比较与torch.nn.BatchNorm2d的差异

查看源文件

torch.nn.BatchNorm2d

class torch.nn.BatchNorm2d(
    num_features,
    eps=1e-05,
    momentum=0.1,
    affine=True,
    track_running_stats=True
)(input) -> Tensor

更多内容详见torch.nn.BatchNorm2d

mindspore.nn.BatchNorm2d

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

更多内容详见mindspore.nn.BatchNorm2d

差异对比

PyTorch:在四维输入(具有额外mini-batch和channel通道的二维输入)上应用批归一化处理,以避免内部协变量偏移。

MindSpore:此API实现功能与PyTorch基本一致,典型区别有两点。MindSpore中momentum参数默认值为0.9,与PyTorch的momentum转换关系为1-momentum,默认值行为与PyTorch相同;训练以及推理时的参数更新策略和PyTorch有所不同,详细区别请参考与PyTorch典型区别-BatchNorm

分类

子类

PyTorch

MindSpore

差异

参数

参数1

num_features

num_features

-

参数2

eps

eps

-

参数3

momentum

momentum

功能一致,但PyTorch中的默认值是0.1,MindSpore中是0.9,与PyTorch的momentum转换关系为1-momentum,默认值行为与PyTorch相同

参数4

affine

affine

-

参数5

track_running_stats

use_batch_statistics

功能一致,不同值对应的默认方式不同,详细区别请参考与PyTorch典型区别-nn.BatchNorm2d

参数6

-

gamma_init

γ 参数的初始化方法,默认值:”ones”。PyTorch无此参数

参数7

-

beta_init

β 参数的初始化方法,默认值:”zeros” 。PyTorch无此参数

参数8

-

moving_mean_init

动态平均值的初始化方法,默认值:”zeros”。PyTorch无此参数

参数9

-

moving_var_init

动态方差的初始化方法,默认值:”ones”。PyTorch无此参数

参数10

-

data_format

MindSpore可指定输入数据格式可为”NHWC”或”NCHW”,默认值:”NCHW”。PyTorch无此参数

输入

单输入

input

x

功能一致,参数名不同

代码示例

PyTorch中,1-momentum后的值等于MindSpore的momentum,都使用mini-batch数据和学习参数进行训练。

# PyTorch
from torch import nn, tensor
import numpy as np

m = nn.BatchNorm2d(num_features=3, momentum=0.1)
input_py = tensor(np.array([[[[0.1, 0.2], [0.3, 0.4]],
                          [[0.5, 0.6], [0.7, 0.8]],
                          [[0.9, 1], [1.1, 1.2]]]]).astype(np.float32))
output = m(input_py)
print(output.detach().numpy())
# [[[[-1.3411044  -0.44703478]
#    [ 0.4470349   1.3411044 ]]
#
#   [[-1.3411043  -0.44703442]
#    [ 0.44703496  1.3411049 ]]
#
#   [[-1.3411039  -0.44703427]
#    [ 0.44703534  1.341105  ]]]]

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

m = nn.BatchNorm2d(num_features=3, momentum=0.9)
m.set_train()
#  BatchNorm2d<num_features=3, eps=1e-05, momentum=0.9, gamma=Parameter (name=gamma, shape=(3,), dtype=Float32, requires_grad=True), beta=Parameter (name=beta, shape=(3,), dtype=Float32, requires_grad=True), moving_mean=Parameter (name=mean, shape=(3,), dtype=Float32, requires_grad=False), moving_variance=Parameter (name=variance, shape=(3,), dtype=Float32, requires_grad=False)>

input_ms = Tensor(np.array([[[[0.1, 0.2], [0.3, 0.4]],
                          [[0.5, 0.6], [0.7, 0.8]],
                          [[0.9, 1], [1.1, 1.2]]]]).astype(np.float32))
output = m(input_ms)
print(output)
# [[[[-1.3411045  -0.4470348 ]
#    [ 0.44703496  1.3411045 ]]
#
#   [[-1.341105   -0.4470351 ]
#    [ 0.44703424  1.3411041 ]]
#
#   [[-1.3411034  -0.44703388]
#    [ 0.44703573  1.3411053 ]]]]