Basic CNN classification model has UnimplementedError: Graph execution error:

Question

I tried a sample code for CNN application on MNIST data classification from the book :

from keras import layers
from keras import models

model = models.Sequential()
model.add(layers.Conv2D(32, (3,3), activation = 'relu', input_shape = (28, 28, 1)))
model.add(layers.MaxPooling2D((2,2)))
model.add(layers.Conv2D(64, (3,3), activation = 'relu'))
model.add(layers.MaxPooling2D((2,2)))
model.add(layers.Conv2D(64, (3,3), activation = 'relu'))
model.add(layers.Flatten())
model.add(layers.Dense(64, activation = 'relu'))
model.add(layers.Dense(10, activation = 'softmax'))
model.summary()

#Test this model on mnist
from keras.datasets import mnist
from tensorflow.keras.utils import to_categorical

(train_images, train_labels), (test_images, test_labels) = mnist.load_data()
train_images = train_images.reshape((60000,28,28,1))
train_images = train_images.astype('float32')/255
test_images = test_images.reshape((10000,28,28,1))
test_images = test_images.astype('float32')/255
train_labels = to_categorical(train_labels)
test_labels = to_categorical(test_labels)
model.compile(optimizer = 'rmsprop', loss = 'categorical_crossentropy', metrics = ['accuracy'])
model.fit(train_images, train_labels, epochs=5, batch_size=64)

The code should be correct, but an error occur when I run the code:

UnimplementedError: Graph execution error:

I think the problem might result from different version of tensorflow (my tensorflow is 2.8, while the sample code was run in tensorflow 2.0). Could anyone tell me how to tackle this problem?

Answer 1

look like you need to perform on shape not by all grains at single numbers.

[ Sample ]:

import tensorflow as tf

import tensorflow_datasets as tfds
from tensorflow.keras.utils import to_categorical

"""""""""""""""""""""""""""""""""""""""""""""""""""""""""
: DataSets
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""
ds = tfds.load('mnist', split='train', shuffle_files=True)
ds = ds.shuffle(1024).batch(64).prefetch(tf.data.experimental.AUTOTUNE)
assert isinstance(ds, tf.data.Dataset)

for example in ds.take(1):
    image, label = example["image"], example["label"]
    #################################################
    image = tf.cast( image, dtype=tf.float32 )
    image = tf.math.divide_no_nan( image, 255 )
    #################################################
    
train_images = image
train_labels = to_categorical(label)
test_labels = to_categorical(label)

"""""""""""""""""""""""""""""""""""""""""""""""""""""""""
: Model Initialize
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""
model = tf.keras.models.Sequential()
model.add(tf.keras.layers.Conv2D(32, (3,3), activation = 'relu', input_shape = (28, 28, 1)))
model.add(tf.keras.layers.MaxPooling2D((2,2)))
model.add(tf.keras.layers.Conv2D(64, (3,3), activation = 'relu'))
model.add(tf.keras.layers.MaxPooling2D((2,2)))
model.add(tf.keras.layers.Conv2D(64, (3,3), activation = 'relu'))
model.add(tf.keras.layers.Flatten())
model.add(tf.keras.layers.Dense(64, activation = 'relu'))
model.add(tf.keras.layers.Dense(10, activation = 'softmax'))
model.summary()

model.compile(optimizer = 'rmsprop', loss = 'categorical_crossentropy', metrics = ['accuracy'])
model.fit(train_images, train_labels, epochs=5, batch_size=64)

[ Output ]: