how does the get_shape function work with placeholders?

[JenkinsY94]

I tried The code exapmle b = tf.placeholder(tf.float32, [None, 10, 32]); shape = get_shape(b) ,
but when I print out the shape, it show tensor objects, rather than the dynamic/static shape as expected.

I wonder how can I use this get_shape function in a session properly in order to get a placeholder's shape?

Thx!

[vahidk]

Try print(tf.Session().run(shape))

[JenkinsY94]

Hi vahidk. It seems not working because get_shape() returns a list of mixture of integer and tensor object.

Here's a working example:

def get_dyn_shape(tensor):
    dynamic_shape = tf.unstack(tf.shape(tensor))
    return dynamic_shape
a = tf.placeholder(tf.float32, [None, 128])
dyn_shape = get_dyn_shape(a)
print(tf.Session().run(dyn_shape, feed_dict={a:np.random.random((3, 128))}))

But When I replace get_dyn_shape with get_shape in your article, it failed. Tensorflow says:

TypeError: Fetch argument 128 has invalid type <class 'int'>, must be a string or Tensor. (Can not 
convert a int into a Tensor or Operation.)

Any suggestions?

Also, I'm confused about when to use this get_shape, as get_dyn_shape returns exactly same result?

[vahidk]

This get_shape() function is not designed to be used for printing shapes. If you want to do that just use session.run(tf.shape(x)). The output of get_shape() is useful for creating other tensors. What it does is that it tries to keep the shape static if possible otherwise dynamic which can be useful at times.

[Shunsuke-1994]

".values().get_shape()" or ".outputs.get_chape()" method worked!
https://www.tensorflow.org/api_docs/python/tf/Operation