tf.compat.v1.sparse_placeholder  |  TensorFlow v2.16.1 (original) (raw)

Inserts a placeholder for a sparse tensor that will be always fed.

View aliases

Compat aliases for migration

SeeMigration guide for more details.

tf.compat.v1.sparse.placeholder

tf.compat.v1.sparse_placeholder(
    dtype, shape=None, name=None
)

Migrate to TF2

This API is not compatible with eager execution and tf.function. To migrate to TF2, rewrite the code to be compatible with eager execution. Check themigration guideon replacing Session.run calls. In TF2, you can just pass tensors directly into ops and layers. If you want to explicitly set up your inputs, also seeKeras functional API on how to use tf.keras.Input to replace tf.compat.v1.sparse_placeholder.tf.function arguments also do the job of tf.compat.v1.sparse_placeholder. For more details please read Better performance with tf.function.

Description

Used in the notebooks

For example:

x = tf.compat.v1.sparse.placeholder(tf.float32)
y = tf.sparse.reduce_sum(x)

with tf.compat.v1.Session() as sess:
  print(sess.run(y))  # ERROR: will fail because x was not fed.

  indices = np.array([[3, 2, 0], [4, 5, 1]], dtype=np.int64)
  values = np.array([1.0, 2.0], dtype=np.float32)
  shape = np.array([7, 9, 2], dtype=np.int64)
  print(sess.run(y, feed_dict={
    x: tf.compat.v1.SparseTensorValue(indices, values, shape)}))  # Will
    succeed.
  print(sess.run(y, feed_dict={
    x: (indices, values, shape)}))  # Will succeed.

  sp = tf.sparse.SparseTensor(indices=indices, values=values,
                              dense_shape=shape)
  sp_value = sp.eval(session=sess)
  print(sess.run(y, feed_dict={x: sp_value}))  # Will succeed.