ValueError: Can't convert non-rectangular Python sequence to Tensor

Question

I want to change list to tensor with tf.convert_to_tensor, data is following:

data=[

array([0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
   0., 0., 0., 0., 0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
   0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
   0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
   0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
   0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
   1., 0., 0.]), 
array([0., 0., 0., 0., 0., 0., 0., 0., 1., 0., 0., 0.]), 
array([0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
   0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
   0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
   0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
   0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
   0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 1.,
   0., 0., 0.]), 
array([0., 0., 0., 0., 0., 0., 1., 0., 0., 0., 0., 0.])
    ]

it didn't work, system says:

ValueError: Can't convert non-rectangular Python sequence to Tensor.

how to solve this problem?

Maifee Ul Asad · Accepted Answer · 2022-04-29 10:24:48Z

35

I'm not sure whether they exist in TensorFlow 1 but TensorFlow 2.0 supports RaggedTensors, which the documentation describes as "... the TensorFlow equivalent of nested variable-length lists."

I think it would be trivial to convert your data to RaggedTensors. It might even be as easy as:

data_tensor = tf.ragged.constant(data)

Example:

>>> a = tf.ragged.constant([[1],[2,3]])
>>> a
<tf.RaggedTensor [[1], [2, 3]]>

edited Apr 29, 2022 at 10:24

Maifee Ul Asad

4,68910 gold badges52 silver badges120 bronze badges

answered Nov 4, 2019 at 13:34

Eric McLachlan

3,6302 gold badges30 silver badges41 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

Dan D Over a year ago

ragged tensor indeed will add padding zeros?

Eric McLachlan Over a year ago

Not by default, no. Check out the documentation: tensorflow.org/guide/ragged_tensor

BlueSun · Accepted Answer · 2019-05-25 15:59:04Z

11

You can't. Like the error message says, TensorFlow arrays can not have different sizes along one dimension. Try to use a list of TensorFlow arrays instead or the dataset api.

answered May 25, 2019 at 15:59

BlueSun

3,5701 gold badge21 silver badges37 bronze badges

1 Comment

tdMJN6B2JtUe Over a year ago

Here is one example I gave of using the tf.data.Dataset API, but the tensors were all the same shape: stackoverflow.com/a/66175008/12763497

Collectives™ on Stack Overflow

ValueError: Can't convert non-rectangular Python sequence to Tensor

2 Answers 2

2 Comments

1 Comment

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

2 Comments

1 Comment

Your Answer

Sign up or log in

Post as a guest

Linked

Related