Keras (tensorflow backend) getting "TypeError: unhashable type: 'Dimension'"

Question

Hi I get a dimension error when fitting this model, would anyone know why?

num_classes = 11
input_shape = (64,64,1)
batch_size = 128
epochs = 12
X_train = tf.reshape(X_train, [-1, 64, 64, 1])
X_test = tf.reshape(X_test, [-1, 64, 64, 1])

model = Sequential()
model.add(Conv2D(32, kernel_size=(3,3), strides=1, activation='relu', input_shape=input_shape)) 
model.add(Flatten())
model.add(Dense(128, activation='relu'))
model.add(Dense(num_classes, activation='softmax'))

model.compile(loss=keras.losses.categorical_crossentropy,
          optimizer=keras.optimizers.Adadelta(),
          metrics=['accuracy'])

model.fit(X_train, y_train,
      batch_size=batch_size,
      epochs=epochs,
      verbose=1,
      validation_data=(X_test, y_test))

The dimension of each variable is

X_train = (27367, 64, 64, 1)
X_test = (4553, 64, 64, 1)
y_train = (164202, 11)
y_test = (27318, 11)

Answer 1

This is because you are using tf.reshape, which returns a Tensor, and the fit method of Keras models don't work well with tensors.

Consider using np.reshape instead, which will do the exact same thing.