Function Differences with tf.compat.v1.scatter_mul

tf.compat.v1.scatter_mul

tf.compat.v1.scatter_mul(
    ref,
    indices,
    updates,
    use_locking=False,
    name=None
) -> Tensor

For more information, see tf.compat.v1.scatter_mul.

mindspore.ops.scatter_mul

mindspore.ops.scatter_mul(
    input_x,
    indices,
    updates
) -> Tensor

For more information, see mindspore.ops.scatter_mul.

Usage

TensorFlow: In-place scatter update for Tensor.

MindSpore: Implement the same function as TensorFlow. TensorFlow can use the use_locking parameter to control whether locking is used when updating the tensor. Locking ensures that the Tensor can be updated correctly in a multi-threaded environment, and the default is False. MindSpore implements unlocked function by default.

Categories	Subcategories	TensorFlow	MindSpore	Differences
Parameters	Parameter1	ref	input_x	Same function, different parameter names
	Parameter2	indices	indices	-
	Parameter3	updates	updates	-
	Parameter4	use_locking	-	MindSpore does not have this parameter and implements unlocked functionality by default.
	Parameter5	name	-	Not involved

Code Example

When use_locking is False in TensorFlow, the two APIs implement the same function.

# TensorFlow
import numpy as np
import tensorflow.compat.v1 as tf
tf.disable_v2_behavior()

ref = tf.Variable(np.array([[1.0, 2.0, 3.0], [1.0, 2.0, 3.0]]), dtype=tf.float32)
indices = tf.constant(np.array([0, 1]),  dtype=tf.int32)
updates = tf.constant(np.array([[1.0, 3.0, 5.0], [2.0, 4.0, 6.0]]), dtype=tf.float32)
op = tf.compat.v1.scatter_mul(ref, indices, updates, use_locking=False)

init = tf.compat.v1.global_variables_initializer()
with tf.compat.v1.Session() as sess:
    sess.run(init)
    out = sess.run(op)
print(out)
# [[ 1.  6. 15.]
#  [ 2.  8. 18.]]

# MindSpore
import numpy as np
import mindspore
from mindspore import Tensor, Parameter
import mindspore.ops as ops

input_x = Parameter(Tensor(np.array([[1.0, 2.0, 3.0], [1.0, 2.0, 3.0]]), mindspore.float32), name="x")
indices = Tensor(np.array([0, 1]), mindspore.int32)
updates = Tensor(np.array([[1.0, 3.0, 5.0], [2.0, 4.0, 6.0]]), mindspore.float32)
output = ops.scatter_mul(input_x, indices, updates)
print(output)
# [[ 1.  6. 15.]
#  [ 2.  8. 18.]]