How to use a tensor to initialize a variable in tensorflow?

Question

I'm trying to use tf.nn.embedding_lookup() and I get the following warning:

UserWarning: Converting sparse IndexedSlices to a dense Tensor of unknown shape. This may consume a large amount of memory.

I read in this link that to avoid this issue we should ensure that the params input to tf.nn.embedding_lookup() is a tf.Variable

But what I pass to tf.nn.embedding_lookup() is already a tensor that is the output of another operation, and I imagine I cannot use it to initialize the tf.Variable.

Is there a way to convert a tensor to tf.Variable or initialize one with another tensor?

Hi, does my answer solve your doubt?

juvchan
– juvchan

2019-06-24 10:03:57 +00:00
Commented Jun 24, 2019 at 10:03 — juvchan
– juvchan, Commented Jun 24, 2019 at 10:03

juvchan · Accepted Answer · 2019-06-23 10:06:39Z

2

A tf.Variable represents a tensor whose value can be changed by running ops on it.

Internally, a tf.Variable stores a persistent tensor.

Source: https://www.tensorflow.org/guide/variables

You can initialize a tf.Variable with a Tensor object: tf.Tensor

# Create a variable.
w = tf.Variable(<initial-value>, name=<optional-name>)

initial_value: A Tensor, or Python object convertible to a Tensor, which is the initial value for the Variable.

The initial value must have a shape specified unless validate_shape is set to False. Can also be a callable with no argument that returns the initial value when called. In that case, dtype must be specified.

(Note that initializer functions from init_ops.py must first be bound to a shape before being used here.)

In summary, you can initialize a tf.Variable with another Tensor object.

edited Jun 23, 2019 at 10:06

answered Jun 23, 2019 at 5:54

juvchan

6,2632 gold badges25 silver badges37 bronze badges

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

Collectives™ on Stack Overflow

How to use a tensor to initialize a variable in tensorflow?

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related