How to cast int32 tensor to float32

Question

How can I cast an int32 tensor to float32 in tensorflow. I don't understand what tf.cast does. It does not seem to do anything.

import tensorflow as tf
import numpy as np

tf.enable_eager_execution()

a = tf.constant([[1, 2, 3, 4], [1, 2, 3, 4]])
b = tf.cast(a, dtype=tf.float32)

print(tf.shape(a))
print(tf.shape(b))

outputs;

tf.Tensor([2 4], shape=(2,), dtype=int32) #a   
tf.Tensor([2 4], shape=(2,), dtype=int32) #b

Answer 1

If you just use;

print(a)
print(b)

you shall get the correct result;

tf.Tensor(
[[1 2 3 4]
 [1 2 3 4]], shape=(2, 4), dtype=int32) #a
tf.Tensor(
[[1. 2. 3. 4.]
 [1. 2. 3. 4.]], shape=(2, 4), dtype=float32) #b

So the tf.cast() works as intended!

---

With tf.shape() you are getting a result that explains the shape details of input.

Returns: A Tensor of type out_type.

out_type: (Optional) The specified output type of the operation (int32 or int64). Defaults to tf.int32

So the dtype of tf.shape() result, is the dtype of the resulting "shape detailing tensor", not of a, or b