NotImplementedError: Cannot convert a symbolic Tensor (truediv_2:0) to a numpy array

[nbro]

System information

• Have I written custom code: Yes
• OS Platform and Distribution: Mac OS Catalina
• TensorFlow installed from: binary
• TensorFlow version (use command below): 2.1.0
• Python version: 3.7.0

Describe the current behavior

I get the following error

NotImplementedError: Cannot convert a symbolic Tensor (truediv_2:0) to a numpy array.

When executing the following code

import tensorflow as tf
import tensorflow_probability as tfp

tf.config.experimental_run_functions_eagerly(True)


def get_mnist_data(normalize=True, categorize=True):
    img_rows, img_cols = 28, 28
    (x_train, y_train), (x_test, y_test) = tf.keras.datasets.mnist.load_data()

    if tf.keras.backend.image_data_format() == 'channels_first':
        x_train = x_train.reshape(x_train.shape[0], 1, img_rows, img_cols)
        x_test = x_test.reshape(x_test.shape[0], 1, img_rows, img_cols)
        input_shape = (1, img_rows, img_cols)
    else:
        x_train = x_train.reshape(x_train.shape[0], img_rows, img_cols, 1)
        x_test = x_test.reshape(x_test.shape[0], img_rows, img_cols, 1)
        input_shape = (img_rows, img_cols, 1)

    x_train = x_train.astype('float32')
    x_test = x_test.astype('float32')

    if normalize:
        x_train /= 255
        x_test /= 255

    if categorize:
        y_train = tf.keras.utils.to_categorical(y_train)
        y_test = tf.keras.utils.to_categorical(y_test)

    return x_train, y_train, x_test, y_test, input_shape


def get_model(input_shape, num_classes=10):
    model = tf.keras.Sequential()
    model.add(tfp.layers.Convolution2DFlipout(6, input_shape=input_shape, kernel_size=3, padding="SAME",
                                              activation=tf.nn.relu))
    model.add(tf.keras.layers.Flatten())
    model.add(tfp.layers.DenseFlipout(num_classes))
    return model


def train():
    x_train, y_train, x_test, y_test, input_shape = get_mnist_data()

    batch_size = 64

    model = get_model(input_shape)

    model.summary()

    model.compile(loss="categorical_crossentropy")

    model.fit(x_train, y_train, batch_size=batch_size, epochs=1)


if __name__ == '__main__':
    train()

This error is caused by the statement tf.config.experimental_run_functions_eagerly(True). However, if I remove that statement, I get another well known, older and extremely annoying problem/bug that is described in this other issue: #33729 that doesn't allow me to do anything for my work.

Describe the expected behavior

NO BUG or ERROR.

See also #38775 and https://stackoverflow.com/q/61388919/3924118.

[Saduf2019]

@nbro
i ran the code shared, please find the gist here

[nbro]

@Saduf2019 You are not using TensorFlow 2.1.0 in that gist.

The new versions of TF are even buggier, so I would like to avoid discussions related to newer versions now (unless they solve my problems, which would be like a miracle).

[Saduf2019]

@nbro
the error on 2.1 is as per this gist

[nbro]

@Saduf2019 You're using tensorflow-probability 0.10.0rc0 in that gist. Use TFP 0.9.0 (the last stable version).

[Saduf2019]

@nbro
I am able to replicate this issue with the specifications requested please find the gist here

I also ran the code shared by you on nightly and do not face any errros, please find the gist here for the same

[google-ml-butler]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you.

[google-ml-butler]

Closing as stale. Please reopen if you'd like to work on this further.

[google-ml-butler]

Are you satisfied with the resolution of your issue?
Yes
No

[Ramkumar47]

Hi, I am also getting the exact same issue, but sorry for not sharing the code as it is on a project.
Is there any solution for this?
i am getting this error when eager execution is enabled, but i can get it to work when it is disabled,
i need eager execution for debugging my code.

[alik604]

so sad that he got ghosted...
I've got the same error

[Saduf2019]

@Ramkumar47

Please create a new issue as this issue is in closed status.

[yparam98]

@Ramkumar47 Any luck? Getting the exact same issue with this:

data_augmentation = keras.Sequential(
    [
        layers.experimental.preprocessing.RandomFlip("horizontal",
                                                     input_shape=(img_height,
                                                                  img_width,
                                                                  3)),
        layers.experimental.preprocessing.RandomRotation(0.1),
        layers.experimental.preprocessing.RandomZoom(0.1),
    ]
)

Error Trace:

Traceback (most recent call last):
  File "/home/yathavan/Documents/python_whiteboard/DRAFT_cvproject/flowers/flowers.py", line 136, in <module>
    data_augmentation = keras.Sequential(
  File "/usr/lib/python3.9/site-packages/tensorflow/python/training/tracking/base.py", line 517, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "/usr/lib/python3.9/site-packages/tensorflow/python/keras/engine/sequential.py", line 144, in __init__
    self.add(layer)
  File "/usr/lib/python3.9/site-packages/tensorflow/python/training/tracking/base.py", line 517, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "/usr/lib/python3.9/site-packages/tensorflow/python/keras/engine/sequential.py", line 223, in add
    output_tensor = layer(self.outputs[0])
  File "/usr/lib/python3.9/site-packages/tensorflow/python/keras/engine/base_layer.py", line 951, in __call__
    return self._functional_construction_call(inputs, args, kwargs,
  File "/usr/lib/python3.9/site-packages/tensorflow/python/keras/engine/base_layer.py", line 1090, in _functional_construction_call
    outputs = self._keras_tensor_symbolic_call(
  File "/usr/lib/python3.9/site-packages/tensorflow/python/keras/engine/base_layer.py", line 822, in _keras_tensor_symbolic_call
    return self._infer_output_signature(inputs, args, kwargs, input_masks)
  File "/usr/lib/python3.9/site-packages/tensorflow/python/keras/engine/base_layer.py", line 863, in _infer_output_signature
    outputs = call_fn(inputs, *args, **kwargs)
  File "/usr/lib/python3.9/site-packages/tensorflow/python/keras/layers/preprocessing/image_preprocessing.py", line 866, in call
    output = control_flow_util.smart_cond(training, random_rotated_inputs,
  File "/usr/lib/python3.9/site-packages/tensorflow/python/keras/utils/control_flow_util.py", line 114, in smart_cond
    return smart_module.smart_cond(
  File "/usr/lib/python3.9/site-packages/tensorflow/python/framework/smart_cond.py", line 54, in smart_cond
    return true_fn()
  File "/usr/lib/python3.9/site-packages/tensorflow/python/keras/layers/preprocessing/image_preprocessing.py", line 861, in random_rotated_inputs
    get_rotation_matrix(angles, img_hd, img_wd),
  File "/usr/lib/python3.9/site-packages/tensorflow/python/keras/layers/preprocessing/image_preprocessing.py", line 757, in get_rotation_matrix
    array_ops.zeros((num_angles, 2), dtypes.float32),
  File "/usr/lib/python3.9/site-packages/tensorflow/python/util/dispatch.py", line 201, in wrapper
    return target(*args, **kwargs)
  File "/usr/lib/python3.9/site-packages/tensorflow/python/ops/array_ops.py", line 2819, in wrapped
    tensor = fun(*args, **kwargs)
  File "/usr/lib/python3.9/site-packages/tensorflow/python/ops/array_ops.py", line 2868, in zeros
    output = _constant_if_small(zero, shape, dtype, name)
  File "/usr/lib/python3.9/site-packages/tensorflow/python/ops/array_ops.py", line 2804, in _constant_if_small
    if np.prod(shape) < 1000:
  File "<__array_function__ internals>", line 5, in prod
  File "/usr/lib/python3.9/site-packages/numpy/core/fromnumeric.py", line 3030, in prod
    return _wrapreduction(a, np.multiply, 'prod', axis, dtype, out,
  File "/usr/lib/python3.9/site-packages/numpy/core/fromnumeric.py", line 87, in _wrapreduction
    return ufunc.reduce(obj, axis, dtype, out, **passkwargs)
  File "/usr/lib/python3.9/site-packages/tensorflow/python/framework/ops.py", line 852, in __array__
    raise NotImplementedError(
NotImplementedError: Cannot convert a symbolic Tensor (random_rotation/rotation_matrix/strided_slice:0) to a numpy array. This error may indicate that you're trying to pass a Tensor to a NumPy call, which is not supported