Loading...
墨滴

希仔

2021/04/07  阅读:25  主题:默认主题

learning curve

import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
from tensorflow.keras import layers, Sequential
from pprint import pprint
cifar10 = tf.keras.datasets.cifar10
(x_train, y_train), (x_test, y_test)=cifar10. load_data ()
Downloading data from https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz
170500096/170498071 [==============================] - 4s 0us/step
x_train, x_test = tf.cast(x_train,dtype=tf.float32)/255.0,tf.cast(x_test,dtype=tf.float32)/255.0
y_train,y_test = tf.cast(y_train,dtype=tf.int32), tf.cast(y_test,dtype=tf.int32)
x_train.shape[1:]
TensorShape([32, 32, 3])
num_train = []
acc_train = []
acc_test = []
for i in range(99,5,-5):
  i = i/100 # i is the validation split rate

  # initialize the model
  model = Sequential([
    # unit 1
    layers.Conv2D(16,kernel_size=(3,3),padding='same',activation=tf.nn.relu, input_shape = x_train.shape[1:]),
    layers.Conv2D(16,kernel_size=(3,3),padding = 'same', activation = tf.nn.relu),
    layers.MaxPool2D(pool_size=(2,2)),
    
    # unit 2
    layers.Conv2D(32,kernel_size=(3,3),padding='same',activation=tf.nn.relu),
    layers.Conv2D(32,kernel_size=(3,3),padding='same',activation=tf.nn.relu),
    
    # unit 3
    layers.Flatten(),
    layers.Dense(128,activation = 'relu'),
    layers.Dense(10,activation='softmax')
  ])
  model.compile(optimizer='adam',loss='sparse_categorical_crossentropy',metrics=['sparse_categorical_accuracy'])
  

  # fit model and add results
  h = model.fit(x_train,y_train,batch_size=64,epochs=15,validation_split=i,verbose=2)
  num_train.append( len(x_train)*(1-i))
  acc_test.append(h.history['val_sparse_categorical_accuracy'][0])
  acc_train.append(h.history['sparse_categorical_accuracy'][0])

Epoch 1/15
8/8 - 3s - loss: 2.2926 - sparse_categorical_accuracy: 0.1140 - val_loss: 2.2506 - val_sparse_categorical_accuracy: 0.1174
Epoch 2/15
8/8 - 2s - loss: 2.1313 - sparse_categorical_accuracy: 0.2180 - val_loss: 2.0844 - val_sparse_categorical_accuracy: 0.2467
Epoch 3/15
8/8 - 2s - loss: 1.9484 - sparse_categorical_accuracy: 0.3280 - val_loss: 2.0849 - val_sparse_categorical_accuracy: 0.2461
Epoch 4/15
8/8 - 2s - loss: 1.8876 - sparse_categorical_accuracy: 0.3440 - val_loss: 2.1378 - val_sparse_categorical_accuracy: 0.2328
Epoch 5/15
8/8 - 2s - loss: 1.7837 - sparse_categorical_accuracy: 0.3680 - val_loss: 1.9796 - val_sparse_categorical_accuracy: 0.2890
Epoch 6/15
8/8 - 2s - loss: 1.5741 - sparse_categorical_accuracy: 0.4580 - val_loss: 2.0069 - val_sparse_categorical_accuracy: 0.2803
Epoch 7/15
8/8 - 2s - loss: 1.4989 - sparse_categorical_accuracy: 0.4980 - val_loss: 1.9629 - val_sparse_categorical_accuracy: 0.3137
Epoch 8/15
8/8 - 2s - loss: 1.3658 - sparse_categorical_accuracy: 0.5260 - val_loss: 1.9063 - val_sparse_categorical_accuracy: 0.3339
Epoch 9/15
8/8 - 2s - loss: 1.2217 - sparse_categorical_accuracy: 0.6040 - val_loss: 1.9664 - val_sparse_categorical_accuracy: 0.3262
Epoch 10/15
8/8 - 2s - loss: 1.0907 - sparse_categorical_accuracy: 0.6120 - val_loss: 2.0264 - val_sparse_categorical_accuracy: 0.3323
Epoch 11/15
8/8 - 2s - loss: 0.9066 - sparse_categorical_accuracy: 0.6980 - val_loss: 2.1427 - val_sparse_categorical_accuracy: 0.3343
Epoch 12/15
8/8 - 2s - loss: 0.8175 - sparse_categorical_accuracy: 0.7140 - val_loss: 2.1413 - val_sparse_categorical_accuracy: 0.3241
Epoch 13/15
8/8 - 2s - loss: 0.7261 - sparse_categorical_accuracy: 0.7600 - val_loss: 2.3974 - val_sparse_categorical_accuracy: 0.3281
Epoch 14/15
8/8 - 2s - loss: 0.5771 - sparse_categorical_accuracy: 0.8180 - val_loss: 2.4342 - val_sparse_categorical_accuracy: 0.3351
Epoch 15/15
8/8 - 2s - loss: 0.4223 - sparse_categorical_accuracy: 0.8880 - val_loss: 2.6841 - val_sparse_categorical_accuracy: 0.3385
Epoch 1/15
47/47 - 3s - loss: 2.1591 - sparse_categorical_accuracy: 0.2003 - val_loss: 1.9854 - val_sparse_categorical_accuracy: 0.2957
Epoch 2/15
47/47 - 2s - loss: 1.8126 - sparse_categorical_accuracy: 0.3510 - val_loss: 1.7213 - val_sparse_categorical_accuracy: 0.3701
Epoch 3/15
47/47 - 2s - loss: 1.5689 - sparse_categorical_accuracy: 0.4417 - val_loss: 1.6557 - val_sparse_categorical_accuracy: 0.4093
Epoch 4/15
47/47 - 2s - loss: 1.4334 - sparse_categorical_accuracy: 0.4920 - val_loss: 1.6091 - val_sparse_categorical_accuracy: 0.4287
Epoch 5/15
47/47 - 2s - loss: 1.3471 - sparse_categorical_accuracy: 0.5267 - val_loss: 1.5869 - val_sparse_categorical_accuracy: 0.4374
Epoch 6/15
47/47 - 2s - loss: 1.1928 - sparse_categorical_accuracy: 0.5777 - val_loss: 1.6040 - val_sparse_categorical_accuracy: 0.4370
Epoch 7/15
47/47 - 2s - loss: 1.1012 - sparse_categorical_accuracy: 0.6000 - val_loss: 1.5934 - val_sparse_categorical_accuracy: 0.4521
Epoch 8/15
47/47 - 2s - loss: 0.9213 - sparse_categorical_accuracy: 0.6797 - val_loss: 1.6152 - val_sparse_categorical_accuracy: 0.4618
Epoch 9/15
47/47 - 2s - loss: 0.7837 - sparse_categorical_accuracy: 0.7323 - val_loss: 1.6569 - val_sparse_categorical_accuracy: 0.4648
Epoch 10/15
47/47 - 2s - loss: 0.6030 - sparse_categorical_accuracy: 0.8063 - val_loss: 1.7982 - val_sparse_categorical_accuracy: 0.4697
Epoch 11/15
47/47 - 2s - loss: 0.4814 - sparse_categorical_accuracy: 0.8443 - val_loss: 1.9082 - val_sparse_categorical_accuracy: 0.4635
Epoch 12/15
47/47 - 2s - loss: 0.3391 - sparse_categorical_accuracy: 0.8987 - val_loss: 2.3030 - val_sparse_categorical_accuracy: 0.4517
Epoch 13/15
47/47 - 2s - loss: 0.2609 - sparse_categorical_accuracy: 0.9220 - val_loss: 2.5185 - val_sparse_categorical_accuracy: 0.4532
Epoch 14/15
47/47 - 2s - loss: 0.1555 - sparse_categorical_accuracy: 0.9593 - val_loss: 2.8269 - val_sparse_categorical_accuracy: 0.4598
Epoch 15/15
47/47 - 2s - loss: 0.1011 - sparse_categorical_accuracy: 0.9777 - val_loss: 3.2102 - val_sparse_categorical_accuracy: 0.4424
Epoch 1/15
86/86 - 3s - loss: 1.9617 - sparse_categorical_accuracy: 0.2746 - val_loss: 1.8060 - val_sparse_categorical_accuracy: 0.3462
Epoch 2/15
86/86 - 2s - loss: 1.5948 - sparse_categorical_accuracy: 0.4197 - val_loss: 1.7242 - val_sparse_categorical_accuracy: 0.3895
Epoch 3/15
86/86 - 2s - loss: 1.4094 - sparse_categorical_accuracy: 0.4930 - val_loss: 1.4933 - val_sparse_categorical_accuracy: 0.4605
Epoch 4/15
86/86 - 2s - loss: 1.2856 - sparse_categorical_accuracy: 0.5401 - val_loss: 1.4513 - val_sparse_categorical_accuracy: 0.4899
Epoch 5/15
86/86 - 2s - loss: 1.1382 - sparse_categorical_accuracy: 0.5990 - val_loss: 1.3950 - val_sparse_categorical_accuracy: 0.5006
Epoch 6/15
86/86 - 2s - loss: 0.9808 - sparse_categorical_accuracy: 0.6574 - val_loss: 1.3657 - val_sparse_categorical_accuracy: 0.5264
Epoch 7/15
86/86 - 2s - loss: 0.8410 - sparse_categorical_accuracy: 0.7036 - val_loss: 1.4060 - val_sparse_categorical_accuracy: 0.5336
Epoch 8/15
86/86 - 2s - loss: 0.6961 - sparse_categorical_accuracy: 0.7570 - val_loss: 1.4891 - val_sparse_categorical_accuracy: 0.5365
Epoch 9/15
86/86 - 2s - loss: 0.5151 - sparse_categorical_accuracy: 0.8212 - val_loss: 1.6269 - val_sparse_categorical_accuracy: 0.5091
Epoch 10/15
86/86 - 2s - loss: 0.3832 - sparse_categorical_accuracy: 0.8720 - val_loss: 1.8404 - val_sparse_categorical_accuracy: 0.5258
Epoch 11/15
86/86 - 2s - loss: 0.2538 - sparse_categorical_accuracy: 0.9229 - val_loss: 2.0340 - val_sparse_categorical_accuracy: 0.5208
Epoch 12/15
86/86 - 2s - loss: 0.1550 - sparse_categorical_accuracy: 0.9540 - val_loss: 2.2580 - val_sparse_categorical_accuracy: 0.5274
Epoch 13/15
86/86 - 2s - loss: 0.0979 - sparse_categorical_accuracy: 0.9736 - val_loss: 2.7749 - val_sparse_categorical_accuracy: 0.5077
Epoch 14/15
86/86 - 2s - loss: 0.1311 - sparse_categorical_accuracy: 0.9609 - val_loss: 2.5498 - val_sparse_categorical_accuracy: 0.5073
Epoch 15/15
86/86 - 2s - loss: 0.0821 - sparse_categorical_accuracy: 0.9753 - val_loss: 3.0587 - val_sparse_categorical_accuracy: 0.5057
Epoch 1/15
125/125 - 2s - loss: 1.8288 - sparse_categorical_accuracy: 0.3336 - val_loss: 1.6540 - val_sparse_categorical_accuracy: 0.3931
Epoch 2/15
125/125 - 2s - loss: 1.5160 - sparse_categorical_accuracy: 0.4538 - val_loss: 1.4609 - val_sparse_categorical_accuracy: 0.4738
Epoch 3/15
125/125 - 2s - loss: 1.3419 - sparse_categorical_accuracy: 0.5151 - val_loss: 1.3959 - val_sparse_categorical_accuracy: 0.5115
Epoch 4/15
125/125 - 2s - loss: 1.2098 - sparse_categorical_accuracy: 0.5586 - val_loss: 1.3536 - val_sparse_categorical_accuracy: 0.5185
Epoch 5/15
125/125 - 2s - loss: 1.0513 - sparse_categorical_accuracy: 0.6273 - val_loss: 1.3741 - val_sparse_categorical_accuracy: 0.5341
Epoch 6/15
125/125 - 2s - loss: 0.9049 - sparse_categorical_accuracy: 0.6768 - val_loss: 1.3025 - val_sparse_categorical_accuracy: 0.5600
Epoch 7/15
125/125 - 2s - loss: 0.7611 - sparse_categorical_accuracy: 0.7322 - val_loss: 1.3330 - val_sparse_categorical_accuracy: 0.5622
Epoch 8/15
125/125 - 2s - loss: 0.6033 - sparse_categorical_accuracy: 0.7881 - val_loss: 1.3982 - val_sparse_categorical_accuracy: 0.5635
Epoch 9/15
125/125 - 2s - loss: 0.4806 - sparse_categorical_accuracy: 0.8378 - val_loss: 1.5365 - val_sparse_categorical_accuracy: 0.5534
Epoch 10/15
125/125 - 2s - loss: 0.3518 - sparse_categorical_accuracy: 0.8805 - val_loss: 1.7565 - val_sparse_categorical_accuracy: 0.5562
Epoch 11/15
125/125 - 2s - loss: 0.2300 - sparse_categorical_accuracy: 0.9246 - val_loss: 2.0413 - val_sparse_categorical_accuracy: 0.5548
Epoch 12/15
125/125 - 2s - loss: 0.1601 - sparse_categorical_accuracy: 0.9481 - val_loss: 2.2164 - val_sparse_categorical_accuracy: 0.5483
Epoch 13/15
125/125 - 2s - loss: 0.1301 - sparse_categorical_accuracy: 0.9603 - val_loss: 2.4700 - val_sparse_categorical_accuracy: 0.5386
Epoch 14/15
125/125 - 2s - loss: 0.0927 - sparse_categorical_accuracy: 0.9722 - val_loss: 2.6786 - val_sparse_categorical_accuracy: 0.5419
Epoch 15/15
125/125 - 2s - loss: 0.0820 - sparse_categorical_accuracy: 0.9736 - val_loss: 2.7324 - val_sparse_categorical_accuracy: 0.5416
Epoch 1/15
165/165 - 3s - loss: 1.7752 - sparse_categorical_accuracy: 0.3474 - val_loss: 1.4984 - val_sparse_categorical_accuracy: 0.4616
Epoch 2/15
165/165 - 2s - loss: 1.4159 - sparse_categorical_accuracy: 0.4903 - val_loss: 1.3456 - val_sparse_categorical_accuracy: 0.5156
Epoch 3/15
165/165 - 2s - loss: 1.2249 - sparse_categorical_accuracy: 0.5618 - val_loss: 1.3063 - val_sparse_categorical_accuracy: 0.5340
Epoch 4/15
165/165 - 2s - loss: 1.0798 - sparse_categorical_accuracy: 0.6156 - val_loss: 1.1932 - val_sparse_categorical_accuracy: 0.5811
Epoch 5/15
165/165 - 2s - loss: 0.9499 - sparse_categorical_accuracy: 0.6645 - val_loss: 1.2333 - val_sparse_categorical_accuracy: 0.5690
Epoch 6/15
165/165 - 2s - loss: 0.8194 - sparse_categorical_accuracy: 0.7121 - val_loss: 1.2479 - val_sparse_categorical_accuracy: 0.5832
Epoch 7/15
165/165 - 2s - loss: 0.6875 - sparse_categorical_accuracy: 0.7569 - val_loss: 1.2018 - val_sparse_categorical_accuracy: 0.6099
Epoch 8/15
165/165 - 2s - loss: 0.5487 - sparse_categorical_accuracy: 0.8091 - val_loss: 1.2963 - val_sparse_categorical_accuracy: 0.5965
Epoch 9/15
165/165 - 2s - loss: 0.4023 - sparse_categorical_accuracy: 0.8633 - val_loss: 1.4263 - val_sparse_categorical_accuracy: 0.5998
Epoch 10/15
165/165 - 2s - loss: 0.2772 - sparse_categorical_accuracy: 0.9081 - val_loss: 1.7235 - val_sparse_categorical_accuracy: 0.5837
Epoch 11/15
165/165 - 2s - loss: 0.1866 - sparse_categorical_accuracy: 0.9410 - val_loss: 1.9007 - val_sparse_categorical_accuracy: 0.5861
Epoch 12/15
165/165 - 2s - loss: 0.1230 - sparse_categorical_accuracy: 0.9609 - val_loss: 2.0680 - val_sparse_categorical_accuracy: 0.5935
Epoch 13/15
165/165 - 2s - loss: 0.0741 - sparse_categorical_accuracy: 0.9790 - val_loss: 2.2622 - val_sparse_categorical_accuracy: 0.5935
Epoch 14/15
165/165 - 2s - loss: 0.0459 - sparse_categorical_accuracy: 0.9881 - val_loss: 2.5476 - val_sparse_categorical_accuracy: 0.5807
Epoch 15/15
165/165 - 2s - loss: 0.0756 - sparse_categorical_accuracy: 0.9761 - val_loss: 2.5671 - val_sparse_categorical_accuracy: 0.5565
Epoch 1/15
204/204 - 3s - loss: 1.7814 - sparse_categorical_accuracy: 0.3475 - val_loss: 1.6958 - val_sparse_categorical_accuracy: 0.4011
Epoch 2/15
204/204 - 2s - loss: 1.3777 - sparse_categorical_accuracy: 0.5059 - val_loss: 1.3519 - val_sparse_categorical_accuracy: 0.5111
Epoch 3/15
204/204 - 2s - loss: 1.1737 - sparse_categorical_accuracy: 0.5839 - val_loss: 1.2351 - val_sparse_categorical_accuracy: 0.5626
Epoch 4/15
204/204 - 2s - loss: 1.0184 - sparse_categorical_accuracy: 0.6362 - val_loss: 1.3537 - val_sparse_categorical_accuracy: 0.5465
Epoch 5/15
204/204 - 2s - loss: 0.8867 - sparse_categorical_accuracy: 0.6871 - val_loss: 1.2940 - val_sparse_categorical_accuracy: 0.5609
Epoch 6/15
204/204 - 2s - loss: 0.7339 - sparse_categorical_accuracy: 0.7408 - val_loss: 1.2384 - val_sparse_categorical_accuracy: 0.5919
Epoch 7/15
204/204 - 2s - loss: 0.5976 - sparse_categorical_accuracy: 0.7898 - val_loss: 1.3451 - val_sparse_categorical_accuracy: 0.5902
Epoch 8/15
204/204 - 2s - loss: 0.4569 - sparse_categorical_accuracy: 0.8401 - val_loss: 1.3964 - val_sparse_categorical_accuracy: 0.6005
Epoch 9/15
204/204 - 2s - loss: 0.3309 - sparse_categorical_accuracy: 0.8873 - val_loss: 1.4236 - val_sparse_categorical_accuracy: 0.6024
Epoch 10/15
204/204 - 2s - loss: 0.2110 - sparse_categorical_accuracy: 0.9303 - val_loss: 1.8078 - val_sparse_categorical_accuracy: 0.5867
Epoch 11/15
204/204 - 2s - loss: 0.1465 - sparse_categorical_accuracy: 0.9539 - val_loss: 2.0099 - val_sparse_categorical_accuracy: 0.5843
Epoch 12/15
204/204 - 2s - loss: 0.0887 - sparse_categorical_accuracy: 0.9717 - val_loss: 2.3220 - val_sparse_categorical_accuracy: 0.5963
Epoch 13/15
204/204 - 2s - loss: 0.0588 - sparse_categorical_accuracy: 0.9817 - val_loss: 2.4513 - val_sparse_categorical_accuracy: 0.5923
Epoch 14/15
204/204 - 2s - loss: 0.0607 - sparse_categorical_accuracy: 0.9812 - val_loss: 2.5488 - val_sparse_categorical_accuracy: 0.5896
Epoch 15/15
204/204 - 2s - loss: 0.0616 - sparse_categorical_accuracy: 0.9814 - val_loss: 2.7336 - val_sparse_categorical_accuracy: 0.5929
Epoch 1/15
243/243 - 3s - loss: 1.6927 - sparse_categorical_accuracy: 0.3820 - val_loss: 1.4618 - val_sparse_categorical_accuracy: 0.4686
Epoch 2/15
243/243 - 2s - loss: 1.3348 - sparse_categorical_accuracy: 0.5193 - val_loss: 1.2778 - val_sparse_categorical_accuracy: 0.5426
Epoch 3/15
243/243 - 2s - loss: 1.1301 - sparse_categorical_accuracy: 0.5993 - val_loss: 1.2291 - val_sparse_categorical_accuracy: 0.5653
Epoch 4/15
243/243 - 2s - loss: 0.9813 - sparse_categorical_accuracy: 0.6555 - val_loss: 1.2109 - val_sparse_categorical_accuracy: 0.5808
Epoch 5/15
243/243 - 2s - loss: 0.8476 - sparse_categorical_accuracy: 0.6997 - val_loss: 1.1179 - val_sparse_categorical_accuracy: 0.6181
Epoch 6/15
243/243 - 2s - loss: 0.7089 - sparse_categorical_accuracy: 0.7532 - val_loss: 1.1044 - val_sparse_categorical_accuracy: 0.6264
Epoch 7/15
243/243 - 2s - loss: 0.5953 - sparse_categorical_accuracy: 0.7923 - val_loss: 1.1835 - val_sparse_categorical_accuracy: 0.6244
Epoch 8/15
243/243 - 2s - loss: 0.4403 - sparse_categorical_accuracy: 0.8466 - val_loss: 1.2643 - val_sparse_categorical_accuracy: 0.6276
Epoch 9/15
243/243 - 2s - loss: 0.3267 - sparse_categorical_accuracy: 0.8892 - val_loss: 1.4060 - val_sparse_categorical_accuracy: 0.6263
Epoch 10/15
243/243 - 2s - loss: 0.2329 - sparse_categorical_accuracy: 0.9201 - val_loss: 1.5897 - val_sparse_categorical_accuracy: 0.6148
Epoch 11/15
243/243 - 2s - loss: 0.1479 - sparse_categorical_accuracy: 0.9513 - val_loss: 1.8417 - val_sparse_categorical_accuracy: 0.6187
Epoch 12/15
243/243 - 2s - loss: 0.1074 - sparse_categorical_accuracy: 0.9659 - val_loss: 1.9831 - val_sparse_categorical_accuracy: 0.6224
Epoch 13/15
243/243 - 2s - loss: 0.0837 - sparse_categorical_accuracy: 0.9734 - val_loss: 2.1743 - val_sparse_categorical_accuracy: 0.6051
Epoch 14/15
243/243 - 2s - loss: 0.0851 - sparse_categorical_accuracy: 0.9714 - val_loss: 2.3730 - val_sparse_categorical_accuracy: 0.6082
Epoch 15/15
243/243 - 2s - loss: 0.0560 - sparse_categorical_accuracy: 0.9820 - val_loss: 2.4987 - val_sparse_categorical_accuracy: 0.6172
Epoch 1/15
282/282 - 3s - loss: 1.7283 - sparse_categorical_accuracy: 0.3680 - val_loss: 1.5407 - val_sparse_categorical_accuracy: 0.4383
Epoch 2/15
282/282 - 2s - loss: 1.4018 - sparse_categorical_accuracy: 0.4963 - val_loss: 1.3136 - val_sparse_categorical_accuracy: 0.5272
Epoch 3/15
282/282 - 2s - loss: 1.1853 - sparse_categorical_accuracy: 0.5748 - val_loss: 1.2490 - val_sparse_categorical_accuracy: 0.5594
Epoch 4/15
282/282 - 2s - loss: 1.0202 - sparse_categorical_accuracy: 0.6382 - val_loss: 1.1421 - val_sparse_categorical_accuracy: 0.5993
Epoch 5/15
282/282 - 2s - loss: 0.8740 - sparse_categorical_accuracy: 0.6927 - val_loss: 1.0946 - val_sparse_categorical_accuracy: 0.6242
Epoch 6/15
282/282 - 2s - loss: 0.7298 - sparse_categorical_accuracy: 0.7444 - val_loss: 1.1117 - val_sparse_categorical_accuracy: 0.6203
Epoch 7/15
282/282 - 2s - loss: 0.5882 - sparse_categorical_accuracy: 0.7929 - val_loss: 1.2033 - val_sparse_categorical_accuracy: 0.6118
Epoch 8/15
282/282 - 2s - loss: 0.4579 - sparse_categorical_accuracy: 0.8398 - val_loss: 1.2449 - val_sparse_categorical_accuracy: 0.6222
Epoch 9/15
282/282 - 2s - loss: 0.3120 - sparse_categorical_accuracy: 0.8919 - val_loss: 1.5632 - val_sparse_categorical_accuracy: 0.6183
Epoch 10/15
282/282 - 2s - loss: 0.2020 - sparse_categorical_accuracy: 0.9329 - val_loss: 1.7444 - val_sparse_categorical_accuracy: 0.6156
Epoch 11/15
282/282 - 2s - loss: 0.1525 - sparse_categorical_accuracy: 0.9488 - val_loss: 1.9240 - val_sparse_categorical_accuracy: 0.6115
Epoch 12/15
282/282 - 2s - loss: 0.1085 - sparse_categorical_accuracy: 0.9644 - val_loss: 2.1295 - val_sparse_categorical_accuracy: 0.5956
Epoch 13/15
282/282 - 2s - loss: 0.0793 - sparse_categorical_accuracy: 0.9746 - val_loss: 2.3004 - val_sparse_categorical_accuracy: 0.6138
Epoch 14/15
282/282 - 2s - loss: 0.0781 - sparse_categorical_accuracy: 0.9723 - val_loss: 2.4158 - val_sparse_categorical_accuracy: 0.6202
Epoch 15/15
282/282 - 2s - loss: 0.0545 - sparse_categorical_accuracy: 0.9824 - val_loss: 2.6454 - val_sparse_categorical_accuracy: 0.6114
Epoch 1/15
321/321 - 3s - loss: 1.7137 - sparse_categorical_accuracy: 0.3712 - val_loss: 1.4600 - val_sparse_categorical_accuracy: 0.4595
Epoch 2/15
321/321 - 2s - loss: 1.3133 - sparse_categorical_accuracy: 0.5276 - val_loss: 1.4174 - val_sparse_categorical_accuracy: 0.5115
Epoch 3/15
321/321 - 2s - loss: 1.1345 - sparse_categorical_accuracy: 0.5956 - val_loss: 1.1512 - val_sparse_categorical_accuracy: 0.5898
Epoch 4/15
321/321 - 2s - loss: 0.9694 - sparse_categorical_accuracy: 0.6587 - val_loss: 1.0891 - val_sparse_categorical_accuracy: 0.6172
Epoch 5/15
321/321 - 2s - loss: 0.8251 - sparse_categorical_accuracy: 0.7107 - val_loss: 1.1251 - val_sparse_categorical_accuracy: 0.6088
Epoch 6/15
321/321 - 2s - loss: 0.6809 - sparse_categorical_accuracy: 0.7637 - val_loss: 1.1800 - val_sparse_categorical_accuracy: 0.6200
Epoch 7/15
321/321 - 2s - loss: 0.5437 - sparse_categorical_accuracy: 0.8115 - val_loss: 1.2330 - val_sparse_categorical_accuracy: 0.6253
Epoch 8/15
321/321 - 2s - loss: 0.3982 - sparse_categorical_accuracy: 0.8631 - val_loss: 1.4454 - val_sparse_categorical_accuracy: 0.6185
Epoch 9/15
321/321 - 2s - loss: 0.2783 - sparse_categorical_accuracy: 0.9050 - val_loss: 1.5374 - val_sparse_categorical_accuracy: 0.6252
Epoch 10/15
321/321 - 2s - loss: 0.1943 - sparse_categorical_accuracy: 0.9336 - val_loss: 1.7108 - val_sparse_categorical_accuracy: 0.6161
Epoch 11/15
321/321 - 2s - loss: 0.1320 - sparse_categorical_accuracy: 0.9575 - val_loss: 1.9338 - val_sparse_categorical_accuracy: 0.6149
Epoch 12/15
321/321 - 2s - loss: 0.1114 - sparse_categorical_accuracy: 0.9621 - val_loss: 2.2133 - val_sparse_categorical_accuracy: 0.6189
Epoch 13/15
321/321 - 2s - loss: 0.0904 - sparse_categorical_accuracy: 0.9685 - val_loss: 2.1788 - val_sparse_categorical_accuracy: 0.6191
Epoch 14/15
321/321 - 2s - loss: 0.0796 - sparse_categorical_accuracy: 0.9753 - val_loss: 2.5125 - val_sparse_categorical_accuracy: 0.6121
Epoch 15/15
321/321 - 2s - loss: 0.0556 - sparse_categorical_accuracy: 0.9815 - val_loss: 2.6856 - val_sparse_categorical_accuracy: 0.6043
Epoch 1/15
360/360 - 3s - loss: 1.6456 - sparse_categorical_accuracy: 0.4079 - val_loss: 1.3725 - val_sparse_categorical_accuracy: 0.5079
Epoch 2/15
360/360 - 2s - loss: 1.2454 - sparse_categorical_accuracy: 0.5547 - val_loss: 1.2569 - val_sparse_categorical_accuracy: 0.5511
Epoch 3/15
360/360 - 2s - loss: 1.0630 - sparse_categorical_accuracy: 0.6233 - val_loss: 1.0922 - val_sparse_categorical_accuracy: 0.6151
Epoch 4/15
360/360 - 2s - loss: 0.9244 - sparse_categorical_accuracy: 0.6754 - val_loss: 1.1153 - val_sparse_categorical_accuracy: 0.6210
Epoch 5/15
360/360 - 2s - loss: 0.7926 - sparse_categorical_accuracy: 0.7208 - val_loss: 1.0941 - val_sparse_categorical_accuracy: 0.6184
Epoch 6/15
360/360 - 2s - loss: 0.6630 - sparse_categorical_accuracy: 0.7665 - val_loss: 1.2105 - val_sparse_categorical_accuracy: 0.6253
Epoch 7/15
360/360 - 2s - loss: 0.5375 - sparse_categorical_accuracy: 0.8133 - val_loss: 1.1186 - val_sparse_categorical_accuracy: 0.6422
Epoch 8/15
360/360 - 2s - loss: 0.4024 - sparse_categorical_accuracy: 0.8594 - val_loss: 1.2765 - val_sparse_categorical_accuracy: 0.6354
Epoch 9/15
360/360 - 2s - loss: 0.2784 - sparse_categorical_accuracy: 0.9037 - val_loss: 1.4534 - val_sparse_categorical_accuracy: 0.6339
Epoch 10/15
360/360 - 2s - loss: 0.1945 - sparse_categorical_accuracy: 0.9337 - val_loss: 1.7332 - val_sparse_categorical_accuracy: 0.6168
Epoch 11/15
360/360 - 2s - loss: 0.1341 - sparse_categorical_accuracy: 0.9556 - val_loss: 1.9224 - val_sparse_categorical_accuracy: 0.6220
Epoch 12/15
360/360 - 2s - loss: 0.1195 - sparse_categorical_accuracy: 0.9589 - val_loss: 2.1189 - val_sparse_categorical_accuracy: 0.6234
Epoch 13/15
360/360 - 2s - loss: 0.0844 - sparse_categorical_accuracy: 0.9724 - val_loss: 2.2274 - val_sparse_categorical_accuracy: 0.6192
Epoch 14/15
360/360 - 2s - loss: 0.0770 - sparse_categorical_accuracy: 0.9742 - val_loss: 2.3697 - val_sparse_categorical_accuracy: 0.6253
Epoch 15/15
360/360 - 2s - loss: 0.0692 - sparse_categorical_accuracy: 0.9766 - val_loss: 2.5001 - val_sparse_categorical_accuracy: 0.6144
Epoch 1/15
399/399 - 3s - loss: 1.5738 - sparse_categorical_accuracy: 0.4305 - val_loss: 1.3370 - val_sparse_categorical_accuracy: 0.5207
Epoch 2/15
399/399 - 2s - loss: 1.1957 - sparse_categorical_accuracy: 0.5727 - val_loss: 1.1411 - val_sparse_categorical_accuracy: 0.5931
Epoch 3/15
399/399 - 2s - loss: 1.0144 - sparse_categorical_accuracy: 0.6421 - val_loss: 1.0977 - val_sparse_categorical_accuracy: 0.6114
Epoch 4/15
399/399 - 2s - loss: 0.8724 - sparse_categorical_accuracy: 0.6902 - val_loss: 0.9870 - val_sparse_categorical_accuracy: 0.6598
Epoch 5/15
399/399 - 2s - loss: 0.7568 - sparse_categorical_accuracy: 0.7355 - val_loss: 1.0057 - val_sparse_categorical_accuracy: 0.6525
Epoch 6/15
399/399 - 2s - loss: 0.6368 - sparse_categorical_accuracy: 0.7743 - val_loss: 1.0603 - val_sparse_categorical_accuracy: 0.6473
Epoch 7/15
399/399 - 2s - loss: 0.5163 - sparse_categorical_accuracy: 0.8199 - val_loss: 1.0683 - val_sparse_categorical_accuracy: 0.6585
Epoch 8/15
399/399 - 2s - loss: 0.3932 - sparse_categorical_accuracy: 0.8617 - val_loss: 1.2022 - val_sparse_categorical_accuracy: 0.6476
Epoch 9/15
399/399 - 2s - loss: 0.2887 - sparse_categorical_accuracy: 0.8985 - val_loss: 1.2980 - val_sparse_categorical_accuracy: 0.6614
Epoch 10/15
399/399 - 2s - loss: 0.1891 - sparse_categorical_accuracy: 0.9352 - val_loss: 1.5628 - val_sparse_categorical_accuracy: 0.6485
Epoch 11/15
399/399 - 2s - loss: 0.1373 - sparse_categorical_accuracy: 0.9533 - val_loss: 1.8157 - val_sparse_categorical_accuracy: 0.6554
Epoch 12/15
399/399 - 2s - loss: 0.1249 - sparse_categorical_accuracy: 0.9577 - val_loss: 1.9810 - val_sparse_categorical_accuracy: 0.6255
Epoch 13/15
399/399 - 2s - loss: 0.0916 - sparse_categorical_accuracy: 0.9695 - val_loss: 2.1326 - val_sparse_categorical_accuracy: 0.6453
Epoch 14/15
399/399 - 2s - loss: 0.0793 - sparse_categorical_accuracy: 0.9736 - val_loss: 2.2679 - val_sparse_categorical_accuracy: 0.6394
Epoch 15/15
399/399 - 2s - loss: 0.0731 - sparse_categorical_accuracy: 0.9749 - val_loss: 2.3751 - val_sparse_categorical_accuracy: 0.6279
Epoch 1/15
438/438 - 3s - loss: 1.5371 - sparse_categorical_accuracy: 0.4459 - val_loss: 1.3170 - val_sparse_categorical_accuracy: 0.5287
Epoch 2/15
438/438 - 2s - loss: 1.1239 - sparse_categorical_accuracy: 0.5978 - val_loss: 1.1441 - val_sparse_categorical_accuracy: 0.5939
Epoch 3/15
438/438 - 2s - loss: 0.9500 - sparse_categorical_accuracy: 0.6667 - val_loss: 1.0117 - val_sparse_categorical_accuracy: 0.6420
Epoch 4/15
438/438 - 2s - loss: 0.8254 - sparse_categorical_accuracy: 0.7099 - val_loss: 0.9872 - val_sparse_categorical_accuracy: 0.6615
Epoch 5/15
438/438 - 2s - loss: 0.6973 - sparse_categorical_accuracy: 0.7550 - val_loss: 0.9843 - val_sparse_categorical_accuracy: 0.6719
Epoch 6/15
438/438 - 2s - loss: 0.5731 - sparse_categorical_accuracy: 0.7985 - val_loss: 0.9952 - val_sparse_categorical_accuracy: 0.6757
Epoch 7/15
438/438 - 2s - loss: 0.4550 - sparse_categorical_accuracy: 0.8395 - val_loss: 1.0767 - val_sparse_categorical_accuracy: 0.6705
Epoch 8/15
438/438 - 2s - loss: 0.3241 - sparse_categorical_accuracy: 0.8862 - val_loss: 1.2085 - val_sparse_categorical_accuracy: 0.6674
Epoch 9/15
438/438 - 2s - loss: 0.2226 - sparse_categorical_accuracy: 0.9227 - val_loss: 1.4018 - val_sparse_categorical_accuracy: 0.6695
Epoch 10/15
438/438 - 2s - loss: 0.1572 - sparse_categorical_accuracy: 0.9452 - val_loss: 1.6021 - val_sparse_categorical_accuracy: 0.6539
Epoch 11/15
438/438 - 2s - loss: 0.1211 - sparse_categorical_accuracy: 0.9574 - val_loss: 1.6904 - val_sparse_categorical_accuracy: 0.6622
Epoch 12/15
438/438 - 2s - loss: 0.0915 - sparse_categorical_accuracy: 0.9693 - val_loss: 2.0730 - val_sparse_categorical_accuracy: 0.6581
Epoch 13/15
438/438 - 2s - loss: 0.0850 - sparse_categorical_accuracy: 0.9713 - val_loss: 2.3288 - val_sparse_categorical_accuracy: 0.6445
Epoch 14/15
438/438 - 2s - loss: 0.0807 - sparse_categorical_accuracy: 0.9719 - val_loss: 2.4051 - val_sparse_categorical_accuracy: 0.6543
Epoch 15/15
438/438 - 2s - loss: 0.0657 - sparse_categorical_accuracy: 0.9774 - val_loss: 2.2795 - val_sparse_categorical_accuracy: 0.6517
Epoch 1/15
477/477 - 3s - loss: 1.5917 - sparse_categorical_accuracy: 0.4240 - val_loss: 1.4124 - val_sparse_categorical_accuracy: 0.4912
Epoch 2/15
477/477 - 2s - loss: 1.1788 - sparse_categorical_accuracy: 0.5790 - val_loss: 1.1191 - val_sparse_categorical_accuracy: 0.6029
Epoch 3/15
477/477 - 2s - loss: 0.9867 - sparse_categorical_accuracy: 0.6529 - val_loss: 1.0490 - val_sparse_categorical_accuracy: 0.6280
Epoch 4/15
477/477 - 2s - loss: 0.8464 - sparse_categorical_accuracy: 0.7024 - val_loss: 1.0151 - val_sparse_categorical_accuracy: 0.6459
Epoch 5/15
477/477 - 2s - loss: 0.7171 - sparse_categorical_accuracy: 0.7472 - val_loss: 1.0114 - val_sparse_categorical_accuracy: 0.6602
Epoch 6/15
477/477 - 2s - loss: 0.5994 - sparse_categorical_accuracy: 0.7905 - val_loss: 1.0498 - val_sparse_categorical_accuracy: 0.6544
Epoch 7/15
477/477 - 2s - loss: 0.4692 - sparse_categorical_accuracy: 0.8380 - val_loss: 1.2099 - val_sparse_categorical_accuracy: 0.6500
Epoch 8/15
477/477 - 2s - loss: 0.3565 - sparse_categorical_accuracy: 0.8761 - val_loss: 1.3065 - val_sparse_categorical_accuracy: 0.6436
Epoch 9/15
477/477 - 2s - loss: 0.2509 - sparse_categorical_accuracy: 0.9130 - val_loss: 1.5099 - val_sparse_categorical_accuracy: 0.6457
Epoch 10/15
477/477 - 2s - loss: 0.1800 - sparse_categorical_accuracy: 0.9378 - val_loss: 1.7733 - val_sparse_categorical_accuracy: 0.6367
Epoch 11/15
477/477 - 2s - loss: 0.1357 - sparse_categorical_accuracy: 0.9534 - val_loss: 1.9868 - val_sparse_categorical_accuracy: 0.6316
Epoch 12/15
477/477 - 2s - loss: 0.1096 - sparse_categorical_accuracy: 0.9623 - val_loss: 2.1402 - val_sparse_categorical_accuracy: 0.6384
Epoch 13/15
477/477 - 2s - loss: 0.1008 - sparse_categorical_accuracy: 0.9648 - val_loss: 2.3269 - val_sparse_categorical_accuracy: 0.6338
Epoch 14/15
477/477 - 2s - loss: 0.0836 - sparse_categorical_accuracy: 0.9717 - val_loss: 2.4145 - val_sparse_categorical_accuracy: 0.6347
Epoch 15/15
477/477 - 2s - loss: 0.0743 - sparse_categorical_accuracy: 0.9756 - val_loss: 2.7177 - val_sparse_categorical_accuracy: 0.6281
Epoch 1/15
516/516 - 3s - loss: 1.5183 - sparse_categorical_accuracy: 0.4518 - val_loss: 1.3323 - val_sparse_categorical_accuracy: 0.5278
Epoch 2/15
516/516 - 2s - loss: 1.1253 - sparse_categorical_accuracy: 0.5990 - val_loss: 1.1145 - val_sparse_categorical_accuracy: 0.6056
Epoch 3/15
516/516 - 2s - loss: 0.9598 - sparse_categorical_accuracy: 0.6599 - val_loss: 1.0262 - val_sparse_categorical_accuracy: 0.6399
Epoch 4/15
516/516 - 2s - loss: 0.8250 - sparse_categorical_accuracy: 0.7116 - val_loss: 0.9610 - val_sparse_categorical_accuracy: 0.6681
Epoch 5/15
516/516 - 2s - loss: 0.7063 - sparse_categorical_accuracy: 0.7547 - val_loss: 0.9006 - val_sparse_categorical_accuracy: 0.6890
Epoch 6/15
516/516 - 2s - loss: 0.5889 - sparse_categorical_accuracy: 0.7941 - val_loss: 0.9425 - val_sparse_categorical_accuracy: 0.6897
Epoch 7/15
516/516 - 2s - loss: 0.4791 - sparse_categorical_accuracy: 0.8340 - val_loss: 0.9778 - val_sparse_categorical_accuracy: 0.6849
Epoch 8/15
516/516 - 2s - loss: 0.3628 - sparse_categorical_accuracy: 0.8723 - val_loss: 1.0768 - val_sparse_categorical_accuracy: 0.6851
Epoch 9/15
516/516 - 2s - loss: 0.2658 - sparse_categorical_accuracy: 0.9078 - val_loss: 1.2290 - val_sparse_categorical_accuracy: 0.6772
Epoch 10/15
516/516 - 2s - loss: 0.1898 - sparse_categorical_accuracy: 0.9351 - val_loss: 1.4627 - val_sparse_categorical_accuracy: 0.6759
Epoch 11/15
516/516 - 2s - loss: 0.1500 - sparse_categorical_accuracy: 0.9467 - val_loss: 1.6295 - val_sparse_categorical_accuracy: 0.6713
Epoch 12/15
516/516 - 2s - loss: 0.1094 - sparse_categorical_accuracy: 0.9631 - val_loss: 1.8116 - val_sparse_categorical_accuracy: 0.6779
Epoch 13/15
516/516 - 2s - loss: 0.0967 - sparse_categorical_accuracy: 0.9663 - val_loss: 1.9396 - val_sparse_categorical_accuracy: 0.6747
Epoch 14/15
516/516 - 2s - loss: 0.0850 - sparse_categorical_accuracy: 0.9703 - val_loss: 1.9334 - val_sparse_categorical_accuracy: 0.6717
Epoch 15/15
516/516 - 2s - loss: 0.0778 - sparse_categorical_accuracy: 0.9746 - val_loss: 2.0476 - val_sparse_categorical_accuracy: 0.6654
Epoch 1/15
555/555 - 3s - loss: 1.5142 - sparse_categorical_accuracy: 0.4523 - val_loss: 1.2940 - val_sparse_categorical_accuracy: 0.5362
Epoch 2/15
555/555 - 2s - loss: 1.1379 - sparse_categorical_accuracy: 0.5955 - val_loss: 1.1008 - val_sparse_categorical_accuracy: 0.6110
Epoch 3/15
555/555 - 2s - loss: 0.9556 - sparse_categorical_accuracy: 0.6618 - val_loss: 1.0257 - val_sparse_categorical_accuracy: 0.6414
Epoch 4/15
555/555 - 2s - loss: 0.8272 - sparse_categorical_accuracy: 0.7096 - val_loss: 0.9552 - val_sparse_categorical_accuracy: 0.6703
Epoch 5/15
555/555 - 2s - loss: 0.7140 - sparse_categorical_accuracy: 0.7494 - val_loss: 0.9567 - val_sparse_categorical_accuracy: 0.6807
Epoch 6/15
555/555 - 2s - loss: 0.6016 - sparse_categorical_accuracy: 0.7873 - val_loss: 1.0207 - val_sparse_categorical_accuracy: 0.6683
Epoch 7/15
555/555 - 2s - loss: 0.4973 - sparse_categorical_accuracy: 0.8233 - val_loss: 0.9903 - val_sparse_categorical_accuracy: 0.6852
Epoch 8/15
555/555 - 2s - loss: 0.3928 - sparse_categorical_accuracy: 0.8625 - val_loss: 1.0634 - val_sparse_categorical_accuracy: 0.6782
Epoch 9/15
555/555 - 2s - loss: 0.2992 - sparse_categorical_accuracy: 0.8953 - val_loss: 1.2339 - val_sparse_categorical_accuracy: 0.6764
Epoch 10/15
555/555 - 2s - loss: 0.2262 - sparse_categorical_accuracy: 0.9212 - val_loss: 1.4272 - val_sparse_categorical_accuracy: 0.6714
Epoch 11/15
555/555 - 2s - loss: 0.1635 - sparse_categorical_accuracy: 0.9438 - val_loss: 1.5400 - val_sparse_categorical_accuracy: 0.6718
Epoch 12/15
555/555 - 2s - loss: 0.1372 - sparse_categorical_accuracy: 0.9512 - val_loss: 1.7355 - val_sparse_categorical_accuracy: 0.6574
Epoch 13/15
555/555 - 2s - loss: 0.1193 - sparse_categorical_accuracy: 0.9578 - val_loss: 1.9467 - val_sparse_categorical_accuracy: 0.6706
Epoch 14/15
555/555 - 2s - loss: 0.1084 - sparse_categorical_accuracy: 0.9617 - val_loss: 2.0482 - val_sparse_categorical_accuracy: 0.6619
Epoch 15/15
555/555 - 2s - loss: 0.0823 - sparse_categorical_accuracy: 0.9719 - val_loss: 2.0680 - val_sparse_categorical_accuracy: 0.6575
Epoch 1/15
594/594 - 3s - loss: 1.5475 - sparse_categorical_accuracy: 0.4391 - val_loss: 1.2962 - val_sparse_categorical_accuracy: 0.5348
Epoch 2/15
594/594 - 2s - loss: 1.1357 - sparse_categorical_accuracy: 0.5983 - val_loss: 1.0586 - val_sparse_categorical_accuracy: 0.6307
Epoch 3/15
594/594 - 2s - loss: 0.9412 - sparse_categorical_accuracy: 0.6689 - val_loss: 0.9429 - val_sparse_categorical_accuracy: 0.6700
Epoch 4/15
594/594 - 2s - loss: 0.8074 - sparse_categorical_accuracy: 0.7157 - val_loss: 0.9194 - val_sparse_categorical_accuracy: 0.6805
Epoch 5/15
594/594 - 2s - loss: 0.6945 - sparse_categorical_accuracy: 0.7578 - val_loss: 0.9045 - val_sparse_categorical_accuracy: 0.6837
Epoch 6/15
594/594 - 2s - loss: 0.5760 - sparse_categorical_accuracy: 0.7980 - val_loss: 0.9418 - val_sparse_categorical_accuracy: 0.6802
Epoch 7/15
594/594 - 2s - loss: 0.4688 - sparse_categorical_accuracy: 0.8373 - val_loss: 1.0164 - val_sparse_categorical_accuracy: 0.6802
Epoch 8/15
594/594 - 2s - loss: 0.3673 - sparse_categorical_accuracy: 0.8721 - val_loss: 1.1062 - val_sparse_categorical_accuracy: 0.6829
Epoch 9/15
594/594 - 2s - loss: 0.2679 - sparse_categorical_accuracy: 0.9075 - val_loss: 1.2754 - val_sparse_categorical_accuracy: 0.6761
Epoch 10/15
594/594 - 2s - loss: 0.1918 - sparse_categorical_accuracy: 0.9350 - val_loss: 1.3776 - val_sparse_categorical_accuracy: 0.6769
Epoch 11/15
594/594 - 2s - loss: 0.1431 - sparse_categorical_accuracy: 0.9506 - val_loss: 1.6755 - val_sparse_categorical_accuracy: 0.6716
Epoch 12/15
594/594 - 2s - loss: 0.1153 - sparse_categorical_accuracy: 0.9597 - val_loss: 1.9183 - val_sparse_categorical_accuracy: 0.6591
Epoch 13/15
594/594 - 3s - loss: 0.1067 - sparse_categorical_accuracy: 0.9629 - val_loss: 1.9378 - val_sparse_categorical_accuracy: 0.6722
Epoch 14/15
594/594 - 2s - loss: 0.0853 - sparse_categorical_accuracy: 0.9710 - val_loss: 2.1068 - val_sparse_categorical_accuracy: 0.6619
Epoch 15/15
594/594 - 2s - loss: 0.0757 - sparse_categorical_accuracy: 0.9737 - val_loss: 2.2108 - val_sparse_categorical_accuracy: 0.6588
Epoch 1/15
633/633 - 3s - loss: 1.5012 - sparse_categorical_accuracy: 0.4590 - val_loss: 1.1979 - val_sparse_categorical_accuracy: 0.5764
Epoch 2/15
633/633 - 3s - loss: 1.0972 - sparse_categorical_accuracy: 0.6088 - val_loss: 1.0397 - val_sparse_categorical_accuracy: 0.6335
Epoch 3/15
633/633 - 3s - loss: 0.9187 - sparse_categorical_accuracy: 0.6767 - val_loss: 0.9740 - val_sparse_categorical_accuracy: 0.6604
Epoch 4/15
633/633 - 2s - loss: 0.8030 - sparse_categorical_accuracy: 0.7194 - val_loss: 0.9366 - val_sparse_categorical_accuracy: 0.6807
Epoch 5/15
633/633 - 2s - loss: 0.6915 - sparse_categorical_accuracy: 0.7605 - val_loss: 0.9130 - val_sparse_categorical_accuracy: 0.6837
Epoch 6/15
633/633 - 3s - loss: 0.5904 - sparse_categorical_accuracy: 0.7937 - val_loss: 0.8983 - val_sparse_categorical_accuracy: 0.7013
Epoch 7/15
633/633 - 2s - loss: 0.4926 - sparse_categorical_accuracy: 0.8276 - val_loss: 0.9430 - val_sparse_categorical_accuracy: 0.6991
Epoch 8/15
633/633 - 3s - loss: 0.3906 - sparse_categorical_accuracy: 0.8641 - val_loss: 1.0589 - val_sparse_categorical_accuracy: 0.6928
Epoch 9/15
633/633 - 3s - loss: 0.3032 - sparse_categorical_accuracy: 0.8929 - val_loss: 1.2086 - val_sparse_categorical_accuracy: 0.6885
Epoch 10/15
633/633 - 3s - loss: 0.2248 - sparse_categorical_accuracy: 0.9210 - val_loss: 1.3717 - val_sparse_categorical_accuracy: 0.6833
Epoch 11/15
633/633 - 3s - loss: 0.1714 - sparse_categorical_accuracy: 0.9389 - val_loss: 1.6122 - val_sparse_categorical_accuracy: 0.6761
Epoch 12/15
633/633 - 3s - loss: 0.1510 - sparse_categorical_accuracy: 0.9455 - val_loss: 1.6442 - val_sparse_categorical_accuracy: 0.6621
Epoch 13/15
633/633 - 3s - loss: 0.1172 - sparse_categorical_accuracy: 0.9592 - val_loss: 1.7635 - val_sparse_categorical_accuracy: 0.6783
Epoch 14/15
633/633 - 3s - loss: 0.0912 - sparse_categorical_accuracy: 0.9688 - val_loss: 2.0338 - val_sparse_categorical_accuracy: 0.6755
Epoch 15/15
633/633 - 3s - loss: 0.0915 - sparse_categorical_accuracy: 0.9681 - val_loss: 2.0622 - val_sparse_categorical_accuracy: 0.6784
Epoch 1/15
672/672 - 3s - loss: 1.4991 - sparse_categorical_accuracy: 0.4599 - val_loss: 1.2377 - val_sparse_categorical_accuracy: 0.5653
Epoch 2/15
672/672 - 3s - loss: 1.0594 - sparse_categorical_accuracy: 0.6277 - val_loss: 0.9675 - val_sparse_categorical_accuracy: 0.6616
Epoch 3/15
672/672 - 3s - loss: 0.8758 - sparse_categorical_accuracy: 0.6925 - val_loss: 0.8828 - val_sparse_categorical_accuracy: 0.6941
Epoch 4/15
672/672 - 3s - loss: 0.7596 - sparse_categorical_accuracy: 0.7335 - val_loss: 0.8410 - val_sparse_categorical_accuracy: 0.7099
Epoch 5/15
672/672 - 3s - loss: 0.6556 - sparse_categorical_accuracy: 0.7690 - val_loss: 0.8629 - val_sparse_categorical_accuracy: 0.6999
Epoch 6/15
672/672 - 3s - loss: 0.5544 - sparse_categorical_accuracy: 0.8045 - val_loss: 0.8670 - val_sparse_categorical_accuracy: 0.7111
Epoch 7/15
672/672 - 3s - loss: 0.4608 - sparse_categorical_accuracy: 0.8375 - val_loss: 0.9435 - val_sparse_categorical_accuracy: 0.7103
Epoch 8/15
672/672 - 3s - loss: 0.3579 - sparse_categorical_accuracy: 0.8730 - val_loss: 1.0184 - val_sparse_categorical_accuracy: 0.6906
Epoch 9/15
672/672 - 3s - loss: 0.2733 - sparse_categorical_accuracy: 0.9026 - val_loss: 1.1449 - val_sparse_categorical_accuracy: 0.7069
Epoch 10/15
672/672 - 3s - loss: 0.2079 - sparse_categorical_accuracy: 0.9275 - val_loss: 1.3158 - val_sparse_categorical_accuracy: 0.6946
Epoch 11/15
672/672 - 3s - loss: 0.1610 - sparse_categorical_accuracy: 0.9420 - val_loss: 1.4644 - val_sparse_categorical_accuracy: 0.6889
Epoch 12/15
672/672 - 3s - loss: 0.1290 - sparse_categorical_accuracy: 0.9548 - val_loss: 1.5539 - val_sparse_categorical_accuracy: 0.7021
Epoch 13/15
672/672 - 3s - loss: 0.1152 - sparse_categorical_accuracy: 0.9599 - val_loss: 1.6512 - val_sparse_categorical_accuracy: 0.6806
Epoch 14/15
672/672 - 3s - loss: 0.0907 - sparse_categorical_accuracy: 0.9692 - val_loss: 2.0270 - val_sparse_categorical_accuracy: 0.6886
Epoch 15/15
672/672 - 3s - loss: 0.0881 - sparse_categorical_accuracy: 0.9691 - val_loss: 2.0357 - val_sparse_categorical_accuracy: 0.6841
Epoch 1/15
711/711 - 3s - loss: 1.4616 - sparse_categorical_accuracy: 0.4759 - val_loss: 1.1891 - val_sparse_categorical_accuracy: 0.5838
Epoch 2/15
711/711 - 3s - loss: 1.0610 - sparse_categorical_accuracy: 0.6241 - val_loss: 1.0229 - val_sparse_categorical_accuracy: 0.6420
Epoch 3/15
711/711 - 3s - loss: 0.8676 - sparse_categorical_accuracy: 0.6933 - val_loss: 0.8819 - val_sparse_categorical_accuracy: 0.6958
Epoch 4/15
711/711 - 3s - loss: 0.7329 - sparse_categorical_accuracy: 0.7420 - val_loss: 0.8553 - val_sparse_categorical_accuracy: 0.7018
Epoch 5/15
711/711 - 3s - loss: 0.6141 - sparse_categorical_accuracy: 0.7839 - val_loss: 0.8605 - val_sparse_categorical_accuracy: 0.7009
Epoch 6/15
711/711 - 3s - loss: 0.4979 - sparse_categorical_accuracy: 0.8253 - val_loss: 0.9075 - val_sparse_categorical_accuracy: 0.7033
Epoch 7/15
711/711 - 3s - loss: 0.3815 - sparse_categorical_accuracy: 0.8675 - val_loss: 0.9478 - val_sparse_categorical_accuracy: 0.7142
Epoch 8/15
711/711 - 3s - loss: 0.2788 - sparse_categorical_accuracy: 0.9028 - val_loss: 1.0833 - val_sparse_categorical_accuracy: 0.7011
Epoch 9/15
711/711 - 3s - loss: 0.1978 - sparse_categorical_accuracy: 0.9307 - val_loss: 1.2711 - val_sparse_categorical_accuracy: 0.7042
Epoch 10/15
711/711 - 3s - loss: 0.1605 - sparse_categorical_accuracy: 0.9439 - val_loss: 1.4073 - val_sparse_categorical_accuracy: 0.6973
Epoch 11/15
711/711 - 3s - loss: 0.1164 - sparse_categorical_accuracy: 0.9596 - val_loss: 1.5774 - val_sparse_categorical_accuracy: 0.7087
Epoch 12/15
711/711 - 3s - loss: 0.0952 - sparse_categorical_accuracy: 0.9671 - val_loss: 1.6625 - val_sparse_categorical_accuracy: 0.6911
Epoch 13/15
711/711 - 3s - loss: 0.0904 - sparse_categorical_accuracy: 0.9690 - val_loss: 1.7850 - val_sparse_categorical_accuracy: 0.6958
Epoch 14/15
711/711 - 3s - loss: 0.0749 - sparse_categorical_accuracy: 0.9742 - val_loss: 1.8189 - val_sparse_categorical_accuracy: 0.6973
Epoch 15/15
711/711 - 3s - loss: 0.0742 - sparse_categorical_accuracy: 0.9740 - val_loss: 2.1127 - val_sparse_categorical_accuracy: 0.6882

plot the learning curve

x = num_train
y1 = acc_train
y2 =acc_test
plt.plot(x,y1, label='Training Accuracy')
plt.plot(x,y2, label='Validation Accuracy')
# plt. xlim (0, 23)
# plt. ylim (20, 70)
plt.xlabel( 'Number of training data', fontsize=12)
plt.ylabel ( 'Accuracy',fontsize = 12)
# plt.title('24 hour temperature and humidity measurement', fontsize=16)
plt. legend () 
plt. show ()


png

get the accuracy on training dataset

# pprint(a.history['sparse_categorical_accuracy'][0])

希仔

2021/04/07  阅读:25  主题:默认主题

作者介绍

希仔