Function Differences with tf.fill

tf.fill

tf.fill(dims, value, name=None) -> Tensor

For more information, see tf.fill.

mindspore.ops.fill

mindspore.ops.fill(type, shape, value) -> Tensor

For more information, see mindspore.ops.fill.

Differences

TensorFlow: ‎ is used to generate a tensor with scalar values.

MindSpore: MindSpore API implements the same function as TensorFlow, and only the parameter names are different.

Categories	Subcategories	TensorFlow	MindSpore	Differences
Parameters	Parameter 1	dims	shape	Same function, different parameter names
	Parameter 2	value	value	-
	Parameter 3	name	-	Not involved
	Parameter 4	-	type	Specify the data type of the output Tensor

Code Example 1

Both APIs implement the same function. MindSpore only has one more parameter specifying the type of output, and the rest of the parameters are used in the same way.

# TensorFlow
import tensorflow as tf
import numpy as np

dims = np.array([2,3])
value = 9
output = tf.fill(dims, value)
output_m = output.numpy()
print(output_m)
#[[9 9 9]
# [9 9 9]]

# MindSpore
import mindspore
import mindspore.ops as ops

type = mindspore.int32
shape = tuple((2,3))
value = 9
output = ops.fill(type, shape, value)
print(output)
#[[9 9 9]
# [9 9 9]]