val_acc: 0.3816
Epoch 172/1000
2334/2334 [==============================] - 5s - loss: 0.5382 - acc: 0.7986 - val_loss: 3.1975 - val_acc: 0.3596
Epoch 173/1000
2334/2334 [==============================] - 5s - loss: 0.5679 - acc: 0.7879 - val_loss: 3.5319 - val_acc: 0.3277
Epoch 174/1000
2334/2334 [==============================] - 5s - loss: 0.5485 - acc: 0.7956 - val_loss: 2.8395 - val_acc: 0.3497
Epoch 175/1000
2334/2334 [==============================] - 5s - loss: 0.5836 - acc: 0.7738 - val_loss: 2.9400 - val_acc: 0.3556
Epoch 176/1000
2334/2334 [==============================] - 5s - loss: 0.5592 - acc: 0.7956 - val_loss: 2.9246 - val_acc: 0.3676
Epoch 177/1000
2334/2334 [==============================] - 5s - loss: 0.5408 - acc: 0.7918 - val_loss: 3.0921 - val_acc: 0.3367
Epoch 178/1000
2334/2334 [==============================] - 5s - loss: 0.5653 - acc: 0.7836 - val_loss: 3.2596 - val_acc: 0.3317
Epoch 179/1000
2334/2334 [==============================] - 5s - loss: 0.5320 - acc: 0.7956 - val_loss: 2.5907 - val_acc: 0.3806
Epoch 180/1000
2334/2334 [==============================] - 5s - loss: 0.5395 - acc: 0.8111 - val_loss: 2.7401 - val_acc: 0.3766
Epoch 181/1000
2334/2334 [==============================] - 5s - loss: 0.5319 - acc: 0.7948 - val_loss: 3.5445 - val_acc: 0.3107
Epoch 182/1000
2334/2334 [==============================] - 5s - loss: 0.5987 - acc: 0.7811 - val_loss: 3.8352 - val_acc: 0.2817
Epoch 183/1000
2334/2334 [==============================] - 5s - loss: 0.5821 - acc: 0.7815 - val_loss: 1.8674 - val_acc: 0.4595
Epoch 184/1000
2334/2334 [==============================] - 5s - loss: 0.8047 - acc: 0.7001 - val_loss: 1.9992 - val_acc: 0.4346
Epoch 185/1000
2334/2334 [==============================] - 5s - loss: 0.8111 - acc: 0.7095 - val_loss: 3.5712 - val_acc: 0.2198
Epoch 186/1000
2334/2334 [==============================] - 5s - loss: 0.7378 - acc: 0.7266 - val_loss: 2.7030 - val_acc: 0.3347
Epoch 187/1000
2334/2334 [==============================] - 5s - loss: 0.7186 - acc: 0.7344 - val_loss: 2.1005 - val_acc: 0.3606
Epoch 188/1000
2334/2334 [==============================] - 5s - loss: 0.6523 - acc: 0.7566 - val_loss: 2.2130 - val_acc: 0.3606
Epoch 189/1000
2334/2334 [==============================] - 5s - loss: 0.5921 - acc: 0.7712 - val_loss: 1.8292 - val_acc: 0.4206
Epoch 190/1000
2334/2334 [==============================] - 5s - loss: 0.5778 - acc: 0.7901 - val_loss: 1.9213 - val_acc: 0.3976
Epoch 191/1000
2334/2334 [==============================] - 5s - loss: 0.5142 - acc: 0.8072 - val_loss: 2.4264 - val_acc: 0.3686
Epoch 192/1000
2334/2334 [==============================] - 5s - loss: 0.5109 - acc: 0.8098 - val_loss: 2.6216 - val_acc: 0.3796
Epoch 193/1000
2334/2334 [==============================] - 5s - loss: 0.5405 - acc: 0.8059 - val_loss: 2.5126 - val_acc: 0.3766
Epoch 194/1000
2334/2334 [==============================] - 5s - loss: 0.5565 - acc: 0.7939 - val_loss: 2.7827 - val_acc: 0.3287
Epoch 195/1000
2334/2334 [==============================] - 5s - loss: 0.5046 - acc: 0.8068 - val_loss: 2.8504 - val_acc: 0.3706
Epoch 196/1000
2334/2334 [==============================] - 5s - loss: 0.4704 - acc: 0.8231 - val_loss: 2.5808 - val_acc: 0.3816
Epoch 197/1000
2334/2334 [==============================] - 5s - loss: 0.4941 - acc: 0.8153 - val_loss: 2.7005 - val_acc: 0.3586
Epoch 198/1000
2334/2334 [==============================] - 5s - loss: 0.4687 - acc: 0.8299 - val_loss: 2.2426 - val_acc: 0.4246
Epoch 199/1000
2334/2334 [==============================] - 5s - loss: 0.4881 - acc: 0.8175 - val_loss: 2.5713 - val_acc: 0.3826
Epoch 200/1000
2334/2334 [==============================] - 5s - loss: 0.4758 - acc: 0.8183 - val_loss: 2.7322 - val_acc: 0.3876
Epoch 201/1000
2334/2334 [==============================] - 5s - loss: 0.4827 - acc: 0.8162 - val_loss: 2.8197 - val_acc: 0.3576
Epoch 202/1000
2334/2334 [==============================] - 5s - loss: 0.4416 - acc: 0.8432 - val_loss: 3.0215 - val_acc: 0.3427
Epoch 203/1000
2334/2334 [==============================] - 5s - loss: 0.4711 - acc: 0.8248 - val_loss: 2.3193 - val_acc: 0.4276
Epoch 204/1000
2334/2334 [==============================] - 5s - loss: 0.5119 - acc: 0.8102 - val_loss: 2.2125 - val_acc: 0.4665
Epoch 205/1000
2334/2334 [==============================] - 5s - loss: 0.5198 - acc: 0.8162 - val_loss: 2.8030 - val_acc: 0.3377
Epoch 206/1000
2334/2334 [==============================] - 5s - loss: 0.5028 - acc: 0.8166 - val_loss: 2.5825 - val_acc: 0.3676
Epoch 207/1000
2334/2334 [==============================] - 5s - loss: 0.4894 - acc: 0.8149 - val_loss: 2.6128 - val_acc: 0.3976
Epoch 208/1000
2334/2334 [==============================] - 5s - loss: 0.4819 - acc: 0.8209 - val_loss: 2.6581 - val_acc: 0.3586
Epoch 209/1000
2334/2334 [==============================] - 5s - loss: 0.4767 - acc: 0.8273 - val_loss: 2.5077 - val_acc: 0.4146
Epoch 210/1000
2334/2334 [==============================] - 5s - loss: 0.4597 - acc: 0.8295 - val_loss: 2.3046 - val_acc: 0.3946
Epoch 211/1000
2334/2334 [==============================] - 5s - loss: 0.4275 - acc: 0.8479 - val_loss: 2.8232 - val_acc: 0.3756
Epoch 212/1000
2334/2334 [==============================] - 5s - loss: 0.4262 - acc: 0.8389 - val_loss: 2.9476 - val_acc: 0.4146
Epoch 213/1000
2334/2334 [==============================] - 5s - loss: 0.4291 - acc: 0.8406 - val_loss: 3.1509 - val_acc: 0.3477
Epoch 214/1000
2334/2334 [==============================] - 5s - loss: 0.4074 - acc: 0.8479 - val_loss: 2.8754 - val_acc: 0.3706
Epoch 215/1000
2334/2334 [==============================] - 5s - loss: 0.3958 - acc: 0.8556 - val_loss: 2.1990 - val_acc: 0.4486
Epoch 216/1000
2334/2334 [==============================] - 5s - loss: 0.4036 - acc: 0.8518 - val_loss: 2.1214 - val_acc: 0.4416
Epoch 217/1000
2334/2334 [==============================] - 5s - loss: 0.4239 - acc: 0.8470 - val_loss: 2.9446 - val_acc: 0.3636
Epoch 218/1000
2334/2334 [==============================] - 5s - loss: 0.4648 - acc: 0.8290 - val_loss: 2.8734 - val_acc: 0.3516
Epoch 219/1000
2334/2334 [==============================] - 5s - loss: 0.4519 - acc: 0.8346 - val_loss: 2.7022 - val_acc: 0.3956
Epoch 220/1000
2334/2334 [==============================] - 5s - loss: 0.4285 - acc: 0.8406 - val_loss: 2.7981 - val_acc: 0.3816
Epoch 221/1000
2334/2334 [==============================] - 5s - loss: 0.4269 - acc: 0.8458 - val_loss: 2.8235 - val_acc: 0.3706
Epoch 222/1000
2334/2334 [==============================] - 5s - loss: 0.4071 - acc: 0.8543 - val_loss: 2.7581 - val_acc: 0.3706
Epoch 223/1000
2334/2334 [==============================] - 5s - loss: 0.3999 - acc: 0.8582 - val_loss: 2.5743 - val_acc: 0.3946
Epoch 224/1000
2334/2334 [==============================] - 5s - loss: 0.3784 - acc: 0.8608 - val_loss: 2.6520 - val_acc: 0.3916
Epoch 225/1000
2334/2334 [==============================] - 5s - loss: 0.3947 - acc: 0.8608 - val_loss: 2.6739 - val_acc: 0.3906
Epoch 226/1000
2334/2334 [==============================] - 5s - loss: 0.4001 - acc: 0.8578 - val_loss: 2.3559 - val_acc: 0.4186
Epoch 227/1000
2334/2334 [==============================] - 5s - loss: 0.3788 - acc: 0.8710 - val_loss: 2.5372 - val_acc: 0.4016
Epoch 228/1000
2334/2334 [==============================] - 5s - loss: 0.3868 - acc: 0.8650 - val_loss: 3.0295 - val_acc: 0.3916
Epoch 229/1000
2334/2334 [==============================] - 5s - loss: 0.3528 - acc: 0.8663 - val_loss: 3.4209 - val_acc: 0.3526
Epoch 230/1000
2334/2334 [==============================] - 5s - loss: 0.3722 - acc: 0.8603 - val_loss: 3.2736 - val_acc: 0.3596
Epoch 231/1000
2334/2334 [==============================] - 5s - loss: 0.3497 - acc: 0.8779 - val_loss: 3.2119 - val_acc: 0.3796
Epoch 232/1000
2334/2334 [==============================] - 5s - loss: 0.3508 - acc: 0.8633 - val_loss: 3.4555 - val_acc: 0.3137
Epoch 233/1000
2334/2334 [==============================] - 5s - loss: 0.3500 - acc: 0.8723 - val_loss: 3.0591 - val_acc: 0.4016
Epoch 234/1000
2334/2334 [==============================] - 5s - loss: 0.3541 - acc: 0.8719 - val_loss: 3.3280 - val_acc: 0.3806
Epoch 235/1000
2334/2334 [==============================] - 5s - loss: 0.4203 - acc: 0.8475 - val_loss: 2.9372 - val_acc: 0.3766
Epoch 236/1000
2334/2334 [==============================] - 5s - loss: 0.3841 - acc: 0.8526 - val_loss: 2.7137 - val_acc: 0.3746
Epoch 237/1000
2334/2334 [==============================] - 5s - loss: 0.3851 - acc: 0.8569 - val_loss: 3.0266 - val_acc: 0.3716
Epoch 238/1000
2334/2334 [==============================] - 5s - loss: 0.3626 - acc: 0.8642 - val_loss: 2.8044 - val_acc: 0.3766
Epoch 239/1000
2334/2334 [==============================] - 5s - loss: 0.3695 - acc: 0.8693 - val_loss: 2.8806 - val_acc: 0.4036
Epoch 240/1000
2334/2334 [==============================] - 5s - loss: 0.3686 - acc: 0.8616 - val_loss: 2.7885 - val_acc: 0.3806
Epoch 241/1000
2334/2334 [==============================] - 5s - loss: 0.3678 - acc: 0.8668 - val_loss: 2.6777 - val_acc: 0.4036
Epoch 242/1000
2334/2334 [==============================] - 5s - loss: 0.3614 - acc: 0.8668 - val_loss: 2.4907 - val_acc: 0.4096
Epoch 243/1000
2334/2334 [==============================] - 5s - loss: 0.3319 - acc: 0.8826 - val_loss: 2.3704 - val_acc: 0.4565
Epoch 244/1000
2334/2334 [==============================] - 5s - loss: 0.3557 - acc: 0.8710 - val_loss: 2.9766 - val_acc: 0.3866
Epoch 245/1000
2334/2334 [==============================] - 5s - loss: 0.3631 - acc: 0.8663 - val_loss: 3.2279 - val_acc: 0.3656
Epoch 246/1000
2334/2334 [==============================] - 5s - loss: 0.3367 - acc: 0.8886 - val_loss: 3.8640 - val_acc: 0.3197
Epoch 247/1000
2334/2334 [==============================] - 5s - loss: 0.3599 - acc: 0.8680 - val_loss: 2.9078 - val_acc: 0.4046
Epoch 248/1000
2334/2334 [==============================] - 5s - loss: 0.3324 - acc: 0.8766 - val_loss: 2.8859 - val_acc: 0.3806
Epoch 249/1000
2334/2334 [==============================] - 5s - loss: 0.3359 - acc: 0.8787 - val_loss: 3.2426 - val_acc: 0.3576
Epoch 250/1000
2334/2334 [==============================] - 5s - loss: 0.3434 - acc: 0.8792 - val_loss: 2.9294 - val_acc: 0.3736
Epoch 251/1000
2334/2334 [==============================] - 5s - loss: 0.3562 - acc: 0.8655 - val_loss: 3.1908 - val_acc: 0.3676
Epoch 252/1000
2334/2334 [==============================] - 5s - loss: 0.3904 - acc: 0.8590 - val_loss: 2.7161 - val_acc: 0.4196
Epoch 253/1000
2334/2334 [==============================] - 5s - loss: 0.3522 - acc: 0.8766 - val_loss: 3.3972 - val_acc: 0.3377
Epoch 254/1000
2334/2334 [==============================] - 5s - loss: 0.3753 - acc: 0.8650 - val_loss: 2.7588 - val_acc: 0.4216
Epoch 255/1000
2334/2334 [==============================] - 5s - loss: 0.3520 - acc: 0.8689 - val_loss: 2.7851 - val_acc: 0.3776
Epoch 256/1000
2334/2334 [==============================] - 5s - loss: 0.3428 - acc: 0.8753 - val_loss: 2.7765 - val_acc: 0.3806
Epoch 257/1000
2334/2334 [==============================] - 5s - loss: 0.3269 - acc: 0.8766 - val_loss: 2.6559 - val_acc: 0.4266
Epoch 258/1000
2334/2334 [==============================] - 5s - loss: 0.3168 - acc: 0.8775 - val_loss: 3.0905 - val_acc: 0.3596
Epoch 259/1000
2334/2334 [==============================] - 5s - loss: 0.3134 - acc: 0.8890 - val_loss: 2.6439 - val_acc: 0.4116
Epoch 260/1000
2334/2334 [==============================] - 5s - loss: 0.3194 - acc: 0.8817 - val_loss: 2.8608 - val_acc: 0.3926
Epoch 261/1000
2334/2334 [==============================] - 5s - loss: 0.3138 - acc: 0.8830 - val_loss: 3.1808 - val_acc: 0.3946
Epoch 262/1000
2334/2334 [==============================] - 5s - loss: 0.3173 - acc: 0.8869 - val_loss: 3.8161 - val_acc: 0.3407
Epoch 263/1000
2334/2334 [==============================] - 5s - loss: 0.3117 - acc: 0.8865 - val_loss: 3.0341 - val_acc: 0.3866
Epoch 264/1000
2334/2334 [==============================] - 5s - loss: 0.3238 - acc: 0.8813 - val_loss: 3.2014 - val_acc: 0.3866
Epoch 265/1000
2334/2334 [==============================] - 5s - loss: 0.3182 - acc: 0.8800 - val_loss: 2.8649 - val_acc: 0.3886
Epoch 266/1000
2334/2334 [==============================] - 5s - loss: 0.3202 - acc: 0.8787 - val_loss: 3.4813 - val_acc: 0.3217
Epoch 267/1000
2334/2334 [==============================] - 5s - loss: 0.3370 - acc: 0.8817 - val_loss: 2.9733 - val_acc: 0.3856
Epoch 268/1000
2334/2334 [==============================] - 5s - loss: 0.3320 - acc: 0.8809 - val_loss: 2.6961 - val_acc: 0.4206
Epoch 269/1000
2334/2334 [==============================] - 5s - loss: 0.2985 - acc: 0.8955 - val_loss: 3.1380 - val_acc: 0.3596
Epoch 270/1000
2334/2334 [==============================] - 5s - loss: 0.2762 - acc: 0.8985 - val_loss: 2.6893 - val_acc: 0.4286
Epoch 271/1000
2334/2334 [==============================] - 5s - loss: 0.2747 - acc: 0.9087 - val_loss: 3.1743 - val_acc: 0.3886
Epoch 272/1000
2334/2334 [==============================] - 5s - loss: 0.2931 - acc: 0.8980 - val_loss: 2.9795 - val_acc: 0.4076
Epoch 273/1000
2334/2334 [==============================] - 5s - loss: 0.2896 - acc: 0.8997 - val_loss: 2.6777 - val_acc: 0.4196
Epoch 274/1000
2334/2334 [==============================] - 5s - loss: 0.2882 - acc: 0.8907 - val_loss: 2.7089 - val_acc: 0.4366
Epoch 275/1000
2334/2334 [==============================] - 5s - loss: 0.3105 - acc: 0.9006 - val_loss: 3.0117 - val_acc: 0.4036
Epoch 276/1000
2334/2334 [==============================] - 5s - loss: 0.2978 - acc: 0.8899 - val_loss: 2.9932 - val_acc: 0.4056
Epoch 277/1000
2334/2334 [==============================] - 5s - loss: 0.3024 - acc: 0.8937 - val_loss: 3.4970 - val_acc: 0.3706
Epoch 278/1000
2334/2334 [==============================] - 5s - loss: 0.3483 - acc: 0.8809 - val_loss: 3.0844 - val_acc: 0.3796
Epoch 279/1000
2334/2334 [==============================] - 5s - loss: 0.2998 - acc: 0.8946 - val_loss: 3.2109 - val_acc: 0.4106
Epoch 280/1000
2334/2334 [==============================] - 5s - loss: 0.3234 - acc: 0.8796 - val_loss: 3.6003 - val_acc: 0.3447
Epoch 281/1000
2334/2334 [==============================] - 5s - loss: 0.2884 - acc: 0.9015 - val_loss: 3.0627 - val_acc: 0.4006
Epoch 282/1000
2334/2334 [==============================] - 5s - loss: 0.3208 - acc: 0.8860 - val_loss: 2.4219 - val_acc: 0.4755
Epoch 283/1000
2334/2334 [==============================] - 5s - loss: 0.3273 - acc: 0.8775 - val_loss: 2.9134 - val_acc: 0.3736
Epoch 284/1000
2334/2334 [==============================] - 5s - loss: 0.3011 - acc: 0.8912 - val_loss: 2.2847 - val_acc: 0.4555
Epoch 285/1000
2334/2334 [==============================] - 5s - loss: 0.3400 - acc: 0.8835 - val_loss: 2.4676 - val_acc: 0.4306
Epoch 286/1000
2334/2334 [==============================] - 5s - loss: 0.3397 - acc: 0.8882 - val_loss: 2.6191 - val_acc: 0.4036
Epoch 287/1000
2334/2334 [==============================] - 5s - loss: 0.3056 - acc: 0.8877 - val_loss: 2.8524 - val_acc: 0.4066
Epoch 288/1000
2334/2334 [==============================] - 5s - loss: 0.2901 - acc: 0.8925 - val_loss: 3.0563 - val_acc: 0.3866
Epoch 289/1000
2334/2334 [==============================] - 5s - loss: 0.2681 - acc: 0.8993 - val_loss: 3.0796 - val_acc: 0.3956
Epoch 290/1000
2334/2334 [==============================] - 5s - loss: 0.2731 - acc: 0.9010 - val_loss: 2.5202 - val_acc: 0.4635
Epoch 291/1000
2334/2334 [==============================] - 5s - loss: 0.2966 - acc: 0.8997 - val_loss: 3.0082 - val_acc: 0.3736
Epoch 292/1000
2334/2334 [==============================] - 5s - loss: 0.2925 - acc: 0.9019 - val_loss: 3.0467 - val_acc: 0.3936
Epoch 293/1000
2334/2334 [==============================] - 5s - loss: 0.2805 - acc: 0.9019 - val_loss: 3.8861 - val_acc: 0.3666
Epoch 294/1000
2334/2334 [==============================] - 5s - loss: 0.2663 - acc: 0.8997 - val_loss: 3.4427 - val_acc: 0.3876
Epoch 295/1000
2334/2334 [==============================] - 5s - loss: 0.2193 - acc: 0.9267 - val_loss: 2.6166 - val_acc: 0.4386
Epoch 296/1000
2334/2334 [==============================] - 5s - loss: 0.2466 - acc: 0.9169 - val_loss: 2.6677 - val_acc: 0.4386
Epoch 297/1000
2334/2334 [==============================] - 5s - loss: 0.2447 - acc: 0.9160 - val_loss: 2.9936 - val_acc: 0.4296
Epoch 298/1000
2334/2334 [==============================] - 5s - loss: 0.2450 - acc: 0.9186 - val_loss: 3.7678 - val_acc: 0.3566
Epoch 299/1000
2334/2334 [==============================] - 5s - loss: 0.2717 - acc: 0.9075 - val_loss: 3.3486 - val_acc: 0.4106
Epoch 300/1000
2334/2334 [==============================] - 5s - loss: 0.2561 - acc: 0.9220 - val_loss: 3.2814 - val_acc: 0.3856
Epoch 301/1000
2334/2334 [==============================] - 5s - loss: 0.2560 - acc: 0.9199 - val_loss: 2.4679 - val_acc: 0.4505
Epoch 302/1000
2334/2334 [==============================] - 5s - loss: 0.2159 - acc: 0.9220 - val_loss: 2.8101 - val_acc: 0.4156
Epoch 303/1000
2334/2334 [==============================] - 5s - loss: 0.2515 - acc: 0.9113 - val_loss: 3.0422 - val_acc: 0.4026
Epoch 304/1000
2334/2334 [==============================] - 5s - loss: 0.2325 - acc: 0.9177 - val_loss: 3.8286 - val_acc: 0.3516
Epoch 305/1000
2334/2334 [==============================] - 5s - loss: 0.2235 - acc: 0.9237 - val_loss: 3.5237 - val_acc: 0.3776
Epoch 306/1000
2334/2334 [==============================] - 5s - loss: 0.2342 - acc: 0.9109 - val_loss: 3.5583 - val_acc: 0.3756
Epoch 307/1000
2334/2334 [==============================] - 5s - loss: 0.2426 - acc: 0.9207 - val_loss: 2.8819 - val_acc: 0.4286
Epoch 308/1000
2334/2334 [==============================] - 5s - loss: 0.2368 - acc: 0.9173 - val_loss: 3.5928 - val_acc: 0.3606
Epoch 309/1000
2334/2334 [==============================] - 5s - loss: 0.2352 - acc: 0.9139 - val_loss: 4.0899 - val_acc: 0.3636
Epoch 310/1000
2334/2334 [==============================] - 5s - loss: 0.2400 - acc: 0.9186 - val_loss: 3.9694 - val_acc: 0.3656
Epoch 311/1000
2334/2334 [==============================] - 5s - loss: 0.2174 - acc: 0.9212 - val_loss: 3.4696 - val_acc: 0.3926
Epoch 312/1000
2334/2334 [==============================] - 5s - loss: 0.2428 - acc: 0.9182 - val_loss: 3.6694 - val_acc: 0.3876
Epoch 313/1000
2334/2334 [==============================] - 5s - loss: 0.2586 - acc: 0.9117 - val_loss: 3.9604 - val_acc: 0.3556
Epoch 314/1000
2334/2334 [==============================] - 5s - loss: 0.2570 - acc: 0.9096 - val_loss: 4.0416 - val_acc: 0.3546
Epoch 315/1000
2334/2334 [==============================] - 5s - loss: 0.2256 - acc: 0.9246 - val_loss: 3.4030 - val_acc: 0.4016
Epoch 316/1000
2334/2334 [==============================] - 5s - loss: 0.2440 - acc: 0.9113 - val_loss: 3.1503 - val_acc: 0.3916
Epoch 317/1000
2334/2334 [==============================] - 5s - loss: 0.1999 - acc: 0.9284 - val_loss: 2.8280 - val_acc: 0.4366
Epoch 318/1000
2334/2334 [==============================] - 5s - loss: 0.2259 - acc: 0.9220 - val_loss: 2.9424 - val_acc: 0.4356
Epoch 319/1000
2334/2334 [==============================] - 5s - loss: 0.2004 - acc: 0.9250 - val_loss: 3.1536 - val_acc: 0.4086
Epoch 320/1000
2334/2334 [==============================] - 5s - loss: 0.2074 - acc: 0.9242 - val_loss: 3.2695 - val_acc: 0.4076
Epoch 321/1000
2334/2334 [==============================] - 5s - loss: 0.2153 - acc: 0.9254 - val_loss: 3.3838 - val_acc: 0.4176
Epoch 322/1000
2334/2334 [==============================] - 5s - loss: 0.2049 - acc: 0.9293 - val_loss: 3.3721 - val_acc: 0.3796
Epoch 323/1000
2334/2334 [==============================] - 5s - loss: 0.1906 - acc: 0.9254 - val_loss: 3.2148 - val_acc: 0.4006
Epoch 324/1000
2334/2334 [==============================] - 5s - loss: 0.2177 - acc: 0.9212 - val_loss: 3.9918 - val_acc: 0.3676
Epoch 325/1000
2334/2334 [==============================] - 5s - loss: 0.1890 - acc: 0.9319 - val_loss: 3.4556 - val_acc: 0.3836
Epoch 326/1000
2334/2334 [==============================] - 5s - loss: 0.2073 - acc: 0.9284 - val_loss: 3.7738 - val_acc: 0.3646
Epoch 327/1000
2334/2334 [==============================] - 5s - loss: 0.2037 - acc: 0.9314 - val_loss: 3.3850 - val_acc: 0.4026
Epoch 328/1000
2334/2334 [==============================] - 5s - loss: 0.2186 - acc: 0.9233 - val_loss: 3.7429 - val_acc: 0.3936
Epoch 329/1000
2334/2334 [==============================] - 5s - loss: 0.1997 - acc: 0.9319 - val_loss: 3.3710 - val_acc: 0.3886
Epoch 330/1000
2334/2334 [==============================] - 5s - loss: 0.2490 - acc: 0.9126 - val_loss: 3.2571 - val_acc: 0.4016
Epoch 331/1000
2334/2334 [==============================] - 5s - loss: 0.3098 - acc: 0.8942 - val_loss: 3.5310 - val_acc: 0.3846
Epoch 332/1000
2334/2334 [==============================] - 5s - loss: 0.2523 - acc: 0.9135 - val_loss: 2.9789 - val_acc: 0.4386
Epoch 333/1000
2334/2334 [==============================] - 5s - loss: 0.2249 - acc: 0.9246 - val_loss: 3.0238 - val_acc: 0.4246
Epoch 334/1000
2334/2334 [==============================] - 5s - loss: 0.2065 - acc: 0.9289 - val_loss: 2.8121 - val_acc: 0.4266
Epoch 335/1000
2334/2334 [==============================] - 5s - loss: 0.2087 - acc: 0.9280 - val_loss: 2.9623 - val_acc: 0.4066
Epoch 336/1000
2334/2334 [==============================] - 5s - loss: 0.1989 - acc: 0.9319 - val_loss: 3.1998 - val_acc: 0.4026
Epoch 337/1000
2334/2334 [==============================] - 5s - loss: 0.2008 - acc: 0.9254 - val_loss: 3.5803 - val_acc: 0.3826
Epoch 338/1000
2334/2334 [==============================] - 5s - loss: 0.2175 - acc: 0.9306 - val_loss: 3.2391 - val_acc: 0.3996
Epoch 339/1000
2334/2334 [==============================] - 5s - loss: 0.1982 - acc: 0.9289 - val_loss: 3.4583 - val_acc: 0.3946
Epoch 340/1000
2334/2334 [==============================] - 5s - loss: 0.1997 - acc: 0.9374 - val_loss: 3.2801 - val_acc: 0.4496
Epoch 341/1000
2334/2334 [==============================] - 5s - loss: 0.2122 - acc: 0.9327 - val_loss: 3.3183 - val_acc: 0.4216
Epoch 342/1000
2334/2334 [==============================] - 5s - loss: 0.2290 - acc: 0.9216 - val_loss: 3.2943 - val_acc: 0.3926
Epoch 343/1000
2334/2334 [==============================] - 5s - loss: 0.2396 - acc: 0.9156 - val_loss: 3.4300 - val_acc: 0.4046
Epoch 344/1000
2334/2334 [==============================] - 5s - loss: 0.2378 - acc: 0.9152 - val_loss: 4.1325 - val_acc: 0.3726
Epoch 345/1000
2334/2334 [==============================] - 5s - loss: 0.3217 - acc: 0.8929 - val_loss: 4.1102 - val_acc: 0.3337
Epoch 346/1000
2334/2334 [==============================] - 5s - loss: 0.2687 - acc: 0.9027 - val_loss: 3.7664 - val_acc: 0.3616
Epoch 347/1000
2334/2334 [==============================] - 5s - loss: 0.2528 - acc: 0.9092 - val_loss: 3.1019 - val_acc: 0.4176
Epoch 348/1000
2334/2334 [==============================] - 5s - loss: 0.2480 - acc: 0.9130 - val_loss: 3.1081 - val_acc: 0.3846
Epoch 349/1000
2334/2334 [==============================] - 5s - loss: 0.2272 - acc: 0.9225 - val_loss: 3.5404 - val_acc: 0.3716
Epoch 350/1000
2334/2334 [==============================] - 5s - loss: 0.2055 - acc: 0.9327 - val_loss: 3.9274 - val_acc: 0.3616
Epoch 351/1000
2334/2334 [==============================] - 5s - loss: 0.2672 - acc: 0.9105 - val_loss: 3.1530 - val_acc: 0.3776
Epoch 352/1000
2334/2334 [==============================] - 5s - loss: 0.1846 - acc: 0.9349 - val_loss: 3.2431 - val_acc: 0.3956
Epoch 353/1000
2334/2334 [==============================] - 5s - loss: 0.2182 - acc: 0.9225 - val_loss: 2.8060 - val_acc: 0.4575
Epoch 354/1000
2334/2334 [==============================] - 5s - loss: 0.2167 - acc: 0.9225 - val_loss: 3.9535 - val_acc: 0.3536
Epoch 355/1000
2334/2334 [==============================] - 5s - loss: 0.2031 - acc: 0.9323 - val_loss: 3.0617 - val_acc: 0.4186
Epoch 356/1000
2334/2334 [==============================] - 5s - loss: 0.1853 - acc: 0.9323 - val_loss: 3.0731 - val_acc: 0.4146
Epoch 357/1000
2334/2334 [==============================] - 5s - loss: 0.1672 - acc: 0.9357 - val_loss: 3.6551 - val_acc: 0.3736
Epoch 358/1000
2334/2334 [==============================] - 5s - loss: 0.1842 - acc: 0.9443 - val_loss: 3.7988 - val_acc: 0.3786
Epoch 359/1000
2334/2334 [==============================] - 5s - loss: 0.1728 - acc: 0.9413 - val_loss: 3.4327 - val_acc: 0.4066
Epoch 360/1000
2334/2334 [==============================] - 5s - loss: 0.2009 - acc: 0.9259 - val_loss: 3.5114 - val_acc: 0.3746
Epoch 361/1000
2334/2334 [==============================] - 5s - loss: 0.1901 - acc: 0.9340 - val_loss: 3.8300 - val_acc: 0.3956
Epoch 362/1000
2334/2334 [==============================] - 5s - loss: 0.1928 - acc: 0.9297 - val_loss: 3.0252 - val_acc: 0.4266
Epoch 363/1000
2334/2334 [==============================] - 5s - loss: 0.2231 - acc: 0.9246 - val_loss: 2.9953 - val_acc: 0.4246
Epoch 364/1000
2334/2334 [==============================] - 5s - loss: 0.2139 - acc: 0.9293 - val_loss: 3.4641 - val_acc: 0.3886
Epoch 365/1000
2334/2334 [==============================] - 5s - loss: 0.1802 - acc: 0.9439 - val_loss: 3.6460 - val_acc: 0.3546
Epoch 366/1000
2334/2334 [==============================] - 5s - loss: 0.1728 - acc: 0.9344 - val_loss: 3.8430 - val_acc: 0.3646
Epoch 367/1000
2334/2334 [==============================] - 5s - loss: 0.1635 - acc: 0.9426 - val_loss: 3.6640 - val_acc: 0.3826
Epoch 368/1000
2334/2334 [==============================] - 5s - loss: 0.1566 - acc: 0.9443 - val_loss: 3.6473 - val_acc: 0.3776
Epoch 369/1000
2334/2334 [==============================] - 5s - loss: 0.1706 - acc: 0.9413 - val_loss: 3.5915 - val_acc: 0.4136
Epoch 370/1000
2334/2334 [==============================] - 5s - loss: 0.2037 - acc: 0.9327 - val_loss: 3.3264 - val_acc: 0.4006
Epoch 371/1000
2334/2334 [==============================] - 5s - loss: 0.1939 - acc: 0.9392 - val_loss: 3.4452 - val_acc: 0.4036
Epoch 372/1000
2334/2334 [==============================] - 5s - loss: 0.1852 - acc: 0.9366 - val_loss: 3.5622 - val_acc: 0.4176
Epoch 373/1000
2334/2334 [==============================] - 5s - loss: 0.1902 - acc: 0.9332 - val_loss: 4.0465 - val_acc: 0.3536
Epoch 374/1000
2334/2334 [==============================] - 5s - loss: 0.1890 - acc: 0.9332 - val_loss: 4.2570 - val_acc: 0.3716
Epoch 375/1000
2334/2334 [==============================] - 5s - loss: 0.1708 - acc: 0.9430 - val_loss: 4.0982 - val_acc: 0.3556
Epoch 376/1000
2334/2334 [==============================] - 5s - loss: 0.1876 - acc: 0.9357 - val_loss: 3.0254 - val_acc: 0.4176
Epoch 377/1000
2334/2334 [==============================] - 5s - loss: 0.1544 - acc: 0.9452 - val_loss: 3.1999 - val_acc: 0.3906
Epoch 378/1000
2334/2334 [==============================] - 5s - loss: 0.1942 - acc: 0.9387 - val_loss: 2.8113 - val_acc: 0.4496
Epoch 379/1000
2334/2334 [==============================] - 5s - loss: 0.1565 - acc: 0.9400 - val_loss: 3.2314 - val_acc: 0.4136
Epoch 380/1000
2334/2334 [==============================] - 5s - loss: 0.1504 - acc: 0.9473 - val_loss: 3.8831 - val_acc: 0.3876
Epoch 381/1000
2334/2334 [==============================] - 5s - loss: 0.1522 - acc: 0.9443 - val_loss: 3.5233 - val_acc: 0.3816
Epoch 382/1000
2334/2334 [==============================] - 5s - loss: 0.1683 - acc: 0.9460 - val_loss: 3.2028 - val_acc: 0.4286
Epoch 383/1000
2334/2334 [==============================] - 5s - loss: 0.1629 - acc: 0.9477 - val_loss: 3.0448 - val_acc: 0.4436
Epoch 384/1000
2334/2334 [==============================] - 5s - loss: 0.1741 - acc: 0.9387 - val_loss: 4.0782 - val_acc: 0.3546
Epoch 385/1000
2334/2334 [==============================] - 5s - loss: 0.2196 - acc: 0.9216 - val_loss: 3.4839 - val_acc: 0.3866
Epoch 386/1000
2334/2334 [==============================] - 5s - loss: 0.2224 - acc: 0.9177 - val_loss: 4.2527 - val_acc: 0.3656
Epoch 387/1000
2334/2334 [==============================] - 5s - loss: 0.2190 - acc: 0.9297 - val_loss: 3.2477 - val_acc: 0.4086
Epoch 388/1000
2334/2334 [==============================] - 5s - loss: 0.2207 - acc: 0.9272 - val_loss: 3.1741 - val_acc: 0.4406
Epoch 389/1000
2334/2334 [==============================] - 5s - loss: 0.2206 - acc: 0.9272 - val_loss: 3.0655 - val_acc: 0.4146
Epoch 390/1000
2334/2334 [==============================] - 5s - loss: 0.1873 - acc: 0.9409 - val_loss: 3.4278 - val_acc: 0.3916
Epoch 391/1000
2334/2334 [==============================] - 5s - loss: 0.1627 - acc: 0.9443 - val_loss: 4.0248 - val_acc: 0.3546
Epoch 392/1000
2334/2334 [==============================] - 5s - loss: 0.1550 - acc: 0.9452 - val_loss: 3.9440 - val_acc: 0.3786
Epoch 393/1000
2334/2334 [==============================] - 5s - loss: 0.1688 - acc: 0.9422 - val_loss: 3.6399 - val_acc: 0.4066
Epoch 394/1000
2334/2334 [==============================] - 5s - loss: 0.1597 - acc: 0.9452 - val_loss: 3.4886 - val_acc: 0.3996
Epoch 395/1000
2334/2334 [==============================] - 5s - loss: 0.1606 - acc: 0.9473 - val_loss: 3.7434 - val_acc: 0.3976
Epoch 396/1000
2334/2334 [==============================] - 5s - loss: 0.1432 - acc: 0.9499 - val_loss: 3.5177 - val_acc: 0.4036
Epoch 397/1000
2334/2334 [==============================] - 5s - loss: 0.1614 - acc: 0.9452 - val_loss: 3.3474 - val_acc: 0.4206
Epoch 398/1000
2334/2334 [==============================] - 5s - loss: 0.1614 - acc: 0.9464 - val_loss: 3.1237 - val_acc: 0.4256
Epoch 399/1000
2334/2334 [==============================] - 5s - loss: 0.1729 - acc: 0.9413 - val_loss: 3.4564 - val_acc: 0.4106
Epoch 400/1000
2334/2334 [==============================] - 5s - loss: 0.1638 - acc: 0.9434 - val_loss: 3.8146 - val_acc: 0.3776
Epoch 401/1000
2334/2334 [==============================] - 5s - loss: 0.1795 - acc: 0.9413 - val_loss: 3.6816 - val_acc: 0.3866
Epoch 402/1000
2334/2334 [==============================] - 5s - loss: 0.1607 - acc: 0.9422 - val_loss: 3.4699 - val_acc: 0.4226
Epoch 403/1000
2334/2334 [==============================] - 5s - loss: 0.1493 - acc: 0.9456 - val_loss: 3.6837 - val_acc: 0.4096
Epoch 404/1000
2334/2334 [==============================] - 5s - loss: 0.1593 - acc: 0.9533 - val_loss: 4.4210 - val_acc: 0.3477
Epoch 405/1000
2334/2334 [==============================] - 5s - loss: 0.1705 - acc: 0.9400 - val_loss: 2.8751 - val_acc: 0.4695
Epoch 406/1000
2334/2334 [==============================] - 5s - loss: 0.1569 - acc: 0.9469 - val_loss: 2.9207 - val_acc: 0.4695
Epoch 407/1000
2334/2334 [==============================] - 5s - loss: 0.1587 - acc: 0.9469 - val_loss: 3.6770 - val_acc: 0.3946
Epoch 408/1000
2334/2334 [==============================] - 5s - loss: 0.1371 - acc: 0.9507 - val_loss: 4.4853 - val_acc: 0.3656
Epoch 409/1000
2334/2334 [==============================] - 5s - loss: 0.1556 - acc: 0.9473 - val_loss: 4.0113 - val_acc: 0.3756
Epoch 410/1000
2334/2334 [==============================] - 5s - loss: 0.1751 - acc: 0.9417 - val_loss: 2.7157 - val_acc: 0.4875
Epoch 411/1000
2334/2334 [==============================] - 5s - loss: 0.1565 - acc: 0.9473 - val_loss: 3.7687 - val_acc: 0.3786
Epoch 412/1000
2334/2334 [==============================] - 5s - loss: 0.1592 - acc: 0.9490 - val_loss: 3.5991 - val_acc: 0.4006
Epoch 413/1000
2334/2334 [==============================] - 5s - loss: 0.1836 - acc: 0.9379 - val_loss: 2.4345 - val_acc: 0.4955
Epoch 414/1000
2334/2334 [==============================] - 5s - loss: 0.2317 - acc: 0.9190 - val_loss: 4.1326 - val_acc: 0.3497
Epoch 415/1000
2334/2334 [==============================] - 5s - loss: 0.2271 - acc: 0.9199 - val_loss: 2.6805 - val_acc: 0.4565
Epoch 416/1000
2334/2334 [==============================] - 5s - loss: 0.1875 - acc: 0.9374 - val_loss: 3.7319 - val_acc: 0.3886
Epoch 417/1000
2334/2334 [==============================] - 5s - loss: 0.1894 - acc: 0.9336 - val_loss: 3.3302 - val_acc: 0.3846
Epoch 418/1000
2334/2334 [==============================] - 5s - loss: 0.1701 - acc: 0.9477 - val_loss: 3.5050 - val_acc: 0.4116
Epoch 419/1000
2334/2334 [==============================] - 5s - loss: 0.2250 - acc: 0.9263 - val_loss: 3.0578 - val_acc: 0.4126
Epoch 420/1000
2334/2334 [==============================] - 5s - loss: 0.1989 - acc: 0.9280 - val_loss: 3.0923 - val_acc: 0.4196
Epoch 421/1000
2334/2334 [==============================] - 5s - loss: 0.1808 - acc: 0.9340 - val_loss: 3.5499 - val_acc: 0.4016
Epoch 422/1000
2334/2334 [==============================] - 5s - loss: 0.1654 - acc: 0.9456 - val_loss: 4.0579 - val_acc: 0.3556
Epoch 423/1000
2334/2334 [==============================] - 5s - loss: 0.1497 - acc: 0.9490 - val_loss: 3.1585 - val_acc: 0.4146
Epoch 424/1000
2334/2334 [==============================] - 5s - loss: 0.1446 - acc: 0.9512 - val_loss: 3.6326 - val_acc: 0.3956
Epoch 425/1000
2334/2334 [==============================] - 5s - loss: 0.1484 - acc: 0.9524 - val_loss: 3.8561 - val_acc: 0.3786
Epoch 426/1000
2334/2334 [==============================] - 5s - loss: 0.1694 - acc: 0.9417 - val_loss: 3.2365 - val_acc: 0.4306
Epoch 427/1000
2334/2334 [==============================] - 5s - loss: 0.1460 - acc: 0.9477 - val_loss: 3.7424 - val_acc: 0.4046
Epoch 428/1000
2334/2334 [==============================] - 5s - loss: 0.1651 - acc: 0.9426 - val_loss: 3.8810 - val_acc: 0.3886
Epoch 429/1000
2334/2334 [==============================] - 5s - loss: 0.1602 - acc: 0.9524 - val_loss: 3.7256 - val_acc: 0.3886
Epoch 430/1000
2334/2334 [==============================] - 5s - loss: 0.1394 - acc: 0.9529 - val_loss: 2.7392 - val_acc: 0.4535
Epoch 431/1000
2334/2334 [==============================] - 5s - loss: 0.1788 - acc: 0.9439 - val_loss: 2.7881 - val_acc: 0.4685
Epoch 432/1000
2334/2334 [==============================] - 5s - loss: 0.1596 - acc: 0.9439 - val_loss: 3.0486 - val_acc: 0.4366
Epoch 433/1000
2334/2334 [==============================] - 5s - loss: 0.1692 - acc: 0.9439 - val_loss: 3.4321 - val_acc: 0.4206
Epoch 434/1000
2334/2334 [==============================] - 5s - loss: 0.1514 - acc: 0.9486 - val_loss: 3.7263 - val_acc: 0.3996
Epoch 435/1000
2334/2334 [==============================] - 5s - loss: 0.1427 - acc: 0.9499 - val_loss: 4.0455 - val_acc: 0.3636
Epoch 436/1000
2334/2334 [==============================] - 5s - loss: 0.1478 - acc: 0.9524 - val_loss: 3.8051 - val_acc: 0.3906
Epoch 437/1000
2334/2334 [==============================] - 5s - loss: 0.1435 - acc: 0.9563 - val_loss: 3.5288 - val_acc: 0.4046
Epoch 438/1000
2334/2334 [==============================] - 5s - loss: 0.1194 - acc: 0.9593 - val_loss: 3.1948 - val_acc: 0.4366
Epoch 439/1000
2334/2334 [==============================] - 5s - loss: 0.1519 - acc: 0.9482 - val_loss: 4.2032 - val_acc: 0.3816
Epoch 440/1000
2334/2334 [==============================] - 5s - loss: 0.1233 - acc: 0.9636 - val_loss: 4.0803 - val_acc: 0.3756
Epoch 441/1000
2334/2334 [==============================] - 5s - loss: 0.1295 - acc: 0.9563 - val_loss: 4.3894 - val_acc: 0.3646
Epoch 442/1000
2334/2334 [==============================] - 5s - loss: 0.1562 - acc: 0.9460 - val_loss: 4.3053 - val_acc: 0.3656
Epoch 443/1000
2334/2334 [==============================] - 5s - loss: 0.1851 - acc: 0.9383 - val_loss: 3.6834 - val_acc: 0.3786
Epoch 444/1000
2334/2334 [==============================] - 5s - loss: 0.1595 - acc: 0.9477 - val_loss: 4.8990 - val_acc: 0.3516
Epoch 445/1000
2334/2334 [==============================] - 5s - loss: 0.1639 - acc: 0.9447 - val_loss: 3.7801 - val_acc: 0.3756
Epoch 446/1000
2334/2334 [==============================] - 5s - loss: 0.1709 - acc: 0.9422 - val_loss: 3.4806 - val_acc: 0.3956
Epoch 447/1000
2334/2334 [==============================] - 5s - loss: 0.1539 - acc: 0.9434 - val_loss: 2.4131 - val_acc: 0.4835
Epoch 448/1000
2334/2334 [==============================] - 5s - loss: 0.1875 - acc: 0.9379 - val_loss: 3.9403 - val_acc: 0.3856
Epoch 449/1000
2334/2334 [==============================] - 5s - loss: 0.1689 - acc: 0.9503 - val_loss: 2.7750 - val_acc: 0.4615
Epoch 450/1000
2334/2334 [==============================] - 5s - loss: 0.2073 - acc: 0.9302 - val_loss: 3.2799 - val_acc: 0.4076
Epoch 451/1000
2334/2334 [==============================] - 5s - loss: 0.1802 - acc: 0.9366 - val_loss: 3.1798 - val_acc: 0.4136
Epoch 452/1000
2334/2334 [==============================] - 5s - loss: 0.1832 - acc: 0.9374 - val_loss: 3.1308 - val_acc: 0.4346
Epoch 453/1000
2334/2334 [==============================] - 5s - loss: 0.1656 - acc: 0.9434 - val_loss: 3.2846 - val_acc: 0.3986
Epoch 454/1000
2334/2334 [==============================] - 5s - loss: 0.1573 - acc: 0.9524 - val_loss: 3.3188 - val_acc: 0.4126
Epoch 455/1000
2334/2334 [==============================] - 5s - loss: 0.1543 - acc: 0.9520 - val_loss: 3.3179 - val_acc: 0.4206
Epoch 456/1000
2334/2334 [==============================] - 5s - loss: 0.1447 - acc: 0.9494 - val_loss: 3.4694 - val_acc: 0.3966
Epoch 457/1000
2334/2334 [==============================] - 5s - loss: 0.1116 - acc: 0.9657 - val_loss: 3.3212 - val_acc: 0.4206
Epoch 458/1000
2334/2334 [==============================] - 5s - loss: 0.1171 - acc: 0.9614 - val_loss: 3.7585 - val_acc: 0.3996
Epoch 459/1000
2334/2334 [==============================] - 5s - loss: 0.1182 - acc: 0.9593 - val_loss: 3.1410 - val_acc: 0.4236
Epoch 460/1000
2334/2334 [==============================] - 5s - loss: 0.1023 - acc: 0.9653 - val_loss: 3.7659 - val_acc: 0.4256
Epoch 461/1000
2334/2334 [==============================] - 5s - loss: 0.1118 - acc: 0.9597 - val_loss: 4.1315 - val_acc: 0.3826
Epoch 462/1000
2334/2334 [==============================] - 5s - loss: 0.1247 - acc: 0.9580 - val_loss: 4.3394 - val_acc: 0.3916
Epoch 463/1000
2334/2334 [==============================] - 5s - loss: 0.1019 - acc: 0.9649 - val_loss: 2.8437 - val_acc: 0.4725
Epoch 464/1000
2334/2334 [==============================] - 5s - loss: 0.1205 - acc: 0.9602 - val_loss: 4.3360 - val_acc: 0.3966
Epoch 465/1000
2334/2334 [==============================] - 5s - loss: 0.0950 - acc: 0.9644 - val_loss: 4.1979 - val_acc: 0.3946
Epoch 466/1000
2334/2334 [==============================] - 5s - loss: 0.1259 - acc: 0.9507 - val_loss: 3.4021 - val_acc: 0.4316
Epoch 467/1000
2334/2334 [==============================] - 5s - loss: 0.1347 - acc: 0.9554 - val_loss: 2.9278 - val_acc: 0.4695
Epoch 468/1000
2334/2334 [==============================] - 5s - loss: 0.1584 - acc: 0.9473 - val_loss: 3.3444 - val_acc: 0.4286
Epoch 469/1000
2334/2334 [==============================] - 5s - loss: 0.1631 - acc: 0.9456 - val_loss: 3.6479 - val_acc: 0.4166
Epoch 470/1000
2334/2334 [==============================] - 5s - loss: 0.1533 - acc: 0.9490 - val_loss: 4.1859 - val_acc: 0.3786
Epoch 471/1000
2334/2334 [==============================] - 5s - loss: 0.1428 - acc: 0.9542 - val_loss: 3.2948 - val_acc: 0.4386
Epoch 472/1000
2334/2334 [==============================] - 5s - loss: 0.1451 - acc: 0.9473 - val_loss: 3.8257 - val_acc: 0.4156
Epoch 473/1000
2334/2334 [==============================] - 5s - loss: 0.1509 - acc: 0.9529 - val_loss: 3.3761 - val_acc: 0.3946
Epoch 474/1000
2334/2334 [==============================] - 5s - loss: 0.1415 - acc: 0.9520 - val_loss: 3.7144 - val_acc: 0.4206
Epoch 475/1000
2334/2334 [==============================] - 5s - loss: 0.1344 - acc: 0.9593 - val_loss: 3.9602 - val_acc: 0.3716
Epoch 476/1000
2334/2334 [==============================] - 5s - loss: 0.1211 - acc: 0.9589 - val_loss: 3.7491 - val_acc: 0.3996
Epoch 477/1000
2334/2334 [==============================] - 5s - loss: 0.1501 - acc: 0.9529 - val_loss: 4.1960 - val_acc: 0.3776
Epoch 478/1000
2334/2334 [==============================] - 5s - loss: 0.1497 - acc: 0.9499 - val_loss: 3.5072 - val_acc: 0.3996
Epoch 479/1000
2334/2334 [==============================] - 5s - loss: 0.1456 - acc: 0.9516 - val_loss: 4.0687 - val_acc: 0.3746
Epoch 480/1000
2334/2334 [==============================] - 5s - loss: 0.1228 - acc: 0.9606 - val_loss: 3.4659 - val_acc: 0.4196
Epoch 481/1000
2334/2334 [==============================] - 5s - loss: 0.1113 - acc: 0.9606 - val_loss: 3.5431 - val_acc: 0.3996
Epoch 482/1000
2334/2334 [==============================] - 5s - loss: 0.1195 - acc: 0.9584 - val_loss: 4.1711 - val_acc: 0.3596
Epoch 483/1000
2334/2334 [==============================] - 5s - loss: 0.1205 - acc: 0.9606 - val_loss: 4.1021 - val_acc: 0.3856
Epoch 484/1000
2334/2334 [==============================] - 5s - loss: 0.1072 - acc: 0.9683 - val_loss: 4.0910 - val_acc: 0.3846
Epoch 485/1000
2334/2334 [==============================] - 5s - loss: 0.1288 - acc: 0.9584 - val_loss: 4.2924 - val_acc: 0.3546
Epoch 486/1000
2334/2334 [==============================] - 5s - loss: 0.1253 - acc: 0.9610 - val_loss: 3.5829 - val_acc: 0.4196
Epoch 487/1000
2334/2334 [==============================] - 5s - loss: 0.1195 - acc: 0.9589 - val_loss: 3.7374 - val_acc: 0.4046
Epoch 488/1000
2334/2334 [==============================] - 5s - loss: 0.1151 - acc: 0.9580 - val_loss: 3.3749 - val_acc: 0.4236
Epoch 489/1000
2334/2334 [==============================] - 5s - loss: 0.1321 - acc: 0.9550 - val_loss: 5.1725 - val_acc: 0.3317
Epoch 490/1000
2334/2334 [==============================] - 5s - loss: 0.1552 - acc: 0.9533 - val_loss: 3.7690 - val_acc: 0.4036
Epoch 491/1000
2334/2334 [==============================] - 5s - loss: 0.1430 - acc: 0.9610 - val_loss: 2.9085 - val_acc: 0.4865
Epoch 492/1000
2334/2334 [==============================] - 5s - loss: 0.1583 - acc: 0.9443 - val_loss: 2.9094 - val_acc: 0.4326
Epoch 493/1000
2334/2334 [==============================] - 5s - loss: 0.1631 - acc: 0.9456 - val_loss: 4.8512 - val_acc: 0.3516
Epoch 494/1000
2334/2334 [==============================] - 5s - loss: 0.1369 - acc: 0.9537 - val_loss: 3.2322 - val_acc: 0.4396
Epoch 495/1000
2334/2334 [==============================] - 5s - loss: 0.1429 - acc: 0.9452 - val_loss: 2.7818 - val_acc: 0.4505
Epoch 496/1000
2334/2334 [==============================] - 5s - loss: 0.1455 - acc: 0.9494 - val_loss: 3.5396 - val_acc: 0.3936
Epoch 497/1000
2334/2334 [==============================] - 5s - loss: 0.1145 - acc: 0.9610 - val_loss: 3.5887 - val_acc: 0.4206
Epoch 498/1000
2334/2334 [==============================] - 5s - loss: 0.1589 - acc: 0.9490 - val_loss: 3.2704 - val_acc: 0.4106
Epoch 499/1000
2334/2334 [==============================] - 5s - loss: 0.1252 - acc: 0.9619 - val_loss: 3.6998 - val_acc: 0.4136
Epoch 500/1000
2334/2334 [==============================] - 5s - loss: 0.1704 - acc: 0.9456 - val_loss: 4.0037 - val_acc: 0.4096
Epoch 501/1000
2334/2334 [==============================] - 5s - loss: 0.1381 - acc: 0.9520 - val_loss: 4.7263 - val_acc: 0.3686
Epoch 502/1000
2334/2334 [==============================] - 5s - loss: 0.1208 - acc: 0.9623 - val_loss: 4.4550 - val_acc: 0.3646
Epoch 503/1000
2334/2334 [==============================] - 5s - loss: 0.1191 - acc: 0.9559 - val_loss: 3.4684 - val_acc: 0.4376
Epoch 504/1000
2334/2334 [==============================] - 5s - loss: 0.1047 - acc: 0.9606 - val_loss: 3.1696 - val_acc: 0.4476
Epoch 505/1000
2334/2334 [==============================] - 5s - loss: 0.1098 - acc: 0.9627 - val_loss: 4.4463 - val_acc: 0.3676
Epoch 506/1000
2334/2334 [==============================] - 5s - loss: 0.1028 - acc: 0.9653 - val_loss: 2.9582 - val_acc: 0.4705
Epoch 507/1000
2334/2334 [==============================] - 5s - loss: 0.1163 - acc: 0.9597 - val_loss: 4.1203 - val_acc: 0.3936
Epoch 508/1000
2334/2334 [==============================] - 5s - loss: 0.1203 - acc: 0.9623 - val_loss: 3.3668 - val_acc: 0.4216
Epoch 509/1000
2334/2334 [==============================] - 5s - loss: 0.1022 - acc: 0.9683 - val_loss: 4.1760 - val_acc: 0.3816
Epoch 510/1000
2334/2334 [==============================] - 5s - loss: 0.0979 - acc: 0.9627 - val_loss: 3.1496 - val_acc: 0.4605
Epoch 511/1000
2334/2334 [==============================] - 5s - loss: 0.0998 - acc: 0.9632 - val_loss: 5.0067 - val_acc: 0.3407
Epoch 512/1000
2334/2334 [==============================] - 5s - loss: 0.1290 - acc: 0.9554 - val_loss: 3.9340 - val_acc: 0.3966
Epoch 513/1000
2334/2334 [==============================] - 5s - loss: 0.1205 - acc: 0.9580 - val_loss: 4.4769 - val_acc: 0.3696
Epoch 514/1000
2334/2334 [==============================] - 5s - loss: 0.1202 - acc: 0.9610 - val_loss: 3.2438 - val_acc: 0.4416
Epoch 515/1000
2334/2334 [==============================] - 5s - loss: 0.1105 - acc: 0.9602 - val_loss: 3.0107 - val_acc: 0.4456
Epoch 516/1000
2334/2334 [==============================] - 5s - loss: 0.1181 - acc: 0.9619 - val_loss: 4.3114 - val_acc: 0.3966
Epoch 517/1000
2334/2334 [==============================] - 5s - loss: 0.1214 - acc: 0.9572 - val_loss: 3.8573 - val_acc: 0.4166
Epoch 518/1000
2334/2334 [==============================] - 5s - loss: 0.1172 - acc: 0.9580 - val_loss: 3.3409 - val_acc: 0.4316
Epoch 519/1000
2334/2334 [==============================] - 5s - loss: 0.1139 - acc: 0.9662 - val_loss: 2.8688 - val_acc: 0.4875
Epoch 520/1000
2334/2334 [==============================] - 5s - loss: 0.1175 - acc: 0.9666 - val_loss: 4.0366 - val_acc: 0.3956
Epoch 521/1000
2334/2334 [==============================] - 5s - loss: 0.0817 - acc: 0.9730 - val_loss: 3.8469 - val_acc: 0.3876
Epoch 522/1000
2334/2334 [==============================] - 5s - loss: 0.1097 - acc: 0.9632 - val_loss: 4.0665 - val_acc: 0.3886
Epoch 523/1000
2334/2334 [==============================] - 5s - loss: 0.0966 - acc: 0.9670 - val_loss: 3.8034 - val_acc: 0.4046
Epoch 524/1000
2334/2334 [==============================] - 5s - loss: 0.1179 - acc: 0.9597 - val_loss: 4.3177 - val_acc: 0.3846
Epoch 525/1000
2334/2334 [==============================] - 5s - loss: 0.1391 - acc: 0.9614 - val_loss: 3.7149 - val_acc: 0.4166
Epoch 526/1000
2334/2334 [==============================] - 5s - loss: 0.1288 - acc: 0.9567 - val_loss: 3.2743 - val_acc: 0.4316
Epoch 527/1000
2334/2334 [==============================] - 5s - loss: 0.1317 - acc: 0.9563 - val_loss: 3.8382 - val_acc: 0.3956
Epoch 528/1000
2334/2334 [==============================] - 5s - loss: 0.1510 - acc: 0.9529 - val_loss: 4.1013 - val_acc: 0.3776
Epoch 529/1000
2334/2334 [==============================] - 5s - loss: 0.1605 - acc: 0.9516 - val_loss: 3.2126 - val_acc: 0.4356
Epoch 530/1000
2334/2334 [==============================] - 5s - loss: 0.1555 - acc: 0.9499 - val_loss: 3.9537 - val_acc: 0.3786
Epoch 531/1000
2334/2334 [==============================] - 5s - loss: 0.1357 - acc: 0.9542 - val_loss: 3.3632 - val_acc: 0.3976
Epoch 532/1000
2334/2334 [==============================] - 5s - loss: 0.1142 - acc: 0.9614 - val_loss: 3.8471 - val_acc: 0.3796
Epoch 533/1000
2334/2334 [==============================] - 5s - loss: 0.1128 - acc: 0.9606 - val_loss: 4.6263 - val_acc: 0.3477
Epoch 534/1000
2334/2334 [==============================] - 5s - loss: 0.1145 - acc: 0.9627 - val_loss: 4.3364 - val_acc: 0.3816
Epoch 535/1000
2334/2334 [==============================] - 5s - loss: 0.1045 - acc: 0.9679 - val_loss: 4.2192 - val_acc: 0.4136
Epoch 536/1000
2334/2334 [==============================] - 5s - loss: 0.1091 - acc: 0.9593 - val_loss: 4.6303 - val_acc: 0.3736
Epoch 537/1000
2334/2334 [==============================] - 5s - loss: 0.1150 - acc: 0.9662 - val_loss: 2.8389 - val_acc: 0.4865
Epoch 538/1000
2334/2334 [==============================] - 5s - loss: 0.0979 - acc: 0.9614 - val_loss: 4.1740 - val_acc: 0.3696
Epoch 539/1000
2334/2334 [==============================] - 5s - loss: 0.1028 - acc: 0.9627 - val_loss: 3.7655 - val_acc: 0.3906
Epoch 540/1000
2334/2334 [==============================] - 5s - loss: 0.0957 - acc: 0.9722 - val_loss: 3.5689 - val_acc: 0.4196
Epoch 541/1000
2334/2334 [==============================] - 5s - loss: 0.1187 - acc: 0.9619 - val_loss: 4.2337 - val_acc: 0.3806
Epoch 542/1000
2334/2334 [==============================] - 5s - loss: 0.0959 - acc: 0.9687 - val_loss: 4.1512 - val_acc: 0.4026
Epoch 543/1000
2334/2334 [==============================] - 5s - loss: 0.1114 - acc: 0.9649 - val_loss: 3.0394 - val_acc: 0.4685
Epoch 544/1000
2334/2334 [==============================] - 5s - loss: 0.1107 - acc: 0.9636 - val_loss: 3.5087 - val_acc: 0.4326
Epoch 545/1000
2334/2334 [==============================] - 5s - loss: 0.1324 - acc: 0.9546 - val_loss: 3.5958 - val_acc: 0.4426
Epoch 546/1000
2334/2334 [==============================] - 5s - loss: 0.1177 - acc: 0.9666 - val_loss: 3.5459 - val_acc: 0.4176
Epoch 547/1000
2334/2334 [==============================] - 5s - loss: 0.1480 - acc: 0.9516 - val_loss: 3.6016 - val_acc: 0.4206
Epoch 548/1000
2334/2334 [==============================] - 5s - loss: 0.1136 - acc: 0.9584 - val_loss: 2.9221 - val_acc: 0.4675
Epoch 549/1000
2334/2334 [==============================] - 5s - loss: 0.1180 - acc: 0.9597 - val_loss: 3.2911 - val_acc: 0.4545
Epoch 550/1000
2334/2334 [==============================] - 5s - loss: 0.1519 - acc: 0.9486 - val_loss: 3.4760 - val_acc: 0.4316
Epoch 551/1000
2334/2334 [==============================] - 5s - loss: 0.1063 - acc: 0.9627 - val_loss: 4.2326 - val_acc: 0.3856
Epoch 552/1000
2334/2334 [==============================] - 5s - loss: 0.1096 - acc: 0.9657 - val_loss: 4.0688 - val_acc: 0.3966
Epoch 553/1000
2334/2334 [==============================] - 5s - loss: 0.1154 - acc: 0.9653 - val_loss: 3.4833 - val_acc: 0.4406
Epoch 554/1000
2334/2334 [==============================] - 5s - loss: 0.0975 - acc: 0.9679 - val_loss: 3.6908 - val_acc: 0.4136
Epoch 555/1000
2334/2334 [==============================] - 5s - loss: 0.1154 - acc: 0.9593 - val_loss: 3.6260 - val_acc: 0.4366
Epoch 556/1000
2334/2334 [==============================] - 5s - loss: 0.1080 - acc: 0.9653 - val_loss: 4.7671 - val_acc: 0.3656
Epoch 557/1000
2334/2334 [==============================] - 5s - loss: 0.0927 - acc: 0.9743 - val_loss: 4.3269 - val_acc: 0.3656
Epoch 558/1000
2334/2334 [==============================] - 5s - loss: 0.0782 - acc: 0.9713 - val_loss: 4.7769 - val_acc: 0.3796
Epoch 559/1000
2334/2334 [==============================] - 5s - loss: 0.1018 - acc: 0.9674 - val_loss: 5.5205 - val_acc: 0.3417
Epoch 560/1000
2334/2334 [==============================] - 5s - loss: 0.0901 - acc: 0.9704 - val_loss: 4.1301 - val_acc: 0.4036
Epoch 561/1000
2334/2334 [==============================] - 5s - loss: 0.0939 - acc: 0.9670 - val_loss: 3.7933 - val_acc: 0.4196
Epoch 562/1000
2334/2334 [==============================] - 5s - loss: 0.0985 - acc: 0.9683 - val_loss: 4.0319 - val_acc: 0.3956
Epoch 563/1000
2334/2334 [==============================] - 5s - loss: 0.0807 - acc: 0.9717 - val_loss: 3.6559 - val_acc: 0.4336
Epoch 564/1000
2334/2334 [==============================] - 5s - loss: 0.0915 - acc: 0.9704 - val_loss: 4.2132 - val_acc: 0.3986
Epoch 565/1000
2334/2334 [==============================] - 5s - loss: 0.1113 - acc: 0.9610 - val_loss: 3.8684 - val_acc: 0.4186
Epoch 566/1000
2334/2334 [==============================] - 5s - loss: 0.1003 - acc: 0.9627 - val_loss: 3.5257 - val_acc: 0.4356
Epoch 567/1000
2334/2334 [==============================] - 5s - loss: 0.1379 - acc: 0.9546 - val_loss: 3.8488 - val_acc: 0.4186
Epoch 568/1000
2334/2334 [==============================] - 5s - loss: 0.0951 - acc: 0.9636 - val_loss: 3.7727 - val_acc: 0.4246
Epoch 569/1000
2334/2334 [==============================] - 5s - loss: 0.1112 - acc: 0.9662 - val_loss: 5.4711 - val_acc: 0.3337
Epoch 570/1000
2334/2334 [==============================] - 5s - loss: 0.1180 - acc: 0.9602 - val_loss: 4.4360 - val_acc: 0.4056
Epoch 571/1000
2334/2334 [==============================] - 5s - loss: 0.1427 - acc: 0.9512 - val_loss: 3.4292 - val_acc: 0.4256
Epoch 572/1000
2334/2334 [==============================] - 5s - loss: 0.1316 - acc: 0.9610 - val_loss: 2.9881 - val_acc: 0.4885
Epoch 573/1000
2334/2334 [==============================] - 5s - loss: 0.1326 - acc: 0.9537 - val_loss: 4.4567 - val_acc: 0.3606
Epoch 574/1000
2334/2334 [==============================] - 5s - loss: 0.1260 - acc: 0.9597 - val_loss: 3.7098 - val_acc: 0.3886
Epoch 575/1000
2334/2334 [==============================] - 5s - loss: 0.1092 - acc: 0.9662 - val_loss: 3.7623 - val_acc: 0.4086
Epoch 576/1000
2334/2334 [==============================] - 5s - loss: 0.0959 - acc: 0.9674 - val_loss: 3.9906 - val_acc: 0.3976
Epoch 577/1000
2334/2334 [==============================] - 5s - loss: 0.0920 - acc: 0.9730 - val_loss: 3.0459 - val_acc: 0.4685
Epoch 578/1000
2334/2334 [==============================] - 5s - loss: 0.1145 - acc: 0.9619 - val_loss: 3.7612 - val_acc: 0.4246
Epoch 579/1000
2334/2334 [==============================] - 5s - loss: 0.0894 - acc: 0.9734 - val_loss: 4.0682 - val_acc: 0.3876
Epoch 580/1000
2334/2334 [==============================] - 5s - loss: 0.0995 - acc: 0.9640 - val_loss: 3.1010 - val_acc: 0.4476
Epoch 581/1000
2334/2334 [==============================] - 5s - loss: 0.1143 - acc: 0.9619 - val_loss: 3.2453 - val_acc: 0.4605
Epoch 582/1000
2334/2334 [==============================] - 5s - loss: 0.1131 - acc: 0.9649 - val_loss: 4.9564 - val_acc: 0.3656
Epoch 583/1000
2334/2334 [==============================] - 5s - loss: 0.0983 - acc: 0.9666 - val_loss: 3.7884 - val_acc: 0.4056
Epoch 584/1000
2334/2334 [==============================] - 5s - loss: 0.0778 - acc: 0.9747 - val_loss: 3.3446 - val_acc: 0.4545
Epoch 585/1000
2334/2334 [==============================] - 5s - loss: 0.0927 - acc: 0.9717 - val_loss: 3.8273 - val_acc: 0.4076
Epoch 586/1000
2334/2334 [==============================] - 5s - loss: 0.0811 - acc: 0.9747 - val_loss: 2.9390 - val_acc: 0.4815
Epoch 587/1000
2334/2334 [==============================] - 5s - loss: 0.0968 - acc: 0.9670 - val_loss: 5.0391 - val_acc: 0.3756
Epoch 588/1000
2334/2334 [==============================] - 5s - loss: 0.1001 - acc: 0.9709 - val_loss: 4.2202 - val_acc: 0.3866
Epoch 589/1000
2334/2334 [==============================] - 5s - loss: 0.0893 - acc: 0.9674 - val_loss: 3.1729 - val_acc: 0.4595
Epoch 590/1000
2334/2334 [==============================] - 5s - loss: 0.1022 - acc: 0.9683 - val_loss: 4.6905 - val_acc: 0.3726
Epoch 591/1000
2334/2334 [==============================] - 5s - loss: 0.1098 - acc: 0.9662 - val_loss: 4.2305 - val_acc: 0.3876
Epoch 592/1000
2334/2334 [==============================] - 5s - loss: 0.1323 - acc: 0.9572 - val_loss: 4.3799 - val_acc: 0.3896
Epoch 593/1000
2334/2334 [==============================] - 5s - loss: 0.1015 - acc: 0.9632 - val_loss: 3.8908 - val_acc: 0.4286
Epoch 594/1000
2334/2334 [==============================] - 5s - loss: 0.0957 - acc: 0.9679 - val_loss: 3.7198 - val_acc: 0.4146
Epoch 595/1000
2334/2334 [==============================] - 5s - loss: 0.1105 - acc: 0.9644 - val_loss: 4.3421 - val_acc: 0.3626
Epoch 596/1000
2334/2334 [==============================] - 5s - loss: 0.0911 - acc: 0.9679 - val_loss: 3.6598 - val_acc: 0.4286
Epoch 597/1000
2334/2334 [==============================] - 5s - loss: 0.1139 - acc: 0.9610 - val_loss: 3.3811 - val_acc: 0.4585
Epoch 598/1000
2334/2334 [==============================] - 5s - loss: 0.1119 - acc: 0.9632 - val_loss: 4.0902 - val_acc: 0.4056
Epoch 599/1000
2334/2334 [==============================] - 5s - loss: 0.1215 - acc: 0.9593 - val_loss: 3.6372 - val_acc: 0.4116
Epoch 600/1000
2334/2334 [==============================] - 5s - loss: 0.1108 - acc: 0.9619 - val_loss: 4.7141 - val_acc: 0.3616
Epoch 601/1000
2334/2334 [==============================] - 5s - loss: 0.1356 - acc: 0.9542 - val_loss: 4.7427 - val_acc: 0.3526
Epoch 602/1000
2334/2334 [==============================] - 5s - loss: 0.1295 - acc: 0.9597 - val_loss: 3.5622 - val_acc: 0.4176
Epoch 603/1000
2334/2334 [==============================] - 5s - loss: 0.0889 - acc: 0.9700 - val_loss: 4.8220 - val_acc: 0.3467
Epoch 604/1000
2334/2334 [==============================] - 5s - loss: 0.0981 - acc: 0.9619 - val_loss: 4.4633 - val_acc: 0.3656
Epoch 605/1000
2334/2334 [==============================] - 5s - loss: 0.1091 - acc: 0.9606 - val_loss: 2.9163 - val_acc: 0.4615
Epoch 606/1000
2334/2334 [==============================] - 5s - loss: 0.0987 - acc: 0.9662 - val_loss: 4.5249 - val_acc: 0.3766
Epoch 607/1000
2334/2334 [==============================] - 5s - loss: 0.0859 - acc: 0.9704 - val_loss: 3.6041 - val_acc: 0.4236
Epoch 608/1000
2334/2334 [==============================] - 5s - loss: 0.0939 - acc: 0.9674 - val_loss: 3.2161 - val_acc: 0.4575
Epoch 609/1000
2334/2334 [==============================] - 5s - loss: 0.1061 - acc: 0.9649 - val_loss: 4.4109 - val_acc: 0.3846
Epoch 610/1000
2334/2334 [==============================] - 5s - loss: 0.1263 - acc: 0.9580 - val_loss: 3.4295 - val_acc: 0.4535
Epoch 611/1000
2334/2334 [==============================] - 5s - loss: 0.1242 - acc: 0.9589 - val_loss: 4.2279 - val_acc: 0.3756
Epoch 612/1000
1000/2334 [===========>..................] - ETA: 2s - loss: 0.1130 - acc: 0.9660

In [ ]:

    

train_loss, train_accuracy = model.evaluate(X_train, y_train, batch_size=500)
train_loss, train_accuracy


    

In [ ]:

    

test_loss, test_accuracy = model.evaluate(X_test, y_test, batch_size=BATCH_SIZE)
test_loss, test_accuracy


    

In [ ]:

    

evaluation_loss, evaluation_accuracy = model.evaluate(evaluation_X, evaluation_y, batch_size=BATCH_SIZE)
evaluation_loss, evaluation_accuracy


    

In [ ]:

    

!mkdir models


    

In [ ]:

    

# model.save('models/conv-vgg-augmented.hdf5')
# model.save('models/conv-vgg-original.hdf5')
model.save('models/conv-vgg-mixed.hdf5')


    

In [ ]:

    

!ls -lh models


    

In [ ]:

    

# !curl --upload-file ./models/conv-vgg-augmented.hdf5 https://transfer.sh/conv-vgg-augmented.hdf5
!curl --upload-file ./models/conv-vgg-original.hdf5 https://transfer.sh/conv-vgg-original.hdf5


    

In [ ]: