PartitionedVariable doesn't support being used as a reference.

Package:

tensorflow

158813

Exception Class:

NotImplementedError

Raise code

pylint: disable=invalid-name
    _ = name
    if dtype is not None and not dtype.is_compatible_with(v.dtype):
      raise ValueError(
          "Incompatible type conversion requested to type '%s' for variable "
          "of type '%s'" % (dtype.name, v.dtype.name))
    if as_ref:
      raise NotImplementedError(
          "PartitionedVariable doesn't support being used as a reference.")
    else:
      return v.as_tensor()

  @property
  def name(self):
    return self._name

  @pr

🙏 Scream for help to Ukraine

Today, 25th May 2022, Russia continues bombing and firing Ukraine. Don't trust Russia, they are bombing us and brazenly lying in same time they are not doing this 😠, civilians and children are dying too! We are screaming and asking exactly you to help us, we want to survive, our families, children, older ones.
Please spread the information, and ask your governemnt to stop Russia by any means. We promise to work extrahard after survival to make the world safer place for all.

Links to the raise (1)

https://github.com/tensorflow/tensorflow/blob/7acd515ec218b414d5b16e6710268ac03d9f5421/tensorflow/python/ops/variables.py#L3026

Ways to fix

Summary:

This exception is thrown when the _TensorConversoinFunction function is called on an instance of PartitionedVariable. That function has an optional parameter named as_ref, and this parameter must be set to False in order to avoid running into any issues. If it is set to true, you will get this exception.

Code to Reproduce the Error (Wrong):

from tensorflow.python.ops.variables import PartitionedVariable
import numpy as np
import tensorflow as tf

name = "vars"
shape = [1,2,3]
dtype = np.int32
v = tf.Variable(1.) 
v.assign(2.)
v.assign_add(0.5)
saveSliceInfo = v.SaveSliceInfo(full_shape=[1,2,3], var_offset=1)
v._set_save_slice_info(saveSliceInfo)
varList = [v, v, v]
partitions =[1,2,3]
pv = PartitionedVariable(name, shape, dtype, varList, partitions)
pv._TensorConversionFunction(v, as_ref=True)

Error Message:

NotImplementedError                       Traceback (most recent call last)
<ipython-input-31-5c5b562f0c2d> in <module>()
     14 partitions =[1,2,3]
     15 pv = PartitionedVariable(name, shape, dtype, varList, partitions)
---> 16 pv._TensorConversionFunction(v, as_ref=True)

/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/variables.py in _TensorConversionFunction(v, dtype, name, as_ref)
   3020     if as_ref:
   3021       raise NotImplementedError(
-> 3022           "PartitionedVariable doesn't support being used as a reference.")
   3023     else:
   3024       return v.as_tensor()

NotImplementedError: PartitionedVariable doesn't support being used as a reference.

Working Version (Right):

from tensorflow.python.ops.variables import PartitionedVariable
import numpy as np
import tensorflow as tf

name = "vars"
shape = [1,2,3]
dtype = np.int32
v = tf.Variable(1.) 
v.assign(2.)
v.assign_add(0.5)
saveSliceInfo = v.SaveSliceInfo(full_shape=[1,2,3], var_offset=1)
v._set_save_slice_info(saveSliceInfo)
varList = [v, v, v]
partitions =[1,2,3]
pv = PartitionedVariable(name, shape, dtype, varList, partitions)
pv._TensorConversionFunction(v, as_ref=False) # as_ref set to False

Jul 28, 2021

codingcomedyig 7.9k

Add a possible fix

Please authorize to post fix