Function Differences with tf.keras.layers.BatchNormalization

tf.keras.layers.BatchNormalization(
    axis=-1,
    momentum=0.99,
    epsilon=0.001,
    center=True,
    scale=True,
    beta_initializer='zeros',
    gamma_initializer='ones',
    moving_mean_initializer='zeros',
    moving_variance_initializer='ones',
    beta_regularizer=None,
    gamma_regularizer=None,
    beta_constraint=None,
    gamma_constraint=None,
    **kwargs
)(inputs, training) -> Tensor

class mindspore.nn.BatchNorm2d(
    num_features,
    eps=1e-5,
    momentum=0.9,
    affine=True,
    gamma_init='ones',
    beta_init='zeros',
    moving_mean_init='zeros',
    moving_var_init='ones',
    use_batch_statistics=None,
    data_format='NCHW'
)(x) -> Tensor

# TensorFlow
import tensorflow as tf

inputx = [[[[1, 2],
           [2, 1]],
          [[3, 4],
           [4, 3]]],
         [[[5, 6],
           [6, 5]],
          [[7, 8],
           [8, 7]]]]
input_tf = tf.constant(inputx, dtype=tf.float32)
output_tf = tf.keras.layers.BatchNormalization(axis=3, momentum=0.1, epsilon=1e-5)
output = output_tf(input_tf, training=False)
print(output.numpy())
# [[[[0.999995  1.99999  ]
#    [1.99999   0.999995 ]]
#
#   [[2.999985  3.99998  ]
#    [3.99998   2.999985 ]]]
#
#
#  [[[4.999975  5.99997  ]
#    [5.99997   4.999975 ]]
#
#   [[6.9999647 7.99996  ]
#    [7.99996   6.9999647]]]]

# MindSpore
from mindspore import Tensor, nn
import numpy as np

m = nn.BatchNorm2d(num_features=2, momentum=0.9)
input_ms = Tensor(np.array([[[[1, 2], [2, 1]],
                          [[3, 4], [4, 3]]],
                          [[[5, 6], [6, 5]],
                          [[7, 8], [8, 7]]]]).astype(np.float32))
output = m(input_ms)
print(output)
# [[[[0.99999493 1.9999899 ]
#    [1.9999899  0.99999493]]
#
#   [[2.9999847  3.9999797 ]
#    [3.9999797  2.9999847 ]]]
#
#
#  [[[4.9999747  5.9999695 ]
#    [5.9999695  4.9999747 ]]
#
#   [[6.9999647  7.9999595 ]
#    [7.9999595  6.9999647 ]]]]