[Bug] `initializer`生成的Tensor行为不符合预期

mindspore==2.3.0rc1

代码如下

from mindspore.common.initializer import initializer
n, m = 5, 3
x = initializer(init='normal', shape=(n, m))
y = x.copy()
print(x)
print(y)

输出如下

[[-9.8994775e-03 -2.8827513e-05 -1.1877520e-02] [-2.8584851e-03 -1.4535164e-02 -1.7329855e-02] [-1.1603085e-03 2.8330029e-04 -8.1528882e-03] [-1.4099754e-02 -1.1020763e-02 3.4321868e-03] [ 1.7921604e-02 1.4062168e-02 -4.5925970e-03]] [[0. 0. 0.] [0. 0. 0.] [0. 0. 0.] [0. 0. 0.] [0. 0. 0.]]

y没有成功复制x而是变成了0。如果把x换成普通的Tensor就没问题，猜测是initializer导致的

demerzel-iv / bmtrain_mindspore

[Bug] `initializer`生成的Tensor行为不符合预期 #2