ValueError: If your data is in the form of symbolic tensors, you should specify the `steps` argument (instead of the `batch_size` argument, because symbolic tensors are expected to produce batches of input data)

[abhisheksrivastava2397]

`import os import tensorflow as tf from generator import Generator from discriminator import Discriminator from keras.applications.vgg16 import VGG16 from keras.preprocessing import image from keras.applications.vgg16 import preprocess_input import numpy as np from keras.layers import Input, Flatten, Dense

class AACNN(): """AACNN model. """ def init(self, sess, FLAGS): """Initialization.

    Args:
        sess: TensorFlow session
        FLAGS: flags object
    """
    # initialize variables
    self.sess = sess
    self.f = FLAGS
    self.l2_weight = FLAGS.l2_weight
    self.global_step_g = tf.Variable(0, trainable=False)
    self.global_step_d = tf.Variable(0, trainable=False)

    # inputs: real (training) images
    images_shape = [self.f.output_size_height, self.f.output_size_wight, self.f.c_dim]
    attributes_shape = [self.f.attribute_size]
    # self.real_images = tf.placeholder(tf.float32, [None] + images_shape, name="real_images")
    
    # inputs: HR images 	
    self.input = tf.placeholder(tf.float32, [self.f.batch_size] + images_shape, name="input")
    self.input_attribute = tf.placeholder(tf.float32, [self.f.batch_size] + attributes_shape, name="input_attribute")

    # initialize models
    generator = Generator(FLAGS)
    discriminator = Discriminator(FLAGS)
    model = VGG16(weights='imagenet',include_top=False)
    # generator network
    self.G = generator(self.input, self.input_attribute)

    if self.f.with_gan:
        ### find out wether self.input is an image or what answer:get_image returns  numpy array to
        # discriminator network for real images
        self.D_real, self.D_real_logits = discriminator(self.input)
        # discriminator network for fake images
        self.D_fake, self.D_fake_logits = discriminator(self.G, reuse=True)
        # feature extraction by using VGG16
        # first unscale the images then preprocess it and then extract features
        #self.input_unscaled = sess.run(self.input)
        #self.G_unscaled = sess.run(self.G)
        #self.img_real = (self.input_unscaled+1)*127.5
        #self.img_fake = (self.G_unscaled + 1)*127.5
        self.img_real = (self.input+1)*127.5
        self.img_fake = (self.G + 1)*127.5
        print("-"*100)
        print(self.img_real.shape)
        print("-"*100)
        #self.img_real=np.expand_dims(self.img_real,axis=0)
        #self.img_fake=np.expand_dims(self.img_fake,axis=0)
        print("-"*100)
        print(self.img_real.shape)
        print("-"*100)
        self.img_real = preprocess_input(self.img_real)
        self.img_fake = preprocess_input(self.img_fake)
        print("-"*100)
        print(self.img_real.shape)
        print("-"*100)
        self.feature_real = model.predict(self.img_real)
        self.feature_fake = model.predict(self.img_fake)
        self.feature_real = tf.convert_to_tensor(self.feature_real , dtype=tf.float32)
        self.feature_fake = tf.convert_to_tensor(self.feature_fake , dtype=tf.float32)
        #
        self.vgg_loss = 2e-6 * tf.cost.mean_squared_error(self.feature_fake, self.feature_real, is_mean=True)
        # losses
        self.d_loss_real = tf.reduce_mean(
            tf.nn.sigmoid_cross_entropy_with_logits(
                logits=self.D_real_logits,
                labels=tf.ones_like(self.D_real))
            )
        self.d_loss_fake = tf.reduce_mean(
            tf.nn.sigmoid_cross_entropy_with_logits(
                logits=self.D_fake_logits,
                labels=tf.zeros_like(self.D_fake))
            )
        self.d_loss = self.d_loss_real + self.d_loss_fake
        self.g_loss_adv = tf.reduce_mean(
            tf.nn.sigmoid_cross_entropy_with_logits(
                logits=self.D_fake_logits,
                labels=tf.ones_like(self.D_fake))
            )
    else:
        self.l2_weight = 1
        
    self.g_loss_l2 = tf.losses.mean_squared_error(
        self.input,
        self.G,
        weights=self.l2_weight,
        scope=None
        )
   
    if self.f.with_gan:
        self.g_loss = self.g_loss_l2 + self.g_loss_adv + self.vgg_loss
    else: 
        self.g_loss = self.g_loss_l2

    # create summaries
    self.__create_summaries()

    # organize variables
    t_vars = tf.trainable_variables()
    if self.f.with_gan:
        self.d_vars = [var for var in t_vars if "d/" in var.name]
    self.g_vars = [var for var in t_vars if "g/" in var.name]
    #print self.g_vars
    # saver
    self.saver = tf.train.Saver()

`

[abhisheksrivastava2397]

here i have a batch size of 64 images image shape is(112,96,3) i want to extract features of self.input and self.G of the entire batch but the model cant take entire batch as input at once

[taehoonlee]

@abhisheksrivastava2397, Can you share your snippets to reproduce the error?

[aGiant]

>>> print(tf.__version__)
2.0.0

Try:

cce = tf.keras.losses.CategoricalCrossentropy()
t = [[1., 0., 0.], [0., 1., 0.], [0., 0., 1.]]
y = [[.9, .05, .05], [.5, .89, .6], [.05, .01, .94]]
loss = cce(t,y)

Error:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/llu/anaconda3/lib/python3.7/site-packages/tensorflow_core/python/keses.py", line 126, in __call__
    losses = self.call(y_true, y_pred)
  File "/home/llu/anaconda3/lib/python3.7/site-packages/tensorflow_core/python/keses.py", line 221, in call
    return self.fn(y_true, y_pred, **self._fn_kwargs)
  File "/home/llu/anaconda3/lib/python3.7/site-packages/tensorflow_core/python/keses.py", line 978, in sparse_categorical_crossentropy
    y_true, y_pred, from_logits=from_logits, axis=axis)
  File "/home/llu/anaconda3/lib/python3.7/site-packages/tensorflow_core/python/kekend.py", line 4503, in sparse_categorical_crossentropy
    output.op.type != 'Softmax'):
AttributeError: 'list' object has no attribute 'op'

And this is standard example from tensorflow: https://www.tensorflow.org/api_docs/python/tf/keras/losses/CategoricalCrossentropy

[chintanp]

@abhisheksrivastava2397 any progress on this?